| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 278, Issue 34, 31814-31824, August 22, 2003
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
**
From the
Department of Pharmaceutical Chemistry,
University of California, San Francisco, California 94143-2280 and
¶Stanford Medical Informatics, Stanford
University, Stanford, California 94305
Received for publication, March 29, 2003 , and in revised form, May 30, 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
,
-fold structure distinct from the highly -helical
structure of the P450 enzymes
(3–6).
Furthermore, the two-stage mechanism for the oxidation of
L-arginine to L-citrulline and nitric oxide (NO) by NOS
is more complex than that of a P450 enzyme, as it requires
tetrahydro-biopterin (H4B) as a transient electron donor
(7). In contrast, the NOS
reductase domain exhibits considerable sequence identity with CPR, and the
electrons, as in P450/CPR, flow from NADPH to the FAD, then to the FMN, and
finally, upon substrate binding, to the heme
(2,
8).
In the cytochrome P450 system, the principal control on inter-protein electron transfer is the requirement for association of the P450 with CPR. The NOS complex also requires interactions between independently transcribed proteins because it is a homodimer in which the reductase domain of one subunit provides electrons to the oxygenase domain of the other (9, 10). However, inter-domain electron transfer in NOS also requires the binding of CaM, and the Ca2+ dependence of CaM binding differs between the constitutive and inducible isoforms (11, 12).
The three NOS isoforms are classified according to their dependence on Ca2+ for CaM binding (2). The inducible isoform (iNOS or NOS II) is regulated primarily at the transcriptional level. It irreversibly binds CaM in a Ca2+-independent manner and exhibits the highest catalytic activity of the three isoforms (11). The constitutively expressed isoforms, eNOS (or NOS III) and nNOS (or NOS I), bind CaM reversibly in a Ca2+-dependent manner (10–12). These differences in activity and regulation reflect the functions of the proteins; a tempered activity is consistent with the roles of the constitutive forms in functions such as blood pressure regulation and neuronal signaling, and a strong burst of activity with that of the inducible form in the immune response to pathogens. Interest in controlling the activities of the individual NOS isoforms currently centers on their roles in inflammatory responses such as arthritis and septic shock as well as in homeostatic diseases such as hypertension (14).
We have sought to define the structural and chemical elements that control electron transfer in the different NOS isoforms by comparing the NOS family with homologous proteins. The redox behavior of the iNOS reductase domain is similar to that of CPR, and iNOS is capable of high electron transfer rates to exogenous acceptors in the absence of CaM. The flavin domain of iNOS is most similar to that of CPR in sequence, whereas those of eNOS and nNOS exhibit slightly lower identity to CPR as a result of the site-specific insertion of additional residues (Ref. 1 and references therein).
The elements within the eNOS and nNOS reductase domains that distinguish them from the iNOS reductase domain and CPR may help to govern the high and low activities of the CaM-bound and CaM-free enzymes, respectively, because CaM binding strictly regulates electron transfer between the reductase and heme domains as well as between flavins within the reductase (15). These differences in electron transfer activity with and without CaM can be observed when the nNOS heme and reductase domains are independently expressed (16, 17). Furthermore, the elements that control the CaM dependence and relatively low rate of electron transfer to the heme domain in eNOS appear to be localized entirely within the reductase domain because NOS chimeras with the eNOS oxygenase domain fused to the iNOS or nNOS reductase domain exhibit high activities (18). Conversely, chimeras in which the eNOS reductase domain is fused to the iNOS or nNOS oxygenase domain have low, eNOS-like activities (18). This CaM regulation of electron transfer rates is postulated to be mainly conformational in nature, because CaM binding has been shown to have no effect on the midpoint potentials of the flavins (19) and no effect on spin-relaxation rates of the flavin radical (20) in nNOS. CaM binding accelerates electron flow within the reductase domain, but electron transfer through the flavins and from them to alternate exogenous acceptors still occurs at low rates even in the absence of CaM (21).
Alignment of the NOS primary sequences identifies differences that could be responsible for the differences in the constitutive and inducible isoforms. The largest sequence difference between the reductase domains of the three NOS isoforms is a 43–45-amino acid insert (at residue 595 of eNOS and 835 of nNOS) that is absent in iNOS. Excision of this autoinhibitory (AI) element from eNOS or nNOS enhances the CaM-dependent NO-synthesizing activity to a level approaching that of iNOS (22, 23). However, complete Ca2+ independence analogous to that of iNOS is not obtained upon deletion of the AI element, so other reductase domain elements must also interact with CaM.
An 
20–40-amino acid extension at the C terminus is a second
feature that distinguishes the NOS enzymes from CPR
(24). Trimming this extension
causes CaM-independent hyper-activity of interflavin electron-transfer in all
three NOS isoforms, but NO production is only slightly increased in iNOS and
is decreased in eNOS and nNOS
(25,
26). However, in contrast to
removal of the AI insert, truncation of the terminal extensions does not alter
the Ca2+ dependence of CaM binding. In eNOS,
phosphorylation of the C terminus provides an alternative route for activation
of electron transfer that is independent of the
Ca2+-dependent regulatory mechanism
(24). Altering the C terminus
may loosen a tightly tuned electron transfer system, so that either deletion
of the unique tails or phosphorylation increases flavin reduction and
consequently NO activity in a CaM-independent manner. Further supporting this
argument, mutation of the FNR-family conserved C-terminal aromatic residue in
nNOS similarly increases electron transfer rates in the absence of CaM
(27).
We have added structural models to the sequence-based analysis of CaM-related regulatory elements. Partial reductase structures of nNOS have been solved (28) or predicted (29), but our models of the full reductase domain allow placement of the various regulatory elements relative to the flavin subdomains. Furthermore, comparison between isoforms identifies a unique loop (CD2A) within the connecting domain that is highly conserved by isoform, but non-conserved between them and CPR (Fig. 1). In eNOS, the NOS isoform with the slowest electron transfer and overall activity rates, this loop is 5 amino acids longer than in iNOS. In our models, this CD2A is extended toward the base of the CaM-binding helix (30), and is able to provide several electrostatic interactions with either the FMN subdomain or CaM itself. The structure of CaM bound to a peptide corresponding to eNOS residues 492–510 was recently solved and includes residues at the N terminus of the eNOS model (31). Given the distance and orientation constraints for the position of the peptide bound by CaM, both the CD2A and the AI elements are modeled to be within contact distance of the CaM-binding region (Fig. 2).
|
|
To test the hypothesis that the CD2A loop provides a control mechanism that links CaM binding to intra-reductase electron flow, we have investigated mutants of both the wild-type and chimeric enzymes in which the AI and CD2A sequences were either deleted or exchanged. We report here that the CD2A loop, in concert with the AI element, cooperatively governs electron transfer within the reductase domains of the constitutive NOS isoforms.
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
-subunit of
sulfite reductase (SiR-FP), and nNOS, all of which had >20% conserved
sequence identity with the NOS reductase domains. The templates were
structurally aligned using COMPARER
(36). To align each NOS reductase to the templates, pair-wise alignments between NOS and 1AMO [PDB] were initially created using PSI-BLAST. These PSI-BLAST alignments were then compared with results from other alignment algorithms, including ClustalW and Pfam, as well as with secondary structure prediction algorithms on the JPred2 server (37–39) and manually resolved. The exon/intron splice sites in the human NOS genes were also identified (40–42). In the C-terminal portion of the NOS reductase for which the rat nNOS structure was solved in 1F20 [PDB] , the human NOS sequences were aligned to 1F20 [PDB] using ClustalW. The final NOS-to-templates alignment was combined with the template alignment from COMPARER and used as input for MODELER (43).
Model Validation—After processing in MODELER, the AMBER simulation package version 6.0 was used to minimize the proteins (44). To do this, the LEAP program of AMBER was used to prepare the structures. The structures were minimized by performing 4000 steps of steepest-descent minimization to resolve moderate and large energy conflicts. After minimization, the models were evaluated and compared with the template structures as well as unminimized models using WHAT IF version 4.99, PROCHECK version 3.5.4, and PROSA version 2.0 (45–49). Sequence similarity was calculated between NOS and the multiple sequences in the modeling alignment using PlotSimilarity from the Wisconsin Package version 9.0 software package (Genetics Computer Group) (50). Finally, the side chain conformations within the cofactor binding sites were compared between each modeled NOS structure and CPR. MidasPlus software (Computer Graphics Laboratory, University of California, San Francisco, CA) was used to align the two structures, and the coordinates from the CPR cofactors were transposed directly onto the NOS models2 without further minimization (51).
Chimera Preparation
Materials—HEPES and agarose were from Fisher. DNA
manipulations were done using enzymes, buffers, and reagents from New England
Biolabs and purification kits from Qiagen, which also supplied the
nickel-nitrilotriacetic acid resin. Oligonucleotide primers were synthesized,
and the DNA sequenced, by the Biomolecular Resource Center (University of
California, San Francisco). Agar was from Invitrogen, H4B from
Alexis (San Diego, CA), isopropyl-1-thio--D-galactopyranoside
from Promega, and LB, yeast extract, and tryptone from Difco. All other
materials were purchased from Sigma.
General PCR Cycling—A Progene thermocycler from Techne (Cambridge, United Kingdom) was employed. Mutagenesis was accomplished by standard overlap extension PCR techniques, utilizing as the template previously constructed NOS chimeras that possess engineered NheI splice sites at amino acids 760–761, 538–539, and 527–528 in nNOS, iNOS, and eNOS, respectively (18). End primers utilized the NheI splice site, the post-termination XbaI cloning site, or preexisting unique sites within the gene (iNOS BsrGI and eNOS KpnI). All PCR reactions utilized 0.5 µM primers, Vent® polymerase, which possesses 3' to 5' proofreading activity, and Vent® polymerase buffer. Me2SO was added when necessary as specified.
Construction of Parental N/I(eCD2A)—The first step toward creating CD2A-swapped mutants was construction of N/I(eCD2A), which was the parent for the /iNOS(eCD2A) reductase module.
A "megaprimer" method was used to fuse sequences from iNOS and eNOS (52). First, flanking primers (forward, 5'-CAAACTGTGT GCCTGGAGGT TCTGGAGAAA GGATCCCCAG GCGGCCCTC; and reverse, 5'-GGCTGAGTGA GCAGGGGGGC AGCCGTGGGT CCCGCACCCA GCTGGGAGGA GGGCCGCCTG GGGATC) produced megaprimer Iecdmega, which possessed external sequences corresponding to iNOS (up to Leu-800 and after Lys-811), an internal sequence corresponding to the eNOS CD2A domain (Glu-835 to Pro-850), and a silent BamHI site to facilitate screening. This megaprimer was used with primer "pre-BsrGI" (5'-CACAGTCCTC TTTGCTAGCG AGACAGGGAA GTCT) to produce PCR product A, and with primer (5'-GCGCGCCGAA GCTTTCAGAG CCTCGTGGCT TTG) to give B. Final PCR product AB was synthesized using 10 ng each of A, B, and the two end primers (A plus B, 3% Me2SO, one cycle of 5 min at 94 °C and 5 min at 72 °C; the end primers were then added, followed by 25 cycles of 1 min at 94 °C, 1 min at 55 °C and 1 min at 72 °C, then one 15-min 72 °C extension). AB was subcloned into the N/I via BsrGI and XbaI sites to produce the N/I(eCD2A) chimera.
Construction of I/I(eCD2A)—I/I(eCD2A) was subcloned from N/I-(eCD2A) into I/E via NheI and XbaI sites.
Construction of E/E(iCD2A) and
E/E(
AI)(iCD2A)—The procedure was identical
to that used for the construction of N/I(eCD2A), except primers EimF
(5'-GTCTGTGGCT GTGGAGCAGC TGGATGAGAG CGGATCCTAC TGGGTCAAAG AC) and EimR
(5'-CGCACGGTGC ACGGGGGCAG CCTCTTGTCT TTGACCCAGT AGGATC) were used to
synthesize the megaprimer, and 5' primer (5'-CCAGACCCCT GGAAAACTAG
TGCGACCAAG GGCGC) and 3' primer (5'-GCCCTTTCGT CTTCAAGCAG
ATCTGAAAAA AAAGCC) were used as end primers. Two-minute steps at 94 °C
were used in place of 1-min 94 °C steps. Subcloning into E/E or
E/E(AI) was accomplished using KpnI and XbaI to
produce E/E(iCD2A) and E/E(AI)(iCD2A), whose eNOS reductase CD2A
sequence (Glu-835 to Pro-850) has been replaced by that of iNOS CD2A (Asp-801
to Lys-811).
Activity Assays—The rate of NO synthesis, determined using the oxyhemoglobin assay, and the cytochrome c reduction rate were both determined at 37 °C according to previously published methods (18, 22). In specific cases, 100 mM KCl was added to determine the effect of salt on activity (22). To determine the Ca2+ dependence of activity, free Ca2+ was calculated at a constant ionic strength of 100 mM KCl using the Kd (Ca2+-EGTA) = 27.9 nM at pH 7.50, 37 °C (22).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
Besides NOS, the cytochrome P450 reductase family specifically includes the
reductase domain of P450BM-3, the reductase domain of sulfite
reductase, and methionine synthase reductase
(34,
35,
54,
55). The N-terminal
FMN-binding domain of the reductase is structurally homologous to flavodoxins
(56,
57), and the C-terminal
FAD/NADPH-binding domain is homologous to ferredoxin reductase (FNR)
(58,
59). Unique to the CPR family
is the disjointed connecting domain of 150 amino acids that is
interrupted by the first 50 amino acids of the FAD/NADPH domain
(8,
59,
60)
(Fig. 2). The secondary
structure predicted using the
JPred2 server for the
NOS reductase domain matches that of the CPR family of reductases nearly
exactly, confirming this structural relationship
(Fig. 3).
|
To model the NOS reductase domain, template structures from the CPR family
were chosen that align with NOS above the 20% threshold required for
accurate alignments (61).
These templates are: 1AMO
[PDB]
, the structure of CPR that includes all the
reductase subdomains (8); 1B1C
[PDB]
,
the FMN domain of CPR (33);
1BVY
[PDB]
, the FMN domain of P450BM-3
(34); 1DDG
[PDB]
, the CD2 plus FNR
domain of SiR-FP (35); and
finally 1F20
[PDB]
, the CD plus FNR domain of rat nNOS
(28). Sequence identity
conservation between the subdomains of these templates and iNOS is shown in
Table I. Our models include
residues 529–1128 of human iNOS, 507–595 and 642–1160 of
human eNOS, and 749–835 and 880–1400 of human nNOS.
|
In generating the NOS-to-template alignments, 1AMO [PDB] was used as the internal register for the N-terminal FMN domain plus part of the CD1, and 1F20 [PDB] for the C-terminal CD1-FNR subdomains. The individual NOS-to-template alignments from ClustalW, PSI-BLAST, Pfam, and JPred2 agreed to within two residue positions throughout the regions of high conserved sequence identity, within the FMN, CD2, and FNR subdomains. The only region of low sequence conservation is within the first portion of the connecting domain (CD1, eNOS residues 709–753; Fig. 3). 1AMO [PDB] is nearly complete in this region, missing only six "hinge" residues (1AMO [PDB] residues 237–242; Fig. 3). However, sequence conservation with iNOS is 6%, well below the "twilight" zone of sequence alignment accuracy (61). 1F20 [PDB] has much better sequence conservation in the CD1, 20 or 35% with iNOS or eNOS, respectively, but the structure begins past the hinge region (at residue 729 of eNOS). Therefore, the short hinge region was unaligned from the templates, and left in the models only for approximation of the backbone position (eNOS residues 719–728; Fig. 3). The 45–47-residue AI element was excluded from the models of eNOS and nNOS, because of the absence of a homologous element within the templates.
Structure Accuracy—The resulting models from MODELER were
evaluated for structural quality relative to the template structures using
PROSA, WHAT IF, and PROCHECK. Plots of PROSA combined energy scores
versus residue number allow recognition of misfolded sequences within
a structure and are regularly used to evaluate homology models
(62). We found strong
correlation between low (good) PROSA combined energy scores and high
conservation of sequence between the models and their structural templates
(Fig. 4, black circle
traces compared with solid white area trace). The gap in the
nNOS (panel A) and eNOS (panel B) traces at residues
835 and 595, respectively, correlates to the position of the AI
element, which has zero sequence conservation with the templates. The models
have similar PROSA energy scores to the 1F20
[PDB]
structure in the C-terminal
domain (Fig. 4, panel A,
black circles compared with white squares). The PROSA energy
scores were greater than zero only in the CD1, as expected from poor sequence
conservation among templates. In light of these poor scores, the CD1 was
treated as only a backbone strand for reference in modeling, and specific CD1
residue interactions are uncertain. The PROSA z-score was also used
as a general quality assessment for compatibility of the NOS sequences with
the reductase fold (Supplemental Table I, available in the on-line version of
this article). When normalized for sequence length into a probability of
goodness, pG, all the models were given highly reliable scores.
|
The more refined qualities such as stereochemistry and packing of residues within the models were also checked, although these scores represent a level of refinement not required for our purposes. WHAT IF Quality Check scores and PROCHECK G-scores were all satisfactory, as calculated for the models before and after energy minimization with AMBER. AMBER minimization reduced minor steric clashes, but did not greatly affect PROCHECK and WHAT IF scores, when compared with scores for unminimized models (data not shown). In no case were the scores outside of an acceptable range. These scores were comparable with the PROCHECK and WHAT IF scores of the templates (Supplemental Table II, available in the on-line version of this article).
For functional reference it was also useful to place the cofactors within these minimized reductase structures. Comparison of the cofactor binding sites between NOS and CPR as well as the templates showed excellent conservation of binding pocket residues, as identified in the CPR structure (8). Thus, the coordinates of the FMN, FAD, and NADPH cofactors could be transposed into the NOS reductase domain models by visual superimposition of the NOS and CPR structures in MIDAS.
Modeled Interactions—Of greatest interest in the present
context are the potential CaM binding interactions with NOS, with reference to
the proposed NOS regulatory elements. Two of the stabilizing salt bridges in
the structure of CPR are between acidic residues at the base of the CD2A loop
and basic residues in the N-terminal helix region
(8). Similar acidic and basic
residues in the CD2A loop and the N-terminal helix are conserved in NOS,
although at the resolution of the models specific salt bridge interactions
cannot be specified with high confidence. The structure of CaM bound to a
peptide corresponding to residues 492–511 of eNOS enables distance and
orientation restrictions to be applied
(31). In this structure, the
last direct CaM interactions are with residue Met-510 of the eNOS peptide, and
the last turn of this peptide is modeled in the reductase models. If the
position of CaM along this canonical recognition helix remains fixed, the
40 x 20-Å elliptical shape of the CaM structure can be used
as a limitation estimate for nearby interactions. The CD2A is within 5–7
Å of the CaM recognition helix at its base. The AI element is inserted
between two points that are within 20 Å of residue Met-510 in eNOS;
thus, it is also within contact distance of CaM, consistent with the ability
of AI-derived peptides to inhibit CaM binding
(29). One of several possible
binding modes for CaM binding to eNOS is drawn in
Fig. 5, showing relevant
distances and orientation.
|
Calcium Dependence—The in vivo modulation of NOS
activity by the free Ca2+ concentration, a primary
mechanism for regulating eNOS and nNOS activity, can be simulated in
vitro by the use of Ca2+/EGTA buffers
(Fig. 6). CD2A-swapped proteins
are labeled according to the isoform that provided the CD2A element, either
eCD2A for eNOS CD2A or iCD2A for the shorter iNOS CD2A. The
wild-type isoforms represent the two extremes of Ca2+
dependence; eNOS has an EC50(Ca2+) of
100 nM, whereas iNOS retains activity even at subnanomolar
Ca2+ levels
(22). Insertion of the iCD2A
element into eNOS in place of the normal eCD2A loop led to a dramatic decrease
of the Ca2+ EC50 to 30 nM, an
effect equal to that obtained by deleting the AI insert
(22). The iCD2A element was
also swapped into the eNOS(AI) construct, yielding
eNOS-(AI)(iCD2A), to determine whether the effect of the CD2A
replacement was sensitive to the absence of the AI element. An
EC50(Ca2+) of 20–30 nM was
measured for both mutants. Thus, whether the AI domain was deleted, the native
eCD2A loop was replaced, or both changes were made at once, the observed
effect was a decrease of the EC50 to 30 nM.
Conversely, insertion of the eCD2A into iNOS did not alter the
Ca2+ dependence significantly from wild type; thus, it
is a function of the CD2A within the context of an eNOS reductase domain that
contributes to Ca2+/CaM dependence of NO activity.
|
The eNOS-derived chimeras with an iCD2A loop also exhibited decreased
activity at Ca2+ levels greater than 100 nM.
At these levels, the amount of added Ca2+-EGTA required
to achieve the desired Ca2+ concentration exceeds 10
mM. It is notable that only the iCD2A-containing chimeras were
sensitive to these high Ca2+-EGTA conditions, whereas
the eNOS(AI) and iNOS(eCD2A) chimeras were unaffected
(Fig. 6). This observation was
consistent with a more general salt inhibition observed in wild-type iNOS
(63). Whereas wild-type iNOS
was not inhibited in the conditions of the Ca2+
dependence assay, the salt sensitivity threshold has been decreased by the
iCD2A swaps. Further observations on the high salt inhibition of NOS activity
in assays done at a fixed 1 mM Ca2+
concentration in the presence or absence of 100 mM KCl are
discussed below.
Overall Activity—The rate of NO production is limited by the activity of the reductase module in NOS heme-reductase chimeras (18, 22, 23). Deletion of the AI element from eNOS enhances activity more than 2-fold, whereas swapping the eNOS AI into iNOS decreases NO activity (23). A similar pattern was observed for the CD2A chimeras, but this effect was modified by the KCl concentration of the assay (Fig. 7, open versus solid bars); therefore, both the autoinhibitory effect and high salt effect are to be analyzed here.
|
Under low salt conditions, the eNOS(iCD2A) mutant was twice as active as
eNOS itself, and this effect was enhanced when combined with the AI deletion
in the double chimera eNOS(AI)(iCD2A). This double chimera was nearly
as active as the E/I chimera, which represents the end point form that would
result from progressive replacements of sequences within the eNOS reductase
with those from iNOS (Fig. 7,
open bars). Thus, replacement of the eNOS CD2A releases some
autoinhibitory function that it had in the context of an eNOS reductase
domain.
When the concentration of KCl was increased to 100 mM in the NO
assay, replacement of the eNOS CD2A with iCD2A had, if anything, a mild
inhibitory effect, consistent with general iNOS-like inhibition by high salt
(compare eNOS(iCD2A) with eNOS, and eNOS(AI)(iCD2A) with
eNOS(AI); Fig. 7,
solid bars). The percentage of activation by addition of KCl is
calculated in Table II to
demonstrate the unique alteration of activity by the iCD2A. From the
calculated values, it is apparent that eNOS activity is normally stimulated by
high salt (31%), and the AI deletion does not greatly alter this stimulation
(13%). However, the iCD2A conferred high salt sensitivity to the eNOS
reductase domain, causing a loss of 45% activity. The combined effect of AI
and iCD2A replacement was also inhibitory (34% decrease), attaining a value
equivalent to the inhibition observed with the E/I chimera (30% decrease)
(Table II).
|
The eNOS CD2A element decreased NO activity when introduced into iNOS, but it did not alter the sensitivity of iNOS to high salt. Similarly, the iNOS(eAI) chimera was also unaltered in its high salt sensitivity when compared with wild type iNOS. The activity of the iNOS(eCD2A) mutant, similar to the iNOS(eAI) mutant, was approximately half as active as wild type iNOS (Fig. 7), and this activity was reduced by 22% in the presence of added KCl (Table II). Interestingly, when placed in the context of the iNOS reductase, which has no AI element, the eNOS CD2A element could lower the overall activity of iNOS to a level similar to that of "activated" eNOS with its AI domain deleted or CD2A replaced. Comparison with the "end point" chimera I/E indicates that still other determinants are required to lower the activity of the iNOS chimeras to the activity level of eNOS, and to switch over to salt stimulation of activity (Table II).
Reductase Activity—The ability of the mutants to reduce cytochrome c was used to measure the intrinsic activity of the reductase domain in the absence (Fig. 8) or presence (Fig. 9) of added salt. CaM stimulation of reductase activity could be tested in all proteins containing an eNOS-heme domain module (Figs. 8 and 9, compare empty with filled bars). Those chimeras containing an iNOS heme domain (and CaM recognition helix) were coexpressed with CaM for stability, and sufficient Ca2+ was present for CaM to remain bound (18). Thus the activities observed for these chimeras reflected CaM binding even when CaM was not added to the assay (gray bars in Figs. 8 and 9). In all cases, the cytochrome c reduction rates for the NOS chimeras exceeded the heme reduction rate calculated from the observed overall activity, a clear indication that electrons are transferred more readily from the NOS reductase domains to cytochrome c than to the adjacent heme domain in the NOS dimer.
|
|
In the absence of KCl, the effect of swapping the iCD2A into eNOS was
minimal on the reductase domain activity
(Fig. 8) under conditions where
NO production had been enhanced by the iCD2A
(Fig. 7). This was the case for
CaM-independent and CaM-dependent cytochrome c activity
(Fig. 8, open and
solid bars). In the presence of KCl, the iCD2A could, however,
increase electron transfer when combined with the AI deletion, and this was
observed for both CaM-free and CaM-dependent reductase activity
(Fig. 9). The
eNOS(AI)(iCD2A) chimera reached reductase activity levels on par with
the E/I chimera under high salt conditions and in the absence of CaM.
CaM-dependent activity of the double chimera even exceeded the activity of the
E/I chimera, suggesting that the AI and iCD2A replacements were sufficient to
attain a parent-like, regulated, high reductase activity
(Fig. 9). Alone the iCD2A was
ineffective, yet it enhanced the AI deletion effect. A further general
observation was that high salt increased the activity of all the eNOS
reductase-containing chimeras (Table
II); thus, the high salt inhibition of NO activity previously
observed in the iCD2A chimeras must result from altered reaction rates after
flavin reduction.
We showed previously that insertion of the eNOS AI domain into iNOS, giving iNOS(eAI), lowered both the reductase and overall activities, although to varying degrees depending on the salt concentration (23). The iNOS(eCD2A) mutant had a low salt activity of two-thirds that of wild-type iNOS, a less dramatic decrease than observed in the iNOS(eAI) chimera, which was one-fifth as active as wild-type iNOS (Fig. 8). When assayed with added KCl, the reductase activity of iNOS(eCD2A) was equivalent to that of wild-type iNOS, which itself was reduced by 80% from the maximal activity observed in the absence of salt (Fig. 9). Thus, the eCD2A had little effect on the activity of iNOS when assayed under low salt conditions, and no effect at all under high salt conditions. Furthermore, in both high and low salt assay conditions, CaM stimulation was minimal, and unaltered from wild-type iNOS.
In general, KCl inhibition of the iNOS(eCD2A) and iNOS(eAI) chimeras was more severe in the cytochrome c assay, whereas NO production was inhibited only slightly by added KCl (Table II). In no case did the iNOS-derived chimeras deviate from the salt dependence of wild-type iNOS (Table II). This indicates that salt inhibition of iNOS occurs on a more general level within the reductase, and that it is independent of the AI or eCD2A alterations. This is an important difference from the iCD2A replacement effects in eNOS, where the iCD2A exerts its effects at the reductase-to-heme electron transfer step or later.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
The CaM recognition helix of the NOS family has been extensively studied
both on the peptide binding level as well as within several NOS chimeras.
Ca2+-bound CaM binds to NOS via its canonical
amphipathic helix, with a hydrophobic (1–5-8–14) binding motif
(67). This 20–26-amino
acid recognition helix includes residues 492–511 of eNOS, 509–534
of iNOS, and 730–752 of nNOS, by human sequence numbering
(31,
68,
69). The low nanomolar
dissociation constant for CaM binding to peptides derived from these eNOS and
nNOS sequences is sufficient to account for tight,
Ca2+-dependent CaM binding to the full-length protein
(30,
68,
69). In a surface
plasmon-resonance study of the equivalent iNOS peptide, irreversible binding
with Kd < 0.1 nM was observed
(68), in contrast with an
alternate gel-shift assay for CaM binding to a slightly longer iNOS-derived
peptide (507–538) that observed weaker,
Ca2+-dependent binding
(30). Circular dichroism
spectroscopic observations show that the CaM recognition element from iNOS
(residues 509–535) actually binds to apo-CaM in a type II -turn
conformation, and its conformation becomes helical when CaM is
Ca2+-bound
(70); therefore, iNOS has at
least two modes in which to bind CaM. NOS chimeras that swap CaM recognition
helices between isoforms have the same Ca2+ dependence
of CaM binding, intermediate between cNOSs and iNOS and only slightly reduced
overall activity when compared with the parent isoform
(30,
71,
72). Thus, only part of the
Ca2+ dependence of CaM binding is the result of the
canonical CaM binding helix itself.
Further interactions in iNOS beyond the canonical CaM recognition helix were shown to contribute to its irreversible, Ca2+-independent affinity for CaM. The Ca2+ dependence of NOS activity could also be reduced in various nNOS chimeras that contain either the iNOS heme or reductase domain along with the iNOS CaM recognition helix, but only the iNOS reductase-containing chimera demonstrated full Ca2+ independence (72). iNOS truncation mutants with successive N- or C-terminal deletions demonstrated a similar requirement of iNOS residues between 490 and 732 for Ca2+-independent CaM binding (71). This evidence has implications for the work presented here, because residues 490–732 include the CaM recognition helix, the FMN, and the CD-1 subdomains, all of which are within direct contact distance of either the AI element or the CD2A loop (Fig. 2). Both the CD2A loop and AI inserts are found only in the constitutive NOSs, and their sequences are highly conserved by isoform, fulfilling the predictive requirements for isoform-specific regulatory elements. We hypothesize that the CD2A element provides a structural connection between CaM binding and regulation of NOS activity, analogous to the function of the AI element.
The iCD2A loop alters the Ca2+ dependence of
CaM-activated NO activity of eNOS in a manner similar to deletion of the AI
element. Attempts to lower the Ca2+ dependence of eNOS
to the Ca2+ independence of iNOS were unable to break
the EC50(Ca2+) limit of 30 nM,
achieved with eNOS(iCD2A) and eNOS(iCD2A)(AI) chimeras
(Fig. 6). This is consistent
with other observations that Ca2+ dependence of NO
activity, and therefore CaM binding, is determined by multiple interactions in
the CaM binding region (30,
71,
72). Progressive replacement
of the eNOS reductase domain segments with sequences from iNOS would
eventually lead to the E/I chimera, which indeed exhibited virtual
Ca2+ independence
(22). The eCD2A did not alter
the Ca2+ independence of iNOS
(Fig. 6), consistent with the
already tight binding of CaM to the CaM recognition helix
(68). The
Kd of CaM for Ca2+ binding
to the last Ca2+ site has been measured as 1.2
nM (13), but
recognition by CaM of its target sequences is likely a sum of cooperative
interactions. The ability of the eNOS(iCD2A) chimeras to bind CaM at 30
nM Ca2+ most likely indicates recognition of
a partial apo/holo-CaM conformation. The CD2A therefore has a role in direct
or indirect interactions that are required for CaM binding.
The eCD2A serves as an autoinhibitory element as well, because replacement
with the iCD2A resulted in increased overall NO production in both the
eNOS(iCD2A) and eNOS(iCD2A)(AI) chimeras. However, at the reductase
activity level, the iCD2A replacement appeared to be masked in the presence of
the AI element when assayed under low salt conditions. This masking was
released in the high salt assay, revealing an additive effect of the iCD2A and
AI deletion on both CaM-dependent and -independent reductase activity. Thus,
the AI is dominant, but the eCD2A also provides autoinhibitory function in
eNOS.
Placing the eCD2A into iNOS had a mixed effect, in that it was visible in the low salt condition favoring iNOS activity, but not in the high salt condition. High salt reduced the activity of iNOS, iNOS(eCD2A), and iNOS(eAI). Insertion of the eCD2A into iNOS reduced both NO and cytochrome c activity; however, this does not indicate it is functioning on an autoinhibitory level. It is more likely that the lost iCD2A interactions would be needed for full iNOS activity. All the conditions assayed here point to nonspecific inhibition in the iNOS(eCD2A) chimera; CaM binding is virtually Ca2+-independent, the CaM binding does not affect the inherent reductase activity, and the low salt preference for iNOS activity is unaltered. Thus, for the eNOS CD2A to function in its autoinhibitory capacity, it needs to be in the context of an eNOS reductase, a conclusion similar to that made for the iNOS(eAI) chimera.
The subtle salt effects of the iCD2A insertion into eNOS are worth further consideration. From Table II it is clear that the iCD2A does not alter the inherent preference of eNOS reductase activity for higher salt, and that high salt inhibition in the iCD2A chimeras arises after the inter-flavin electron transfer steps. A logical interpretation of this salt inhibition of the reductase-to-heme electron transfer step includes the participation of electrostatic interactions. If the eNOS(iCD2A) chimera has electrostatic interactions that are disrupted by high salt concentrations, this implies that the electrostatic interactions provided by the iCD2A still are able to form within the context of an eNOS protein, and that these interactions have become the limiting factor. Other limiting factors that can be stimulated by high salt must exist in wild-type eNOS, at the reductase level. It is extremely interesting to consider conserved electrostatic interactions within the eNOS and iNOS reductase domains contributed by the flanking amino acids near the CD2A. Given their poised locations near the FMN-CD subdomain interface and connections to the CaM recognition helix, it is highly likely that these same interactions connect CaM binding to regulation of electron transfer.
We have shown the following facts. (i) The eCD2A element contributes to the
Ca2+ dependence of CaM-bound NO activity, and this
contribution is equal to that observed for the AI element. However, the CD2A
does not alone determine the Ca2+ dependence of CaM
binding, because complete Ca2+ independence is not
observed in the eCD2A deletion, nor is Ca2+ dependence
conferred when the eCD2A is introduced into iNOS. (ii) The eCD2A also plays an
autoinhibitory role in the control of NO production and electron transfer
activity. The AI element serves as the main autoinhibitory switch that masks
the effect of the eCD2A deletion in the eNOS(iCD2A) mutant, but the two
elements are revealed to be additive when both substitutions are made,
eNOS(iCD2A)(AI). (iii) The CD2A is involved in determining the salt
dependence of NO activity but not reductase activity, possibly through
electrostatic interactions connecting the CaM-binding helix with the CD-FMN
interface. In conclusion, our models for the interplay between CD2A loop
interactions with the CaM-binding helix and with CaM itself provide a
meaningful structural basis for the isoform-specific functional effects of the
CD2A on CaM-dependent regulation of NOS activity.
|
FOOTNOTES |
|---|
The on-line version of this article (available at
http://www.jbc.org)
contains Supplemental Tables I and II.
Supported in part by a fellowship from the Medicinal Chemistry Division of
the American Chemical Society sponsored by Pfizer Global Inc.
|| Supported by National Institutes of Health Grant AR47720-01 (to T.
Klein).
** To whom correspondence should be addressed. Fax: 415-502-4728; E-mail: ortiz{at}cgl.ucsf.edu.
1 The abbreviations used are: NOS, nitric-oxide synthase; nNOS, eNOS, and
iNOS, neuronal, endothelial, and inducible nitric-oxide synthase,
respectively; P450 reductase, NADPH-cytochrome P450 reductase or CPR; CaM,
Ca2+-dependent calmodulin; SiR-FP, -subunit of
sulfite reductase of E. coli; P450BM-3, CYP102 from
Bacillus megaterium; E/I chimera, eNOS heme domain connected to iNOS
reductase domain; AI, autoinhibitory element, residues 595–642 from eNOS
or 835–880 from nNOS; iCD2A or eCD2A, residues 809–813 of iNOS or
834–843 of eNOS, respectively; eNOS(AI)(iCD2A), eNOS with
deletion of the autoinhibitory element and the CD2A replaced by that of iNOS;
heme, iron protoporphyrin IX; H4B,
(6R)-5,6,7,8-tetrahydro-L-erythrobiopterin.
2 Coordinates of the nNOS, eNOS, and iNOS reductase domain models are
available upon request.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
