(Received for publication, September 20, 1996, and in revised form, February 11, 1997)
From the The active form of endothelial nitric-oxide
synthase (eNOS) is a homodimer. The activity of the enzyme is regulated
in vivo by calcium signaling involving the binding of
calmodulin (CAM), which triggers the activation of eNOS. We have
examined the possible role of calcium-mediated CAM binding in promoting
dimerization of eNOS through the oxygenase domain of the enzyme. A
recombinant form of the oxygenase domain of human eNOS was expressed in
a prokaryotic expression system. This recombinant domain contains the
catalytic cytochrome P-450 site for arginine oxidation by molecular
oxygen as well as the binding sites for tetrahydrobiopterin and
Ca2+-CAM but lacks the reductase domain and associated FAD,
FMN, and NADPH binding sites. Binding of Ca2+-CAM caused an
association of monomeric eNOS oxygenase domain as determined by changes
in fluorescence, both intrinsic and extrinsic, and by gel filtration,
chemical cross-linking, and particle-sizing. Dimerization of the domain
was not dependent on the presence of the substrate, arginine, or the
cofactor, tetrahydrobiopterin. A truncated form of the eNOS oxygenase
domain lacking the Ca2+-CAM binding region did not undergo
self-association to form dimers. These results show that the eNOS
reductase domain is not required for Ca2+-CAM-induced
dimerization of eNOS and suggest that this dimerization may be a
primary event in the activation of eNOS by Ca2+.
The family of nitric-oxide synthases
(NOSs)1 can be divided into two classes:
those that are synthesized constitutively and activated by the binding
of Ca2+-CAM (endothelial and neuronal isoforms) and those
that are induced through the action of specific cytokines and endotoxin
and possess tightly bound Ca2+-CAM (macrophage isoform).
Calmodulin (CAM) is a calcium-activated regulatory protein of 148 amino
acids that binds to and regulates a number of enzymes known to respond
to changes in intracellular calcium levels (1). As a calcium-triggered
switch, Ca2+-CAM recognizes specific basic, amphiphilic
The active form of all isoforms of NOS appears to be the homodimer (8,
17-21). Another common structural feature of the three isoforms of NOS
is a bidomain structure comprised of an N-terminal oxygenase domain and
a C-terminal reductase domain connected by a Ca2+-CAM
binding region, which presumably facilitates electron transfer between
the reductase and oxygenase domains (9-11, 22-25). In the case of
iNOS, subunit association is through the oxygenase domain and is
dependent on BH4, arginine, and heme (8, 11-13, 18). Neither the reductase domain nor the CAM binding region are required for dimerization. Dimerization of nNOS also is promoted by
BH4 and requires heme (20). Dimerization of eNOS, in
contrast, does not appear to be influenced by BH4 (26), and
both the N-terminal and C-terminal domains may be involved in assembly
(21). Thus, the dimerization of eNOS may be unique and may reflect
differences in regulation because oligomerization is a possible mode of
regulation for the NOSs.
The functioning of calmodulin as the molecular switch for activation of
eNOS has been known for some time but the potential role of
Ca2+-CAM binding in homodimer formation has not been
addressed. The bidomain structure of NOS permits the molecular
dissection of the molecule into separate domains, which facilitates
molecular studies aimed at establishing important structure-function
relationships associated with the respective domains. To study the
dimerization step of eNOS activation in relation to
Ca2+-CAM binding, we utilized a prokaryotic expression
system to generate the oxygenase domain of human eNOS, which contains
the Ca2+-CAM regulatory domain, substrate and
BH4 binding sites, and the catalytic heme binding site.
Conformational changes in the recombinant oxygenase domain were
observed as quenching or enhancement of fluorescence both from
intrinsic sources (primarily tryptophan) and from extrinsically tagged
(dansyl) CAM. Association of monomers separate from or concomitant with
Ca2+-CAM binding was tested by gel filtration
chromatography, chemical cross-linking followed by SDS-PAGE analysis,
and light scatter particle sizing. Our results provide the first
experimental evidence that Ca2+-CAM promotes dimerization
of human eNOS through the oxygenase domain of the enzyme. A preliminary
report of these findings has been published (27).
The human endothelial
nitric-oxide synthase cDNA (28) used in this study was the kind
gift of Dr. Kenneth D. Bloch (Harvard Medical School). An
EcoRI-EagI fragment containing the N-terminal 1635 nucleotides of the coding region (NTF) was generated from this
cDNA and included the catalytic and regulatory domains of the
enzyme but not the flavoprotein reductase portion. The N-terminal oxygenase domain was cloned into pET23b (Novagen) to form the plasmid,
pNTF, which was then transformed into the expression host,
Escherichia coli BL21(DE3). The recombinant peptide was verified as eNOS using an eNOS-specific monoclonal antibody (gift of
Dr. J. Pollock, Abbott Labs). A truncated form of the NTF cDNA consisting of the first 1430 nucleotides, which deleted the CAM-binding region, was also generated and transformed into E. coli
BL21(DE3) as plasmid, pNTFcm Recombinant BL21(DE3)-pNTF cells were grown to an
A600 of 0.5 to 0.6 in Luria broth,
aminolevulinate was added to a final concentration of 0.2 mM, and expression was induced by the addition of
isopropylthiogalactoside to a final concentration of 0.4 mM. Incubation was continued for 6-18 h, after which cells
were harvested and eNOS oxygenase domain was purified by a method
adapted from McMillan and Masters (22). Cell pellets were resuspended
in 0.01 vol of S buffer composed of 20 mM Hepes (pH 7.4), 2 mM CaCl2, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 7 µM pepstatin A, 10 µM leupeptin, and 10% glycerol. Lysozyme
was added to 1 mg/ml, and the suspension was allowed to stand on ice
for 30 min. Cells were lysed by 3-6 min of pulsed sonication (Branson 450, 30 watts on 50% duty cycle) on ice, and the lysate was
centrifuged for 60 min at 30,000 × g. The supernatant
was mixed with 0.5 vol of calmodulin agarose (Sigma) containing 0.2 mg
of calmodulin/ml of matrix equilibrated with S buffer. The supernatant
and beads were mixed on a rotator at 4 °C for 2 h. The mixture
was then poured into a column, washed with 50 column volumes of S
buffer plus 0.2 M NaCl, and then washed with 50 column
volumes of S buffer plus 0.5 M NaCl. Bound proteins were
eluted in 20 ml of S buffer without CaCl2 and containing 5 mM EGTA. Peak protein fractions were pooled, concentrated
by ultrafiltration (Amicon, 30k NMWCO), and further purified by
Superose 12 gel filtration chromatography. Recombinant eNOS fractions
were verified by Western blotting and concentrated by ultrafiltration.
Protein concentrations were determined by the BCA reaction (Pierce)
using bovine serum albumin as standard. Recombinant protein from
BL21(DE3)pNTFcm All measurements were made at
room temperature (21 °C) using a Jasco FP770 spectrofluorometer and
a quartz cuvette with 3-mm path length. Extrinsic fluorescence was used
to measure Ca2+-CAM binding to recombinant eNOS oxygenase
domain. A dansylated derivative of calmodulin (Sigma) was used to
provide the external fluor, and solutions were freshly prepared just
before use. The extent of dansylation was calculated using the
extinction coefficient, For measurement of intrinsic fluorescence changes, purified eNOS
oxygenase domain was incubated in the same buffer as used for extrinsic
measurements with the substitution of CAM for dansyl-CAM. In
experiments to measure spontaneous association, incubation was allowed
to continue for 30 min. EGTA was present at a final concentration of 5 mM to block CAM binding. For assaying
Ca2+-CAM-mediated changes, incubation was for 10 min. The
measurements were done using an excitation wavelength of 295 nm with a
slit width of 10 nm and an emission wavelength of 330 nm at a slit width of 5 nm. Calibration was done on buffer blanks. Baseline fluorescence for monomer was determined by heating the samples 10 min
at 55 °C. Calmodulin lacks tryptophan, and tyrosine does not
contribute significantly to the intrinsic fluorescence under these
conditions. Changes in fluorescence emission intensity are expressed as
the percent of change relative to the monomer value.
Recombinant eNOS
oxygenase domain was incubated in assay mixture (see recipe under
"Fluorescence Measurements") with or without calcium and sampled at
various times or for 30 min with various concentrations of recombinant
domain at room temperature. After incubation, the reactions were made
100 mM in phosphate buffer (pH 8.0), and cross-linking was
done by incubating the samples for 1 h at room temperature with
0.1 mM dimethylsuberimidate (Pierce). Reaction was
terminated by the addition of 1 M Tris-HCl (pH 6.7) to a
final concentration of 1 mM. Samples were then analyzed by SDS-polyacrylamide gel electrophoresis. For protein visualization, gels
were stained with Coomassie Brilliant Blue. In some experiments proteins were blotted to nitrocellulose filters, and the eNOS oxygenase
domain was detected using a specific antibody and an alkaline
phosphatase-based colorimetric method (Bio-Rad).
Aliquots of eNOS oxygenase domain were
treated as described for cross-linking and analyzed by fast protein
liquid chromatography using a Superose 12 sizing column (Pharmacia).
The column was calibrated with protein molecular mass standards as
follows: myosin, 205 kDa; An aliquot of eNOS oxygenase domain was
incubated at a final concentration of 1 mg/ml in 20 mM
Hepes (pH 7.4), 0.13 M NaCl, 1 mM
CaCl2, 0.3 mM L-arginine, 10 µM BH4 with or without 150 nM CAM
for 10 min at 21 °C. Samples were analyzed using a DynaPro 801 Molecular Sizing Instrument (Protein Solutions, Inc., Charlottesville, VA). Bovine serum albumin was run as a standard.
The domain structure of human eNOS is shown in Fig.
1 with the cloned regions for NTF and
NTFcm
Purification of the recombinant eNOS oxygenase domain was based on the
use of CAM affinity chromatography, an extensively used and very
effective method for isolating CAM-binding proteins (29). The expressed
peptide had a molecular mass of about 60 kDa as determined by SDS-PAGE
analysis and Western blotting. The yield of purified recombinant
protein was approximately 0.5 mg/liter of cell culture. CAM-agarose
affinity chromatography provides a nearly single-step purification of
the recombinant protein from the bulk of bacterial proteins, few of
which bind to CAM because the organism does not produce this protein.
The fast protein liquid chromatography-purified recombinant protein is
>90% pure as judged by SDS-PAGE analysis and has a reduced carbon
monoxide spectrum characteristic of a cytochrome P-450-type hemeprotein
(data not shown).
Because the oxygenase
domain alone is incapable of NO production, the "activity" of the
recombinant eNOS oxygenase domain was defined in terms of its capacity
to bind Ca2+-CAM and/or to undergo self-association.
Binding of Ca2+-CAM to its target proteins or peptides may
be determined with great sensitivity by using CAM to which a
fluorescent moiety such as dansyl has been attached. Dansyl-CAM retains
the physical and biological properties of native CAM, has virtually the
same dissociation constants, and can be detected in the low nanomolar
range (30). Fig. 2A shows spectra of
dansyl-CAM under various conditions. Dansyl-CAM undergoes a large
increase in fluorescent intensity and a blueshift from about 520-495
nm upon binding of calcium (compare spectra a and
b). This can be explained by the known alteration in
conformation of CAM when it binds calcium, which in this case results
in a change in the environment of the dansyl moiety to a more
hydrophobic state, increasing the quantum yield, and shifting the
emission wavelength (30). Addition of eNOS oxygenase domain (Fig.
2A, spectra c-f) to Ca2+-dansyl-CAM
results in a further increase in fluorescent intensity and shift in
wavelength of maximum emission to about 475 nm associated with the
binding of Ca2+-dansyl-CAM to the recombinant oxygenase
domain. Omission of the substrate for NOS, arginine, had no apparent
effect on fluorescent intensity, whereas omission of the cofactor,
BH4, appeared to slightly reduce the level of fluorescent
intensity (data not shown). Both arginine and BH4 were
included in the incubation mixtures to stabilize the recombinant
oxygenase domain.
Fig. 2B shows titration of 0.1 µM dansyl-CAM
with the recombinant oxygenase domain. The concentration that gave
half-maximal change in fluorescence emission intensity was about 1 nM. The truncated recombinant domain, which lacks the CAM
binding site, shows little or no enhancement of fluorescence emission
intensity. The stoichiometry of binding of eNOS oxygenase domain to
Ca2+-CAM is approximately 1:1 (Fig. 2C).
Measurements of changes in intrinsic fluorescence of recombinant eNOS
oxygenase domain (Fig. 3), which under the parameters used reflect primarily tryptophan fluorescence, demonstrated a Ca2+-CAM-dependent quenching, suggesting a
reorientation of the peptide to expose tryptophan residues to a more
hydrophilic or charged environment. Some quenching also occurs after
extended incubation (30 min) and at high (>100 nM)
concentrations of recombinant oxygenase domain in the absence of
Ca2+-CAM binding, which may be attributed to a spontaneous
concentration- and time-dependent association of monomers.
The omission of BH4 from the incubation mixture appears to
slightly alter the extent of complex formation, but a transfer of
fluorescent energy mediated by BH4 cannot be ruled out as a
cause of the changed fluorescent intensity.
Activation of eNOS is known to involve both
Ca2+-CAM binding and homodimerization of the enzyme (21).
To characterize the Ca2+-CAM-eNOS oxygenase domain complex
implicit in our dansyl-CAM fluorescence enhancement results, we
examined the association by gel filtration (Fig. 4) and
particle sizing (Table I). In Fig. 4, aliquots of eNOS
oxygenase domain were incubated with CAM in the presence (panel
B) or the absence (panel C) of Ca2+ and
then chromatographed on a size exclusion column equilibrated with the
appropriate buffer. Peaks corresponding to both monomeric and dimeric
forms were observed in the presence of Ca2+, indicating
some instability of the complex under these conditions. In the absence
of Ca2+ (+EGTA), there was a minor peak corresponding to a
dimer, which was presumably due to the concentration- and
time-dependent association of monomers observed in the
absence of Ca2+. The elution volume of this minor peak
corresponds to a lower molecular weight than the major peak observed in
the presence of Ca2+ due to the absence of bound CAM. Fig.
4A gives the sizes of protein standards run under the same
conditions. A standard curve of molecular mass versus
retention was plotted by linear regression, and the sizes of eNOS
oxygenase domain monomer and dimer were calculated as approximately 55 and 120 kDa, respectively.
Table I.
Quaternary structure of recombinant eNOS oxygenase domain
Department of Biochemistry and Molecular
Biology,
-helical domains in target proteins to which it binds with high
affinity (2, 3). The constitutive nitric-oxide synthases type I
(neuronal (nNOS)) and type III (endothelial (eNOS)) possess these
amphiphilic, helical regions (4-6) and are expressed as inactive
apoenzymes that only become functional upon binding
Ca2+-CAM. The inducible NOS (iNOS, type II) differs
markedly in that Ca2+-CAM is so tightly bound as to become
virtually a subunit of the iNOS monomer, thus obviating the need for
calcium as a switch to turn on activity (7-12). The activity of
constitutive NOSs is regulated primarily by intracellular calcium
levels through the binding of calcium-charged calmodulin. The
mechanism(s) involved in this CAM-dependent activation are
not known, but by analogy to other CAM-dependent enzymes
could involve a conformational change, oligomerization, or both
(14-16).
Preparation of NTF Expression Clones
.
was expressed and purified in a similar
manner except that a histidine-binding nickel column was used according
to the manufacturer's instructions (Novagen) to bind the expressed,
polyhistidine-tagged fusion protein.
320 = 3400 mol
1
cm
1 and found to be 0.8 mol dansyl group/mol CAM (29).
Incubations were done for 10 min at room temperature in a 200-µl
assay mixture of 20 mM Hepes (pH 7.4), 0.1 M
NaCl, 0.1 mM CaCl2, 0.3 mM
L-arginine, 10 µM BH4, and 1 µM dansyl-CAM. Longer incubation times did not produce
any change in the reading. Excitation was at 345 nm with slit width of
10 nm, and emission was read at 480-500 nm with slit width of 5 nm.
The recombinant eNOS oxygenase domain showed negligible emission under
these conditions. Changes in fluorescent emission due to association
with the recombinant protein were expressed as the percentage of change
relative to dansyl-CAM alone after correcting for background
fluorescence.
-galactosidase, 116 kDa; phosphorylase B,
97 kDa; bovine serum albumin, 67 kDa; and, ovalbumin, 47 kDa.
Structure and Purification of Recombinant eNOS Oxygenase
Domain
indicated. The oxygenase portion of the enzyme
contains the heme coordination site, the sites for
L-arginine and BH4 binding, and the
Ca2+-CAM regulatory region.
Fig. 1.
Domain structure of eNOS. The regions
included in the NTF and NTFcm constructs are
indicated.
[View Larger Version of this Image (15K GIF file)]
Fig. 2.
Fluorescence analysis of dansyl-CAM binding
to recombinant eNOS oxygenase domain. A, fluorescence
emission spectra of 0.1 µM dansyl-CAM. Fluorescent
intensity is given in arbitrary units. Spectrum a, no
additions; spectrum b, with 1 mM EGTA;
spectrum c, with 0.01 µM recombinant domain;
spectrum d, with 0.02 µM recombinant domain;
spectrum e, with 0.05 µM recombinant domain;
spectrum f, with 0.10 µM recombinant domain.
B, titration of 1 µM dansyl-CAM with complete
or truncated recombinant eNOS oxygenase domains. C,
stoichiometry of dansyl-CAM (dCAM) binding to recombinant
oxygenase domain. FE, fluorescence emission intensity.
[View Larger Version of this Image (27K GIF file)]
Fig. 3.
Changes in intrinsic fluorescence of
recombinant eNOS oxygenase domain as a function of concentration.
Incubations were done for 30 min at room temperature in an assay mix
containing 1 µM CAM with or without 0.1 mM
CaCl2 or 10 µM BH4 (solid
line, +BH4; dashed line,
BH4). The change in fluorescent intensity is expressed
relative to that obtained from monomer.
[View Larger Version of this Image (21K GIF file)]
Fig. 4.
Gel filtration molecular sizing of
recombinant eNOS oxygenase domain. Aliquots of recombinant domain
were incubated for 30 min in complete reaction mixture with or without
CaCl2 then analyzed by fast protein liquid chromatography
on a sizing column of Superose 12. The column was calibrated with
protein standards as described under "Materials and Methods."
A, protein size standards. B, with
CaCl2. C, with EGTA.
[View Larger Version of this Image (20K GIF file)]
Condition
Apparent massa
kDa
Controlb
149 ± 11c
Minus
BH4
150 ± 29
Minus CAM
80 ± 10
Bovine
serum albumin standard
79 ± 3
a
Apparent molecular masses were determined at 23 °C
with a DynaPro 801 Molecular Sizing Instrument.
b
The control mixture contained 1 mg/ml of recombinant eNOS
oxygenase domain in Hepes (pH 7.4) containing arginine, BH4,
CaCl2, calmodulin, and dithiothreitol.
c
95% confidence limits.
To further test the self-association of eNOS oxygenase domain monomers as a function of Ca2+-CAM binding, we incubated the recombinant domain in the presence and the absence of Ca2+-CAM and measured complex size with a laser light scatter particle-sizing apparatus. The results, which are shown in Table I, indicate a complex of 149 kDa formed when Ca2+-CAM was present. This complex did not form in the absence of CAM but did form in the absence of BH4. The size of the complex would suggest association of two recombinant domains with two Ca2+-CAMs. This would agree with the 1:1 stoichiometry result seen previously.
SDS-PAGE Analysis of Cross-linked DomainsTo determine
whether the eNOS oxygenase domain undergoes self-association in the
absence of calmodulin binding, increasing concentrations of recombinant
domain were incubated in assay mixture without calcium (with 5 mM EGTA) and then subjected to chemical cross-linking and
SDS-PAGE analysis. Fig. 5A shows that
spontaneous association can occur under these conditions but requires
relatively high concentrations and longer times. Panel B
compares rates of dimerization with and without binding of
Ca2+-CAM. Whereas dimerization in the presence of
Ca2+-CAM is essentially complete within 5 min, a
substantial amount of monomer remained in the assay without calcium
even after 60 min. Although some rearrangement in peptide conformation
seems likely as a result of dimerization alone, it cannot be determined from these data exactly what the respective contributions of
Ca2+-CAM binding and dimerization are to the molecular
switch that turns on NOS activity. It seems clear, however, that
Ca2+-CAM binding serves to promote association of eNOS
subunits. The apparent spontaneous concentration- and
time-dependent association of recombinant eNOS oxygenase
domain monomers in the absence of Ca2+ may suggest that
dimerization alone is not sufficient for the activation of eNOS.
The activation of a number of enzymes is dependent on either oligomerization of the enzyme or on binding of an allosteric effector molecule such as calmodulin. In some cases, such as myosin light chain kinase (7, 14), erythrocyte Ca2+-ATPase (14-16), and the NOSs, both CAM-binding and oligomerization may be involved in enzyme activation. The role of calmodulin in promoting dimerization of eNOS may be unique to this isoform of NOS and could be a part of the signal transduction pathway that regulates NO production in endothelial cells. As pointed out by Lee et al. (21), the NOS isoforms share many features but have different physiological roles and appear to be regulated by different mechanisms. These authors suggested that oligomerization may be a possible mechanism for the modulation of NOS activity based on their results and those of others indicating that the active form of all NOS isoforms is a homodimer and that dimer formation is influenced by substrates and cofactors (8, 11-13, 17-21). Ghosh and Stuehr (11) showed that the N-terminal oxygenase domain of iNOS participates in dimer formation and that this association is dependent on the cofactor, tetrahydrobiopterin. This association did not require calmodulin. The dimerization of nNOS also appears to occur through the oxygenase domain, requires heme, and is stabilized by BH4 and arginine (20). A possible requirement for calmodulin in the dimerization of nNOS was not addressed. Rodriguez-Crespo et al. (26), using recombinant bovine eNOS, reported that tetrahydrobiopterin was required for activity but did not appear to affect the extent of dimerization as determined by SDS-PAGE. An elegant study by Lee et al. (21) demonstrated that oligomerization of eNOS is important for eNOS activity in vivo. Their results suggested that both the N-terminal and C-terminal domains are involved in eNOS dimerization. However, the constructs used to produce the truncated N-terminal and C-terminal eNOS mutants contained overlapping sequences (residues 512-732 in the eNOS sequence), making interpretation of their results difficult. Some of this shared sequence (residues 512-545) is also contained in the recombinant eNOS oxygenase domain used in the present study.
Our results clearly demonstrate that the binding of calmodulin to a recombinant oxygenase domain of human eNOS promotes dimerization and that the C-terminal reductase domain is not required for dimer formation. Our results also show that the substrate, arginine, and the cofactor, tetrahydrobiopterin, are not required for eNOS dimer formation, in agreement with the the results of Rodriguez-Crespo et al. (26), but our results do suggest that tetrahydrobiopterin may stabilize the dimeric state of eNOS. These results are in contrast to the results reported by Stuehr's group, which demonstrated an essential role for tetrahydrobiopterin in the dimerization of iNOS (11). Thus, these two isoforms of NOS clearly differ in this regard. Interestingly, iNOS activity is not regulated by calcium signaling because calmodulin remains tightly bound to the enzyme (continuously activated). Because the calcium-dependent binding of calmodulin is required for the activation of eNOS, calmodulin-induced dimerization may play an important role in the modulation of the activity of this key enzyme in vasoregulation.