Structural and functional characterization of the Zn(II) site in dimethylargininase-1 (DDAH-1) from bovine brain. Zn(II) release activates DDAH-1.

L-N(omega),N(omega)-dimethylarginine dimethylaminohydrolase-1 (DDAH-1) is a Zn(II)-containing enzyme that, through hydrolysis of side-chain methylated l-arginines, regulates the activity of nitric-oxide synthase. Herein we report the structural and functional properties of the Zn(II)-binding site in DDAH-1 from bovine brain. Activity measurements of the native and metal-free enzyme have revealed that the endogenously bound Zn(II) inhibits the enzyme. Native DDAH-1 could be fully or partially activated using various concentrations of phosphate, imidazole, histidine, and histamine, a process that is paralleled by the release of Zn(II). The slow activation of the enzyme by the bulky complexing agents EDTA and 1,10-phenantroline suggests that the Zn(II)-binding site is partially buried in the protein structure. The apparent Zn(II)-dissociation constant of 4.2 nm, determined by 19F NMR using the chelator 5F-BAPTA (1,2-bis(2-amino-5-fluorophenoxy)ethane-N,N,N',N'-tetraacetic acid), lies in the range of intracellular free Zn(II) concentrations. These results suggest a regulatory role for the Zn(II)-binding site. The coordination environment of the Zn(II) in DDAH-1 has been examined by Zn K-edge x-ray absorption spectroscopy. The extended x-ray absorption fine structure observed is consistent with Zn(II) being coordinated by 2 S and 2 N (or O) atoms. The biological implications of these findings are discussed.

The two side-chain methylated derivatives of L-Arg, L-Nmethylarginine (monomethylated L-Arg (MMA) 1 ) and L-N ,Ndimethylarginine (asymmetric dimethylated L-Arg (ADMA)) were first found as a result of post-translational modifications of Arg residues, mainly in DNA-and RNA-binding proteins (1). Further studies established that methylated L-Arg molecules also occur freely in body fluids (2)(3)(4)(5) and various tissues (6,7). Although degradation of Arg-methylated proteins has been shown to be a source of free MMA and ADMA (8,9), a direct methylation of L-Arg via a so far unknown pathway has also been suggested (10,11). The y ϩ membrane transporter system is apparently involved in the uptake or release of MMA and ADMA (11,12).
Nitric-oxide synthase (NOS), which generates the free radical NO from L-Arg, exists in at least three isoforms that are involved in different biological processes. The two constitutive isoforms, nNOS (neuronal) and eNOS (endothelial), produce NO for neurotransmission and cardiovascular regulation, respectively, and are predominantly regulated by cytosolic Ca 2ϩ levels (13). It is well established that MMA and ADMA are endogenous competitive inhibitors of NOS (9,11,14). The intracellular L-Arg levels and the corresponding K m value of NOS are such that the production of NO should not depend on L-Arg supplementation (15). However, in biological studies, the addition of L-Arg increased the production of NO (16), a phenomenon called the "Arg paradox." In this context, it may be noted that recent in vitro studies have shown that the concomitant presence of varying, yet physiologically relevant, levels of MMA and ADMA dramatically affects NOS activity (17).
The increased levels of MMA and ADMA in plasma and urine have been found in several diseases that are linked to vascular dysfunction characterized by low NO levels, e.g. uremia, atherosclerosis, hypercholesterolemia, diabetes mellitus, hypertension, and homocysteinemia (27). Moreover, increased levels of ADMA in plasma have been discussed as a risk factor for atherosclerosis (3,4,9). In contrast, NO overproduction apparently leads to diseases such as septic shock and migraine. In these cases, clinical studies using MMA as a drug have been or are currently being conducted (28,29).
The above studies clearly indicate the importance of DDAH in NO metabolism. Previously, we have demonstrated that DDAH-1 from bovine brain is a monomeric (31.2 kDa) Zn(II)containing protein (30) and that the Zn(II) is not involved in the * This work was supported in part by the Olga Mayenfisch-Stiftung (to M. K.) and by Swiss National Science Foundation Grant 31-58858.99. 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 U.S.C. Section 1734 solely to indicate this fact.
The present work was undertaken with the aim of gaining a better understanding of the role of Zn(II) in mammalian DDAH-1. To facilitate these investigations, a substantially improved isolation procedure for DDAH-1 from bovine brain has been developed. The nature of the Zn(II)-binding site in DDAH-1 has been examined by extended x-ray absorption fine structure (EXAFS) associated with the Zn K-edge. The competition between the metal chelator 5F-BAPTA and Zn(II)-DDAH-1 has been used to determine the Zn(II) binding constant by 19 F NMR spectroscopy. Measurements of enzyme activity have revealed that in a number of common buffer systems DDAH-1 is inactive, but it becomes active when the Zn(II)-free (or apo-) form is generated. The latter effect could be reversed by a subsequent metal reincorporation, suggesting that the bound Zn(II) inhibits the enzyme. The possible biological implications of these findings are discussed.

MATERIALS AND METHODS
Aprotinin from bovine lung and leupeptin were purchased from Sigma and MMA⅐HOAc from Alexis. Tris-(2-carboxyethyl)-phosphine⅐HCl, soybean trypsin inhibitor, L-His, D-His, and histamine⅐2 HCl were obtained from Fluka. K 4 5F-BAPTA is a product of Molecular Probes Inc. Dialysis tubing (molecular weight cut-off; 12-14 kDa, Spectrum) was rendered metal-free as described (33). Bovine brains were obtained from a slaughterhouse and stored at Ϫ20°C until used.

Metal and Protein Quantification
The zinc concentration was determined by atomic absorption spectroscopy (IL Video 12) and that of DDAH-1 by absorption spectroscopy (Cary 300, Varian), using the molar extinction coefficient ⑀ 280 nm ϭ 14,420 M Ϫ1 cm Ϫ1 (24).

Activity Measurements
The product of the enzymatic reaction, L-Cit, was determined by our colorimetric method in 96-well microtiter plates (34). If not otherwise stated, upon substrate addition (6.7 mM MMA) the samples were incubated for 30 min at 37°C. All activity measurements were performed in triplicate. One unit of DDAH-1 is defined as the amount of enzyme that produces 1 mol of L-Cit/min under the conditions given.

Protein Purification
All purification steps (a-f) except L-Arg affinity chromatography were performed at 4°C. During the protein isolation, the activity of DDAH-1 was determined in 250 mM imidazole/HCl, pH 7.0. The protein concentration was estimated by taking one absorbance unit at 280 nm to correspond to ϳ1 mg of protein ml Ϫ1 .
Step a: Tissue Homogenization-Two frozen bovine brains were thawed overnight. The tissue was cut into pieces and washed with 20 mM Tris/HCl, 150 mM NaCl, 2 mM EDTA, pH 8.0. Subsequently, the tissue was homogenized in a food processor together with a triple volume of 50 mM TEA/HCl, 5 mM benzamidine, 1 mM phenylmethanesulfonyl fluoride, 10 g ml Ϫ1 trypsin inhibitor, 0.1 g ml Ϫ1 aprotinin, 0.5 g ml Ϫ1 leupeptin, pH 8.0. The mixture was then stirred for 30 min and centrifuged for 60 min at 5,000 ϫ g.
Step b: (NH 4 ) 2 SO 4 Precipitation-The supernatant (a) was fractionated by (NH 4 ) 2 SO 4 , and the precipitate formed between 30 and 80% saturation was collected by centrifugation. The precipitate was resuspended in about 200 ml of 5 mM TEA/HCl, pH 8.2.
Step c: Hydrophobic Interaction Chromatography-The protein sample (b) was applied to a Phenyl-Sepharose 6 Fast Flow column (5.5 ϫ 16 cm) (High Sub, Amersham Pharmacia Biotech). The column was equilibrated in 1.8 mM (NH 4 ) 2 SO 4 /TEA, pH 7.8. Proteins were eluted with a linear salt gradient from 1.8 mM (NH 4 ) 2 SO 4 /TEA to 20 mM TEA/HCl at pH 7.8 and the active fractions were combined. The solution was dialyzed two times against 5 liters of 1 mM TEA/HCl, pH 7.8. Prior to dialysis, the following protease inhibitors were added: 10 g ml Ϫ1 trypsin inhibitor, 0.1 g of ml Ϫ1 aprotinin, 0.5 g ml Ϫ1 leupeptin, 1 mM phenylmethanesulfonyl fluoride.
Step d: Anion Exchange Chromatography-The conductivity of the protein sample (c) was adjusted with H 2 O to Յ 15 m⍀ Ϫ1 cm Ϫ1 and then applied to a Source 15Q column (2.6 ϫ 15 cm, Pharmacia) equilibrated with 10 mM TEA/HCl, pH 7.6. The protein was eluted with a linear salt gradient from 0 to 200 mM KCl and the active fractions were pooled.
Step e: L-Arg Affinity Chromatography-L-Arg-Sepharose was prepared from NHS-activated Sepharose 4 Fast Flow (Amersham Pharmacia Biotech) as described (35) including the following changes. In the coupling reaction, a solution of 500 mM L-Arg/HCl, 200 mM H 3 BO 3 , 500 mM NaCl, pH 8.2, was used. Subsequently, the resin was washed three times, alternately with 500 mM ethanolamine/HCl, 500 mM NaCl, pH 9.0, and 250 mM NaOAc/HOAc, 500 mM NaCl, pH 4.0. A coupling efficiency of Ն99% was found by amino acid analysis.
The conductivity of the protein sample (d) was adjusted with H 2 O to Յ 5 m⍀ Ϫ1 cm Ϫ1 and applied to a L-Arg-Sepharose column (2.6 ϫ 13 cm) equilibrated with 10 mM TEA/HCl, pH 7.6 at room temperature. The resin was washed with 100 ml of 10 mM TEA/HCl, pH 7.6 and the protein eluted with 100 ml of 10 mM TEA/HCl, 30 mM L-Arg, pH 7.6.
Step f: Size Exclusion Chromatography-The protein solution (e) was concentrated in Centriprep-10 (Millipore) to 3 ml. The final protein sample was applied to a HiLoad 16/60 Superdex 75 column (Pharmacia) equilibrated with 10 mM TEA/HCl, 100 mM KCl, pH 7.4. The active fractions were combined.

Depletion and Reintroduction of Zn(II)
All dialysis steps were carried out at 4°C and the outer solutions were steadily bubbled with argon. Buffers were rendered metal-free as previously described (24). To remove the Zn(II) from native DDAH-1, 200 l of the protein sample (ϳ50 M) was dialyzed against 100 ml of 100 mM imidazole/HCl, 10 mM DTT, pH 6.8. The dialysis buffer was changed 4 times (every 2 h) and the dialysis completed overnight. To remove imidazole and DTT, the protein was dialyzed against 2 changes of 100 ml of 10 mM MES/NaOH, 100 mM KCl, pH 6.2. The re-introduction of Zn(II) into apoDDAH-1 was accomplished for 200 l of the apoprotein (ϳ50 M); this was first dialyzed against 100 ml of 10 mM Tris/HCl, 50 mM KCl, 10 mM 2-mercaptoethanol, 0.5 mM Tris-(2-carboxyethyl)-phosphine, pH 8.0 (2 h), followed by dialysis overnight against 100 ml of 10 mM Tris/HCl, 50 mM KCl, 1 mM ZnSO 4 , pH 7.4. To remove unbound Zn(II), the sample was passed through a Fast Desalting Column HR 10/10 (Amersham Pharmacia Biotech) equilibrated in 10 mM MES/NaOH, 100 mM KCl, pH 6.2. Protein-containing fractions were collected and concentrated using Centricon-10 (Millipore).

Dependence of the Enzymatic Activity on Various Buffers and pH Values
Native DDAH-1 (0.91 M) or apoDDAH-1 (0.5 M) was incubated in 60 l of various 100 mM buffers at different pH values due to the corresponding buffer capacities (Fig. 1). The pH values of these buffers were determined at 37°C.

Dependence of the Enzymatic Activity on the Buffer Concentration
The activity of 0.79 M native DDAH-1 and apoDDAH-1, respectively, was determined in 60 l of either 10, 20, 50, 100, or 400 mM Tris/HCl, imidazole/HCl, or HEPES/NaOH buffers. The pH of buffer solutions containing MMA was adjusted to 7.4.

Influence of the L-His Concentration on the Zn(II)
Content of DDAH-1

Influence of Salt Concentration and Various Derivatives of Imidazole on the Enzymatic Activity
Native DDAH-1 (1.1 M) was incubated in 100 mM TEA/HCl, pH 7.4, containing (at 50 mM each) imidazole, L-His, D-His, or histamine, and the activity was determined. The activity of a control sample, containing DDAH-1 in 100 mM TEA/HCl, pH 7.4, was also measured. Similar measurements were performed in the presence of additional 50 or 500 mM KCl.

Preparation of XAS Sample and XAS Data Collection and Analysis
Native DDAH-1 was concentrated to 0.8 mM using a Centricon-10 microconcentrator, exchanging the buffer simultaneously to argon-saturated 25 mM TEA/HCl, 15% (v/v) glycerol, 100 mM KCl, pH 7.4. The solution was injected into an aluminum sample cell with Mylar windows, frozen in liquid nitrogen and the sample was maintained at 77 K in a cryostat with Mylar windows. The x-ray absorption spectrum was measured at Zn K-edge on station 8.1 at the CLRC Daresbury Synchrotron Radiation Source operating at 2 GeV with an average current of 150 mA. An Si(220) double crystal monochromator was used, detuned to 70% of the maximum intensity to minimize any residual harmonic contamination after the focusing mirror.
The monochromator energy was calibrated by recording the K-edge profile of a zinc foil before and after the collection of spectra for native DDAH-1; the position of the K-edge of the zinc foil (9669 eV) did not change during data collection. The Zn K-edge XAS of native DDAH-1 was obtained in fluorescence mode using a Canberra 13-element solidstate detector; 24 scans were recorded and averaged. The extended x-ray absorption fine structure (EXAFS) was analyzed using the Daresbury program EXCURV98 (39), involving exact curved wave theory (40, 41). The phase shifts were derived from ab initio calculations using Hedin-Lundqvist potentials and von Barth ground states (42). The EXAFS simulations commenced from the Fourier transform of the experimental data and involved placing of shells of back-scatterering atoms around the Zn(II), at chemically sensible distances, and refining (i) the Fermi energy (E f ), (ii) the absorber-scatterer distances, (iii) the occupation number, and (iv) the Debye-Waller type factors (2 2 ) of each shell, to obtain optimum agreement between the experimental and simulated data.

RESULTS
Purification of DDAH-1 from Bovine Brain-An expression system for recombinant human DDAH-1 in Escherichia coli has recently been established by Kimoto et al. (1998) (31) and kindly provided to us. Despite great efforts of this group and our group, to date expression in E. coli has not yielded sufficient amounts of soluble protein (Ͻ1 mg l Ϫ1 culture) for extensive biochemical studies. In addition, our attempts to refold over-expressed protein from inclusion bodies have been unsuccessful. Previously, we could isolate satisfactory amounts of the native protein from bovine brain (30), however, the reported method was rather time consuming. The substantially improved isolation procedure of bovine brain DDAH-1 described herein yields significantly more protein, i.e. 1-1.5 mg/100 g tissue (Table I). On basis of the DDAH-1 activity in crude extract, the concentration of the enzyme at 4 -5 mg/100 g brain tissue is rather high. In brief, in the first purification step the soluble protein fraction obtained after tissue homogenization was subjected to a (NH 4 ) 2 SO 4 fractionation. This step was followed by hydrophobic interaction chromatography. The protein fractions containing DDAH-1 were then desalted by dialysis, and the protein mixture purified further by an anion exchange chromatography. The latter step was very effective in removing a large number of other proteins (Table I). Because millimolar concentrations of the substrate analog L-Arg inhibit the activity of DDAH-1 (18), we introduced an L-Arg affinity column. The advantage of using an L-Arg over a MMA affinity column (18), is that L-Arg is not degraded to L-Cit by DDAH. Although the reported K i value for L-Cit is also in millimolar range (18), in our hands the enzyme did not show a sufficient binding to an L-Cit affinity column. The purification was completed by an additional gel filtration chromatography. CD, ESI-MS, enzymatic activity, atomic absorption spectroscopy, and SDS-PAGE have been used to assess the protein purity and its structural features. Based on these criteria, the isolated enzyme was homogenous and showed the same chemical and structural characteristics previously reported for DDAH-1 from bovine brain (24,30). The Zn(II) content was between 0.85-0.95 mol equivalents.
Effect of CH 3 NH 2 on DDAH-1 Activity-A possible inhibitory effect of the reaction products CH 3 NH 2 and (CH 3 ) 2 NH on the enzyme activity has not previously been investigated. Therefore, the DDAH-1 activity in the presence of 1, 5, and 10 mM CH 3 NH 2 was examined. The corresponding activities observed, referenced to that obtained in the absence of CH 3 NH 2 (set to 100%), were 90%, 87%, and 79%. Thus, the product inhibition by CH 3 NH 2 is only minor. However, to minimize the inhibition by both L-Cit and CH 3 NH 2 , the enzymatic reactions were terminated at submillimolar product concentrations.
Dependence of the DDAH-1 Activity on the Type and pH of the Buffer-In the first characterization of DDAH-1 from rat kid- ney, the enzymatic activities were determined in 100 mM phosphate buffer at its pH maximum of 6.5 (18,43). The same conditions were also employed in the characterization of the enzyme isolated from human liver (31), in recombinant human DDAH-1 and DDAH-2 (19,31), and in cell culture studies (10,44). The maximum enzymatic activity of bovine brain DDAH-1 in 100 mM phosphate buffer was found at pH 6.2 (Fig. 1A).
In the present study, the enzymatic activity of native bovine DDAH-1 was determined in various 100 mM buffers between pH 4.1 and 9.5 (Fig. 1A). Surprisingly, the buffer type dramatically affected the enzymatic activity. Thus, in imidazole/HCl compared with phosphate buffer a substantially increased activity was observed over the whole pH range employed with a maximum at pH 6.8 (ϳ25% higher) (Fig. 1A). In marked contrast to the protein behavior in phosphate buffer and imidazole/ HCl, in cacodylic acid/NaOH, NaOAc/HOAc, and BisTris/HCl substantially lower enzymatic activities with the pH maxima shifted to lower pH values (between pH 5.1 and 5.4) were detected. Measurements at alkaline pH values, using Tris/HCl and H 3 BO 3 /NaOH, yielded low and pH independent enzymatic activities (Fig. 1A). A similar behavior was also discerned in MES/NaOH, HEPES/NaOH, and TEA/HCl (not shown).
Because the activity of holoDDAH-1 is strongly dependent on the type of buffer, the effect of different buffer concentrations was also examined. Fig. 2A illustrates the activity of holo-DDAH-1 at pH 7.4 in various concentrations of imidazole/HCl, Tris/HCl, or HEPES/NaOH. Although the activity in imidazole/ HCl, as well as in the presence of comparable concentrations of L-His (Fig. 3A), increases with increasing buffer concentration in a hyperbolic manner, in the other buffers the enzymatic activity remained almost unaffected. Similar studies of the effect of the buffer concentration were also conducted at pH 6.2. In this case, a hyperbolic increase of enzymatic activity was observed in phosphate buffer that resembles the behavior in imidazole/HCl, whereas variations in the concentrations of cacodylic acid/NaOH and BisTris/HCl had no effect (data not shown).
Given the variation of the activity of the holoenzyme with pH and buffer type, similar studies were conducted for the apoenzyme (Fig. 1B). It was found to be independent of the buffer system employed (100 mM NaOAc/HOAc, cacodylic acid/NaOH, Tris/HCl, and H 3 BO 3 /NaOH) with a maximum at about pH 7.8.
The effect of buffer concentration on the activity of apoDDAH-1 was also investigated at pH values 7.4 (Fig. 2B) and 6.2 (not shown). In all of the buffers investigated, the activity was observed to be similar, concentration-independent, and at a level comparable with that of holoDDAH-1 in 400 mM imidazole/HCl (Fig. 2A). Taken together, contrary to apoDDAH-1, the   FIG. 1. Effect of various buffers (100 mM) and pH values on the enzymatic activity of (A) holo-and (B) apoDDAH-1. OE, NaOAc/ HOAc; E, cacodylic acid/NaOH; ‚, BisTris/HCl; q, phosphate buffer; f, imidazole/HCl; ࡗ, Tris/HCl; Ⅺ, H 3 BO 3 /NaOH. Activities were measured after a 30-min sample incubation with 6.7 mM MMA at 37°C (for details see "Materials and Methods"). activity of Zn(II)-DDAH-1 is strongly dependent on the type of the buffer and its concentration. Therefore, we have addressed the question as to whether or not the different behavior of Zn(II)-DDAH-1 in various buffers is due to a release of Zn(II).

Influence of L-His on the Zn(II) Content of Native DDAH-1-
The initial experiments with holoDDAH-1 (ϳ1 mg ml Ϫ1 ) showed that after dialysis against 1:100 (v/v) 100 mM imidazole/HCl or phosphate buffer for only 30 min, about 30% of the total amount of Zn(II) was removed (not shown). This was a rather surprising finding, as the removal of Zn(II) by the strong complexing agents EDTA or 1,10-phenantroline was very slow (1-2 days) (24). Further investigations employed the more physiologically relevant L-His instead of imidazole. Native DDAH-1 and 10 mM MMA were added and incubated at 37°C. The reaction time course shows that in the absence of L-His, the formation of L-Cit was very slow, and after 160 min 67% of MMA was still present (Fig. 3A). We ascribe this result to a residual activity of the enzyme due to the presence of small amounts of the apoprotein (see above). In contrast, the addition of L-His resulted in a sharp increase of the amount of L-Cit produced. For instance, in 100 mM L-His ϳ45 min was sufficient to convert all MMA to L-Cit (Fig. 3A). A plot of the initial rates versus the concentration of L-His results in a hyperbolic curve (Fig. 3B). To examine whether this behavior correlates with the Zn(II) content of DDAH-1, after a 160-min incubation the released Zn(II) was removed through dialysis, and the concentration of protein-bound Zn(II) was determined. A plot of the remaining Zn(II) content of the enzyme against the L-His concentration revealed an exponential decrease (Fig. 3C). This behavior provides direct evidence for a concentration-dependent removal of Zn(II) from DDAH-1 by L-His. In addition, a linear correlation between the starting velocity and the Zn(II) content of DDAH-1 (inset, Fig. 3C; correlation coefficient r ϭ Ϫ0.9604), with no activity for the fully metal loaded protein, demonstrates that apoDDAH-1 is the active form of the enzyme.

Specificity of the Zn(II) Depletion and the Influence of Ionic Strength-
To learn more about the specificity of the Zn(II) depletion, imidazole and its derivatives, L-His, D-His, and histamine, were compared (Fig. 4). The activating ability of these compounds follows the order histamine Ͻ Ͻ imidazole Ͻ Ͻ L-His Ͻ D-His. In addition, the effect of differing salt concentrations was found to be negligible. Thus, the enzyme activa-tion observed with various compounds reflects their ability to remove the bound Zn(II).
As mentioned above, substantial enzyme activation was also found in phosphate buffer, a chemically very different compound. To exclude the possibility that this effect mimics that of endogenously occurring phosphoesters, the enzyme activity was determined in the presence of 500 M ATP, ADP, cyclic AMP, GTP, or NADPH. However, no enzyme activation was found (not shown).
Reconstitution of DDAH-1 with Zn(II)-In our previous studies on DDAH-1, the apoprotein was obtained upon an extensive dialysis against 2 mM 1,10-phenathroline. However, we failed to reintroduced Zn(II) into the protein (24). Herein we report a method that allows a reversible preparation of metal-free and metal-containing DDAH-1. The preparation of the apoprotein takes advantage of the facile release of Zn(II) in imidazole/HCl (see above). The Zn(II)-depleted protein contained only a small amount of residual Zn(II) (0.10 Ϯ 0.03 mol equivalent). The subsequent incorporation of Zn(II) was successfully accomplished when the sample was reduced prior to dialysis. The enzymatic activity was taken as a measure of Zn(II) incorporation. In this case, the reconstituted holoenzyme was inactive in TEA/HCl but became active in imidazole/HCl. The effect of bound Zn(II) on the protein structure was also examined by CD spectroscopy in 10 mM Tris/H 3 BO 3 , pH 7.4. A comparison of the CD profiles of the apo-and Zn(II)-reconstituted protein revealed only small differences mainly in the high energy region (below 220 nm) (not shown). In the previous studies a similar effect of Zn(II) on the protein structure was also found (24). Consequently, the identity of the CD spectra of native and reconstituted DDAH-1 supports the correct metal incorporation.
Measurement of the Apparent Zn(II) Binding Constant-In view of the enzyme activation by removal of Zn(II), the affinity of DDAH-1 for Zn(II) was determined. We employed our recently developed 19 F NMR method, which is based on the competition between 5F-BAPTA and a metalloprotein (36). The introduction of two homotopic 19 F nuclei in the structure of the metal chelator BAPTA enables the concomitant detection of free 5F-BAPTA and its 1:1 complexes with various metal ions because of a change in the 19 F chemical shift position of the chelator upon metal binding (45).
The spectrum showed two 19 F signals that were well separated (by 2.2 ppm). The signals were assigned based on the comparison with 19 F NMR spectra of free and Zn(II) loaded 5F-BAPTA. The presence of both signals in the NMR spectrum of the DDAH-1/5F-BAPTA mixture reflects the partitioning of Zn(II) between the protein and the chelator. The same spectrum acquired after an additional 20-h sample incubation established that the thermodynamic equilibrium had already been reached. The reported Zn(II)-binding constant of 5F-BAPTA was recalculated for our conditions, using the program CHELATOR (38), to be K Zn(II)-5F-BAPTA ϭ 2.6 ϫ 10 8 M Ϫ1 . Using this value, the apparent Zn(II)-binding constant for DDAH-1 of K Zn(II)-DDAH-1 ϭ 2.4 ϫ 10 8 M Ϫ1 at 20°C was obtained.
X-ray Absorption Spectroscopy of the Zn(II)-binding Site-The x-ray absorption spectrum associated with a metal absorption edge, in particular the EXAFS, provides information regarding the nature of the local environment of the metal (46). The Zn K-edge EXAFS recorded for native DDAH-1 is shown in Fig. 5A and the associated Fourier transform in Fig. 5B; the profile of the latter contains only one major peak centered at ϳ2.1 Å. The EXAFS of native DDAH-1 can be simulated successfully by the back-scattering from 2 S atoms at ϳ2.24 Å and 2 N (or 2 O) atoms at ϳ1.95 Å. The uncertainty in the values of these distances is considered Ն 0.03 Å. These interpretations involved the individual shells with associated Debye-Waller parameters of 0.006 Å 2 for the S shell and 0.005 Å 2 for the N (or O) shell; the R-factor was 29.0 (or 29.3).
The proposed environment for Zn(II) in DDAH-1 resembles that in some zinc-finger proteins (47,48), which are considered to involve coordination of the Zn(II) by 2 Cys and 2 His residues, i.e. Zn(II)His 2 Cys 2 . Thus, the EXAFS study of transcription factor IIIA from Xenopus laevis indicated Zn-S and Zn-N distances of ϳ2.30 and 2.00 Å, respectively. A similar result was also obtained for the Zn(II) environment in the Fur protein from E. coli (49).
The position of the Zn K-edge of DDAH-1 (9661.5 Ϯ 0.5 eV) is consistent with Zn(II) bound to 2 S and 2 N atoms (49,50). The Zn-S and Zn-N (or Zn-O) distances obtained in this study are each slightly shorter than the corresponding value for transcription factor IIIA and Fur protein. Also, in the Fourier transform of EXAFS recorded for DDAH-1 (Fig. 5B), there is no clear evidence for the back-scattering from the outer shells of the imidazole group (46). Therefore, although the S ligands of the Zn(II) in DDAH-1 most likely derive from Cys residues, the identity of the N (or O) donor ligands is not clear.

DISCUSSION
In this work we have demonstrated that the isolated Zn(II)-DDAH-1 is inactive. Several other enzymes, including glyceraldehyde-3-phosphate dehydrogenase (51) and caspase-3 (52), and receptors like N-methyl-D-aspartate receptor and ␥-aminobutyric acid receptor A (53) are inhibited by Zn(II). However, to our knowledge, the isolation of a Zn(II)-containing enzyme, in which Zn(II) is an inhibitor, has so far not been reported. The observed activation of DDAH-1 upon release of Zn(II) suggests a regulatory role for this metal-binding site.
The activity of the apoenzyme displays a broad pH-profile with maximum activity at about pH 7.8 (Fig. 1B). The activation of the holoenzyme by the release of Zn(II) was demonstrated in phosphate and imidazole buffers, for which a maximum activity was observed at pH 6.2 and 6.8, respectively (Fig.  1A). In the other buffers examined, the maximum activity occurred at around pH 5.3. The differences observed in the degree of DDAH-1 activation and in the pH maxima for various buffers essentially follow the variations of the activity of the apoenzyme with pH (Fig. 1). Thus, these results indicate that the observed behavior reflects changes in the amount of the active apoform brought about by differences in metal affinities of these buffers. In buffers that lack an appreciable metal binding affinity, the obtained enzymatic activity reflects the presence of a small amount of apoenzyme in native DDAH-1. In view of these data, we suggest that in the initial characterization of DDAH-1 from rat, a Zn(II)-containing form has also been studied since a pH maximum at 6.5 in 100 mM phosphate buffer has been reported (18).
As judged by the comparison of the specific activities of the apo-and holoenzyme, full metal activation of the holoenzyme would require Ն400 mM imidazole (Fig. 3). Taking advantage of the enzyme activation in imidazole/HCl and the absence of this effect in TEA/HCl, we examined the activity of DDAH-1 in crude bovine brain extract, using 250 mM buffer concentrations. An ϳ30% higher DDAH-1 activity is observed in imidazole/HCl compared with TEA/HCl, suggesting that only about 70% of the enzyme was present in the active Zn(II)-free form. This implies that if the true activity is sought, e.g. in cellular extracts, non-activating buffers should be used.
Both phosphate and imidazole buffers have been shown to be much more effective in generating apoprotein than bulky metal-chelating agents such as EDTA or 1,10-phenantroline. We ascribe this effect to an inaccessibility of the Zn(II)binding site to these rather bulky chelators. Thus, it would appear that the Zn(II)-binding site is partially buried in the protein structure. The enzyme activation by phosphate or imidazole presumably reflects both the ability of these agents to access the bound Zn(II) and their metal-complexing ability. However, their marked chemical and electrostatic differences may imply that the mechanism of enzyme activation differs. The differences in the Zn(II)-DDAH-1 activation observed with both stereoisomers of His suggest a certain specificity toward the structure of the metal-depleting compound. However, in all instances the enzyme activation required high concentrations of the activating compounds, suggesting that they are not the physiological enzyme-activating molecules. Because the DDAH from Pseudomonas aureus exists as a dimer (32), it was of interest to examine whether the activation of DDAH-1 through Zn(II) depletion could lead to its dimerization. Gel filtration studies of apoDDAH-1 revealed that this is not the case (not shown).
The concentration of "labile" or free Zn(II) in cells has been estimated to be in the nanomolar range (53,54). The apparent Zn(II) dissociation constant determined for DDAH-1 of 4.2 nM lies in the range of intracellular free Zn(II) concentrations. This result together with the herein reported enzyme activation by various compounds supports a regulatory role for this metal binding site.
The Zn K-edge EXAFS study suggests that the Zn(II) coordination environment consists of 2 Cys and 2 N (or O) donor ligands. Although the imidazole group of His residues could be responsible for the N ligands, the lack of back-scattering from the outer shells of this residue, usually discerned in the EXAFS and its Fourier transform, was not apparent in this study. A comparison of the Zn-S and Zn-N (or Zn-O) distances of 2.24 and 2.00 Å, respectively, in DDAH-1 with those reported from crystallographic studies of Zn(II)-binding proteins containing Cys and His ligands is strongly consistent with a tetrahedral coordination of the metal (49,55). The sequences of the mammalian DDAH-1 from human (31), rat (56), mouse (constructed from cDNA fragments of dbEST), and also in part from bovine brain (24) are currently available. All of these amino acid sequences contain between 5 and 7 Cys residues and show a high degree (92%) of identity. However, examination of the DDAH-1 sequences for the known Zn(II)-binding motifs identified thus far failed to reveal any matches. This is perhaps not surprising, because there is only limited knowledge about the nature of regulatory Zn(II) sites in proteins.
Possible biological implications of a regulatory Zn(II) site in DDAH-1 should also be discussed. The contribution of DDAH-1 to the control of NOS and NO in vivo is evident from studies demonstrating that both DDAH-1 and nNOS mRNAs are upregulated in neurons following nerve injury (26). Moreover, it has been speculated that a coordinated up-regulation of the DDAH1 and NOS1 genes presumably enhances NO production. A moderate amount of NO formation has been proposed to be important for neuronal survival, regeneration, and plasticity (57). Oxidative stress has been implicated as the cause of a number of neurodegenerative disorders, including Alzheimer's disease (AD) (58). It is well documented that under these pathological conditions the levels of free Zn(II) are increased in the brain (59 -62). Increased intracellular levels of free Zn(II) would substantially reduce the activity of DDAH-1, resulting in increased MMA and ADMA concentrations, which in turn would inhibit NOS. Indeed, reduced DDAH-1 activity and elevated ADMA levels have been observed in in vivo and in vitro studies of endothelial cells under oxidative stress conditions, and this effect has been prevented by the use of antioxidants (9). In our recent immunohistochemical studies (63), marked changes in the distribution of DDAH-1 have been seen in AD brains as compared with controls lacking AD pathology. In AD brains the protein was co-localized with the intracellular neurofibrillary tangles, a hallmark of AD. Thus, it would appear that under the pathological conditions of AD DDAH-1 is inhibited not only by increased Zn(II) levels but also, presumably, by its immobilization through binding to neurofibrillary tangles composed of hyperphosphorylated tau protein. It should be noted that, depending on its concentration, NO could act as a pro-or antioxidant. Low levels of NO are apparently beneficial in oxidative stress, as it intercepts reactive oxygen species, such as the HO ⅐ radical, converting them to less damaging and more easily detoxified products (64).
In conclusion, this work presents the first structural and functional characterization of the Zn(II) site in DDAH-1. We have shown that the endogenously bound Zn(II) inhibits this enzyme; the metal binding affinity lies in the nanomolar range. This and the enzyme activation by release of Zn(II) by several low molecular weight compounds such as phosphate or imidazole suggest a regulatory role for this site in DDAH-1. The EXAFS data are consistent with Zn(II) bound to 2 S and 2 N (or O) atoms. The physiological compound that activates DDAH-1 by removing the Zn(II) remains to be established.