These data indicate that a key factor for efficient oxidation of a guanidine by iNOS to NO is the ability of the corresponding N-hydroxyguanidine to bind to the active site without being
Trang 1Relationship between the structure of guanidines and
N-hydroxyguanidines, their binding to inducible nitric
oxide synthase (iNOS) and their iNOS-catalysed oxidation
to NO
David Lefe`vre-Groboillot1,2, Jean-Luc Boucher1, Dennis J Stuehr2and Daniel Mansuy1
1 Laboratoire de Chimie et Biochimie Pharmacologiques et Toxicologiques, Universite´ Paris 5, France
2 Department of Immunology, Lerner Research Institute, Cleveland, OH, USA
Keywords
binding kinetic; guanidines
N-hydroxyguanidines; nitric oxide synthase;
UV ⁄ Vis difference spectroscopy
Correspondence
J-L Boucher, Laboratoire de Chimie et
Biochimie Pharmacologiques et
Toxicologiques, UMR 8601 CNRS,
Universite´ Paris 5, 45 rue des Saints-Pe`res,
75270 Paris Cedex 06, France
Fax: +33 1 42 86 83 87
Tel: +33 1 42 86 21 91
E-mail: boucher@biomedicale.univ-paris5.fr
(Received 18 February 2005, revised
20 April 2005, accepted 25 April 2005)
doi:10.1111/j.1742-4658.2005.04736.x
The binding of several alkyl- and aryl-guanidines and N-hydroxyguanidines
to the oxygenase domain of inducible NO-synthase (iNOSoxy) was studied
by UV⁄ Vis difference spectroscopy In a very general manner, monosubsti-tuted guanidines exhibited affinities for iNOSoxy that were very close to those of the corresponding N-hydroxyguanidines The highest affinities were observed for the natural substrates, l-arginine and Nx
-hydroxy-l-arginine (Kd at the lm level) The deletion of either the CO2H or the
NH2 function of their amino acid moiety led to dramatic decreases in the affinity However, alkylguanidines with a relatively small alkyl chain exhib-ited interesting affinities, the best being observed for a butyl chain (Kd¼
20 lm) Arylguanidines also bound to iNOSoxy, however, with lower affinit-ies (Kd> 250 lm) Many N-alkyl- and N-aryl-N¢-hydroxyguanidines are oxidized by iNOS with formation of NO, whereas only few alkylguanidines led to significant production of NO under identical conditions, and all the arylguanidines tested to date were unable to lead to the production of NO The kcatvalues of NO production from the oxidation by iNOS of the stud-ied N-hydroxyguanidines were found to vary independently of their affinity for the protein The kcat values determined for the two-step oxidation of alkylguanidines to NO were not clearly related to the Kdof these substrates toward iNOSoxy However, there is a qualitative relationship between these
kcatvalues and the apparent rate constants of dissociation of the complex between iNOSoxy and the corresponding N-alkyl-N¢-hydroxyguanidine (koffapp) that were determined by stopped-flow UV⁄ Vis spectroscopy These data indicate that a key factor for efficient oxidation of a guanidine by iNOS to NO is the ability of the corresponding N-hydroxyguanidine to bind to the active site without being too rapidly released before its further oxidation This explains why 4,4,4-trifluorobutylguanidine is so far the best non-a-amino acid guanidine substrate of iNOS with formation of NO, because the koffapp of the corresponding N-hydroxyguanidine is particularly low This suggests that the rational design of guanidines as new NO donors
Abbreviations
BH 4 , (6R)-5,6,7,8-tetrahydro- L -biopterin; BuGua, n-butylguanidine; BuNOHG, N-(n-butyl)-N hydroxyguanidine; BzNOHG, N-benzyl-N ¢-hydroxyguanidine; ClPhNOHG, N-(4-chlorophenyl)-N¢-¢-hydroxyguanidine; FPhGua, 4-fluorophenylguanidine; FPhNOHG,
N-(4-fluorophenyl)-N ¢-hydroxyguanidine; HexGua, n-hexylguanidine; HexNOHG, N-(n-hexyl)-N ¢-hydroxyguanidine; homo- L -Arg, homo- L -arginine; homo-NOHA,
N x -hydroxy-homo- L -arginine; HS, high spin; ImH, imidazole; L -Arg, L -arginine; LS, low spin; NOHA, N x -hydroxy- L -Arginine; Nor- L -Arg, nor- L -Arginine; NOHAgma, Nx-hydroxyagmatine; NOHGPA, Nx-hydroxyguanidinopentanoic acid; NOS, nitric oxide synthase; NOS oxy , oxygenase domain of NOS; PentylGua, n-pentylguanidine; PentylNOHG, N-(n-pentyl)-N¢-hydroxyguanidine; ProGua, n-propylguanidine; ProNOHG, N-(n-propyl)-N ¢-hydroxyguanidine; TFBGua, 4,4,4-trifluorobutylguanidine; TFBNOHG, N-(4,4,4-trifluorobutyl)-N ¢-hydroxyguanidine.
Trang 2Nitric oxide synthases (NOS) catalyse the oxidation of
l-arginine (l-Arg) into l-citrulline and NO, with the
intermediate formation of Nx-hydroxy-l-arginine
(NOHA) [1–3] This reaction ideally consumes 1.5 mol
of NADPH and 2 mol of O2 It occurs in the
homo-dimeric N-terminal domain of the protein called NOS
oxygenase domain (NOSoxy) that contains two
cofac-tors per monomer, the heme (iron-protoporphyrin IX)
and (6R)-5,6,7,8-tetrahydro-l-biopterin (BH4)
Elec-trons from NADPH are provided to heme by flanking
C-terminal reductase domains NOSs are heme-thiolate
monooxygenases comparable with cytochrome P450
Whereas proteins of the cytochrome P450 family are
known to be able to bind and oxidize a very large
number of compounds of various structures, until
recently NOSs were only known to be able to oxidize
l-Arg and a very small number of its close a-amino
acid analogues Recent reports have shown that NOSs
are able to produce NO from the oxidation of many
non-a-amino acid monosubstituted
N-hydroxyguani-dines, including N-alkyl-N¢-hydroxyguanidines and
N-aryl-N¢-hydroxyguanidines, provided that the alkyl
or aryl substituent is neither too small nor too bulky
[4–7] NOS-catalysed oxidation of some of these
com-pounds showed kcat values as high as 80% that
obtained with NOHA, and some proved to be selective
for one of the three isoforms vs the others [5,7] More
recently, NO production has also been observed from
the oxidation of several non-a-amino acid
alkylguani-dines by purified iNOS or by activated mouse
macro-phages, opening the way to the design of stable
exogenous NOS substrates of pharmacological interest
[8,9]
Apart from some equilibrium and kinetic constants
related to the binding of l-Arg [10–14] and NOHA
[10], nothing is known about the thermodynamics and
kinetics of the binding of guanidines and
N-hydroxy-guanidines to iNOS Removal of the a-amino or
a-carboxylate moiety of l-Arg has important effects on
the ability of the resulting compounds to affect the
heme iron spin equilibrium, and to trigger NADPH
consumption and NO production [14–16] Interestingly,
it has been shown that several binding modes exist for
N-hydroxyguanidines in the heme pocket of NOSs [17–
20] Also, the fact that isoform-selective substrates for
NOS [5,7,9] were characterized is striking, given the
high level of similarity between the crystal structures of
the oxygenase domains of the three isoforms [18,19,21]
This study was undertaken to determine structural fac-tors that are important for a guanidine or N-hydroxy-guanidine to be well recognized by the NOS active site, and to be efficiently oxidized with NO formation For that purpose, the dissociation constants of several com-plexes of the oxygenase domain of iNOS (iNOSoxy) with various alkyl- and aryl-guanidines and N-hydroxygua-nidines were determined by UV⁄ Vis difference spectros-copy, according to a previously described technique [22] The kinetics of the binding of some of these substrates
to iNOSoxy was also studied by UV⁄ Vis spectroscopy using stopped-flow techniques [23,24] The correspond-ing thermodynamic and kinetic bindcorrespond-ing constants were then compared with the kinetic constants of NO forma-tion from iNOS-catalysed oxidaforma-tion of guanidine and N-hydroxyguanidine substrates Our results suggest that
a key factor in the efficient oxidation of a guanidine to
NO by iNOS could be the ability of the corresponding N-hydroxyguanidine to bind to the active site without being too rapidly released before its further oxidation Our results may help in the further rational design of guanidines as new NO precursors
Results
Study of the binding of guanidines and N-hydroxyguanidines to iNOSoxyby UV⁄ Vis difference spectroscopy
Purified recombinant iNOSoxy showed a wide Soret band with a maximum absorption wavelength around
400 nm, indicating that the heme-iron existed in equi-librium between a hexacoordinated low-spin (LS) state and a pentacoordinated high-spin (HS) state, the major fraction being in the HS state As previously described, addition of l-Arg leads to conversion of the minor population of heme centres being in the LS state into the HS state and to the appearance of a difference spectrum [22,25–27] However, the intensity of this dif-ference spectrum is small, because the spin state of the major fraction of the protein is not affected Imidazole (ImH) was thus used to completely convert iNOSoxy into a LS state iNOSoxy–Fe(III)–ImH complex allow-ing one to more easily follow the bindallow-ing of guanidines
or N-hydroxyguanidines to the iNOSoxy substrate binding site [12,22,28] iNOSoxy(1 lm) in the presence
of 400 lm ImH was first titrated with l-Arg A differ-ence spectrum displaying a peak at 392 nm and a
upon in situ oxidation by NOSs should take into account both thermody-namic and kinetic characteristics of the interaction of the protein not only with the guanidine but also with the corresponding N-hydroxyguanidine
Trang 3trough at 430 nm (Fig 1) resulting from the
conver-sion of the LS NOS–Fe(III)–ImH complex to the
HS NOS–Fe(III)–l-Arg complex was observed
Inhibi-tion of the binding of l-Arg to iNOSoxyby ImH, and of
the iNOS-catalysed conversion of l-Arg into l-citrulline
has previously been shown to be competitive [11]
Equa-tion (1) was thus used to calculate corrected
equi-librium constants, Kd, for the iNOSoxy–substrate
complexes from apparent constants Kapp[12–14,23,24]
Kapp=Kd¼ 1 þ ½ImH=KImH ð1Þ
With the ImH concentration used in this study
(400 lm), Eqn (1) became
Variations in the amplitudes of the difference spectra
with the concentrations of l-Arg were in agreement
with a single binding site model (see Experimental
pro-cedures) and Kapp value of 26 ± 2 lm was found for
the apparent equilibrium constant for the dissociation
of the iNOSoxy–l-Arg complex, in good agreement
with a previously reported value (28 ± 4 lm) obtained
with the same ImH concentration [25] At the end of
the titration, the absolute spectrum of the iNOSoxy
solution containing 400 lm ImH and 1 mm l-Arg
showed a maximum absorption wavelength at
395 ± 3 nm (not shown)
Similar titrations of iNOSoxy in the presence of
400 lm ImH were then performed with a large number
of guanidines and N-hydroxyguanidines previously
evaluated as iNOS substrates [4–9,29] The positions of the peaks and troughs of the difference spectra observed during these titrations were similar to those observed when l-Arg was used (Fig 1) Variation
in the amplitude of the observed difference spectra with the concentration of the studied guanidines or N-hydroxyguanidines was always in reasonable agree-ment with a single binding site model The apparent equilibrium constants derived from these experiments are shown in Table 1
NOHA was found to bind to iNOS with a Kapp value slightly lower than that of l-Arg (18 ± 7 lm, Table 1) Homo-L-Arg and homo-NOHA, the l-Arg and NOHA analogues bearing one extra methylene group in the alkyl side-chain, were found to bind to iNOSoxywith higher Kapp values than l-Arg (80 ± 13 and 150 ± 40 lm, respectively) Finally, a much higher Kapp value (2.4 mm) was found for nor-l-Arg, the analogue bearing one methylene fewer than l-Arg Removal of either the a-COOH or the a-NH2group
of NOHA led to a dramatic decrease in the affinity of the resulting compounds, the Kapp values measured for
Nx-hydroxyagmatine (NOHAgma), and Nx -hydroxy-guanidino-pentanoic acid (NOHGPA), being > 1 mm (2 and > 4 mm, respectively; Table 1) However, the simultaneous removal of both the a-NH2and a-COOH functions of NOHA led to N-(n-butyl)-N¢-hydroxygu-anidine (BuNOHG), which showed a much lower Kapp value of 160 ± 40 lm (Table 1) Replacement of the terminal CH3 group of the n-butyl chain by a CF3 group, leading to N-(4,4,4-trifluorobutyl)-N¢-hydroxy-guanidine (TFBNOHG), resulted in a sixfold increase
in the Kappvalue Shorter nonfunctionalized analogues N-(n-propyl)-N¢-hydroxyguanidine (ProNOHG) and longer ones N-(n-pentyl)-N¢-hydroxyguanidine (Pentyl-NOHG) and N-(n-hexyl)-N¢-hydroxyguanidine (Hex-NOHG) showed higher Kapp values than the N-(n-butyl) compound (270, 900 and >1000 lm, respectively) Finally, N-benzyl-N¢-hydroxyguanidine (BzNOHG) and the three para-substituted aryl-deriva-tives N-(4-fluoro-, 4-methyl- and
ClPhNOHG), showed Kappvalues > 2 mm
A study of the binding of the corresponding non-functionalized alkylguanidines to iNOSoxy led to very similar conclusions In the studied series, the alkylgu-anidine exhibiting the highest affinity for iNOS was n-butylguanidine (BuGua), with a Kapp value of
140 ± 20 lm (Table 1) Trifluorination of the terminal methyl group of the n-butyl chain, leading to 4,4,4-tri-fluorobutylguanidine (TFBGua), increased the Kapp value by 10-fold The longer nonsubstituted n-pentyl-and n-hexylguanidines (PentylGua n-pentyl-and HexGua) also
Fig 1 Difference spectrum obtained upon addition of increasing
concentrations of L -Arg to iNOSoxyin the presence of ImH iNOSoxy
and ImH concentrations were 1 and 400 l M , respectively (Inset)
Plot of 1 ⁄ DA vs 1 ⁄ [ L -Arg].
Trang 4showed higher Kapp values (600 lm and > 4 mm for
the n-pentyl and n-hexyl derivatives, respectively) The
shorter n-propylguanidine (ProGua) showed a Kapp
value similar to that found for BuGua Finally, the
arylguanidines 4-fluorophenyl- and
4-methylphenyl-guanidines (FPhGua and TolGua) interacted with
iNOS with Kappvalues>2 mm
Relationship between the equilibrium constants
measured for the binding of guanidines and
N-hydroxyguanidines to iNOSoxyand the kinetic
constants measured for their iNOS-catalysed
oxidation to NO
In previous studies, we have identified some
N-alkyl-and N-aryl-N¢-hydroxyguanidines, N-alkyl-and alkylguanidines
as NO donors following their oxidation catalysed by
iNOS containing all its cofactors [4,5,8,9] Table 2 gives the Km and kcat values measured for the oxidation of seven N-hydroxyguanidines leading to the highest pro-duction of NO in the presence of iNOS, together with
Kmand kcatvalues for the oxidation of the correspond-ing guanidines [4,5,8,9] The seven N-hydroxyguani-dines NOHA, homo-NOHA, BuNOHG, TFBNOHG, PentylNOHG, FPhNOHG and TolNOHG were oxid-ized with formation of NO with similar high kcatvalues ranging from 58 to 100% of that found for NOHA They showed widespread Km⁄Kd ratios, generally >1 and that varied from 1 to 20 (Table 2) In that ser-ies, the kcat value for the production of NO from the oxidation of the N-hydroxyguanidines varied by less than a factor 2, whereas the kcatvalue for the produc-tion of NO from the oxidaproduc-tion of the guanidines varied
a great deal from 0 to 100% of the kcatvalue obtained
Table 2 Kinetic constants for the formation of NO from the oxidation of guanidines and N-hydroxyguanidines by recombinant iNOS See Table 1 for the structure of compounds Kmand kcatvalues are taken from previous publications [4,5,8,9,29] kcatvalues are expressed per NOS dimer The corrected dissociation equilibrium constants (K d ) for the binding of guanidines and N-hydroxyguanidines to iNOS oxy were obtained by dividing K app values (taken from Table 1) by 8.7.
Compounds
K m (l M ) N-Hydroxy
k cat (min)1) N-Hydroxy
K m ⁄ K d N-Hydroxy
a
The rates of the production of NO from the oxidation by iNOS of 1 m M FPhGua or TolGua were lower than 2 min)1.
Table 1 Apparent equilibrium constants (Kapp) for the binding of N-hydroxyguanidines R-NH-C(¼ NOH)-NH 2 and guanidines R-NH-C(¼ NH 2
)-NH2to iNOSoxy Titrations were performed by UV ⁄ Vis difference spectroscopy in the presence of 25 l M BH4, 1 m M dithiothreitol and
400 l M ImH K app values were calculated as described in Experimental procedures Values ± SD from three different experiments n.d., not determined.
R
N-Hydroxyguanidines
Guanidines
Trang 5with l-Arg, with the order l-Arg >
homo-l-Arg TFBGua > PentylGua > BuGua >> FPhGua
TolGua (Table 2)
The Kmand kcatvalues calculated for NO formation
from iNOS-catalysed oxidation of guanidines are
com-plex parameters as they correspond to a two-step
reaction with intermediate formation of
N-hydroxygu-anidines The data are difficult to correlate with kinetic
or thermodynamic constants clearly describing
individ-ual reactions, such as Kd (or Kapp) The situation
should be less complex for Kmand kcatfor NO
forma-tion from N-hydroxyguanidines that are more closely
related to a one-step enzymatic reaction
It is actually well known that, for enzymes having
high kcatvalues, the Kmvalues can be markedly higher
than the Kd values, as indicated by the classical
rela-tion given here [30]
Km¼ Kdþ kcat=kon or
Km=Kd¼ 1 þ kcat=koff This equation implies that Km⁄ Kd will increase as kcat
increases and koff decreases Because the kcat value
found for these seven N-hydroxyguanidines varied by
less than a factor 2, it was tempting to investigate a
possible relationship between Km⁄ Kd and koff The
fol-lowing experiments were performed as a first approach
to find the variation in koff as a function of the iNOS
substrate structure
Kinetics of the binding of guanidines and
N-hydroxyguanidines to iNOSoxymeasured by
stopped-flow UV⁄ Vis spectroscopy
An iNOSoxysolution containing 400 lm ImH was
rap-idly mixed with a solution of the studied ligand
contain-ing the same concentration of ImH Postmixcontain-ing ligand
concentrations corresponded to pseudo-first-order
con-ditions Absorption variations were monitored at 430
and 392 nm (Fig 2), allowing one to follow,
respect-ively, the disappearance of the NOS–Fe(III)–ImH
com-plex and the appearance of the high-spin NOS–Fe(III)
species The calculated kinetic constants kobswere
plot-ted against the ligand concentration and satisfactorily
fitted with a linear function
kobs¼ koffappþ koffapp½L
where L is the guanidine or N-hydroxyguanidine used
(Fig 3), in agreement with a competitive model for the
interaction between ImH and the studied guanidine or
N-hydroxyguanidine [23,24,28] It has previously been
shown that displacement of ImH from the NOS heme-iron by l-Arg or its analogues is a two-step process [23,28] and might involve an intermediate and transient ternary complex between the protein, ImH and the
l-Arg analogue [23] The konapp and koffapp values are thus apparent association and dissociation rate con-stants of the guanidine or N-hydroxyguanidine with the protein in the presence of 400 lm ImH
Three guanidines and the corresponding N-hydroxy-guanidines were studied l-Arg and NOHA were used
0.33 0.32 430 nm
392 nm
0.31
0.29 0.28 0.27 0.26 0.25
Time (s)
0.3
Wavelength (nm)
A
B
0.063 0.113 0.163 0.213 0.263 0.313
Fig 2 Spectral transitions observed as a function of time upon the fast addition of BuNOHG to iNOS oxy in the presence of ImH ImH concentration was 400 l M Final heme and BuNOHG concentra-tions were 5 l M and 2 m M , respectively (A) Rapid-scanning stopped-flow spectra recorded during the reaction (B) Cross-sec-tion of (A) variaCross-sec-tion in absorbance at 430 and 392 nm as a funcCross-sec-tion
of time.
Trang 6as reference compounds and two pairs of non-a-amino acid compounds, BuGua⁄ BuNOHG and TFBGua ⁄ TFBNOHG were also studied The determined values
of konapp and koffapp are reported in Table 3 In the studied range of concentrations, the kobs values were higher for a guanidine than for the corresponding N-hydroxyguanidine (Fig 3) The konapp values for the guanidines were found to be 5–10 higher than those for the corresponding N-hydroxyguanidines, and the
koffappvalues for the guanidines were 25–60 times higher than those for the corresponding N-hydroxy-guanidines (Table 3) The konapp values for the non-a-amino acid guanidines BuGua and TFBGua were found to be 7- and 25-fold lower than that for l-Arg, and those for the non-a-amino acid N-hydroxyguani-dines BuNOHG and TFBNOHG were 3- and 12-fold lower than that for NOHA The koffapp values for BuGua and TFBGua were found to be 10 and 6 times higher than that for l-Arg, and those for BuNOHG and TFBNOHG were 20 and 5 times higher than that for NOHA Interestingly, konapp values for the
3.8- and 3.6-fold lower than those for their nonfluori-nated analogues BuGua and BuNOHG, respectively, and the koffapp values for TFBGua and TFBNOHG are 1.6 and 4 times lower than those for BuGua and BuNOHG, respectively
Discussion
Binding of guanidines and N-hydroxyguanidines
to iNOSoxy
In the series of guanidines and N-hydroxyguanidines studied here, the ratio between the Kappor Kd (calcula-ted using Eqn 1¢) of a guanidine and that of its corres-ponding N-hydroxyguanidine was always found to be between 0.5 and 2 (Table 1) This was true for pairs
of compounds showing dissociation constants in the micromolar range (l-Arg⁄ NOHA) and pairs of compounds showing Kd in the millimolar range (FPhGua⁄ FPhNOHG) The difference between the
Kapp values for the guanidines and those for the cor-responding N-hydroxyguanidines was, in most cases, small and barely significant However, we found that the Kapp value for NOHA is slightly lower than that for l-Arg (Table 1), and because such an observation has also been previously reported by several authors with nNOS [12,22,23] and iNOS [10], this difference is probably significant By contrast, we found that the
Km value for NOHA is higher than that for l-Arg (Table 2), also in accordance with the literature data
on the three isoforms [29,31,32] In the studied series,
Fig 3 Plots of the rates of spectral transitions observed upon the
addition of guanidines or N-hydroxyguanidines to iNOSoxy in the
presence of ImH vs the postmixing concentration of the studied
guanidine or N-hydroxyguanidine Best linear fits are shown.
(A) L -Arg and NOHA, (B) N-hydroxyguanidines BuNOHG and
TFBNOHG, (C) guanidines BuGua and TFBGua See Table 1 for the
structure of compounds.
Trang 7binding of an N-hydroxyguanidine moiety in the iNOS
heme pocket thus roughly involves the same binding
energy as binding of the guanidine moiety, and the
equilibrium constants are mainly determined by the
alkyl or aryl substituents of the compounds
The crystal structures of mouse iNOSoxy–l-Arg and
bovine eNOSoxy–l-Arg complexes [33,34] have shown
that the guanidine moiety of l-Arg makes a salt bridge
with the side chain of a conserved glutamate residue
(E371 in mouse iNOS), and H-bonds with a backbone
carbonyl oxygen atom (W366 in mouse iNOS) The
crystal structures of iNOSoxy–NOHA complexes
showed identical positioning of the
N-hydroxyguani-dine moiety of NOHA, with additional contacts
between the N-hydroxyguanidine hydroxy group and
the amide nitrogen of a conserved glycine (G365 in
mouse iNOS) [18,19,21] The crystal structures of
eNOSoxy–ClPhNOHG and nNOSoxy–BuNOHG
com-plexes showed similar positionings of the
N-hydroxy-guanidine moiety of ClPhNOHG and BuNOHG
involving: (a) a salt bridge between a glutamate side
chain and the two nonhydroxylated nitrogens of the
N-hydroxyguanidine, and (b) a nonbonded contact
between the hydroxy group and a glycine nitrogen
[17,19] It thus seems that in such a positioning,
chan-ging the N-hydroxyguanidine moiety into a guanidine
moiety does not strongly modify the energy of binding
to iNOS Preliminary data show that this is also true
for nNOS (D Lefe`vre-Groboillot, unpublished data)
The structure–affinity relationship for
alkylguani-dines and N-alkyl-N¢-hydroxyguanialkylguani-dines bearing
non-functionalized linear alkyl chains (ProGua, BuGua,
PentylGua, HexGua, ProNOHG, BuNOHG,
Pentyl-NOHG and HexPentyl-NOHG) showed that the binding
affinity is maximal for the compounds bearing a butyl
chain, i.e BuNOHG and BuGua, with Kd values
around 20 lm (Table 1) Compounds bearing a
n-pentyl chain (PentylNOHG and PentylGua) still bind well to the iNOS active site (Kd around 100 lm) but compounds bearing an n-hexyl chain (HexNOHG and HexGua) interact with iNOS with low affinities (Kd> 150 lm) The crystal structure of the nNOSoxy– BuNOHG complex showed that the butyl chain of BuNOHG interacts with the side chain of a conserved valine residue (V567), a conserved proline (P565) and the amide moiety of a conserved glutamine (Q478) [19] Because BuNOHG has previously been reported
to be similarly efficiently oxidized into NO by both iNOS and nNOS [4,6,9], the binding modes of this compound for the two isoforms is expected to be sim-ilar We measured the Kdfor the binding of BuNOHG
to nNOSoxy and found a somewhat higher value of
100 lm (data not shown), suggesting that the binding
of BuNOHG to iNOS is favoured slightly over its binding to nNOS The Km value for the oxidation of BuNOHG by iNOS and nNOS were also found to fol-low the order iNOS < nNOS [4,6,9] It appears that the hydrophobic contacts such as those observed between the butyl chain and the protein in the nNOSoxy–BuNOHG crystal structure are sufficient to allow compounds BuNOHG and BuGua to bind to the active site of iNOS and nNOS with Kd values in the 20–100 lm range Interestingly, the crystal struc-ture of the nNOSoxy–BuNOHG complex revealed that upon binding of BuNOHG the side chain of residue Q257 has to shift from its position observed in other complexes (including the nNOSoxy–NOHA complex),
in order to accommodate the terminal methyl group
of BuNOHG [19] This is in agreement with the fact that longer compounds such as PentylNOHG or HexNOHG showed lower affinities for the iNOS active site, because their binding may require an important reorganization of the protein environment
The introduction of both an amino function and a carboxylate function on the terminal methyl group of BuGua and BuNOHG in a configuration leading to the natural substrates, l-Arg and NOHA, led to a 10-fold decrease in the observed equilibrium constants (Table 1) The positioning of the a-amino acid moiety
of NOHA or l-Arg analogues appears to be critical for binding to iNOSoxy Indeed, the Kd values for
l-Arg and NOHA were found to be in the 2–4 lm range, in agreement with previously reported data [10,11,13], whereas those for the longer analogues homo-l-Arg and homo-NOHA were found to be in the 10–20 lm range and that for the shorter analogue Nor-l-Arg was found to be > 300 lm This indicates that the alkyl chains of l-Arg (or NOHA) optimally position their guanidine (or N-hydroxyguanidine) and a-amino acid moieties relative to each other in the
Table 3 Apparent association and dissociation rate constants (k onapp
and k offapp) for the binding of guanidines and N-hydroxyguanidines
to iNOS oxy in the presence of 400 l M ImH See Table 1 for the
structure of compounds The rates of spectral transitions (Fig 2)
were fitted vs the postmixing concentrations of the studied
guani-dine or N-hydroxyguaniguani-dine with a linear function, as shown on
Fig 3 koffapp was defined as the y intercept and konapp as the slope.
offapp(s)1)
Trang 8NOS active site This also indicates that adding one
methylene in the l-Arg chain does not impede efficient
binding, whereas removal of one methylene group is
detrimental for the interaction between the protein and
the substrate The crystal structure of the eNOSoxy–
homo-l-Arg complex (PDB entry 1DM7, C.S Raman
et al 1999) actually showed that homo-l-Arg interacts
with the active site of eNOSoxy in a manner similar to
l-Arg, involving roughly identical positionings of the
guanidine and a-amino acid moieties However, the
longer alkyl chain of homo-l-Arg forms a small bulge
between the two heme propionates in contact with the
heme and the side chain of a conserved valine (V338)
The decrease of the affinity of l-Arg and
homo-NOHA compared with l-Arg and homo-NOHA (Table 1)
could be linked to this unfavourable bulging
confor-mation of the alkyl chain of homo-l-Arg
The simultaneous presence of both the a-amino and
a-carboxylate moieties appears to be necessary because
the NOHA analogue bearing only an a-amino moiety
(NOHAgma) interacted with iNOSoxy with an affinity
(Kd 250 lm) much lower than that found for
BuNOHG, and the NOHA analogue bearing only an
a-carboxylate moiety (NOHGPA) did not interact with
iNOS (Kd> 500 lm) This suggests that the a-amino
and a-carboxylate groups cooperate to provide
favour-able binding enthalpy for the formation of the complex
between NOHA and the protein The crystal structures
of l-Arg or NOHA in NOS active sites actually
showed that the a-amino acid moiety of these
com-pounds interacts with the protein via an H-bond
net-work involving one or two water molecules that links
the a-amino and a-carboxylate moieties one to each
other, and to protein residues [18,19,21]
Finally, six compounds bearing an aryl moiety,
BzNOHG, FPhGua and ClPhGua (Table 1), exhibited
Kd values > 250 lm The crystal structure of the
eNOSoxy–ClPhNOHG complex showed that the phenyl
ring of ClPhNOHG is in close contact with the side
chain of the conserved valine, V338, and with a
pro-pionate of the heme [17] The chlorine atom is also
involved in nonbonded contacts with the conserved
methionine M341 Because ClPhNOHG was previously
reported to be an iNOS-specific substrate [5,6], we also
measured the equilibrium constants for the binding of
these compounds to eNOS and nNOS (data not
shown) ClPhNOHG actually displayed significantly
higher affinities for the two constitutive isoforms than
for iNOS: the Kd values for its binding to nNOSoxy
and eNOSoxywere found to be around 50 and 95 lm,
respectively, whereas that for its binding to iNOSoxywas
found to be close to 350 lm Similar higher affinities
for nNOSoxy compared with iNOSoxy were also observed for TolNOHG, which is also an iNOS specific substrate [5,6], and for FPhNOHG, which is a substrate highly selective for iNOS [5] These results obtained with guanidines or hydroxyguanidines bear-ing an N-aryl moiety recall the well-documented selec-tivity of N-arylamidines for inhibition of nNOS vs iNOS [35]
Relationship between the structure of N-hydroxy-guanidines, their affinity for iNOSoxyand their oxidation by iNOS with formation of NO Previous data showed that a very large number of monosubstituted N-hydroxyguanidines
R-NH-C(¼NOH)-NH2 bearing an alkyl or aryl substituent R, neither too small nor too bulky, led to the detectable production
of NO in the presence of iNOS [4–7,9,29] Formation of
NO from the oxidation of an N-hydroxyguanidine by iNOS is thus not specific to NOHA and can occur with many N-hydroxyguanidines
The rates of NO formation from the oxidation of a great number of N-alkyl- and N-aryl-N¢-hydroxyguan-idines by iNOS were found to be highly dependent
on their structure [4–7,9] However, the kcat values found for NO formation upon iNOS-catalysed oxida-tion of the seven N-hydroxyguanidines menoxida-tioned in Table 2 varied by less than a factor 2, whereas their
Kd values varied by a factor 200 (Table 1) It thus appears that the kcat of NO formation is not simply related to the affinity of the substrate for iNOS For instance, the kcat of NO formation from FPhNOHG oxidation is 83% of that found for NOHA, whereas the Kd of this substrate is 130 times higher than that
of NOHA
As mentioned above and shown in Table 2, very dif-ferent N-hydroxyguanidines leading to similar kcat val-ues (58–100% of that found for NOHA) showed widespread Km⁄ Kdratios (from 1 to 20) This vari-ation may be related to that in koff, as expected by considering the relation Km⁄ Kd¼ 1 + kcat⁄ koff From
a qualitative point of view, this is in agreement with the variation in koffapp for NOHA (0.1 s)1), TFBNOHG (0.5 s)1) and BuNOHG (2 s)1) (Table 3), which is inversely related to that of Km⁄ Kd for these N-hydroxyguanidines (19.3, 8.1 and 3.0 for NOHA, TFBNOHG and BuNOHG, respectively) Rigorous and quantitative correlations could not be done imme-diately, as Km and kcat, Kapp, konapp and koffapp values were measured under different conditions for experi-mental reasons (different temperatures or the presence
of imidazole) However, our data provide a first gen-eral basis to understand the structural factors that are
Trang 9necessary for guanidines and N-hydroxyguanidines to
efficiently bind to iNOS
Criteria for the formation of NO from the
oxidation of a guanidine by iNOS
Contrary to what is observed for the
N-hydroxyguani-dines, not all the guanidines that bind to iNOS lead to
the production of NO [5–9] For example, all
arylguani-dines assayed to date, among them FPhGua and
TolGua, have failed to lead to any detectable amount
of NO, although their affinity for iNOS is not lower than
that for the corresponding
N-aryl-N¢-hydroxyguani-dines that lead to kcatvalues of formation of NO as high
as 83 and 69% that obtained for NOHA (Table 2)
As in the case of the N-hydroxyguanidines, the kcat
values of NO formation from the oxidation of
guani-dines do not appear to be linked to the affinity of the
compounds for iNOS For example, compound
TFB-Gua led to a kcatvalue of NO formation of 55% that
obtained with l-Arg (Table 2) even though it bound to
iNOS with a Kappvalue 50 times higher (Table 1)
Interestingly, the kcatof production of NO by
oxida-tion of the studied guanidines followed the same order
l-Arg > homo-l-Arg TFBGua > PentylGua > BuGua
as that found for the Km⁄ Kd ratio of the
correspond-ing N-hydroxyguanidines: NOHA > homo-NOHA
TFBNOHG > PentylNOHG > BuNOHG (Table 2)
This suggests that the variations in the kcat values
found for NO formation from the guanidines could
be related to those of the koff of the corresponding
N-hydroxyguanidines Accordingly, the order l-Arg >
TFBGua > BuGua found for the kcatof production of
NO from oxidation of these guanidines corresponds
well to the order NOHA > TFBNOHG > BuNOHG
found for 1⁄ koffapp of the corresponding
N-hydroxygu-anidines (Table 3) These results may suggest that a key
factor for a guanidine to lead to NO formation in the
presence of iNOS could be the ability of the
corres-ponding N-hydroxyguanidine to bind to the active site
without being released before being further oxidized
They could explain why the compound TFBGua is so
far the best non a-amino acid NO precursor upon
oxi-dation by iNOS (Table 2), because the koffapp value of
the corresponding N-hydroxyguanidine TFBNOHG is
particularly low (Table 3) In a more general manner,
our data suggest that changes in the NOS–substrate
complex structure (changes of the substrate structure,
but also mutation or post-translational modification of
the protein) could likely lead to a shift of the activity of
NOS from NO synthesis to N-hydroxyguanidine
syn-thesis Further investigations are currently underway to
test these hypotheses Our results also suggest that the
rational design of guanidines as new NO donors upon
in situ oxidation by NOSs should take into account both thermodynamic and kinetic characteristics of the interaction of the protein not only with the guanidine, but also with the corresponding N-hydroxyguanidine
Experimental procedures
Chemicals and reagents
Paris, France) and l-Arg and homo-l-Arg were from Sigma (Saint-Quentin Fallavies, France) Alkylguanidines were obtained by reaction of the corresponding amine with pyrazole-1-carboxamidine hydrochloride in the presence of diisopropylethylamine following a previously described pro-tocol [8] Arylguanidines were obtained by reaction of the amine with N,N¢-bis(tert-butyloxycarbonyl)pyrazole-1-carb-oxamidine followed by acidic deprotection as previously des-cribed [8] N-Hydroxyguanidines were obtained, as well as small amounts of the corresponding ureas, by the addition of hydroxylamine hydrochloride to intermediate cyanamides in anhydrous ethanol [4–7] Cyanamides were obtained from the amines by addition of BrCN in methanol containing anhydrous sodium acetate [4,5] The physicochemical charac-teristics of N-(n-propyl)-N¢-hydroxyguanidine,
N-(4-fluorophenyl)-N¢-hydroxyguanidine, N-(4-chlorophenyl)-N¢-N-(4-fluorophenyl)-N¢-hydroxyguanidine, N-(4-methylphenyl)-N¢-hydroxyguanidine, N-benzyl-N¢-hydro-xyguanidine, n-butylguanidine, 4,4,4-trifluorobutylguanidine and (4-methyl)phenylguanidine have been published previ-ously [5,6,8] NOHA and homo-NOHA were synthesized as previously reported [29] Other chemicals were from Aldrich (Saint-Quentin Fallavies, France), Sigma or Across (Noisy le Grand, France) unless otherwise indicated and were of the highest purity commercially available
Protein preparation
at its C-terminus was overexpressed in Escherichia coli and
The enzyme concentration was determined from the 444 nm absorbance of its ferrous–CO complex by using an extinction
Assessment of NO formation
using the classical spectrophotometric oxyhemoglobin assay
Trang 10for NO [38] under conditions described previously [4,5,8].
In some assays, the level of NO formation was measured
by electron paramagnetic resonance spectroscopy following
the formation of the paramagnetic ferrous mononitrosyl
diethyldithiocarbamate complex under previously described
conditions [8,39]
Determination of the dissociation constants for
the complexes between BH4-containing iNOSoxy
and guanidines or N-hydroxyguanidines
Studies were carried out at room temperature in an
UVIKON 942 spectrophotometer (Kontron Biotek), in a
1-cm path length cuvettes (150 lL total volume) Each
dithio-threitol The amplitude of the observed difference spectra
with a linear function The two fits gave dissociation
experiments, indicating that a single binding site model
found, very close to the values reported by others [11–13]
Maximum amplitude of the difference spectrum was
The study of the binding of guanidines and
of 400 lm ImH, a situation that allows the monitoring by
com-plex between the protein and compounds which bind to the
substrate binding site [11,12,14,22,25,36,40–42] The studied
guanidine or N-hydroxyguanidine (dissolved in buffer) was
added stepwise to the sample cuvette, and equivalent
vol-umes of buffer were added to the reference cuvette All
experiments were carried out under conditions where the
concentration of bound ligand was much smaller than the
calcula-tion, the amplitude of the observed difference spectra
N-hydroxy-guanidine concentration [ligand] with a hyperbolic function
[lig-and] with a linear function The two fits always gave
complexes that were less different than the values obtained
from two identical experiments This indicated that a single
binding site model satisfactorily accounted for the observed
spectral changes The maximum amplitude of the difference
signifi-cantly with the structure of the guanidine or
N-hydroxy-guanidine
Rapid kinetic studies of the binding
of guanidines and N-hydroxyguanidines to iNOSoxy
instrument equipped with a rapid-scanning diode array detector (Hi-Tech MG 6000) and following a protocol
400 lm ImH was mixed with a solution of guanidine (or N-hydroxyguanidine) also containing 400 lm ImH Post-mixing heme concentration was 5 lm The reaction was monitored by following the absorbance at 430 and 392 nm Variations of the absorbances at these two wavelengths were fitted with monoexponential functions Observed rate
rate constants measured at the two wavelengths over 5–10 shots
Acknowledgements
The authors thank Sylvie Dijols (UMR 8601 CNRS, Paris) for the synthesis of the guanidines and N-hydroxyguanidines used in this study DL-G thanks Zhi-Qiang Wang, Chin-Chuan Wei and Koustubh Panda (Cleveland Clinic Foundation) for their help with the stopped-flow experiments, and Jeroˆme Santolini (CEA Saclay, France) for his help in the preparation
of proteins and helpful discussions This work was supported by the French Ministry of Research (fellow-ship grant to DL-G), and by National Institutes of Health (grant CA53914 to DJS)
References
1 Stuehr DJ (1999) Mammalian nitric oxide synthases Biochim Biophys Acta 1411, 217–230
2 Pfeiffer S, Mayer B & Hemmens B (1999) Nitric oxide: chemical puzzles posed by a biological messenger Angew Chem Int Ed 38, 1714–1731
3 Alderton WK, Cooper CE & Knowles RG (2001) Nitric oxide synthases: structure, function and inhibition Biochem J 357, 593–615
4 Dijols S, Perollier C, Lefevre-Groboillot D, Pethe S, Attias R, Boucher J-L, Stuehr DJ & Mansuy D (2001)
NO-syn-thase: the simple, non amino acid N-butyl-N’-hydroxy-guanidine is almost as efficient an NO precursor as
5 Renodon-Corniere A, Dijols S, Perollier C, Lefevre-Groboillot D, Boucher J-L, Attias R, Sari M-A, Stuehr
D & Mansuy D (2002) N-Aryl-N’-hydroxyguanidines, a new class of NO-donors after selective oxidation by nitric oxide synthases: structure–activity relationship
J Med Chem 45, 944–954