Role of Bound Zinc in Dimer Stabilization but Not Enzyme Activity of Neuronal Nitric-oxide Synthase*

Nitric-oxide synthases (NOS) are homodimeric proteins and can form an intersubunit Zn(4S) cluster. We have measured zinc bound to NOS purified from pig brain (0.6 mol/mol of NOS) and baculovirus-expressed rat neuronal NOS (nNOS) (0.49 ± 0.13 mol/mol of NOS), by on-line gel-filtration/inductively coupled plasma mass spectrometry. Cobalt, manganese, molybdenum, nickel, and vanadium were all undetectable. Baculovirus-expressed nNOS also bound up to 2.00 ± 0.58 mol of copper/mol of NOS. Diethylenetriaminepentaacetic acid (DTPA) reduced the bound zinc to 0.28 ± 0.07 and the copper to 0.97 ± 0.24 mol/mol of NOS. Desalting of samples into thiol-free buffer did not affect the zinc content but completely eliminated the bound copper (≤0.02 mol/mol of NOS). Most (≥75%) of the bound zinc was released from baculovirus-expressed rat nNOS byp-chloromercuriphenylsulfonic acid (PMPS). PMPS-treated nNOS was strongly (90 ± 5%) inactivated. To isolate functional effects of zinc release from other effects of PMPS, PMPS-substituted thiols were unblocked by excess reduced thiol in the presence of DTPA, which hindered reincorporation of zinc. The resulting enzyme contained 0.12 ± 0.05 mol of zinc but had a specific activity of 426 ± 46 nmol of citrulline.mg−1.min−1, corresponding to 93 ± 10% of non-PMPS-treated controls. PMPS also caused dissociation of nNOS dimers under native conditions, an effect that was blocked by the pteridine cofactor tetrahydrobiopterin (H4biopterin). H4biopterin did not affect zinc release. Even in the presence of H4biopterin, PMPS prevented conversion of NOS dimers to an SDS-resistant form. We conclude that zinc binding is a prerequisite for formation of SDS-resistant NOS dimers but is not essential for catalysis.

nine and oxygen (reviewed in (1)). Three isozymes are known: neuronal (nNOS), inducible (iNOS), and endothelial (eNOS). The reaction occurs at the oxygenase domain of NOS, which contains a cysteine-thiolate-ligated heme and forms the Nterminal half of the protein. Reducing equivalents for the reaction are passed from NADPH to the heme via an integral FMN-and FAD-containing reductase domain, which forms the C-terminal half of the protein, under the control of Ca 2ϩ /calmodulin binding near the FMN site. Active NOS is a homodimer, with the structural dimeric interface being formed between the two oxygenase domains.
Unlike other heme enzymes, NOS is completely dependent on the pteridine cofactor tetrahydrobiopterin (H 4 biopterin), which binds to the oxygenase domain. Why the enzyme needs H 4 biopterin has not yet been determined unambiguously, although evidence is mounting for a novel, one-electron redox cycling of the pteridine (2)(3)(4)(5). Apart from this emerging redox function, H 4 biopterin has effects on the structure of the protein, of which the most easily observed are a shift from low spin to high spin heme (6,7) and stabilization of the dimer (8,9). The latter effect is so marked that H 4 biopterin-saturated NOS migrates in dimeric form during SDS-PAGE if the samples are not boiled before loading onto the gel (8). The molecular basis of this unusual SDS resistance has not been completely elucidated, although the dimeric interface features a large buried surface area, and the H 4 biopterin binding sites are intimately involved with the dimeric contact region.
A new factor relevant to NOS dimerization was revealed in crystal structures of eNOS oxygenase dimers (3). This was a Zn(4S) cluster in which two cysteine-thiolate ligands were contributed by each subunit. The two cysteines are conserved in all known NOS sequences, with the more N-terminal of the pair about 30 -35 residues from the N-terminal oxygenase domain boundary. Not all NOS oxygenase dimer structures have contained zinc at this site, the first structure of murine iNOS oxygenase contained a symmetric intersubunit disulfide bridge formed by the more C-terminal of the two pairs of cysteines (10). Since then, pairs of zinc-containing and zinc-free structures of iNOS oxygenase dimers have been solved (11,12). As to the possible importance of the zinc for enzyme activity, relevant mutants of the protein were already known. A site-directed mutant of the more C-terminal of the zinc-ligand cysteines (to alanine) had been studied in all three isozymes (13)(14)(15). The activity of the freshly isolated enzyme differed between groups, but all detected significant activity, in one case after prolonged preincubation with L-arginine (L-Arg) (15) and in others only at high concentrations of H 4 biopterin (13,14). This mutant (in nNOS) was recently re-examined with respect to its zinc content and found to contain about 0.05 zinc per dimer and could be reactivated by preincubation with L-Arg (16). Thus, an intact zinc site may protect NOS against inactivation without being essential for activity.
If the zinc is not necessary for NOS activity, it may not be incorporated into NOS under all physiological circumstances. Because of this, and because of a claim of non-heme iron involvement in NOS catalysis (17), several groups have re-examined the metal content of NOS. For zinc, values around 0.35-0.5 mol per subunit have been found in human iNOS expressed in Escherichia coli, in nNOS and eNOS expressed in E. coli (16,18), and in human eNOS expressed in yeast (19).
In the present study, we aimed first to examine nNOS from both a baculovirus expression system and natural tissue, for bound zinc. Second, we wanted to study its contribution to structure and function of wild-type rather than mutant NOS: If it is inessential, we reasoned, it might be possible to remove it without destroying the enzyme activity. Finally, because such an intersubunit connection would be expected to contribute to dimer stability, we wanted to check whether, and how, removal of the zinc would affect the dimerization of the enzyme.
NOS Expression and Purification-nNOS from pig cerebellum was isolated according to Mayer et al. (21). Rat nNOS was expressed in baculovirus-infected Sf9 cells and purified by sequential affinity chromatography on 2Ј,5Ј-ADP-Sepharose and then calmodulin-Sepharose, as described previously (22). Because the normal enzyme storage buffer contained 12 mM 2-mercaptoethanol and 2 mM EGTA, NOS samples for experiments involving zinc release were stripped of these compounds by gel filtration. NOS samples (3-5 mg) with phenylmethylsulfonyl fluoride (1 mM) and hen egg white trypsin inhibitor (1 g/ml) were concentrated to about 15 mg.ml Ϫ1 in a dialysis tube (Spectra/Por, 8-mm flat width, cut-off 15 kDa) laid in solid sucrose, before injection onto a Superose 6 HR 10/30 gel filtration column connected to an Ä KTA Purifier 10 chromatography system (Amersham Pharmacia Biotech, Vienna, Austria). The elution buffer was 0.05 M Tris-HCl, pH 7.4, containing 0.15 M NaCl. For convenience we refer to the resulting nNOS as "thiol-free." Protein was determined by the method of Bradford (23), using bovine serum albumin as a standard protein.
Analysis of Enzyme-bound Metals-Metals bound to NOS enzymes were analyzed by gel filtration with on-line detection by inductively coupled plasma mass spectrometry (ICP-MS). Before analysis, the ICP-MS was tuned to give typically ϳ22,000 counts per second (cps) (background 3 cps) for lithium and ϳ32000 cps (background 2 cps) for yttrium at a doubly charged ratio Ce 2ϩ /Ce ϳ1.6% and an oxide ratio CeO/Ce ϳ0.2% (at 10 g/liter lithium, yttrium, and cesium). Gel filtration columns were connected to a Hewlett-Packard HP1100 ChemStation HPLC system incorporating a UV monitor set at 280 nm. The outlet of the UV detector was connected directly to the Babington-type nebulizer of a Hewlett-Packard HP4500 ICP-MS. In initial experiments, the following isotopes were monitored: 51 V, 55 Mn, 57 Fe, 59 Co, 60 Ni, 64 Zn, Cu, 66 Zn, and 68 Zn were measured. 10 -40 pmol of myoglobin, carbonic anhydrase, and azurin were run to calibrate the peak areas of the ICP-MS traces for iron, zinc, and copper, respectively. For each standard protein, the respective metal was found to be fully bound to the protein.
For ICP-MS experiments, we used either a Superose 6 HR 10/30 column (Amersham Pharmacia Biotech), with 0.05 M Tris-HCl, pH 7.4, containing 0.15 M NaCl, at a flow rate of 0.3 ml.min Ϫ1 , or a SigmaChrom GFC1300 column (30 cm ϫ 7.5 mm; Supelco) run at 0.5 ml.min Ϫ1 in 0.05 M Tris-HCl, pH 7.4, containing 0.2 M NH 4 Cl. We sometimes (as specified below) injected samples of DTPA (90 l, 2 mM) to purge metal ions from the chromatography system. Photometric Assay for Zinc Release-Release of zinc from thiol-free NOS was monitored using the absorbance change of the metal-binding dye PAR at 500 nm (24,25). PAR was added to 200-l samples to give a final concentration of 100 M. The response was calibrated from 0 to 10 M using solutions of zinc and/or copper acetate in the same buffer present in the enzyme samples (0.05 M Tris-HCl, pH 7.4, containing 0.15 M NaCl). The slope of the absorbance increase was 76 mM Ϫ1 for zinc and 43 mM Ϫ1 for copper. The chelator NTA, when added to a final concentration of 2 mM, was found to completely abolish the response to zinc, whereas the signal for copper was only reduced to 33 mM Ϫ1 .
The PAR reaction was used to measure release of zinc from NOS on treatment with PMPS. PAR was added to samples of NOS (3-5 M), and the absorbance spectrum was recorded. After adding 4 l of PMPS (final concentration 0.16 mM), the absorbance at 500 nm was measured at intervals of 10 s for 200 s.
ICP-MS Analysis and NOS Assay of Zinc-depleted NOS-Samples of NOS (5 M; 400 l), kept on ice, were treated with the following combinations of PMPS (0.15 mM), DTPA (2 mM), and L-cysteine (10 mM). To a control sample (control) were added DTPA and L-cysteine; this sample was incubated for 3 min. To a second sample (PMPS-only), PMPS was added, followed 1 min later by DTPA (2 mM) and 2 min of further incubation. To a third sample (PMPS/cysteine), PMPS was added, followed 1 min later by DTPA and after a further 2 min by cysteine. Immediately after these incubations, three 110-l portions of each sample were frozen in liquid N 2 for ICP-MS analysis. A further 10-l portion was taken from each sample for NOS assays.
Of each 110-l sample for ICP-MS, 90 l was injected onto the SigmaChrom GPC 1300 column. Samples were injected in the following order: first, two samples of 2 mM DTPA, then standards in the range 0 -10 M, then 2 ϫ 2 mM DTPA, then two injections of 2 mM DTPA plus 0.3 mM PMPS, and then the three PMPS-only replicate samples. Next, the system was re-equilibrated for 1 h in buffer containing 10 mM L-cysteine followed by 2 ϫ 2 mM DTPA, and the control samples were applied. A single injection of DTPA was made between these and the PMPS/cysteine samples.
Gel Filtration-NOS dimerization was analyzed by gel filtration with a Superose 6 HR 10/30 column on an Ä KTA chromatography apparatus, at 10°C. The flow rate was set to 0.3 ml.min Ϫ1 and the buffer used was 0.05 M Tris-HCl, pH 7.4, containing 0.2 M NH 4 Cl. Samples of thiol-free NOS (400 l; 1.6 M) were treated with PMPS (0.15 mM) for 3 min on ice then either directly injected onto the column or re-reduced by adding L-cysteine (10 mM). Before PMPS-only samples were injected, a sample of PMPS (400 l; 0.3 mM) was passed through the column. For samples to which L-cysteine had been added, the column was first re-equilibrated in buffer supplemented with 10 mM L-cysteine. For samples that had been preincubated with H 4 biopterin, the column was equilibrated in buffer containing 10 M H 4 biopterin.
Gel Electrophoresis-Formation of SDS-resistant NOS dimers was investigated using the low temperature PAGE method of Klatt et al. (8). Enzyme samples were preincubated in a total volume of 70 l at an NOS concentration of 1.6 M in 50 mM Tris-HCl, pH 7.4. Some of the samples were preincubated with H 4 biopterin (0.2 mM) for 20 min on ice; of these, some were further treated with PMPS (0.15 mM) on ice for 3 min, before adding sample buffer. From each 70-l preincubation, two portions of 32 l each were taken and mixed with 8 l of 5-fold Laemmli buffer, with ("reducing gel") or without ("non-reducing gel") 2-mercaptoethanol. The other components of the 5-fold sample buffer were 0.32 M Tris-HCl (pH 6.8), 0.5 M glycine, 10% SDS, 50% glycerol, and 0.03% bromphenol blue (27). Samples containing 8 g of NOS were subjected to SDS-PAGE on 5.5% SDS gels, using the Mini Protean II system from Bio-Rad (Vienna, Austria). Gels and buffers were equilibrated at 4°C, and the buffer tank was cooled during electrophoresis in an ice bath. Proteins were visualized by Coomassie Brilliant Blue R staining.

RESULTS
Zinc Content of NOS-Metal content of NOS samples was measured initially by on-line gel-filtration/ICP-MS with a Superose 6 column. Iron, zinc, and copper were detected in rat nNOS expressed in baculovirus-infected Sf9 cells and in NOS isolated from pig brain (Fig. 1). The samples were also monitored for cobalt, manganese, molybdenum, nickel, and vanadium, but none of these metals were detected. The pig brain NOS contained 0.60 mol of zinc and 0.35 mol of copper/mol of NOS; the baculovirus-expressed rat nNOS contained 0.49 Ϯ 0.13 mol of zinc/mol of NOS and 1.35 Ϯ 0.18 mol of copper/mol of NOS (means Ϯ S.E. of three preparations).
Values for the baculovirus-expressed rat nNOS, obtained with the SigmaChrom GPC 1300 column, are shown in Table I. We were surprised to find so much copper, and set out to characterize how tightly the metals were bound.
First, we checked the effect of flushing our chromatography system with DTPA. In the initial samples, the small-molecule fraction at the end of the chromatograms had contained peaks of all metals (with no UV signal) 20-to 50-fold larger than those associated with NOS. Repeated injections of DTPA reduced these peaks to a ϳ10-fold smaller value that was stable from the third injection onwards. After treating the column in this way, considerably less copper eluted with the NOS samples (Table I). When the column was flushed with DTPA as well as adding DTPA to the samples, the bound copper was reduced further and the bound zinc was almost halved (Table I). When NOS samples had been desalted to remove 2-mercaptoethanol prior to analysis, the zinc was not affected but bound copper was reduced to undetectable levels (Table I).
Zinc Release from nNOS by PMPS-Because a considerable fraction of the zinc bound to NOS was not removed by chelators and a previous study reported that use of oxidants was necessary to obtain release of the zinc (16), we tested the mercurial reagent PMPS, which is able to oxidize thiol groups reversibly, as a means of releasing zinc from NOS. Initially, we used the PAR assay to detect metal release. A maximal increase in absorbance at 500 nm was reached at about a 15-fold molar excess of PMPS over nNOS (data not shown); the absorbance changes were complete within 2 min (Fig. 2). For baculovirusexpressed rat nNOS, we obtained an estimate of 0.49 Ϯ 0.02 mol of zinc released per NOS subunit (mean Ϯ S.E. of three preparations). We also measured release of zinc from NOS by PMPS using on-line gel-filtration/ICP-MS. Three-quarters of the bound zinc was released (from 0.28 Ϯ 0.03 to 0.07 Ϯ 0.01 mol of zinc/mol of NOS; means Ϯ S.E. of three preparations).
Effects of PMPS on Dimerization of nNOS-We examined the effects of PMPS on dimerization of nNOS by gel filtration. Thiol-free nNOS treated with PMPS underwent substantial dissociation, from 83 Ϯ 2% to 40 Ϯ 3% dimer (means Ϯ S.E. of three preparations) (Fig. 4, A and B). The monomers generated by PMPS treatment exhibited a peak in the 398-nm trace (dotted line in Fig. 4B). Addition of L-cysteine to the PMPSoxidized enzyme resulted in a significant reversal of the dissociation, reaching a final value of 50 Ϯ 4% dimer. L-Cysteine alone was found to cause a slight dissociation (to 65 Ϯ 1% dimer).
Because H 4 biopterin has been shown to stabilize NOS dimers, this experiment was repeated with nNOS that had been preincubated with a saturating concentration of H 4 biopterin before adding PMPS, and using an elution buffer supplemented with H 4 biopterin (Fig. 4, C and D). Under these conditions, PMPS treatment caused no significant dissociation (81 Ϯ 3% to 79 Ϯ 3% dimer). In the PAR assay, preincubation of thiol-free nNOS with H 4 biopterin did not affect the extent of zinc release. However, preincubation with H 4 biopterin did not prevent inactivation of the enzyme (to 7.6 Ϯ 3.6% of controls; mean Ϯ S.E. of three preparations).

Redox and PMPS Effects on Formation of SDS-resistant NOS Dimer-
The ability of thiol-free nNOS to form SDS-resistant dimers was examined by low temperature SDS-PAGE (Fig. 5). Samples that had been preincubated with or without H 4 biopterin and/or PMPS were split and subjected to electrophoresis under reducing (5% 2-mercaptoethanol (0.6 M) in the sample buffer) or non-reducing (thiol-free sample buffer) con- Proteins were chromatographed on a Superose 6 HR column with on-line UV and ICP-MS detection, as described under "Experimental Procedures." A, baculovirus-expressed rat nNOS; B, pig brain NOS. The ICP-MS signals have been offset for clarity. The ratio of the left and right y axis scales is the same in both panels. For baculovirus-expressed rat nNOS, the content of these metals was calculated based on the amount of protein injected, determined by the Bradford assay. The pig brain NOS sample contained a significant protein contaminant smaller than NOS, so that the total protein concentration of these samples could not be used as a basis for calculating its metal content; instead, this was estimated relative to the peak area of the 280-nm trace, assuming the same extinction coefficient as for baculovirus-expressed enzyme. ditions. On reducing gels, control samples to which no additions had been made contained a small proportion of SDSresistant dimer (8.2 Ϯ 3% (mean Ϯ S.E. of three preparations)), which increased significantly in the samples preincubated with H 4 biopterin (to 38 Ϯ 5%). This is the familiar behavior observed in earlier studies (8). The same samples applied to non-reducing gels exhibited the same pattern, except that the basal formation of SDS-resistant dimer without added H 4 biopterin was more pronounced (20 Ϯ 2%), rising to 38 Ϯ 2% on preincubation with H 4 biopterin. Essentially the same behavior was observed for thiol-free nNOS. On reducing gels, H 4 biopterin increased the fraction of dimer present from 8.4 Ϯ 2% to 41 Ϯ 2%, and on non-reducing gels from 9.6 Ϯ 5% to 23 Ϯ 7%. However, on the non-reducing gels an additional band was observed above the "normal" SDSresistant dimer. This species was resistant to heating at 95°C ("boiled" sample on non-reducing gel) but disappeared when the same samples were electrophoresed in the presence of 2-mercaptoethanol.
When thiol-free nNOS that had been preincubated with H 4 biopterin was treated with PMPS, the "normal" SDS-resistant dimer band disappeared completely. The additional band of lower mobility was not affected. When a PMPS-treated sample was applied to a reducing gel, the SDS-resistant dimer content recovered to 23 Ϯ 9%. "Normal" nNOS, which already contained 12 mM 2-mercaptoethanol, was not affected by PMPS. Thus, as observed under native conditions, the effect of PMPS on SDS-resistant dimerization was antagonized by excess thiol.

DISCUSSION
Occurrence of Bound Zinc in nNOS-NO synthase has long been known to contain iron (28 -30); interest in other metals has been more recent. Several groups have found zinc in NOS (as isolated, in eNOS or nNOS (3,16,19) or inserted in vitro into iNOS (11,12,31)). The present results show that zinc is bound to baculovirus-expressed rat nNOS and, for the first time, to NOS isolated from a natural tissue (pig brain).
The amount of zinc bound to NOS was substantially reduced by flushing the chromatography system with DTPA and adding DTPA to the NOS samples. The latter condition being more stringent, we suppose that the lower zinc values most accurately represent the zinc bound to high affinity sites such as that linking the subunits. The results obtained without this precaution suggest that NOS (unsurprisingly for a large protein) can also bind metals adventitiously. A comparison of Fig.  1 with Fig. 3 also suggests that the more stringent conditions removed nearly all metal from the NOS monomers (the sample in Fig. 3A contained about 30% monomer). The Fe:Zn ratio using the more stringent method was 2.7:1, suggesting that three-quarters of the intact nNOS dimers with bound heme also contained zinc. Similarly, the large and variable amount of copper found in our initial experiments may well have resulted from the NOS picking up copper on its way through the chromatography system. Use of DTPA gave a value close to 1 mol/mol. Exposure of the NOS to slightly oxidizing conditions (removal of 2-mercaptoethanol) caused complete loss of copper. Perry and Marletta (17) also found close to 1 mol of copper/mol of NOS, which was easily removed.
Release of Zinc by PMPS-PMPS is a mercurial reagent that can reversibly oxidize thiol groups to mercaptides, in the process releasing metals bound to the thiols (24,25,32). The amounts of zinc it released from NOS, as detected by the PAR assay (ϳ0.5 mol/mol of NOS), were greater than those detected by gel filtration/ICP-MS of thiol-free NOS samples (ϳ0.3 mol/ mol of NOS). A component of the PAR signal seems to be zinc that is not tightly bound to NOS and separates from the enzyme on the column. Thus caution is needed in interpreting the PAR results in terms of occupancy of the intersubunit site.
When we used NTA to mask the zinc signal in the PAR assay, we never detected any release of copper, agreeing with the ICP-MS result that the copper was completely lost from the thiol-free enzyme.
Zinc Depletion and NOS Activity-The inactivation of the enzyme by PMPS was not surprising, because the stabilizing effect of reduced thiols on NOS activity, most easily explained by an essential role of one or more reduced cysteines, is well known (33)(34)(35). NOS contains several cysteines besides the zinc ligands, so that the inactivation is not necessarily related to the zinc site. Having found that excess reduced thiol restored virtually full activity to the PMPS-modified enzyme, we looked for conditions that would keep the zinc out of its binding site despite the reversal of the PMPS modification. We found that the presence of DTPA was sufficient to limit the zinc binding to the re-reduced enzyme to 43 Ϯ 16% of the starting value. The resulting enzyme regained 93 Ϯ 10% of its original activity (before addition of PMPS), reaching a specific activity of over 400 nmol of citrulline.mg Ϫ1 .min Ϫ1 with a zinc content of only 0.12 mol/mol of NOS. This is a substantial divergence of NOS activity from zinc content.
The samples used for enzyme assays were diluted from those used for ICP-MS, and it could be objected that this might tilt the conditions toward greater zinc binding in the NOS assay than was measured by ICP-MS. To exclude this, we raised the molar ratio of DTPA to NOS from 400:1 in the samples analyzed for zinc to 2000:1 in the diluted stocks and the enzyme assays. DTPA is compatible with the NOS assay because its affinity for Zn 2ϩ is about 10 8 -fold higher than for Ca 2ϩ (36). Thus we could provide sufficient free Ca 2ϩ to support NOS activity without impairing the ability of the DTPA to bind zinc. We also used the highest practicable NOS concentration (ϳ0.1 M) in the assays, to maximize the ratio of NOS to possibly available zinc.
The only previous experiment to test whether zinc is needed for NOS activity was the reactivation of the zinc-deficient C331A mutant by extended preincubation with L-Arg, reported by Miller et al. (16). In that case, the zinc content of the enzyme was, apparently, only determined before the preincubation, and no evidence was shown to exclude that zinc bound to the enzyme during the preincubation and assay. It is difficult to reduce background zinc to levels below the usual enzyme concentrations used in NOS assays (typically 5-20 nM). Thus the interpretation of this result as evidence against an essential role of zinc seems to rest on the assumption that the mutant enzyme would not bind zinc. Although this seems fairly plausible, the affinity of the wild-type zinc site is such that it can compete partially with a 400-fold excess of DTPA (the dissoci-ation constant for the DTPA-zinc complex is around 10 Ϫ18 M (36)), and even the mutated site with only two ligands might still have considerable affinity. Thus we find that our use of a protein-chemical approach with the wild-type enzyme is an important complement to this result.
Effects of PMPS on Dimerization of NOS-In parallel with the release of zinc, PMPS caused substantial dissociation of the NOS dimer, which could be reversed by adding reduced thiol. We have not shown definitively that this effect is due to the disruption and reconstitution of the zinc site. However, it is not obvious what other kind of dimer-stabilizing interaction except for a metal site could depend on keeping a protein thiol in a reduced state. In NOS there is one other such site, namely, the proximal cysteine thiolate ligand of the heme. PMPS did not affect the intensity of the heme Soret band (data not shown), nor did it cause any significant loss of iron (measured by ICP-MS). Zinc release thus appears to be the most likely explanation for PMPS effects on dimerization.
The PMPS-induced dissociation is untypical for nNOS in two respects. First, the monomers formed in this fashion clearly retain some heme. Before now, we generally found that monomers of nNOS were heme-free, and that heme binding implied dimerization (20,37). Second, the dissociation was prevented by H 4 biopterin. Previously, we have not observed large effects of H 4 biopterin on the net dimerization of nNOS, rather, the predominant effect was a stabilization of already existing dimer. In contrast, for iNOS both heme-containing monomers and pteridine-dependent conversion of monomers to dimers are well documented (9, 38 -40). Interestingly, iNOS, as isolated by two out of three groups, was zinc-free and only bound zinc after extra treatment with reductants (11,12). Thus, artificially depleting nNOS of zinc seems to convert it to a similar mode of dimerization to iNOS, which may be naturally zinc-deficient.
Formation of SDS-resistant Dimer-Although PMPS did not cause net subunit dissociation in the presence of H 4 biopterin, it did block conversion of the NOS dimer into the SDS-resistant form. Before now, this conversion has been consistently found SCHEME 1. Summary of NOS dimerization.
to depend on pteridine binding (8,19,41,42). The present result suggests that, in addition to bound H 4 biopterin, an intact zinc bridge between the subunits is essential to make the dimer resistant to dissociation by SDS. Indeed, this type of linkage would fulfill exactly the known characteristics of the dimer observed in low temperature SDS-PAGE, namely, resistance to SDS (the zinc effectively forms a covalent bridge) and to 2-mercaptoethanol (which would disrupt disulfide linkages, but stabilize the zinc site by keeping the cysteines reduced). Why should such a structure require additional stabilization through H 4 biopterin binding? The key might be an effect of H 4 biopterin on the degree of exposure of the zinc site to solvent. Simply by excluding solvent from its own binding site, H 4 biopterin might significantly protect the zinc site from oxidants or from potential chelators. This could reconcile the observed dependence of SDS resistance on H 4 biopterin with the finding that pteridine binding does not involve large conformational changes of the protein (3).
We have attempted to integrate these and previous findings on NOS dimerization in Scheme 1. We previously studied hemedependent dimerization of heme-free nNOS (species I in Scheme 1) (20,37); based on the present results and on the positive effect of dithiothreitol on heme-reconstituted NOS (37), we now suspect that this process may have been assisted by binding of background zinc. Species III is fully active, native NOS. Removal of thiol from the medium and treatment with PMPS leads to the inactive, PMPS-substituted enzyme (Scheme 1, species IV), which is susceptible to dissociation to heme-containing monomers (species V). The dissociation can, however, be prevented by H 4 biopterin and thus resembles the equilibrium between heme-containing monomers and pteridine-bound dimers, observed for iNOS (9, 38 -40). In the absence of available zinc, reduction of the PMPS-modified enzyme by thiol results in an active, zinc-free form (Scheme 1, species VII). Exposure of NOS to mildly oxidizing conditions may lead to gradual loss of zinc and formation of an intersubunit disulfide bond (Scheme 1, species VI), which may be the additional dimer band on our non-reducing gels, and has also been observed in crystal structures (10,12).
If zinc binding is necessary for conversion of NOS dimers into an SDS-resistant form, but not for NOS activity, then the SDS-resistant state is not essential for NOS catalysis. The role of the zinc site seems to be stabilization of the enzyme on a time-scale of many catalytic cycles. Exposing NOS to a thiolfree environment caused the appearance of a new band on the low temperature SDS-gels, which was resistant to boiling but sensitive to 2-mercaptoethanol, suggesting that it was dependent on a disulfide bond. Intersubunit disulfide links have been observed in crystal structures of zinc-free NOS. Thus, under some in vitro conditions, conversion between reduced, zinclinked and oxidized, disulfide-linked NOS dimers seems possible. Whether NOS ever encounters sufficiently oxidizing conditions in vivo to cause zinc release, is an interesting question, particularly in two respects. First, NOS itself is suspected to contribute to oxidative stress in a variety of ischemic or inflammatory states, and thus the release of zinc could offer a lastresort mechanism to destabilize and ultimately inactivate the enzyme. Second, it is tempting to link the lack of bound zinc in iNOS to the fact that this isozyme must routinely operate under relatively oxidizing conditions, for example in activated macrophages.