Structural analysis of isoform-specific inhibitors targeting the tetrahydrobiopterin binding site of human nitric oxide synthases.

Structural Analysis of Isoform-Specific Inhibitors Targeting the
Tetrahydrobiopterin Binding Site of Human Nitric Oxide Synthases
Hans Matter,† H. S. Arun Kumar,*,‡ Roman Fedorov,§,X Armin Frey,⊥ Peter Kotsonis,⊥ Elisabeth Hartmann,§
Lothar G. Fro¨hlich,⊥ Andreas Reif,⊥ Wolfgang Pfleiderer,| Peter Scheurer,# Dipak K. Ghosh,3
Ilme Schlichting,§,X and Harald H. H. W. Schmidt⊥,‡,¢
Sanofi-Aventis, Chemical Sciences, Drug Design, Building G 878, D-65926 Frankfurt am Main, Germany, Max Planck Institut
fu¨r Molekulare Physiologie, Abt. Biophysikalische Chemie, Otto-Hahn-Strasse 11, D-44227 Dortmund, Germany, Institut fu¨r
Pharmakologie und Toxikologie, Bayerische Julius-Maximilians-Universita¨t Wu¨rzburg, Versbacher Strasse 9, D-97078
Wu¨rzburg, Germany, Fakulta¨t fu¨r Chemie, Universita¨t Konstanz, D-78434 Konstanz, Germany, Vasopharm BIOTECH GmbH,
Winchester Strasse 2, D-35398 Giessen, Germany, Duke University and VA Medical Center, Durham, NC 27713, Max Planck
Institut fu¨r Medizinische Forschung, Abt. Biomolekulare Mechanismen, Jahnstrasse 29, D-69120 Heidelberg, Germany,
Rudolf-Buchheim-Institut fu¨r Pharmakologie, Justus-Liebig-Universita¨t, Frankfurter Strasse 107, 35392 Giessen, Germany,
Department of Pharmacology, Monash University, Melbourne, Victoria 3800, Australia
Received January 5, 2005
Nitric oxide synthesized from L-arginine by nitric oxide synthase isoforms (NOS-I-III) is
physiologically important but also can be deleterious when overproduced. Selective NOS
inhibitors are of clinical interest, given their differing pathophysiological roles. Here we describe
our approach to target the unique NOS (6R,1′R,2′S)-5,6,7,8-tetrahydrobiopterin (H4Bip) binding
site. By a combination of ligand- and structure-based design, the structure-activity relationship
(SAR) for a focused set of 41 pteridine analogues on four scaffolds was developed, revealing
selective NOS-I inhibitors. The X-ray crystal structure of rat NOS-I dimeric-oxygenase domain
with H4Bip and L-arginine was determined and used for human isoform homology modeling.
All available NOS structural information was subjected to comparative analysis of favorable
protein-ligand interactions using the GRID/concensus principal component analysis (CPCA)
approach to identify the isoform-specific interaction site. Our interpretation, based on protein
structures, is in good agreement with the ligand SAR and thus permits the rational design of
next-generation inhibitors targeting the H4Bip binding site with enhanced isoform selectivity
for therapeutics in pathology with NO overproduction.
Introduction
NO synthases (NOSs) are multidomain proteins con-
sisting of a heme-containing catalytic oxygenase do-
main, a calmodulin-binding linker, and a NADPH
reductase domain that catalyze the formation of NO
from L-arginine, oxygen, and electrons.1 The three
isoforms of NOS are involved in several pathological
processes, including Alzheimer’s disease and stroke
(NOS-I or nNOS), septic shock, arthritis, and inflam-
mation (NOS-II or iNOS), and edema formation and
reperfusion injury (NOS-III or eNOS).1 Hence, the need
for the discovery of isoform-selective NOS inhibitors is
envisaged for specific therapeutic outcomes.2
All NOS isoforms are structurally similar with highly
conserved active sites, as revealed by X-ray crystal
structures.3-9 Computational10 and experimental stud-
ies9,11 revealed differences at only one active site residue
involved in the recognition of the L-arginine carbonyl
oxygen (NOS-I Asp597 versus NOS-III Asn368),12 which
is a ligand motif in constrained peptidomimetic NOS-I
inhibitors.12-14 Consequently, most arginine-site di-
rected inhibitors show only minimal isoform selectivity
(except for a series of dipeptide amides and constrained
peptidomimetics)12-14 and long-term cross-reactivity
with other heme-containing or arginine-binding en-
zymes cannot be ruled out a priori.
An alternative strategy arises from the fact that NOSs
are the only heme-containing enzymes that require
(6R,1′R,2′S)-5,6,7,8-tetrahydrobiopterin (H4Bip) for maxi-
mal activation. The binding site within the oxygenase
domain, differs structurally compared to other pterin-
dependent enzymes such as aromatic amino acid hy-
droxylases15 and, importantly, has noticeable amino acid
differences among the NOS family. Since H4Bip is
essential for catalysis, dimer stabilization, or both, the
H4Bip binding site is a promising target for isoform-
specificity and thus reduction of the side effects.
Many pteridine-based NOS inhibitors were previously
shown to bind specifically and reversibly to the H4Bip
binding site.15-20 In animal models of disease, these
ligands were demonstrated to be beneficial. To mention
a few, in allograft rejection, survival was prolonged with
similar efficacy as high doses of cyclosporin A,21 while
in septic shock, 4-amino-H4Bip is superior to the sub-
strate-based inhibitor NG-monomethyl-L-arginine in
improving survival.22 The analysis of structural prereq-
* Correspondence to Dr. H. S. Arun Kumar, D.B.A., D.V.M., Ph.D.,
Rudolf-Buchheim-Institute for Pharmacology, Frankfurter Str. 107,
D-35392 Giessen, Germany. Tel: 0049 641 99-47631. Fax: 0049 641
99-47619. E-mail: Kumar.Arun@pharma.med.uni-giessen.de.
† Sanofi-Aventis.
‡ Justus-Liebig-Universita¨t.
§ Max Planck Institut fu¨r Molekulare Physiologie.
⊥ Bayerische Julius-Maximilians-Universita¨t Wu¨rzburg.
| Universita¨t Konstanz.
# Vasopharm BIOTECH GmbH.
3 Duke University and VA Medical Center.
X Max Planck Institut fu¨r Medizinische Forschung.
¢ Monash University.
4783J. Med. Chem. 2005, 48, 4783-4792
10.1021/jm050007x CCC: $30.25 © 2005 American Chemical Society
Published on Web 06/25/2005

uisites for NOS inhibitor binding23 prompted us to
examine the determinants for isoform selectivity of this
series with respect to other NOS isoforms.
We here apply two strategies to identify and validate
selective inhibitors targeting NOS-I using ligand and
protein structure-based approaches. First, the structure-
activity relationship of a focused set of 41 pteridine
analogues based on four different scaffolds was estab-
lished for all three recombinant human NOS isozymes.
Systematic variations at positions 4, 5, 6, and 7 of these
chemotypes revealed substitutions with up to 58-fold
selectivity for NOS-I compared to other isoforms in these
assays.
To complement the identification of selective inhibi-
tors and focus on structural reasons for selectivity, the
X-ray crystal structure of the rat NOS-I oxygenase
dimer, (unknown at the start of this project), was
determined initially to a resolution of 2.0 Å with bound
H4Bip, then at 2.5 Å with bound H4Bip and L-arginine
substrate. Both experimental structures served to con-
struct a homology model for the dimeric oxygenase
domain of the human NOS-I isoform.23
Understanding the structural differences among NOS
isoforms in the H4Bip binding site is paramount in
guiding rational design of selective inhibitors. Conse-
quently the structural information derived from the
human and mammalian NOS isoforms was subjected
to chemometrical analysis of favorable protein-ligand
interactions using the GRID/CPCA strategy.24 This
approach based on the GRID force field25-29 highlights
regions among all H4Bip binding sites showing signifi-
cant differences in their potential recognition of ligands.
The method was also successfully applied for the NOS
L-arginine binding site,14 serine proteases,24,31 cyto-
chrome P450s,32 kinases,33 and matrix metalloprotein-
ases.34 In general, the computed binding site profiles
derived from the GRID force field are analyzed using
principal component analysis (PCA) to simplify the
complex information, while consensus PCA (CPCA)30 is
a recent development to handle protein-ligand interac-
tions from different field types (i.e., polar versus hydro-
phobic). The combined interpretation of the data from
the focused ligand series, X-ray crystal structures, and
comparative binding site analysis may be of value in
guiding the design of the next-generation H4Bip inhibi-
tors with greater selectivity.
Results and Discussion
Structure-Activity Relationship of Selective
Inhibitors. Forty-one pteridine analogues were tested
for isoform-selectivity against three human NOS iso-
forms (hNOS) using known assays.15,35 Table 1 sum-
marizes the biological data for the most active com-
pounds. Previously we observed that varying 6- and
Table 1. IC50 Values of Selected Pteridine-Based NOS-I Inhibitorsa
pteridine scaffold and substituents IC50 (íM) selectivityb
PHS no. scaffold R4 R5 R6 R7 R9 hNOS-I hNOS-II hNOS-III II/I III/I III/II
2 A2 O CH2OCH2-
CH3
H H >500 >500 >500
341 A2 O 1-naphthyl H H 85.10 >500 350.70 >5.8 4.12 <0.70
339 A2 O 4-FPh H H 65.51 >500 253.00 >7.6 3.86 <0.51
183 A2 O CH2OCO-
[4-(NdN)-
(CF3)CPh]
H H 33.46 374.20 208.60 11.18 6.23 0.56
303 A1 N(CH2Ph)2 Ph H H 129.30 451.80 343.40 3.49 2.66 0.76
305 A1 N(CH2Ph)2 4-CH3OPh H H 34.72 400.00 160.00 11.52 4.61 0.40
28 T2 O H CH2NHCH3 H H 3.90 10.57 4.12 2.71 1.06 0.39
30 T2 O H CH2OCO-
[CH(NH2)-
(CH3)]
H H 116.90 >500 >500 >4.3 >4.3 <4.28
32 T2 O COOCH2-R6 cyclic R5-R6 H H 22.35 37.44 10.05 1.68 0.45 0.27
56 T2 O COPh H H H >500 >500 >500
72 T2 O CO-3-
pyridinium-
1-CH2Ph
Ph Ph COCH(CH3)2 111.10 122.70 253.20 1.10 2.28 2.06
331 T1 N(C3H7)2 H Ph H H 15.72 229.80 48.14 14.62 1.15 0.21
332 T1 N(C3H7)2 H 4-CH3OPh H H 9.28 280.50 54.64 30.23 3.91 0.19
203 T1 NH2 H CHOHCH-
OHCH3
H H 1.26 1.70 5.46 1.35 4.33 3.21
330 T1 piperidino H 4-ClPh H H 10.56 106.20 14.71 10.06 1.39 0.14
333 T1 N(C2H5)2 H 4-ClPh H H 13.97 307.40 56.66 22.01 4.06 0.18
334 T1 NH(CH2-
C6H11)
H 4-ClPh H H 3.68 214.20 31.71 58.20 8.62 0.15
a Incubation time was 30 min in the presence of 2 íM H4Bip; Antipterins were used in the concentration range from 1 to 1000 íM
(seven data points, all fits r2 > 0.95). Control values were determined without antipterins. For experimental details, see Experimental
Section. A denotes aromatic scaffold; T denotes tetrahydro scaffold; R represents substituents; Ph denotes phenyl. b Selectivity is defined
as the ratio of the two IC50 values (e.g., II/I represents (IC50 for hNOS-II)/(IC50 for hNOS-I), that is, the substance is x-fold more selective
for hNOS-I over hNOS-II).
4784 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 15 Matter et al.

7-positions of the reduced antipterins with oxo (A2 or
T2) or amino (A1 or T1) groups on position 4 of a
tetrahydro (T) or aromatic (A) pteridine scaffold resulted
in potent porcine NOS-I (pNOS) inhibitors.19,23,36 This
is confirmed for human NOS-I, where also substitution
in positions 4, 5, 6, and 7 of aromatic and reduced
pteridines was essential for affinity and selectivity.
Importantly, bulky, hydrophobic substituents at R5 or
R6 and alkylation of the 4-amino group with hydro-
phobic groups increased hNOS-I selectivity compared
to hNOS-II or -III.
Replacing small groups at R6 by bulky, hydrophobic
ones such as phenyl (PHS 341/339) or oxycarbonyl (PHS
183) resulted in antipterins with 30-100 íM IC50 values
and hNOS-I selectivity. In particular, PHS 183 was 11-
and 6-fold selective for hNOS-I vs -II and -III.
Replacing the 4-oxo group by substituted amines led
to aromatic 2,4-diaminopteridines (scaffold A1; PHS
303, 305). Binding affinity was increased by dibenzy-
lation of the 4-amino group combined with phenyl (PHS
303) or 4-methoxyphenyl (PHS 305) at R6. Specific
4-amino modifications and favorable changes at position
6 led to the most selective inhibitors. PHS 305 has an
IC50 of 35 íM for hNOS-I and selectivity of 12 (hNOS-
I/-II), whereas PHS 303 has an hNOS-I IC50 of 129 íM
and a selectivity of 3 (hNOS-I/-II/-III).
Favorable substitutions were identified for reduced
4-oxo pteridines (scaffold T2; PHS 28, 30, 32, 56, 72).
PHS 28 with a methylaminomethyl group at R6 had an
IC50 value of 5 íM for all isoforms, while an acetyloxy
substituent (PHS 30) at R6 is detrimental (PHS 30 IC50
) 117 to >500 íM). Acylation at N-5 with benzoyl or
benzoylnicotinoylium is also detrimental (PHS 72, IC50
> 100 íM), but PHS 32 with a cyclic urethane is a
moderate, nonselective inhibitor (hNOS-I IC50 ) 22 íM).
Finally, alkylated 4-amino-tetrahydropteridines (scaf-
fold T1; PHS 331, 332, 203, 330, 333, 334) were potent
and selective NOS inhibitors. Although PHS 203 is the
best inhibitor (IC50 ) 1-5 íM for all isoforms), it lacks
efficacy, while full enzyme inhibition (to 0% of Vmax) was
observed for 4-amino alkylated pteridines with hydro-
phobic, electron-rich phenyl (PHS 331), 4-methoxyphe-
nyl (PHS 332), or 4-chlorophenyl (PHS 330, 333 and
334) substituents at R6 (IC50 ) 4-16 íM). These types
of inhibitors were selective up to 58-fold for hNOS-I
versus hNOS-II (PHS 334). In particular, the 4-methoxy
substitution of the R6 phenyl ring influenced affinity
(PHS 331 vs PHS 332; hNOS-I IC50 ) 16 vs 9 íM) and
selectivity (hNOS-I/-II 15-30). Hence, steric bulk and
physicochemical properties of alkyl groups attached to
4-amino pteridines with R6 para-chlorophenyl (PHS
330, PHS 333 and PHS 334) were crucial for affinity
and isoform selectivity. Remarkably, one of the most
potent and selective hNOS-I inhibitors was obtained by
adding a hydrophobic cyclohexylmethyl group at the
amine in R4 (PHS 334; hNOS-I IC50 ) 3.68 íM;
selectivity hNOS-I/-II 58.2, hNOS-I/-III 8.6). Thus, bulky
and hydrophobic substituents at R5 or R6 combined
with appropriate alkylation of the 4-amino group pref-
erably by hydrophobic substituents increased inhibition
and selectivity to hNOS-I compared to NOS-II and -III,
thus demonstrating the effect of hydrophobicity and
steric bulk in these regions of the NOS-I H4Bip binding
site.
Crystal Structures of Rat NOS-I Oxygenase
Domains. To understand the structural basis for NOS
isoform specificity by pteridine analogues, the ligand
SAR is complemented with a comparative analysis of
different NOS binding sites. Structural information for
NOS-I was not available at the start of the project; we
determined the crystal structure of oxygenase domain
of rat NOS-I in the presence/absence of substrate
L-arginine (Table 2). As expected, the overall structure
corresponds to those of NOS-II and -III with the H4Bip
binding site topology (Figure 1) being very similar to
rat NOS-I oxygenase dimer.8,9,12,14 The main interac-
tions in the rat NOS-I H4Bip binding site linked to
binding affinity indicated differences in hydrogen bond
interactions, distances between heavy atoms (Å), and
amino acid sequence compared to those in NOS-I and
-III (Figure 2). Finally, differences within a radius of 8
Å around H4Bip in its experimental position are indi-
cated.
The cofactor H4Bip is deeply buried in the cavity and
not accessible to bulk solvent, it is oriented proximal
and perpendicular to heme. The main protein-ligand
interaction as in other pterin-protein complexes, occurs
between the planar H4Bip ring and the Trp678 indole,
stacked at 3.6 Å distance.37,38 In general, the hydrogen-
bond pattern corresponds to H4Bip bound to NOS-II or
-III.3-5 The 5,6,7,8-tetrahydropteridine interacts with
heme carboxylate (O4 via solvent, N3 directly); the
structurally conserved water is present in related X-ray
structures. The C4-carbonyl oxygen is hydrogen-bonded
to Arg596 guanidine from the substrate binding helix.
For the NOS-II monomer, this part is disordered,
explaining that only dimeric NOS with properly oriented
Arg596 binds H4Bip. The entrance to the H4Bip binding
region is occupied by Met336 and Leu337 side chains,
the latter is present in rat NOS-I but replaced by
Table 2. Data and Refinement Statistics of the Crystal
Structure Determination of NOS-Ioxy
NOS-Ioxy complexed with
H4Bip,
no substrate
H4Bip,
L-arginine
PDB code 1ZVI 1ZVL
Crystal Parameters
space group C2221 P212121
cell parameters: a, b, c (Å) 43.2, 107.5, 163.0 52.4, 111.3, 165.2
Data Collection
ESRF beamline ID14-1 ID14-2
wavelength (Å) 0.934 0.933
resolution of data (Å) 2.0 2.5
no. of obsns/unique reflnsa 72678/24173 181445/32016
completeness
(total/high) %a
92.3/79.5 93.4/79.6
〈I/ó(I)〉 (total/high)a 11.0/4.4 14.3/4.5
Rsym (total/high) %a,b 5.7/23.9 9.2/34.4
Refinement Statistics
resolution range (Å) 8.0-2.0 8.0-2.5
included amino acids 297-716 297-716
no. of protein atoms 3395 2 � 3419
no. of waters 150 324
Rwork/Rfree %c 19.9/25.5 20.1/24.9
rms deviation for
bonds (Å)/angles (deg)
0.008/1.2 0.009/1.3
a Completeness, Rsym, and 〈I/ó(I)〉 are given for all data and for
the highest resolution shell. Substrate-free: 2.0-1.9 Å. L-Arg:
2.5-2.4 Å. b Rsym ) ∑jI - 〈I〉j/∑I. c Rwork ) ∑jFobsdj - kjFcalcdj/∑jFobsdj;
5% of randomly chosen reflections were used for the calculation
of Rfree.
Isoform-Specific Inhibitors of Human NOS Journal of Medicinal Chemistry, 2005, Vol. 48, No. 15 4785

histidine in the human NOS-I ortholog. These 2-amino-
groups interact with the heme carboxylate and the
backbone carbonyl group of Trp678. Finally N8 is in
close hydrogen bonding with the carbonyl group of
Val677, which is replaced by Ala448 in NOS-III. The
H4Bip dihydroxypropyl side chain occupies a pocket
formed by side chains of Met336 (Val106 in NOS-III),
TrpB676, TrpB306, and other backbone amide bonds.
This polar chain interacts with Ser334 and PheB691
carbonyl oxygens and with HisB692 via structural
water, present in other structures but not this X-ray
structure.
Docking and Homology Modeling. To address our
objective to design therapeutically useful hNOS-I in-
hibitors, we employed the rat NOS-I oxygenase X-ray
structure to derive a homology model for the human
ortholog. This model was generated using the program
Composer39-41 in analogy to a previous model.23 Despite
high sequence identity, there are changes close to the
H4Bip binding site, in particular the Leu337 versus His
mutation in hNOS-I. This model and the rat NOS-I
structure were used for docking studies. Figure 3 shows
the binding mode for PHS 334 using QXP42 in compari-
son to the experimental binding mode of H4Bip in the
Figure 1. Stereoview of Fobsd - Fcalcd difference electron density map contoured at 3ó showing the H4Bip binding site in the
oxygenase domain of rat NOS-I. Hydrogen bonds are indicated by dotted lines and water molecules by cyan spheres; atoms are
colored by elements: carbon, gray; oxygen, red; nitrogen, blue. F691, marked by an asterisk, belongs to the other molecule in the
biological dimer.
Figure 2. Schematic of the interaction of H4Bip in the rat NOS-I binding site and the amino acid differences among all three
NOS isoforms. The rat NOS-I residue numbering is given in black with NOS-III numbering (grey) in comparison.
4786 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 15 Matter et al.

rat NOS-I X-ray structure. Hydrogens are omitted for
clarity; the protein binding pocket is indicated using a
solvent-accessible surface43,44 colored by cavity depth.45,46
Since the Monte Carlo based docking could reproduce
H4Bip orientation only with structural water molecules,
these were further included. The postulated binding
mode of PHS 334 is similar in both human and rat
NOS-I cavities and agrees with that of H4Bip with direct
and potentially water-mediated interactions to heme
propionic acid side chain. The 2-amino group is hydro-
gen bonded to heme and Trp678 CdO, while N1
interacts with the same residue via water. Replacing
the C4 carbonyl oxygen by NH results in rearrangement
of Arg596 or additional interaction involving water. The
NH in position 8 is hydrogen bonded to Val677 CdO.
The p-chlorophenyl substituent at C6 is directed toward
a more hydrophobic pocket. This pocket, formed by side
chains of Met336, Trp306B, Trp676B, and Glu694 and
some backbone amide groups, accommodates the hy-
drophobic substituent. The side chain of Met336 plays
a key role by interacting with the cyclohexylmethyl
substituent at N4, thus contacting and stabilizing two
hydrophobic groups in different subpockets. Replace-
ment by the smaller Val106 in NOS-III resulted in less
favorable interactions, which might partially explain the
selectivity difference.
Chemometrical Analysis of Selectivity Differ-
ences. To investigate isoform-specific binding site
interactions in NOS, we applied the GRID/CPCA
strategy24-28,33 to the rat and human NOS-I oxygenase
structures described above and to 11 representative
structures of other isoforms deposited in the PDB at the
time of this study (Table 3). Favorable protein-ligand
interactions were analyzed by PCA to understand
similarities in protein binding sites. In addition to
showing how different two binding sites are from the
perspective of a ligand, PCA also unveils structural
reasons for the discrimination. Different models were
generated, while we will discuss the model using three
GRID probes and a cutout region of 4 Å around PHS
334. This dataset served to derive submodels for pockets
filled with PHS 334 cyclohexyl and p-chlorophenyl
substituents, respectively.
The PCA score plot for the NOS family is shown in
Figure 4. The first principle component (x-axis in Figure
4) reveals separation of NOS-I and -II on the right
(positive PC1 scores) from NOS-III, including human
(3nos) and bovine isotypes. The second component
(y-axis) separates NOS-I (positive PC2 scores) from
other isoforms. The contributions of amide nitrogen,
Figure 3. Comparison of the experimental H4Bip binding
mode in rat NOS-I and the best docking mode for PHS 334.
Water molecules from the X-ray structure are indicated by
cyan spheres; hydrogens are omitted for clarity. The NOS-I
binding site is indicated by its solvent-accessible surface, color
coded by cavity depth.
Table 3. Protein Structures Used for NOS Target Family
GRID/CPCA Analysisa
structure organism method
resolution
(Å) ligand
NOS-I (nNOS)
nnox rat X-ray 2.0 H4Bip
nno2 human homologyb
NOS-II (iNOS)
1nsi human X-ray 2.50 H4Bip
2nsi human X-ray 3.00 H4Bip
4nos humanc X-ray 2.20 H4Bip
NOS-III (eNOS)
1d1v bovine X-ray 1.90 H4Bip
1dmi bovine X-ray 2.00 H4Bip
1dmj bovine X-ray 2.30 5,6-cyclic tetrahydro-
pteridine PHS 32
1dmk bovine X-ray 1.90 4-amino-6-phenyl-
tetrahydropteridine
1fop bovine X-ray 2.30 H4Bip
1nse bovine X-ray 1.90 H4Bip
2nse bovine X-ray 2.30 H4Bip
3nos human X-ray 2.40 H4Bip
a All proteins except nno2 and nnox were retrieved from
Research Collaboratory for Structural Bioinformatics (RCSB).
Ligands, metals, and counterions were removed except for heme
next to the H4Bip binding site. See RCSB entries for references.
Dimers were constructed using the matrix given in the PDB
header. b Built in according to ref 23 using nnox as templates. This
model is based on the Genbank accession code P29475 after
truncations of the N- and C- terminal parts. c Different N- and
C-terminal truncations plus one mutation compared to 1nsi and
2nsi.
Figure 4. Score plot for the GRID/PCA analysis with three
NOS isoforms and a 4 Å region around PHS 334. The
clustering of binding sites in these score plots results only from
similar protein-ligand 3D interaction patterns and not from
1D sequence similarity.
Isoform-Specific Inhibitors of Human NOS Journal of Medicinal Chemistry, 2005, Vol. 48, No. 15 4787