L-Thiocitrulline A STEREOSPECIFIC, HEME-BINDING INHIBITOR OF NITRIC-OXIDE SYNTHASES”

Nitric-oxide synthase (NOS) catalyzes the oxidation of L-arginine to citrulline and nitric oxide (NO). The en- zyme is inhibited by a variety of Nu-monosubstituted L-arginine analogs, and some of these compounds are useful in reversing pathologies associated with the overproduction of NO (e.g. the hypotension of septic shock). We report here that L-thiocitrulline (y-thioureido-L-nor-valine) is a potent, stereospecific inhibitor of the consti- tutive brain and endothelial isoforms of NOS as well as the isoform induced in vascular smooth muscle cells by lipopolysaccharide and interferon-?. Steady state kinetic studies show L-thiocitrulline inhibition is competitive with L-arginine (Ki - 42W0 of Kkg), indicating that initial binding is as a substrate/product analog. In contrast to L-arginine and Ne-methyl-L-arginine, the proto- typic NOS inhibitor, L-thiocitrulline binding elicits a ’Type 11” difference spectrum, indicating a high spin to low spin transition of the iron in the heme cofactor. This finding suggests that L-thiocitrulline is contributing the sixth ligand to heme iron, probably through the thioureido sulfur. Such interaction with heme iron neither stimulates nor inhibits the direct flavin-mediated cytochrome c reduction activity of the enzyme, but it does inhibit heme-dependent superoxide formation. In uiuo, L-thiocitrulline is a potent pressor agent in both normal and endotoxemic rats, the latter finding suggesting utility in treating the hypotension of septic shock. the iliac artery was connected to a pressure transducer and used to monitor blood pressure. For studies in septic rats, abdomens were opened under anesthesia, and the cecum was isolated and ligated with a silk suture. A small (-4 mm) incision was made proximal to the ligation and fecal contents were expressed, allowing development of peritonitis (48). L-arginine and then oxidation of NOH-Arg to conversion of L-thiocitrulline to citrulline (37); this result has not been Note that in preliminary studies we saw time- and NOS-dependent reproducible. Although L-thiocitrulline is stable in conventional buffers exhibit some nonenzymatic degradation to citrulline (and other prod- and is not affected by dithiothreitol, EDTA, Ca”, or calmodulin, it does ucts) with redox active cofactors (FAD, FMN). Such degradation re- quires several hours at 25 “C, however, and would seem to be too slow to account for our previous result. We are continuing to examine the stability of L-thiocitrulline.

Although there is only modest sequence homology among the three NOS isoforms (50-60%) (7, 81, all have a C-terminal domain that is somewhat homologous to cytochrome P-450 reductase (7, 9) and an N-terminal domain that is hypothesized to have cytochrome P-450-like activity (10). Consistent with this view, the C-terminal domain has sequences suggesting binding sites for NADPH and the required cofactors FAD and FMN; the N-terminal domain binds heme and arginine (9)(10)(11). Mechanistic studies suggest that catalysis by all isoforms involves two successive monooxygenase reactions, the first a ferry1 hemedependent hydroxylation of a guanidino nitrogen of arginine yielding NOH-Arg and the second a nucleophilic attack by peroxy heme on the guanidino carbon of NOH-Arg or on the one electron oxidation product of NOH-Arg (1,2,(12)(13)(14). Both steps find precedent in cytochrome P-450 reductase/cytochrome P-450 biochemistry (1, [13][14][15]. Electron flow is believed to follow the sequence NADPH + FAD 4 FMN + heme + 0, (13,14); in nNOS, the FMN 4 heme transfer is dependent on Ca2+/ calmodulin (16).
Reports over the past 7 years have implicated NO production by constitutive NOS isoforms in a wide variety of physiological processes including blood pressure homeostasis (17,181, control of organ perfusion (191, inhibition of platelet adhesion and aggregation (201, and neurotransmission including control of gastrointestinal sphincters (21,221. nNOS may also have a role in subtle central nervous system effects such as long term potentiation and long term depression (7, 23). iNOS-mediated NO production, which can be quantitatively substantial, is implicated in the cytotoxickytostatic activity of macrophages (24,251 and perhaps other cells (26). NOS-inhibitory N"-monosubstituted L-arginine derivatives (e.g. N"-methyl-L-arginine (L-NMA) (1, 24) and Nu-nitro-L-arginine (L-NNA) (1, 27)) have been extremely valuable in elucidating the NOS dependence of physiological phenomena studied in cells and whole organisms, It has also been found that NOS inhibitors are of potential therapeutic value in suppressing the overproduction of NO that accounts at least in part for the postischemic reperfusion injury of stroke (281, the hypotension of cytokine-induced and septic shock (29,301, and the tissue damage associated with inflammatory conditions, including arthritis (31,32).
The research and clinical utility of NOS inhibitors is dependent on both their specificity and their potency of inhibition. To date, NOS specificity has been best achieved with arginine analogs, some of which are tight-binding inhibitors (e.g. L-NNA, Refs. 33 and 34) and some of which are mechanism-based inhibitors (e.g. L-NMA, Refs. 35 and 36). In the present studies, we have examined L-thiocitrulline (373, a novel amino acid that is recognized by the argininehitrulline binding site of NOS and that appears to interact as an axial sixth ligand of the heme cofactor.
Rat nNOS was isolated from stably transfected kidney 293 cells (9) as described (10). Studies with iNOS were carried out with a crude homogenate of cultured rat aortic smooth muscle cells grown in the presence of lipopolysaccharide and interferon-y to induce iNOS expression. The properties of this enzyme have been reported (42).
Reversibility of NOS inhibition was tested by preincubating nNOS with inhibitor, NADPH, and cofactors and then removing portions of that mixture at intervals and assaying for residual activity on the basis of NO-mediated conversion of oxyhemoglobin to methemoglobin (43).
NADPH-dependent reduction of cytochrome c was determined spec-Na' HEPES buffer, pH 7.5,50 p~ EDTA, 50 p NADPH, 50 1.1~ bovine trophotometrically (IO); the 500-pl reaction mixtures contained 50 mM heart cytochrome c, 0-1 m~ L-thiocitrulline, and 2.4 pg of nNOS (specific activity = 200-220 nmol of cytochrome c reduced per min.mg). Cytochrome c reduction was monitored a t 550 nm ( E = 2.1 x 104/~). In some studies, 2.0 m~ CaCI, and 10 pg/ml calmodulin were added to the reaction mixtures to allow reduction of heme iron and superoxide formation (11,16). Possible product formation was examined in 200-pl reaction mixtures containing 50 mM Na' HEPES buffer, pH 7.5,IOO p~ EDTA, 2 mM CaCI,, 10 pg/ml calmodulin, 25 PM FAD, 25 p~ FMN, 50 p~ THB, 100 p~ dithiothreitol, 100 pg/ml bovine serum albumin, 500 p NADPH, and 100 p potential substrate (L-thiocitrulline, L-arginine, or L-NMA). Reaction was initiated by the addition of 36 pg of nNOS, and the reaction mixtures were placed a t 25 "C. At various times, a 40 F.1 portion was removed, placed in a boiling water bath for 1 min., and then frozen in Dry Ice/methanol. For analysis, the frozen solutions were thawed and centrifuged and a 20-pl portion of the supernatant was derivatized with o-phthalaldehyde (44) and fractionated by reverse phase HPLC as described (45). In this system, citrulline, thiocitrulline, and ornithine elute at 10, 13.5, and 25 min, respectively (flow rate = 1.7 ml/min). In replicate studies, the supernatants were analyzed by conventional ion exchange chromatography (Beckman model 6300 amino acid analyzer) using the elution scheme designed for protein hydrolysates.
The effect of NOS inhibitors on methacholine-induced aortic ring relaxation was determined as described (46). Briefly, thoracic aortic rings (2-3 mm) from male New Zealand white rabbits were suspended in 15-ml organ baths containing Kreb's bicarbonate buffer (118 mM NaCl, 4 mM KCl, 3 m~ CaCI,, 24 mM NaHCO,, 1.4 m~ KH,PO,, 1.2 m~ MgSO,, and 11 mM glucose, pH 7.4) at 37 "C and continuously gassed with a 95% 0,, 5% CO, mixture. Isometric tension was measured with force-dependent transducers (Grass Instrument Co, Quincy, MA) and recorded on a polygraph interfaced with an analog-to-digital convertor. Resting tension was adjusted to a length tension maximum of 2 g, and the rings were equilibrated for 1 h. After confirming responsivity to 40 mM KC1 (3.7 * 1.2 g of contraction), rings were washed repeatedly with Kreb's buffer and precontracted with 1 p norepinephrine. M e r stabilizing 5-10 min, a dose-response curve to methacholine (10-9-10-5 M) was performed, and percent relaxation to each dose was recorded. The rings were then rinsed with Kreb's buffer and equilibrated with Lthiocitrulline or L -N M A (1-300 1.1~) for 10 min. The rings were then contracted with 1 p norepinephrine, and the dose-response curve for methacholine was repeated. Each ring served as its own control; each concentration of inhibitor was examined three times.
Titration of nNOS by optical difference spectrophotometry was performed as described previously (47) using a Shimadzu 2101 Dual-Beam UV/visible spectrophotometer equipped with a Peltier temperature controller. Samples (0.5 ml) of nNoS (prepared in 50 mM Tris-HCl buffer, 10% glycerol, 0.1 m~ EDTA, pH 7.5) were placed in the sample and reference cuvettes, and the absorbance difference was adjusted to base line. All compounds used in the titrations were dissolved in 50 mM Tris-HC1 buffer, 10% glycerol, pH 7.5, and did not possess intrinsic absorbance at wavelengths used in the difference titrations. Titrations were conducted at 15 "C. Additions of perturbant to the sample cuvette were made using a Hamilton syringe, followed by mixing and recording of the resultant difference spectrum. The total change in sample volume was ~2 % .
Spectra were normalized to zero absorbance as described in the figure legends.
Male Sprague-Dawley rats (200 and 450 g) were used for all in vivo studies. Anesthesia was induced with ketamine (50 mgkg intramuscularly) and acepromazine (5 mgkg intramuscularly) and maintained throughout the study by additional intramuscular injections of the mixture. Rats were placed on a heated table thermostatically controlled to 37 "C. Venous cannula were inserted either into the iliac or internal jugular vein for administration of inhibitors. A cannula in the iliac artery was connected to a pressure transducer and used to monitor blood pressure. For studies in septic rats, abdomens were opened under anesthesia, and the cecum was isolated and ligated with a silk suture. A small (-4 mm) incision was made proximal to the ligation and fecal contents were expressed, allowing development of peritonitis (48). For comparison, 100 pM L-NM,A inhibited about 90%. With both isoforms, product increased almost linearly with time, offering no suggestion of significant irreversible inhibition during the 10.5-min time course examined ( Fig. 1, 10 p~ L-thiocitrulline). As reported by others (4% L-citrulline (100 p~) was not an inhibitor of either NOS isoform.

Kinetic Characterization of nNOS and iNOS Inhibition by L-Thiocitrulline and Related Compounds-In
To examine the possibility that small thioureas might inhibit, N-methylthiourea (100 p~ and 1 mM) was also tested; it did not inhibit (not shown).
Initial-rate kinetic studies were carried out over a range of ~-['~C]arginine and L-thiocitrulline concentrations to determine the nature of nNOS and iNOS inhibition by L-thiocitrulline ( Fig.  2, a and b, respectively). As shown, inhibition was competitive  Reversibility of Inhibition by L-Thiocitrulline-Although initial inhibition of NOS by L -N M A is competitive with arginine and fully reversible, NOS undergoes mechanism-based irreversible inhibition by L-NMAwith time (35,361. Similar studies were carried out with L-thiocitrulline. As shown in Fig. 3, nNOS is moderately unstable when incubated with NADPH and cofactors, but the addition of L-thiocitrulline does not accelerate the rate of inactivation. In contrast, L-NMA substantially in- Inhibition of eNOS-dependent Aortic Ring Contraction by L-Thiocitrulline-nNOS and iNOS are soluble isoforms that are readily available via overexpression (nNOS) or cytokine-mediated induction (iNOS). In contrast, eNOS is membrane-bound and present in very low abundance in the COS expression system available to us (8); detailed kinetic studies were not possible. On the other hand, methacholine-induced relaxation of aortic rings precontracted with norepinephrine is an eNOSdependent phenomenon that is readily quantitated. Studies were carried out using rabbit aortic rings exposed for 10 min to 1-300 p~ L-thiocitrulline and then relaxed by cumulative addition of 0.003-10 p methacholine (9 concentrations); percent relaxation was assessed after each addition. Similar studies were carried out with rings exposed to L-NMA. Percent relaxation was calculated based on results with the same rings prior to inhibitor treatment.
At all concentrations tested, L-thiocitrulline and L-NMA were approximately equipotent in blocking relaxation; results for 10 p~ methacholine are shown in Fig.  4. Although L-thiocitrulline was slightly more potent than L-NMA at all concentrations tested, the differences were not statistically significant. We note that inhibition presumably depends to some extent on the presently unknown rates of inhibitor transport into the endothelial cells of the rings and, for this reason, rigorous kinetic comparison of L-thiocitrulline and L-NMA must await an adequate supply of eNOS.
Spectral Evidence for Interaction of L-Thiocitrulline with NOS-McMillan and Masters (47) have previously shown that the heme iron of nNOS as isolated is mainly high spin. However, titration of the enzyme with L-arginine, NOH-Arg, or L-NMA results in a Type I difference spectrum (a peak a t -380 nm, a trough at -420 nm, and an isosbestic point at -405 nm), Inhibitor (pM)

FIG. 4. Effect of NOS inhibitors on methacholine-induced aortic ring relaxation.
Rabbit aortic rings in organ baths were prepared and contracted with norepinephrine as described under "Methods." After determining percent relaxation to increasing doses of methacholine, the rings were equilibrated with 10-300 PM L-thiocitrulline (M) or L-NMA(O), and the dose-response curve with methacholine was repeated. The extent to which the NOS inhibitor blocked relaxation was calculated for each concentration of methacholine tested. The results for 10 PM methacholine are shown; qualitatively similar results were obtained at all other methacholine doses tested (ie. L-NMA was a slightly less effective inhibitor than L-thiocitrulline). The data are shown as means -c S.D. for triplicate measurements.
suggesting that binding of these substrates (or inhibitor) results in a low spin to high spin heme iron transition in that subpopulation of the enzyme that was not initially high spin. Because low spin heme iron is associated with the hexacoordinate state, the transition was inferred to occur by displacement of an amino acid side chain or water molecule as the sixth axial ligand of heme iron (47, 50). Fig. 5 shows the results of a spectral titration of 7.7 p~ nNOS with L-thiocitrulline (final concentrations = 0.4-9.0 p~). These difference spectra were normalized to zero absorbance at 400 nm for presentation. An isosbestic point of 427 nm for the L-thiocitrulline (200 p~)perturbed difference spectrum was obtained by arithmetic subtraction of the absolute spectra using 8.8 PM nNOS (data not shown). The resulting difference spectra reflect a mixture of high spin and low spin species produced by this single compound. Since L-thiocitrulline, but not L-arginine or L-citrulline, produces this effect, the implication is that the thiol group interacts directly with the heme-iron, i.e. heme-ligation (see "Discussion"). As shown, L-thiocitrulline causes a perturbation in the absorbance spectrum of the heme Soret transition band characterized by a trough at -392 nm, a shoulder at -413 nm, and a peak at 435 nm. Spectral perturbation was stereospecific; D-thiocitrulline did not elicit heme spectral changes at concentrations up to 10 VM (data not shown). The inset of Fig. 5 shows a double reciprocal plot of the AA (A,,, -A39,) uersus perturbant concentration that was used to estimate a spectral binding constant of 3 (uncorrected for bound ligand, which was <5% of total). In three separate determinations, the resulting spectral binding constants ranged from 1-3 p.
For comparison, Fig. 6 shows the difference spectra obtained upon addition to 3. imidazole, which is a typical Type I1 difference spectrum characterized by a peak at 432 nm and a trough at 393 nm. L-Homothiocitrulline, which contains an additional methylene group, elicits a Type I difference spectrum, suggesting no direct sulfur-iron ligation. The inset of Fig. 6 shows the difference spectrum obtained with the nitrogenous ligand, imidazole, which represents a typical Type I1 interaction with a peak at 432 nm and a trough at 393 nm and, hence, the L-thiocitdline spectral perturbation is best described as a modified Type 11.
To further elucidate L-thiocitrulline binding, L-arginine and imidazole were used as additional perturbants (Fig. 7 ) . Difference spectra were obtained with 8.8 VM nNOS following the addition of L-thiocitrulline to a final concentration of 10 VM (Fig.  7 A , curue A ) , followed by additions of L-arginine to 100 VM (curue B ) and then L-thiocitrulline to 1010 p (curue C ) . As shown, L-thiocitrulline produces a modified Type I1 difference spectrum with a shoulder at -410 nm and a bathochromicshifted peak at -435 nm. Upon the addition of L-arginine, a typical Type I difference spectrum is obtained, while the final addition of L-thiocitrulline accentuates the 410-nm shoulder and restores the modified Type I1 difference spectrum. In the studies shown in Fig. 7 B , nNOS was adjusted t o a final concentration of 1000 PM imidazole to shift most of the enzyme to low spin heme iron. The addition of 300, 700, and 1100 p~ t-thiocitrulline to this sample produced red-shifted Type I spectra. Estimation of the spectral binding constant yielded a value of -27 VM (uncorrected for bound ligand); the -10-fold increase in K, over that seen with L-thiocitrulline alone presumably reflects the competition between L-thiocitrulline and imidazole for access t o the heme-containing region of the active site.
Effect of L-Thiocitrulline on Cytochrome c Reduction by nNOS-It has been suggested that Ca2+/calmodulin serves to align the reductase and heme-containing oxidase domains of nNOS, facilitating transfer of electrons from flavins to heme (16). In the absence of Ca'+/calmodulin, heme is not reduced, and nNOS catalyzes the flavin-mediated, NADPH-dependent reduction of cytochrome c (11,51). As described under "Methods,'' we carried out studies to determine if the presence of a heme-binding inhibitor (L-thiocitrulline) affected cytochrome c reduction. In the absence of Ca2+/calmodulin, there was no effect; rates of cytochrome c reduction in the absence of L-thiocitrulline or in the presence of 10 1.1~ or 100 VM L-thiocitrulline were 205 3 5 , 207 * 5 , and 208 2 7 nmol/min/mg of nNOS (means 2 S.D. for three to five determinations). In the presence of Ca2+/calmodulin, NOS-dependent reduction of cytochrome c is enhanced >lO-fold, and cytochrome c reduction is mediated largely by superoxide released from heme (11,52). L-Thiocitrulline (100 1 . 1 ) inhibited cytochrome c reduction in the presence of Ca2+/calmodulin by 82 2 3% ( n = 4); D-thiocitrulline at the same concentration was without effect. Possible Metabolism of L-Thiocitrulline-Because L-NMA has been shown t o be a poor substrate and mechanism-based inhibitor of NOS (35,36), we examined the possible conversion of L-thiocitrulline to citrulline or ornithine by NOS. Incubation of 36 pg of nNOS with 500 PM NADPH, 100 PM L-thiocitrulline, and necessary cofactors (see "Experimental Procedures") caused no diminution of the L-thiocitrulline present and no detectable product formation when the reaction mixtures were examined by HPLC or ion-exchange amino acid analysis. Formation of 1% citrulline or ornithine would have been easily detected. Under similar conditions, 100 VM L-arginine was converted completely to citrulline, and 100 J~M L-NMA yielded 90 p~ ~i t r u l l i n e .~ In Vivo Activity of L-Thiocitrulline and Related Compounds-Administration of L-thiocitrulline to anesthetized rats caused a nearly immediate and profound increase in blood pressure (Fig.   a). At a dose of 20 mg/kg, the pressor effect of L-thiocitrulline was about 20% greater in magnitude and comparable in duration to that elicited by L-NMA (Fig. 8, B and C ) . Note that 20 mgkg is a maximally effective dose for both L-thiocitrulline and L-NMA. At lower doses, L-thiocitrulline was also slightly more potent than L-NMA. The maximum pressor response to 5, 10, and 15 mgkg of L-thiocitrulline was about 40, 50, and 85 mm Hg (systolic).
L-Thiocitrulline was also a n effective pressor agent in hypotensive septic rats; a typical result is presented in Fig. 9. As shown, a rat with surgically induced peritonitis developed hypotension over a 6-h period. Infusion of normal saline caused only a small improvement in blood pressure, but L-thiocitrulline (20 mgkg) returned pressure to normal levels for 20-30 min.

DISCUSSION
All available data indicate that each NOS monomer contains a single arginine/citrulline binding site that accommodates Larginine, NOH-Arg (generated in situ or provided exogenously, Ref. 2) and a variety of N'"-monosubstituted inhibitory L-arginine analogs. Initial binding of the L-arginine analogs invariably shows competitive kinetics versus L-arginine. The amino acid binding site is stereospecific for L-enantiomers and is thought to place one guanidino nitrogen of L-arginine in sufficiently close proximity to heme iron to allow two successive heme-mediated monooxygenation reactions to occur (formation of NOH-Arg from L-arginine and then oxidation of NOH-Arg to conversion of L-thiocitrulline to citrulline (37); this result has not been Note that in preliminary studies we saw time-and NOS-dependent reproducible. Although L-thiocitrulline is stable in conventional buffers exhibit some nonenzymatic degradation to citrulline (and other prod-and is not affected by dithiothreitol, EDTA, Ca", or calmodulin, it does ucts) with redox active cofactors (FAD, FMN). Such degradation requires several hours at 25 "C, however, and would seem to be too slow to account for our previous result. We are continuing to examine the stability of L-thiocitrulline. The pressor effect of thiocitrulline is significantly greater than that of L-NMA at either dose ( p < 0.05).

FIG. 9. Effect of tthiocitrulline on
blood pressure in septic, hypotensive rats. Male Sprague-Dawley rats were anesthetized and instrumented as described under "Methods." Peritonitis and hypotension developed over a 6-h period, and, at that time, a saline bolus (10 ml/kg) was administered. At 5 rnin following the saline bolus, L-thiocitrulline (20 mg/kg) was given intravenously. A second bolus injection of saline was administered 35 min after L-thiocitrulline. citrulline and NO). For Nw-monosubstituted arginine substrates, it is the substituted guanidino nitrogen that is bound near heme; thus, the hydroxylated nitrogen of NOH-Arg is further oxidized to NO (21, and L-M is processed to N"hydroxy-Nw-methylarginine rather than to N"-hydroxy-N"'methylarginine (35).
Prior to the present work, several lines of evidence suggested that the NOS argininelcitrulline binding site had high affinity mainly for cationic arginine analogs. L-Arginine (33-36,42,47), L-NMA (35, 36), Nu-amino-L-arginine (53, 54), and N'-iminoethyl-L-ornithine (55) are all tightly bound and contain highly basic side chains. In contrast, L-canavanine, which is isosteric with L-arginine but has a side chain pK, of 6, is a relatively weak competitive inhibitor (1,55).' Similarly, the NOS reaction presumably leaves L-citrulline transiently bound t o the amino acid binding site, but when added exogenously, this neutral amino acid does not inhibit even when present at high concentration relative to arginine (49, present work). Nevertheless, our finding that L-thiocitrulline competes with L-arginine for its binding site and exhibits a Ki substantially lower than the K, for arginine or the K, for L-NMA demonstrates that a neutral citrulline analog can be tightly bound and raises the question of what factors contribute to its binding affinity.
The observation that L-thiocitrulline elicits a Type 11-like difference spectra indicates that its binding is accompanied by a shift of heme iron from a high spin to a low spin state consistent with a transition from pentacoordinate to hexacoordinate heme. Although our studies do not allow unambiguous assignment of the sixth heme ligand, the sulfur atom of thiocitrulline, present as either the thiono (-NH-CS-NH,) or the thiol (-NH-C(SH)=NH) tautomer, is an attractive candidate. The spectral results shown in Figs. 5-7 provide further understanding of the nature and extent of heme-ligand interaction. As shown in Fig. 5, L-thiocitrulline elicits an optical absorption difference spectrum with nNOS that resembles a Type I1 difference spectrum (see Fig. 6, inset, for imidazole-induced spectrum for a typical Type I1 spectrum). The family of spectra shown in Fig. 5 , however, reveals a distinct spectral signature that suggests additional interactions with the nNOS protein not previously observed with any other substrate or inhibitor of this enzyme. The difference spectra obtained with L-thiocitrulline after the addition of imidazole (Fig. 7B), resulting in the appearance of typical Type I spectrum and a >lO-fold increase in A absorbance, suggest that the citrulline-like structure of this inhibitor interacts in the L-arginine binding site producing a difference spectrum similar t o that obtained with L-arginine.

Time (minutes)
On the other hand, the addition of L-thiocitrulline, followed by L-arginine, produces a Type I difference spectrum with L-arginine that, upon the subsequent addition of L-thiocitrulline, reverts to the modified Type I1 spectrum with an even more pronounced peak at -408 nm. As discussed below, these results can be explained by the reversible binding of each of these substratelinhibitors in its own individual orientation.
The use of the optical difference spectroscopy approach has the advantage of magnifying minor spectral changes observed in the absolute spectra because the absorbance background becomes the base line. However, in interpreting these spectra, it must be appreciated that only relative changes are measured, and the appearance of a peak may be the difference between two troughs in the absolute spectrum, which in the sample cuvette is less negative than in the reference cuvette. Nevertheless, all of the spectra obtained with L-thiocitrulline can be modeled by Gaussian curves for each of the spin states of nNOS. The anomalous spectra seen upon binding of L-thiocitrulline in the L-arginine binding site of native nNOS represent a shifting of mostly high spin native nNOS (47) to Lthiocitrulline-liganded nNOS (Scheme I, reaction I). The latter is an equilbrium mixture of low and high spin forms suggesting that the sulfur atom of L-thiocitrulline is not perfectly aligned with the open axial position of heme iron and achieves a statistical equilibrium, sometimes acting as a sixth ligand (low spin) and sometimes being perhaps near heme iron but not close enough to perturb the (I electron orbitals of iron (high spin). In either the high or low spin state, L-thiocitrulline apparently prevents reduction and/or oxygen binding by heme since superoxide formation is reduced by 80-85%. Although our present data do not allow precise calculation of 'the NOS-Lthiocitrulline high spidow spin equilibrium ratio, the fraction in the low spin form must be greater than the -10% low spin form found in native nNOS since L-thiocitrulline binding is accompanied by a Type 11-like spectral shift (i.e. increasing low spin form). As shown previously, addition of L-arginine to nNOS greatly diminishes the fraction of low spin iron, perhaps to zero (47). The results shown in Fig. 7A can then be understood in terms of reactions I, 2, and 3 (Scheme I) occurring in sequence as L-thiocitrulline was added to nNOS (modified Type I1 spectral shift), L-arginine was added to that complex (Type I spectral shift), and, finally, excess L-thiocitrulline was added to displace L-arginine and restore the modified Type I1 spectra.
The results from the study starting with the hexacoordinate, low spin NOS-imidazole complex (Fig. 7 B ) I, reaction 4). Thus the displacement of the imidazole by L-thiocitrulline is accompanied by a Type I spectral shift attributable to formation of high spin NOS-Lthiocitrulline. Note that formation of low spin NOS-L-thiocitrulline also occurs when imidazole is displaced by L-thiocitrulline, but this species is spectrally silent since, in this study, the iron was already in a low spin state. Conversely, when L-thiocitrulline binds to native nNOS, the high spin species in the NOS-L-thiocitrulline equilibrium is spectrally silent.
Because NOS catalyzes the formation of NO and homocitrulline from L-homoarginine (24)4 and is inhibited by N"-methyl-L-lysine ("homo-NMA"), we anticipated that L-homothiocitrulline would inhibit. It does so, albeit with K, values 12-50-fold higher than the K, values for L-thiocitrulline. In our view, the higher K, values are due only in part to the inherently lower affinity of NOS for L-homoarginine and its analogs; the relatively poor afflnity also reflects the fact that this derivative apparently does not interact strongly with heme iron. Thus, L-homothiocitrulline binding to native nNOS elicits a Type I rather than Type I1 difference spectrum indicating that the high spidow spin statistical equilibrium for NOS-L-homothiocitrulline contains less low spin form than is present in native nNOS ( i e . < -10% (47)). This finding suggests that the folding necessary to accommodate L-homothiocitrulline in the arginine/ citrulline binding site moves the sulfur atom away from the heme iron, thereby decreasing its ability to alter the spin state of iron. If the reactive guanidino nitrogen of L-homoarginine is similarly displaced, such displacement may account for its relatively slow NOS-mediated oxidation. The fact that L-homothiocitrulline is a moderately strong inhibitor despite its limited, possibly absent, direct interaction with heme iron indicates that other interactions between NOS and the thioureido inhibitors may contribute to binding. To investigate these interactions further, we examined the effects of thiourea itself on the nNOS heme spectra. At concentrations of 1-10 mM, thiourea caused a spectral perturbation similar to that of L-thiocitrulline with a peak of 440 nm, a trough at 396 nm, and a shoulder at 410 nm (data not shown). Additional studies are necessary to fully elucidate the interactions of nNOS with thioureas.
Our data indicate that inhibition by L-thiocitrulline is reversible and is not dependent on metabolism of the inhibitor; in these respects, inhibition by L-thiocitrulline differs from the mechanism-based inhibition seen with L-NMA (35, 36). Inhibition by L-thiocitrulline more closely resembles that of L-NNA, another neutral, tight binding substrate analog (33, 34). Thus, L-NNA is reported to dissociate slowly from NOS (33, 341, and we similarly find a short, but distinct lag (-30 s) in restoration of activity following incubation of nNOS with L-thiocitrulline. Analogy between L-NNA and L-thiocitrulline is, however, not complete; reversibility of L-NNA binding is substantially slower K. Narayanan and 0. W. Griffth, unpublished observations. (-30 min) than that with L-thiocitrulline, and L-NNA gives a Type I rather than Type I1 difference spectrum, suggesting it does not interact directly with the heme c~factor.~ In vivo studies indicate that L-thiocitrulline is an effective pressor agent blocking formation of vasoactive NO by both eNOS (normotensive rat studies) and iNOS (hypotensive, septic rat studies). L-Thiocitrulline may thus be of therapeutic value in treating the hypotension of cytokine-induced and septic shock. Whether or not L-thiocitrulline is more effective than L-NMA for this purpose remains to be determined, but it has several potential advantages including a lower K, with iNOS.
Furthermore, in contrast to L-NMA, L-thiocitrulline is not metabolized to L-citrulline, a product that is converted to L-arginine in vivo and may thereby possibly sustain overproduction of NO (57,58). Finally, L-NMA is transported as a basic amino acid (mainly by the y+ and Bo.+ systems, Refs. 59 and 601, whereas L-thiocitrulline is neutral and likely to be transported as such. Tissue distribution of L-thiocitrulline may thus be different than that of L-NMA. Where complete inhibition of NOS is desired, co-administration of L-NMA and L-thiocitrulline may be particularly effective in achieving NOS blockade in a wide range of tissues.