866
Endothelium-Derived Relaxing Factor From
Pulmonary Artery and Vein Possesses
Pharmacologic and Chemical Properties
Identical to Those of Nitric Oxide Radical
Louis J. Ignarro, Russell E. Byrns, Georgette M. Buga, and Keith S. Wood
Downloaded from http://circres.ahajournals.org/ by guest on June 15, 2017
The objective of this study was to elucidate the close similarity in properties between endotheliumderived relaxing factor (EDRF) and nitric oxide radical (NO). Whenever possible, a comparison was
also made between arterial and venous EDRF. In vascular relaxation experiments, acetylcholine and
bradykinin were used as endothelium-dependent relaxants of isolated rings of bovine intrapulmonary
artery and vein, respectively, and NO was used to relax endothelium-denuded rings. Oxyhemoglobin
produced virtually identical concentration-dependent inhibitory effects on both endotheliumdependent and NO-elicited relaxation. Oxyhemoglobin and oxymyoglobin lowered cyclic guanosine
monophosphate (cGMP) levels, increased tone in unrubbed artery and vein, and abolished the marked
accumulation of vascular cGMP caused both by endothelium-dependent relaxants and by NO. The
marked inhibitory effects of oxyhemoglobin on arterial and venous relaxant responses and cGMP
accumulation as well as its contractile effects were abolished or reversed by carbon monoxide. These
observations indicate that EDRF and NO possess identical properties in their interactions with
oxyhemoproteins. Both EDRF from artery and vein and NO activated purified soluble guanylate
cyclase by heme-dependent mechanisms, thereby revealing an additional similarity in heme
interactions. Spectrophotometric analysis disclosed that the characteristic shift in the Soret peak for
hemoglobin produced by NO was also produced by an endothelium-derived factor released from
washed aortic endothelial cells by acetylcholine or A23187. Pyrogallol, via the action of superoxide
anion, markedly inhibited the spectral shifts, relaxant effects, and cGMP accumulating actions
produced by both EDRF and NO. Superoxide dismutase enhanced the relaxant and cGMP
accumulating effects of both EDRF and NO. Thus, EDRF and NO are inactivated by superoxide in
a closely similar manner. We conclude, therefore, that EDRF from artery and vein is either NO or
a chemically related radical species. (Circulation Research 1987;61:866-879)
T
he mechanisms by which acetylcholine and
vasodilators that release nitric oxide radical
(NO) cause vascular smooth muscle relaxation
show many similarities. The major dissimilarity is that
arterial relaxation elicited by acetylcholine1"3 and
venous relaxation produced by bradykinin4 5 are dependent on the presence of endothelium, while relaxation caused by NO and nitrogen oxide-containing
vasodilators such as glyceryl trinitrate and sodium
nitroprusside occurs by endothelium-independent
mechanisms. 67 Both types of relaxants, however,
produce a time- and concentration-dependent stimulation of vascular cyclic guanosine monophosphate
(cGMP) accumulation that precedes the onset of
smooth muscle relaxation.*"15 Although the muscarinic
receptor antagonist atropine selectively inhibits
acetylcholine-elicited responses, the soluble guanylate
cyclase inhibitor methylene blue inhibits GMP forma-
From the Department of Pharmacology, University of California
Los Angeles, School of Medicine, Los Angeles, Calif.
Supported in part by National Institutes of Health Grant HL35014
and a grant from the Laubisch Fund for Cardiovascular Research.
Address for reprints: Dr. Louis J. Ignarro, Department of
Pharmacology, UCLA School of Medicine, 23-278 CHS, Los
Angeles, CA 90024.
Received December 26, 1986; accepted August 11, 1987.
tion and relaxation produced by acetylcholine, bradykinin, and NO.4-5-8"10141617
Hemoproteins related to hemoglobin and myoglobin, which are well known to bind NO with high
affinity, were shown to inhibit or abolish the arterial
relaxant and cGMP accumulating actions of NO.8'9'1618
Recently, hemoglobin and myoglobin were reported to
also inhibit acetylcholine-elicited, endotheliumdependent arterial responses.1419 The effects of hemoproteins on endothelium-dependent and -independent
relaxation of veins, however, have not been studied.
NO is a potent activator of soluble guanylate
cyclase,20"22 including enzyme purified from vascular
smooth muscle,9 and enzyme activation is abolished by
hemoproteins and methylene blue.23 Similarly, unrubbed arterial and venous rings, respectively, contain
an unstable acetylcholine- and bradykinin-releasable
factor that activates purified soluble guanylate cyclase
in a methylene blue-inhibitable and hemoglobininhibitable manner.5 The similar inhibitory action of
hemoglobin on both NO and what appears to be
endothelium-derived relaxing factor (EDRF), together
with the knowledge that hemoglobin avidly binds and
sequestors NO, suggest that EDRF possesses chemical
properties that are similar to those of NO.
The objective of the present approach was to
compare and contrast certain pharmacologic, biochem-
Ignarro el al
Vascular EDRF and Nitric Oxide Radical
Downloaded from http://circres.ahajournals.org/ by guest on June 15, 2017
ical, and chemical properties of EDRF and NO. The
aims of this study were severalfold. A detailed comparison of the pattern of inhibitory effects of increasing
concentrations of oxyhemoglobin on arterial and venous relaxation produced by EDRF and NO was made.
Carbon monoxide (CO), NO, and oxygen all bind to
a common heme site on hemoglobin and, therefore,
have the potential of competition for a common
binding site. Carboxyhemoglobin was recently reported to be slightly less active than hemoglobin in
inhibiting EDRF-mediated arterial relaxation, but the
instability of carboxyhemoglobin in oxygenated medium precluded meaningful interpretation of the data."
In the present study, the effects of continuous delivery
of CO into bathing media on relaxation and cGMP
accumulation in response to arterial and venous EDRF
and to NO were compared. As guanylate cyclasebound heme is required for enzymatic activation by
JSJQ 22.24-27 j t w a s of importance to determine whether a
similar requirement was true for the direct activation
of guanylate cyclase by EDRF.5 In view of the
knowledge that NO reacts with the heme of hemoglobin
to form the corresponding NO-heme adduct,2829 a
spectral analysis was made to ascertain whether EDRF
released from washed aortic endothelial cells could
react similarly with hemoglobin. Finally, the unstable
nature of EDRF and NO were compared, and the effects
of superoxide anion on relaxation, cGMP accumulation, and hemoglobin spectral shifts produced by both
EDRF and NO were evaluated.
Materials and Methods
Preparation of Rings of Bovine Intrapulmonary
Artery and Vein
Bovine lungs were obtained from cows 5 years of age
or older and transported to the laboratory as described
previously.30 Intrapulmonary arterial and venous
branches were rapidly isolated; gently cleaned of
parenchyma, fat, and adhering connective tissue; and
placed in cold preoxygenated Krebs-bicarbonate solution. Segments of the 2nd arterial branch and underlying 2nd venous branch extending into the larger lobe
were isolated. Outside diameters were 4 - 6 mm (artery)
and 6-8 mm (vein). Vessel segments were sliced into
rings (4 mm wide) using a specially designed
microtome.10 Rings prepared in this manner possessed
an intact or functional endothelium as assessed by
80-100% relaxation responses to 10" 7 -10" 6 M acetylcholine (arteries) or 10"8 M bradykinin (veins).
These vascular rings are referred to in the text as
unrubbed. Endothelial cells were largely removed from
vessel segments prior to slicing by gently passing a
moistened stick from a cotton swab into the lumen and
rubbing against the intima for 30-40 seconds without
stretching the vessel walls. These endotheliumdenuded rings sharply contracted in response to 10"6 M
acetylcholine (arteries) or 10"8 M bradykinin (veins).
Mounting Rings and Recording of Muscle Tension
Arterial and venous rings were mounted by means
of fine nichrome wires in jacketed, 25-ml-capacity,
867
drop-away chambers containing Krebs-bicarbonate
solution (37° C) gassed with 95% O2-5% CO2.10 The
upper nichrome wire of each ring was attached to a
force-displacement transducer (model FT03C, Grass
Instrument Co., Quincy, Mass.), and changes in
isometric force were recorded on a Grass polygraph
(model 79D). Length-tension relations were determined initially for unrubbed and endothelium-denuded
rings of artery and vein. Tension was adjusted to the
optimal length for maximal isometric contractions to
potassium by progressively stretching the rings and
repeatedly obtaining contractile responses to 80 mM
KC1, with washing and 15 minutes of equilibration
between each contractile response.30 Arterial rings
generally required a greater optimal tension than
venous rings. Optimal tensions did not vary significantly as a function of intimal rubbing. The optimal
tensions determined in these initial experiments were
employed in all subsequent experiments. Optimal
tensions and maximal contractile tensions developed in
response to potassium chloride at optimal lengths,
respectively, were 6 g and 20-24 g for arterial rings and
4 g and 18-22 g for venous rings.
Arterial and venous rings were routinely depolarized
by addition of 120 mM KC1 following 2 hours of
equilibration at optimal tension and were subsequently
washed and allowed to equilibrate for 45 minutes prior
to initiating any given protocol. This procedure increases and stabilizes any subsequent submaximal
precontractile responses to phenylephrine and related
contractile agents, presumably by loading the smooth
muscle cells with calcium. This procedure has been
employed routinely for bovine pulmonary vessels in
this laboratory.51030-32
Phenylephrine, a selective a,-adrenergic receptor
agonist, was employed to elicit precontractile responses in arterial rings whereas U46619, a thromboxane A2-mimetic eicosanoid, was used to precontract
venous rings. The reason for using U46619 to contract
veins is that contractile responses were highly reproducible and well maintained. In contrast, contractile
responses of venous rings to phenylephrine varied in
magnitude and were inconsistent. This was particularly
evident with unrubbed venous rings and may be
attributed to the presence of a very active EDRF that
tends to override the contractile effects of phenylephrine but not potassium.30 In this regard, U46619
behaved in part like potassium. However, U46619
differed from potassium in that endothelium-dependent
relaxant responses to bradykinin were fully evident in
veins that were precontracted with U46619 but were
markedly attenuated in veins that were precontracted
with potassium. The mechanism by which U46619
mobilizes calcium and causes vascular smooth muscle
contraction is unknown.
Determination of Cyclic Nucleotide Levels
cGMP and cyclic adenosine monophosphate
(cAMP) determinations were made in arterial and
venous rings that had been equilibrated under tension
and depolarized with potassium chloride. Tone was
868
monitored until the time of freeze-clamping. The use
of drop-away bath chambers, freeze-clamping of rings,
preparation and extraction of tissues for cyclic nucleotide determinations, and radioimmunoassay procedures were described previously.8"103132 cGMP and
cAMP levels were determined in aliquots from the
same ring extract. None of the test agents added to bath
chambers interfered directly with antigen-antibody
binding in the radioimmunoassay procedures. Recoveries of standard amounts of added cyclic nucleotides
were determined periodically, and the values ranged
from 92 to 104%. Therefore, no corrections for sample
recoveries were made.
Downloaded from http://circres.ahajournals.org/ by guest on June 15, 2017
Chemicals, Solutions, and Delivery of NO and CO
Acetylcholine chloride, bradykinin triacetate,
A23187, phenylephrine hydrochloride, potassium cyanide, methemoglobin, myoglobin, catalase, hemin,
protoporphyrin IX, sodium dithionite, dithiothreitol,
pyrogallol, superoxide dismutase (bovine liver), and
Sephadex G-25 were obtained from Sigma Chemical
Co., St. Louis, Mo. The crystalline, purified metalloporphyrins tin-, cobalt-, and nickel-protoporphyrin
IX were obtained from Porphyrin Products, Logan,
Utah. U46619 ([15S]-hydroxy-lla,9o: [epoxymethano]prosta 5Z,13E dienoic acid) was provided by The
Upjohn Co., Kalamazoo, Mich., and was dissolved in
absolute ethanol at a concentration of 5 mg/ml.
Dilutions were prepared in cold distilled water just
before use. Solutions of hygroscopic acetylcholine
chloride were prepared, aliquoted, and stored frozen as
described previously.10 Bradykinin, pyrogallol, and
superoxide dismutase are unstable and were prepared
fresh in distilled water just before use.
Nitric oxide (NO; 99% pure) was obtained from
Matheson Gas, Cucamonga, Calif., and was delivered
into bath chambers in the following manner. A saturated solution of NO in distilled water (1-1.5 mM) was
prepared by injecting 50 ml NO as a fine stream of
bubbles into 2 ml ice-cold water that had been
previously deoxygenated by vacuum evacuation for 20
minutes and then flushed with oxygen-free nitrogen
(Matheson Gas) for 20 minutes. The water was
contained in small test tubes sealed with rubber serum
caps so that syringe needles could be inserted for
evacuation and delivery of gases. Serial dilutions were
made in oxygen-free cold water with the aid of gastight Hamilton microliter syringes. Appropriate aliquots of NO solutions were added to bath chambers
using conventional automatic pipettes immediately
after removing the airtight seals. An estimate of the
concentration of NO in saturated and diluted solutions
was obtained by employing the sensitive colorimetric
procedure developed by Bell et al.33 The concentration
of NO in saturated solutions was consistently 1-1.5
mM. The above procedure entailing the use of oxygenfree solutions of NO resulted in vascular relaxant
responses that were very similar to the responses
obtained when gaseous dilutions of NO were bubbled
into the bathing medium, as we had described
previously.81618
Circulation Research Vol 61, No 6, December 1987
Carbon monoxide (CO; 99%) was obtained from
Matheson Gas and was delivered as the gas into bath
chambers as follows. The CO gasline was connected
to the main O2-CO2 gasline that supplies the tissue bath
chambers through fritted glass disks located at the
bottom of the chambers. Approximately equal volumes
of CO and the O2-CO2 mixture were delivered. Delivery
of CO into oxygenated Krebs-bicarbonate solution by
this procedure did not alter the pH of the medium (pH
7.4). Failure to maintain adequate delivery of the
O2-CO2 mixture, however, caused the pH of the medium to decline rapidly and the medium to turn cloudy.
Crystalline methemoglobin was used as supplied.
Deoxyhemoglobin was prepared from methemoglobin
by reduction with dithionite in deoxygenated Krebsbicarbonate solution.26 Deoxymyoglobin was prepared
from the commercially available myoglobin, which is
approximately 75% metmyoglobin, by the same procedure. Oxyhemoglobin and oxymyoglobin solutions
were prepared by slowly bubbling oxygen into the
deoxyhemoprotein solutions for 20 minutes at 4° C.
Cyanohemoglobin was prepared by dissolving methemoglobin in 10 mM triethanolamine-HCl, pH 7.4,
containing a fiftyfold molar excess of KCN at 25° C.
After 30 minutes, the excess KCN was removed by gel
filtration chromatography using a 0.7 X 5 cm column
of Sephadex G-25 that had been preequilibrated with
10 mM triethanolamine-HCl, pH 7.4, containing 0.1
M NaCl. The void volume was determined by precalibrating the column using blue dextran. The identification and concentration of cyanohemoglobin, as well
as the verification of oxyhemoglobin and oxymyoglobin, were determined by visible absorption
spectroscopy.26 Ferro-protoporphyrin IX (heme) was
prepared from crystalline hemin by dissolving the
hemin in 0.1 N NaOH containing 50 mM dithionite.
Dilutions were made in 10 mM triethanolamine-HCl,
pH 7.4, just before use. Tin-, cobalt-, and nickelprotoporphyrin IX were dissolved in 0.1 N NaOH, and
dilutions were made in 50 mM triethanolamine-HCl,
pH 7.6. All concentrations indicated in the text are
expressed as final bath concentrations.
Krebs-bicarbonate solution consisted of (in mM):
NaCl 118, KC1 4.7, CaCl2 1.5, NaHCO3 25, MgSO4
1.1, KH2PO4 1.2, and glucose 5.6. Depolarizing
potassium chloride solution had a composition similar
to Krebs-bicarbonate solution except the sodium chloride was replaced with an equimolar concentration of
potassium chloride. The salt solution in which the
bovine lungs were transported from the slaughterhouse
to the laboratory (30 minutes) had the following
composition (in mM): Tris-HCl (pH 7.4) 23.8, NaCl
125, KC1 2.7, CaCl2 1.8, and glucose 11. Lungs were
excised from animals and immediately submerged in
ice-cold salt solution contained in a plastic bag. The
bag was sealed, placed in a cooler packed with ice, and
brought to the laboratory.
Spectrophotomethc Analyses
Fresh bovine aortic endothelial cells were employed
as the source of EDRF and were isolated by a
Ignarro et al Vascular EDRF and Nitric Oxide Radical
Downloaded from http://circres.ahajournals.org/ by guest on June 15, 2017
modification of the collagenase digestion procedure
reported previously.34" Fresh segments of bovine
thoracic aorta (30-40 cm in length) were rinsed with
phosphate-buffered saline (pH 7.4) at 25° C and small
arterial branches were ligated to reduce leakage. A
solution of collagenase (protease free; 0.2% wt/vol in
buffered saline) was added into the lumen of each aorta
(sealed at one end) and maintained for 15-20 minutes
with periodic agitation. The media were collected and
the above procedure was repeated four times. Pooled
media were centrifuged, and cell sediments were
resuspended and washed twice in buffered saline and
resuspended in unoxygenated Krebs-bicarbonate solution to yield a cell concentration of approximately 106
endothelial cells/ml. Cell suspensions were kept at 37°
C for 1-3 hours before use.
A solution of 5 fiM (0.325 mg/ml) deoxyhemoglobin in pH 7.4 buffered (25 mM Tris-HCl) Krebsbicarbonate solution was prepared and scanned at 25°
C using an LKB Ultrospec II Model 4050 recording
spectrophotometer. A sharp absorbance peak in the
Soret region at 433 nm (OD = 0.6; E = 133/mM/cm)
was consistently observed. NO was delivered by
injection of 1 ml or 10 ml of purified NO gas into a
quartz cuvette containing 4 ml of deoxyhemoglobin,
and the solution was immediately scanned. A sharp
absorbance peak at 406 nm (OD = 0.5; E=120/
mM/cm) was observed when excess NO (10 ml) was
reacted with the deoxyhemoglobin. An aliquot of aortic
endothelial cells (1 ml containing either 2 x 105 cells or
106 cells) was added to 3 ml of 6.25 /JLM deoxyhemoglobin in the absence or presence of added test agents,
and the suspension was mixed gently at 25° C for 90
seconds. The cell-free liquid component was aspirated
into a Pasteur pipette fitted at the tip with a small piece
of Nytex nylon cloth, which served to exclude the cells.
The cell-free solution was transferred to a cuvette and
immediately scanned.
Purification and Assay of Bovine Lung Soluble
Guanylate Cyclase
Soluble guanylate cyclase from bovine lung was
purified to apparent homogeneity in either the hemecontaining or heme-deficient form exactly as described
previously.25 Before use of enzyme in each experiment,
enzyme fractions were subjected to gel filtration
chromatography to remove dithiothreitol and glycerol,
both of which inactivate EDRF.5 Guanylate cyclase
activity was determined by measuring the formation of
[32P]cyclic GMP from a-[32P]GTP as described.5 The
effects of EDRF released from isolated rings of bovine
intrapulmonary artery or vein into reaction mixtures
containing guanylate cyclase were assessed as recently
described.5
Calculations and Statistical Analysis
Relaxation was measured as the decrease in tension
below the elevated tension elicited by precontracting
arterial or venous smooth muscle with phenylephrine
or U46619, respectively, as indicated. All values in the
text are expressed as the mean±SEM and represent
869
unpaired data. Comparisons were made using either the
Duncan's multiple range test36 where comparisons with
a common control were made (Figure 5) or the
Student's t test for unpaired values for all other
comparisons. The level of statistically significant
difference was p < 0 . 0 5 .
Results
Characteristics of Endothelium-Dependent and
NO-Elicited Relaxation and cGMP Accumulation
in Artery and Vein
Acetylcholine and bradykinin were employed as
endothelium-dependent relaxants of isolated rings of
bovine intrapulmonary artery and vein, respectively.
Acetylcholine relaxes phenylephrine-precontracted arterial rings and causes a concomitant increase in tissue
levels of cGMP but not cAMP, whereas venous rings
are further contracted by acetylcholine.1032 Therefore,
endothelium-dependent relaxation of veins was assessed by addition of bradykinin, which relaxes precontracted venous rings and causes tissue accumulation
of cGMP but not cAMP.4 Bradykinin was not used to
relax arteries because bradykinin stimulates both
cGMP and cAMP formation in bovine pulmonary
artery, and EDRF appears to relax vascular smooth
muscle through an action involving only cGMP.37 NO
was used to elicit endothelium-independent relaxation
of both artery and vein. Endothelium-denuded rings
were used to eliminate any possible influence of
endothelium on intrinsic tone, cGMP levels, and
relaxant responses to NO. Acetylcholine and bradykinin elicited maximal relaxant responses at 10"6 M and
10"7 M, respectively. Higher concentrations often
caused contractile responses because of a direct action
on the underlying vascular smooth muscle, and were
therefore avoided. The limit of solubility of NO in
water as well as the procedure used to deliver NO into
bath chambers precluded the testing of bath concentrations in excess of 10"6 M.
Effects of Hemoproteins and Metalloporphyrins
on Endothelium-Dependent and NO-Elicited
Relaxation of Artery and Vein
Oxyhemoglobin produced a concentration-dependent, rightward shift of the concentration-relaxation
curves to acetylcholine, bradykinin, and NO (Figures
1 and 2). Lower concentrations of oxyhemoglobin
(below 3 X 10"8 M) produced a parallel shift of the
concentration-relaxation curves without depressing
maximal responses, while higher concentrations markedly depressed maximal responses. Oxymyoglobin
produced inhibitory effects that were very similar to
those of oxyhemoglobin (not shown). The deoxyreduced forms of hemoglobin and myoglobin, which
presumably become oxygenated in the tissue bathing
medium, also produced inhibitory effects, but preliminary experiments showed that such hemoproteins were
about tenfold less potent than their preoxygenated
reduced forms. Therefore, the deoxy-reduced hemoproteins were converted to the oxy forms just prior to
use. Ferro-protoporphyrin IX (heme) partially antag-
Circulation Research
870
ioo
100 •
(E
^_
Vol 61, No 6, December 1987
9
8
7
10
6
£? ioo
9
8
7
Bradykinin (-log M)
Acetylcholine (-Jog M)
ioo
Downloaded from http://circres.ahajournals.org/ by guest on June 15, 2017
9
8
7
6
NO (-log M)
9
8
7
6
NO (-log M)
FIGURE 1. Influence of increasing concentrations of oxyhemoglobin on endothelium-dependent and NO-elicited relaxation
of artery. Unrubbed (+ endothelium) and endothelium-denuded
(— endothelium) arterial rings were mounted under 6 g of tension, allowed to equilibrate for 2 hours, and depolarized with
potassium chloride. After washing and 45 minutes of equilibration, oxyhemoglobin was added at a final concentration of
3 X 10-' M (a), 10-* M (b), 3 X W8 M (c), 10~' M (d), or W
M (e)for 10 minutes. Control rings (•) received no oxyhemoglobin . Phenylephrine was then added to rings at concentrations
that elicited contractile responses equivalent to 65-75% of
maximal. At peak contractions, cumulative additions of acetylcholine or NO were made as indicated. Values represent the
mean ± SEM using 15 ringsfrom 5 animals (3 rings per animal).
FIGURE 2. Influence of increasing concentrations of oxyhemoglobin on endothelium-dependent and NO-elicited relaxation
of vein. Unrubbed (+ endothelium) and endothelium-denuded
( — endothelium) venous rings were mounted under 4 g of tension, allowed to equilibrate for 2 hours, and depolarized with
potassium chloride. After washing and 45 minutes of equilibration, oxyhemoglobin was added at a final concentration of
3 X W'9 M (a), 10-" M (b), 3 x W M (c), W M (d), or 10'6
M (e)for 10 minutes. Control rings (•) received no oxyhemoglobin. U46619 was then added to rings at concentrations that
elicited contractile responses equivalent to 65-75% of maximal.
At peak contractions, cumulative additions of bradykinin or NO
were made as indicated. Values represent the mean ± SEM using
15 rings from 5 animals (3 rings per animal).
onized relaxant responses whereas noniron metalloporphyrins such as tin-, cobalt-, and nickel-protoporphyrin IX were inactive, as was protoporphyrin IX
itself (not shown). As reported previously for hemoglobin and myoglobin, l4 " the inhibitory effects of
oxyhemoglobin and oxymyoglobin were completely
reversed after washing the ring preparations several
times with Krebs-bicarbonate solution. Although hot
illustrated, neither oxyhemoglobin nor oxymyoglobin
antagonized the contractile responses of acetylcholine
or bradykinin in endothelium-denuded artery and vein.
In contrast, oxyhemoproteins actually enhanced arterial and venous contractile responses to acetylcholine
or bradykinin. Therefore, oxyhemoproteins did not
inhibit endothelium-dependent relaxation by directly
binding to acetylcholine or bradykinin.
Reversal by CO of Inhibitory Effects
of Oxyhemoglobin on Endothelium-Dependent
and NO-Elicited Relaxation of Artery and Vein
The inhibitory effect of hemoproteins on NO-el icited
vascular smooth muscle relaxation is attributed to the
direct binding of NO to the hemoprotein in the bathing
medium 89 " 18 because the affinity of hemoglobin and
myoglobin for NO is very high. The close similarity
between the inhibitory effect of oxyhemoproteins on
relaxant responses to NO and to acetylcholine or
bradykinin suggests that EDRF and NO possess similar
properties with respect to binding to hemoproteins.
CO, which displays a very high binding affinity for
hemoglobin or oxyhemoglobin, was tested to determine whether the inhibitory action of hemoproteins on
relaxant responses to NO, acetylcholine, and brady-
Ignarro et al
kinin could be similarly reversed or blocked. Oxyhemoproteins were added to bathing media, and CO was
immediately delivered for the duration of a given
experimental protocol. Figures 3 and 4 illustrate that
continuous CO delivery markedly attenuated the inhibitory effects of oxyhemoglobin on both endothelium-dependent relaxant responses of artery and
vein to acetylcholine and bradykinin, respectively, and
endothelium-independent relaxant responses to NO.
CO also abolished the capacity of oxyhemoglobin to
enhance contractile responses in unrubbed rings of
artery and vein. In the absence of added hemoprotein,
CO failed to alter smooth muscle tone, contractile
responses, or relaxant responses. Cyanohemoglobin
was synthesized and tested because this form of
hemoglobin, with its fifth coordinate ligand occupied,
cannot bind to O2, NO, or CO. Cyanohemoglobin (10 ~8
M-10" 6 M) failed to alter endothelium-dependent or
endothelium-independent relaxation of artery and vein
(not shown). These observations suggest that EDRF
binds to and is sequestered by hemoglobin in a manner
that is similar to that for NO.
Relation Between cGMP Accumulation and Relaxation
in Absence and Presence of Hemoproteins
Tables 1 and 2 illustrate the influence of oxyhemoproteins, cyanohemoglobin, and CO on arterial and
venous levels of cGMP in control rings and after
addition of acetylcholine, bradykinin, or NO. cGMP
levels were consistently about threefold higher in
100
100 -
25
"X5
o
Rel
Downloaded from http://circres.ahajournals.org/ by guest on June 15, 2017
o
871
Vascular EDRF and Nitric Oxide Radical
0
9
8
7
6
Acetylcholine (-logM)
c
10
9
•
8
7
Bradykinin (-logM)
0)
o
o>
100
0.
9
8
7
6
NO (-log M)
FIGURE 3. Reversal by CO of oxyhemoglobin blockade of
endothelium-dependent and NO-elicited relaxation of artery.
Unrubbed (+ endothelium)
and
endothelium-denuded
(- endothelium) arterial rings were mounted under 6 g of tension, allowed to equilibrate for 2 hours, and depolarized with
potassium chloride. After washing and 45 minutes of equilibration, JO'7 M (A, A ) or 10'6 M (•, a) oxyhemoglobin was
added for 10 minutes. CO was delivered (A, •) as a fine stream
of bubbles and was started at the time of addition of oxyhemoglobin . Control rings (*) received neither oxyhemoglobin nor
CO. Phenylephrine was then added to rings at concentrations
that elicited contractile responses equivalent to 65-75% of
maximal. At peak contractions, cumulative additions of acetylcholine or NO were made as indicated. Values represent the
mean ± SEM using 24 ringsfrom 4 animals (6 rings per animal).
8
7
NO (-log M)
6
FIGURE 4. Reversal by CO of oxyhemoglobin blockade of
endothelium-dependent and NO-elicited relaxation of vein.
Unrubbed (+ endothelium) and endothelium-denuded ( — endothelium) venous rings were mounted under 4 g of tension,
allowed to equilibrate for 2 hours, and depolarized with
potassium chloride. After washing and 45 minutes of equilibration, 10'7 M (A, A) or 10'6 M (•, a) oxyhemoglobin was
addedfor 10 minutes. CO was delivered (A, •) as a fine stream
of bubbles and was started at the time of addition of oxyhemoglobin. Control rings (•) received neither oxyhemoglobin nor
CO. U46619 was then added to rings at concentrations that
elicited contractile responses equivalent to 65-75% of maximal.
At peak contractions, cumulative additions of bradykinin or NO
were made as indicated. Values represent the mean ± SEM using
24 rings from 4 animals (6 rings per animal).
872
Circulation Research
Vol 61, No 6, December 1987
Table 1. Influence of Hemoproteins and CO on Arterial Levels of cGMP in the Absence and Presence of
Acetylcholine or NO
cGMP (pmol/g tissue)
Controls
Downloaded from http://circres.ahajournals.org/ by guest on June 15, 2017
NO (10" 6 M)
ACh (10~6 M)
+E
-E
Additions
-E
+E
11+2.3
362 ±45
230 ±37
37±4.6
None
10±1.8
21 ± 1.7*
12±1.8*
9.2 + 2.1*
Oxyhemoglobin, 10~6 M
10±2.6
26±3.3*
15±2.1*
13±2.7*
Oxymyoglobin, 4 x 10~6 M
13±2.5
376 ±48
271 ±33
40 ±3.9
Cyanohemoglobin, 10~6 M
14±1.8
296 ±38
217±25
31 ±4.1
Oxyhemoglobin, 10~6 M plus CO
Unrubbed ( + E) and endothelium-denuded ( —E) arterial rings were mounted under 6 g of tension, allowed to
equilibrate for 2 hours, and depolarized with potassium chloride. After washing and 45 minutes of equilibration,
hemoproteins were added for 10 minutes as indicated. Phenylephrine was then added to all rings at concentrations that
elicited contractile responses equivalent to 65-75% of maximal. Carbon monoxide (CO) was delivered as afinestream of
bubbles and was started at the time of addition of oxyhemoglobin. Control rings were quick-frozen at peak contractile
responses. Acetylcholine (ACh) and nitric oxide (NO) were added as indicated to precontracted rings, and rings were
quick-frozen 60 seconds later. Values represent the mean ± SEM using 9 to 12ringsisolated from 3 or 4 animals (3 rings
per animal). cGMP, cyclic guanosine monophosphate.
•Significantly different (p<0.01) from corresponding values obtained in the absence of added hemoprotein (None).
unrubbed than in endothelium-denuded arterial and
venous rings. Acetylcholine and NO, each at 10~6 M,
caused a tenfold and twentyfold increase in cGMP
levels in unrubbed and endothelium-denuded artery,
respectively, after 60 seconds (Table 1). Bradykinin
(10~8 M) and NO (10~6 M) caused a sevenfold and
22-fold increase in cGMP levels in unrubbed and
endothelium-denuded vein, respectively, after 60 seconds (Table 2). Oxyhemoglobin and oxymyoglobin,
but not cyanohemoglobin, lowered the intrinsic cGMP
levels in unrubbed artery and vein to values that
approximated those of endothelium-denuded rings.
cGMP levels in endothelium-denuded rings were unaltered by oxyhemoproteins. CO markedly reversed the
inhibitory effect of oxyhemoglobin and restored cGMP
levels back to the normal range. Oxyhemoglobin and
oxymyoglobin, but not cyanohemoglobin, nearly abolished the tissue accumulation of cGMP caused by
acetylcholine and NO in artery and by bradykinin and
NO in vein. CO markedly reversed the inhibitory effect
of oxyhemoglobin and restored cGMP accumulation
back toward the normal range. None of the hemoproteins tested significantly altered vascular cAMP levels
(not shown).
Heme-Dependent Activation of Soluble Guanylate
Cyclase by NO and EDRF
EDRF released from arterial or venous rings into
enzyme reaction mixtures containing acetylcholine or
bradykinin, respectively, was recently shown to directly activate soluble guanylate cyclase purified from
bovine lung.5 NO was previously shown to activate
purified bovine lung guanylate cyclase by hemedependent mechanisms.25 The purpose of this experiment was to determine whether heme is also required
for enzyme activation by EDRF. As described
previously,5 rings were mounted in bath chambers,
equilibrated under tension, and phenylephrineprecontracted rings were assessed for the presence of
functional endothelium by determining relaxant re-
Table 2. Influence of Hemoproteins and CO on Venous Levels of cGMP in the Absence and Presence of
Bradykinin or NO
cGMP (pmol/g tissue)
Controls
BKN (10" 8 M)
NO (10" 6 M)
-E
+E
-E
+E
Additions
12±1.6
None
276 ±38
262±31
40±5.3
10±1.2
12+1.1*
26±2.1*
Oxyhemoglobin, 10~6 M
19±1.4*
6
11±2.1
Oxymyoglobin, 4x 10~ M
18±2.5*
24±1.8*
22±1.3*
14±1.4
40±3.7
Cyanohemoglobin, 10~6 M
241 ±22
251 ±36
232 ±34
33±4.1
14+1.9
Oxyhemoglobin, 10~6 M plus CO
301 ±26
Unrubbed ( + E) and endothelium-denuded ( — E) venous rings were mounted under 4 g of tension, allowed to
equilibrate for 2 hours, and depolarized with potassium chloride. After washing and 45 minutes of equilibrium,
hemoproteins were added for 10 minutes as indicated. U46619 was then added to all rings at concentrations that elicited
contractile responses equivalent to 65-75% of maximal. CO was delivered as afinestream of bubbles and was started at
the time of addition of oxyhemoglobin. Controlringswere quick-frozen at peak contractile responses. Bradykinin (BKN)
and NO were added as indicated to precontracted rings, and rings were quick-frozen 60 seconds later. Values represent
the mean ± SEM using 9 to 12 rings isolated from 3 or 4 animals (3 rings per animal).
*Significantly different (p<0.01) from corresponding values obtained in the absence of added hemoprotein (None).
Ignarro et al
Vascular EDRF and Nitric Oxide Radical
sponses to acetylcholine or bradykinin. After rinsing
and equilibration, rings were added to guanyiate
cyclase reaction mixtures in the presence or absence of
added acetylcholine, bradykinin, NO, and methylene
blue. Endothelium-intact, but not denuded, rings
caused a twofold to threefold activation of hemecontaining guanyiate cyclase that was markedly enhanced by acetylcholine or bradykinin and abolished
by methylene blue (Table 3). Heme-deficient enzyme
was not activated significantly by rings in the absence
or presence of acetylcholine or bradykinin. NO caused
a direct heme-dependent activation of guanyiate cyclase that was independent of the presence of vascular
rings and was inhibited by methylene blue. Oxyhemoglobin (1 ;u.M) was also tested only in the presence
of arterial rings and was found to produce effects on
acetylcholine- and NO-elicited enzyme activation that
were very similar to the effects of methylene blue.
Downloaded from http://circres.ahajournals.org/ by guest on June 15, 2017
Effects of Pyrogallol and Superoxide Dismutase on
Arterial Relaxation and cGMP Accumulation in
Response to Acetylcholine and NO
Pyrogallol, which generates superoxide anion in
oxygenated medium,38 and superoxide dismutase were
recently reported to inactivate and stabilize, respectively, EDRF released from cultured aortic endothelial
cells.39 In the present study using isolated arterial rings,
pyrogallol was found to inhibit markedly the relaxant
873
and cGMP accumulating effects of acetylcholine in
unrubbed arterial rings and of NO in endotheliumdenuded arterial rings (Figure 5). In contrast, superoxide dismutase enhanced these vascular responses and
also caused a small degree of endothelium-dependent
relaxation and cGMP accumulation when tested alone
(Figure 5). In these experiments, concentrations of
acetylcholine and NO were selected that produced
submaximal vascular responses. The principal action
of superoxide dismutase is to inactivate or scavenge
superoxide radical. High concentrations of hydroxyl
radical scavengers and spin trap reagents, including
mannitol, tyrosine, tryptophan, dimethyl pyrroline
N-oxide, and several substituted phenyl- and butylnitrones, were tested and found to produce no appreciable effects on the vascular responses to acetylcholine, bradykinin, or NO (not shown).
Comparison of Reactivity of Hemoglobin with NO
and EDRF
Hemoglobin reacts characteristically with NO to
form nitrosyl- or NO-hemoglobin, which can be
monitored spectrophotometrically. The addition of
excess NO (10 ml) to a dilute solution of deoxyhemoglobin in pH 7.4 buffered Krebs-bicarbonate caused
a complete shift in the Soret absorption peak (Figure
6) from 433 nm (deoxyhemoglobin) to 406 nm (NOhemoglobin). Addition of a less than excess amount of
Table 3. Heme-Dependent Activation of Purified Soluble Guanyiate Cyclase by EDRF and NO
Guanyiate cyclase activity (nmol cGMP/min/mg)
Artery
Additions
None
Acetylcholine, 10~6 M
+ Meth. blue, 10~4 M
Bradykinin, 10~7 M
+ Meth. blue, 10~4 M
NO, 10"7 M
+ Meth. blue, 10"4 M
-E
32 + 2.1
30±1.7
26 ±2.5
633±45t
41+5.2
Vein
+E
-E
Heme-containing guanyiate cyclase
84 ±6.4*
34 + 3.1
+E
65 ±5.2*
442±28*t
34±3.3t
30±2.4
27±3.2
614±41t
690±54t
38±4.4
43±3.2t
Heme-deficient guanyiate cyclase
65 ±6.1
63 + 5.3
71 ±'8.4
64 + 4.5
96±7.7
87±8.8
338±25*t
31±4.0t
658±53t
43±2.7t
61 ±7.4
60 ±4.7
None
58±3.1
Acetylcholine, 10" 6 M
Bradykinin, 10" 7 M
66±5.6
7
94±9.7
90±6.8
NO, 10" M
Unrubbed (+ E) and endothelium-denuded ( - E) arterial and venousringswere mounted in bath chambers, equilibrated under tension, and tested for endothelial integrity by determining relaxant responses to acetylcholine (artery) or
bradykinin (vein). After rinsing and equilibration, rings were added to enzyme reaction mixtures (0.4 ml) containing 40
mM triethanolamine HC1, pH 7.4, 0.1 mM GTP, 2 mM MgCl2, 0.3 mM CaCl2, 0.1 mM EDTA, 4 ng bovine serum
albumin, 10 mg glucose, 40 mM NaCl, 0.1 fig purified heme-containing or heme-deficient guanyiate cyclase, and the
agents indicated under Additions. Reaction mixtures excluding enzyme were preequilibrated at 4° C with 95% O2-5%
CO2 for 30 minutes, warmed to 37° C, and enzyme was added just prior to insertion of rings. Reaction mixtures were
intubated at 37° C for 10 minutes following addition of rings.5 Basal or unstimulated enzyme activities were 30 + 2.3 and
62 ± 6.7 nmol cGMP/min/mg for heme-containing and heme-deficient guanyiate cyclase, respectively. Values represent
the mean ± SEM using 6ringsisolated from 3 animals (2ringsper animal). Meth. blue signifies methylene blue; EDRF,
endothelium-derived relaxing factor.
•Significantly different (/?<0.01) from corresponding values obtained for - E .
tSignificantly different (p<0.01) from corresponding values obtained for None.
874
Circulation Research
Vol 61, No 6, December 1987
line, 1 JU,M A23187 produced a complete shift of the
Soret peak. Neither cells alone nor acetylcholine or
A23187 alone altered the spectral properties of hemoglobin (Figures 6 and 7). The knowledge that pyrogallol inactivates EDRF39 prompted an analysis of the
influence of pyrogallol on the hemoglobin spectral
shifts produced by A23187 and NO. Pyrogallol nearly
abolished the characteristic spectral shift produced by
both A23187 and NO (Figure 8). Neither superoxide
dismutase nor the oxygen radical scavengers and spin
trap reagents discussed above could be tested in the
spectrophotometric assays because of their marked
interference with absorbance at the wavelengths used.
Downloaded from http://circres.ahajournals.org/ by guest on June 15, 2017
FIGURE 5. Influence of pyrogallol and superoxide dismutase
on endothelium-dependent and NO-elicited arterial relaxation
andcGMP accumulation. Unrubbed (+ endothelium) andendothelium-denuded (— endothelium) arterial rings were mounted
under 6 g of tension, allowed to equilibrate for 2 hours, and
depolarized with potassium chloride. After washing and 45
minutes of equilibration, arterial rings were precontracted by
phenylephrine to 65-75% of maximal, and 10~4 M pyrogallol
(Pyr) or 100 units/ml of superoxide dismutase (SOD) was added
for 60 seconds where indicated. Acetylcholine (10~7 M) or NO
(10~7M) was then addedfor 60 seconds where indicated. Rings
were quick-frozen at 60 seconds after the additions indicated
above. Control rings (C) were quick-frozen at the time of peak
contraction to phenylephrine or 60 seconds after addition of
acetylcholine or NO as indicated. Values represent the
mean ± SEM using 12 rings isolatedfrom 3 animals (4 rings per
animal). * Significantly different (p<0.01)from corresponding
control values.
NO (1 ml) resulted in only a partial spectral shift to 406
nm, with a distinct absorption peak remaining at 433
nm. Because the observations made in this study
suggest that the properties of EDRF closely resemble
those of NO, an attempt was made to conduct an
analogous reaction between EDRF and hemoglobin.
Freshly washed, bovine aortic endothelial cells were
employed to generate sufficient amounts of EDRF for
subsequent reaction with hemoglobin. In the presence
of acetylcholine, the addition of endothelial cells to
deoxyhemoglobin solutions caused a partial shift in the
Soret peak from 433 to 406 nm (Figure 6). The
magnitude of the spectral shift was dependent on the
number of cells and the presence of acetylcholine and
probably, therefore, the amount of EDRF released for
reaction with hemoglobin. The calcium ionophore
A23187, a potent endothelium-dependent relaxant,
also produced a concentration-dependent shift in the
Soret absorption peak of hemoglobin in the presence of
endothelial cells (Figure 7). Unlike 10 fiM acetylcho-
Discussion
The marked inhibitory effect of various hemoproteins on coronary arterial relaxation elicited by NO was
first reported in 1979.16 Hemoproteins antagonized the
effects of NO by binding to and sequestering the NO
in the extracellular compartment and thereby preventing entry of NO into the smooth muscle cells. In that
study, hemoproteins showed little or no antagonism of
the relaxant effects of nitroso compounds or nitroglycerin because the latter vasodilators first penetrate
smooth muscle cells and subsequently release NO,
which produces cGMP accumulation and relaxation.9•40
Hemoproteins are unable to penetrate cells and, therefore, bind NO only in the extracellular compartment.
More recently, Martin and coworkers1419 demonstrated
that reduced hemoproteins antagonized the relaxant
and cGMP accumulating effects of endothelium-
0.6
400
425 450 400 425
Wavelength (nm)
450
FIGURE 6. Spectrophotometric analysis of the reaction between reduced hemoglobin and NO or aortic endothelial cells
plus acetylcholine. Panel A:
, 5 fiM deoxyhemoglobin
solution without any additions;
, 5 fiM deoxyhemoglobin solution bubbled with 10 ml NO;
, 5 fiM deoxyhemoglobin solution bubbled with 1 ml NO. Panel B:
5 fiM
deoxyhemoglobin solution containing 10 /xA/ acetylcholine;
, 5(iM deoxyhemoglobin solution having been reacted
with 10 ixM acetylcholine plus 10" aortic endothelial cells;
,
5 ixM deoxyhemoglobin solution having been reacted with 10
yM acetylcholine plus 2 x 10s aortic endothelial cells. Refer to
"Materials and Methods" for experimental details. The data
illustrated arefrom 1 of3 separate experiments, each employing
a different batch offresh aortic endothelial cells.
Ignarro et al
875
Vascular EDRF and Nitric Oxide Radical
400 425 450 400 425 450
Wavelength (nm)
Downloaded from http://circres.ahajournals.org/ by guest on June 15, 2017
FIGURE 7. Spectrophotometric analysis of the reaction between reduced hemoglobin and aortic endothelial cells plus
A23I87. Panel A:
, 5 yM deoxyhemoglobin solution
without any additions;
, 5 yM deoxyhemoglobin solution
having been reacted with 106 aortic endothelial cells;
,5 yM
deoxyhemoglobin solution having been reacted with 1 yM
, 5 yM
A23187plus 10s aortic endothelial cells. Panel B:
deoxyhemoglobin solution containing 1 yM A23187;
,5
yM deoxyhemoglobin solution having been reacted with 0.1 yM
, 5 yM deoxyA23187 plus 106 aortic endothelial cells;
hemoglobin solution having been reacted with 1 yM A23187
plus JO6 aortic endothelial cells. For experimental details, see
"Materials and Methods." The data illustrated are from 1 of 3
separate experiments, each employing a different batch of fresh
aortic endothelial cells.
dependent arterial relaxants. Oxidized hemoproteins
were partially inhibitory only at higher concentrations.
These investigators postulated that hemoproteins inhibited endothelium-dependent relaxation by binding
to, and thereby interfering with, the relaxant action of
arterial EDRF. Very few studies have been published
that address endothelium-dependent venous smooth
muscle relaxation, and the influence of hemoproteins
on venous tone has not been described. Moreover, a
direct comparison between EDRF and NO in the same
vessel with respect to the effects of hemoproteins has
not been previously made.
The results of the present study illustrate the close
similarity between endothelium-dependent arterial and
venous smooth muscle relaxants and NO with respect
to their actions on isolated rings of bovine intrapulmonary artery and vein. Oxyhemoglobin and oxymyoglobin antagonized the smooth muscle relaxant and
cGMP accumulating effects of acetylcholine, bradykinin, and NO in a nearly identical manner, as indicated
by the remarkably similar rightward shifts of
concentration-relaxation curves in the presence of
increasing concentrations of oxyhemoproteins. The
inhibitory actions of oxyhemoproteins on both smooth
muscle relaxation and cGMP accumulation were similarly and markedly inhibited or reversed by excess
carbon monoxide, presumably because of the continuous formation of carboxyhemoproteins, which have
a lower affinity than oxyhemoproteins for NO and
perhaps EDRF. Although a previous study indicated
that preformed carboxyhemoglobin was slightly less
effective than hemoglobin as an inhibitor of
endothelium-dependent arterial relaxation, the transient nature of this effect, due to the instability of
carboxyhemoglobin in oxygenated medium, precluded
a more in-depth study." In the present study, the
continuous delivery of carbon monoxide along with
oxygen and carbon dioxide into the bathing media
produced consistent reversal of the inhibitory effects of
oxyhemoproteins. These observations suggest that
EDRF and NO bind to a common site on hemoglobin
and myoglobin. This view is consistent with the
findings that not only NO but also EDRF from both
artery and vein required binding to the heme of purified
soluble guanylate cyclase to cause enzyme activation.
More direct evidence for an interaction between EDRF
and heme derived from spectral studies, which revealed
that EDRF reacted with hemoglobin to generate a
hemoprotein complex that was indistinguishable from
the complex formed in the reaction of hemoglobin with
NO under identical conditions. Other similarities
between EDRF and NO were their chemical instability
in oxygenated media, further inactivation by superoxide anion, and stabilization by superoxide dismutase. Collectively, these observations suggest that
EDRF is chemically either NO or some closely related
radical species.
The very high affinity of the reduced forms of
hemoglobin, myoglobin, and iron-protoporphyrin IX
for NO is well known.41 These substances react readily
with NO to form the corresponding NO- or nitrosylheme complexes. The oxidized forms of these substances also bind NO but higher concentrations of NO
are required to first generate the reduced heme iron
0.6
0.5
q
0.4
d 0.3
A
B
1
-\ \
1
'
s
- \/y\
'•£^
0.2 t
" A
\
\
L
\ \
0.1
\ -
•
r
\/
\ '
\
Y
0
400 425 450 400 425 450
Wavelength (nm)
FIGURE 8. Effect of pyrogallol on the reaction between reduced hemoglobin and either NO or aortic endothelial cells
plus A23187. Panel A:
, 5 yM deoxyhemoglobin solution without any additions;
, 5 yM deoxyhemoglobin
solution bubbled with 10 ml NO;
, 5 yM deoxyhemoglobin solution bubbled with 10 ml NO in the presence of 3 yM
pyrogallol. Panel B:
, 5 yM deoxyhemoglobin solution
containing 3 yM pyrogallol;
, 5 yM deoxyhemoglobin
solution having been reacted with 1 yM A23187 plus 106 aortic
endothelial cells;
, 5 yM deoxyhemoglobin solution having
been reacted with 1 yM A23187plus 105 aortic endothelial cells
in the presence of 3 yM pyrogallol. Refer to "Materials and
Methods" for experimental details. Thedata illustrated are from
Iof2 experiments, each employing a different batch of fresh
aortic endothelial cells.
876
Downloaded from http://circres.ahajournals.org/ by guest on June 15, 2017
species.2829 Oxyhemoglobin and oxymyoglobin bind
NO just as avidly as their deoxy-reduced forms because
the affinity of these hemoproteins for NO exceeds by
several orders of magnitude their affinity for oxygen.
Protoporphyrin IX, which lacks iron, cyanohemoglobin, and certain noniron metalloporphyrins such as
tin-, cobalt-, or nickel-protoporphyrin IX are unable to
bind appreciable amounts of either NO or oxygen. The
present observations on the effects of hemoproteins,
metalloporphyrins, and protoporphyrin IX on the
responses of arterial and venous ring preparations to
NO are highly consistent with the known properties of
these substances. In addition, hemoproteins are known
to inhibit the activating effect of NO on soluble
guanylate cyclase.'6-23 The effects of the various forms
of hemoproteins, metalloporphyrins, and protoporphyrin IX on vascular responses to endothelium-dependent
vasodilators were virtually identical to their effects on
responses to NO, thus indicating the close similarity
between EDRF and NO. The present observations
represent a confirmation and extension of those of
Martin et al,1419 who reported that reduced hemoproteins inhibited endothelium-dependent arterial relaxation. The latter investigators, however, did not study
venous relaxant responses, the relaxant effects of NO,
or the effects of noniron metalloporphyrins. The present findings also support those of Griffith et al," who
conducted perfusion-superfusion experiments and suggested that hemoglobin inhibits acetylcholine-elicited
relaxation by combining with and inactivating EDRF.
The inhibitory effects of the oxyhemoproteins on
endothelium-dependent relaxation are not attributed to
any direct binding to acetylcholine or bradykinin as
contractile responses of endothelium-denuded rings to
the latter agonists were unaltered by oxyhemoproteins.
Moreover, the consistent findings that oxyhemoproteins themselves contracted arteries and veins, lowered
intrinsic cGMP levels, and enhanced contractile responses of arteries to phenylephrine and veins to
U46619, all in an endothelium-dependent manner,
argue in favor of the view that the hemoproteins bind
to, and thereby sequester, EDRF. These observations
are consistent with previous reports that certain reduced
hemoproteins augmented arterial tone in rabbit and rat
aorta1419 and lowered cGMP levels in rat aorta42 by
endothel ium-dependent mechanisms.
The concentration-dependent inhibitory effects of
oxy hemoglobin on endothelium-dependent relaxant
responses of artery and vein to acetylcholine and
bradykinin, respectively, were remarkably similar to
the inhibitory effects on endothelium-independent relaxant responses of artery and vein to NO. The
rightward parallel shift in the concentration-relaxation
curves produced by lower concentrations (3 X 10~9 M
to 3 x 10~8 M) of oxyhemoglobin are suggestive of a
competitive antagonistic effect. In addition, the
marked depression of maximal relaxant responses
produced by higher concentrations of oxyhemoglobin
suggests that noncompetitive antagonism is a component of the overall inhibitory action of oxyhemoglobin.
A classical interpretation of these inhibitory effects as
Circulation Research
Vol 61, No 6, December 1987
competitive or noncompetitive, however, may not be
appropriate as oxyhemoglobin appears to elicit its
inhibitory action by binding directly to either NO or
EDRF, thereby eliciting a chemical antagonism. Raising the concentration of NO or EDRF (presumably by
increasing the concentration of acetylcholine or bradykinin) would be expected to overcome the inhibitory or
binding effect of oxyhemoglobin. At high concentrations of oxyhemoglobin, however, the limiting factor
becomes the maximal concentration of NO or EDRF
attainable in the bath chambers. Moreover, in vitro
experiments are complicated by the frequent observation that elevating bath concentrations of acetylcholine
above 10~6 M or bradykinin above 10~7 M results in
ensuing contractile responses due to the direct action
of these agonists on the underlying smooth muscle.
Carbon monoxide reacts with hemoglobin or oxyhemoglobin to form carboxyhemoglobin. Excess carbon monoxide would be expected to compete with NO
or any other ligand that binds similarly to the heme
moiety. This mechanism probably explains the present
findings that carbon monoxide reversed the inhibitory
effect of oxyhemoglobin on the relaxant and cGMP
accumulating responses of artery and vein to NO. In
experiments designed to compare the relaxant responses of NO with those of EDRF, it was clear that
carbon monoxide similarly prevented the inhibitory
effects of oxyhemoglobin on relaxant and cGMP
accumulating responses of artery to acetylcholine and
of vein to bradykinin. These observations suggest
strongly that EDRF and NO interact with a common
binding site on heme. The inability of carboxyhemoglobin to bind large amounts of EDRF likely accounts
for the findings that carboxyhemoglobin, unlike oxyhemoglobin, failed to contract unrubbed vascular rings
or to augment tone elicited by other contractile agents.
Recent studies from this laboratory showed that an
endothelium-derived substance from artery and vein
activated heme-containing guanylate cyclase and that
enzyme activation was abolished by the guanylate
cyclase inhibitor methylene blue. 5 The present study
reveals that such enzyme activation was hemedependent. Activation of soluble guanylate cyclase by
NO is now well recognized to be heme-dependent.22-24"27
The reason for this is that the activating species is
NO-heme, a paramagnetic complex that alters the
molecular configuration of guanylate cyclase and
thereby facilitates the catalytic conversion of GTP to
cGMP.2543 The present observation that guanylate
cyclase-bound heme was required for enzyme activation by EDRF provides strong evidence for the closely
similar properties of EDRF and NO. Moreover, excess
heme in the form of hemoglobin abolished the actions
of both EDRF and NO.
Recent studies have suggested that EDRF can be
inactivated by oxygen radicals such as superoxide
anion.3944 Pyrogallol, which generates superoxide in
oxygenated medium,38 was shown to inactivate EDRF,39
whereas superoxide dismutase protected EDRF
against inactivation.3943 In the present study, pyrogallol
inhibited both endothelium-dependent arterial relax
Ignarro et al
Vascular EDRF and Nitric Oxide Radical
Downloaded from http://circres.ahajournals.org/ by guest on June 15, 2017
ation and cGMP accumulation in response to acetylcholine. In addition, pyrogallol similarly inhibited
endothelium-independent arterial responses to NO.
Moreover, superoxide dismutase enhanced both arterial responses to acetylcholine and NO in a similar
manner. These observations indicate that EDRF and
NO possess virtually identical properties of chemical
instability in the presence of oxygen or superoxide.
The rapid reaction between hemoglobin and NO to
generate the NO-heme adduct of the hemoprotein can
be monitored spectrophotometrically as a shift in the
Soret absorption peak from 433 nm to 406 nm.2MS This
spectral property is highly characteristic of reactions
between NO and many hemoproteins.28'29 The addition
of freshly washed, bovine aortic endothelial cells to
small volumes of deoxyhemoglobin solution containing acetylcholine or A23187 caused a spectral shift that
was characteristic of that produced by NO. This
spectral shift was dependent on the number of endothelial cells added and was more pronounced when
A23187 rather than acetylcholine was used to release
EDRF from the cells. Both NO and A23187 produced
concentration-dependent and identical shifts in the
absorption spectrum of deoxyhemoglobin. The dependence of the magnitude of the spectral shift on the
concentration of A23187 or NO and the absence of a
spectral change in the presence of cells alone indicates
that the spectral shifts observed were not attributed to
formation of oxyhemoglobin due to the introduction of
oxygen into the solutions. Moreover, the Soret absorption maximum for oxyhemoglobin was 417 nm, while
that for NO-hemoglobin was 406 nm. Formation of
oxyhemoglobin from hemoglobin solutions under the
experimental conditions employed was negligible.
Thus, acetylcholine and A23187 stimulated the release
of an endothelium-derived substance, presumably
EDRF, that closely resembles NO in its reaction with
hemoglobin. The observations that pyrogallol prevented the characteristic spectral shift produced both by
NO and by A23187 plus endothelial cells provide
further evidence for the commonality in the reactions
of hemoglobin with NO and EDRF.
Hemoproteins and oxygen, respectively, are well
known to scavenge or bind and to oxidize NO. In the
present study, certain hemoproteins as well as superoxide anion, generated by pyrogallol, were observed to
inhibit the vascular actions or reactivity of both NO and
EDRF. Several hydroxyl radical scavengers and spin
trap reagents known not to bind to or react with NO
were tested and found to elicit no effects on vascular
responses to acetylcholine, bradykinin, or NO. These
radical scavengers included mannitol, tyrosine, tryptophan, dimethyl pyrroline N-oxide, and several substituted phenyl- and butyl-nitrones. Superoxide dismutase, on the other hand, actually enhanced the
vascular effects of NO and endothelium-dependent
relaxants, presumably by scavenging or inactivating
superoxide and preventing its destructive effect on
EDRF and NO. Thus, it is noteworthy that the only
agents that inhibited endothelium-dependent vascular
responses were those agents that scavenged or de-
877
stroyed NO and also inhibited NO-elicited relaxant
responses. Therefore, it is unlikely that EDRF is an
oxygen radical species chemically related to superoxide, hydroxyl radical, or singlet oxygen. Indeed, the
present data suggest that EDRF is a radical species
chemically related to NO radical.
In conclusion, the present study provides indirect
pharmacologic and direct chemical evidence that
EDRF from both intrapulmonary artery and vein very
closely resembles NO radical. The evidence is that both
EDRF and NO 1) require guanylate cyclase-bound
heme for enzyme activation, 2) bind to a common site
on hemoglobin and myoglobin, 3) are inactivated by
superoxide anion and protected by superoxide dismutase, and 4) react with hemoglobin to form the same
heme adduct, which appears to be nitrosyl-hemoglobin. Procedures for the isolation and additional
chemical identification of NO, which is a highly
unstable substance, could be difficult to perform by
conventional techniques. In addition, the question of
the formation or source of such a species in vascular
tissue must be addressed. A recent report indicates that
certain oxides of nitrogen such as nitrate are synthesized independently of microfiora in the germfree rat.46
Although nitrate synthesis appears, therefore, to be a
mammalian process, it is unknown whether nitrite
formation can likewise occur. Assuming that nitrite
formation by mammalian cells can occur, the conversion of nitrite to NO should proceed quite readily in
muscle cells because of the presence of myoglobin.
Nitrite reacts with myoglobin or oxymyoglobin to
generate NO-myoglobin by either nonenzymatic4748 or
enzymatic49 mechanisms. Reducing agents such as
ascorbate, monothiols such as cysteine, and reduced
iron all accelerate or enhance the formation of NOmyoglobin.50 Thiols can accelerate the denitration of
organic nitrate esters such as nitroglycerin to yield
nitrite, 951 " and the nitrite would be available for
reaction with myoglobin. NO-myoglobin complexes
readily dissociate with the release of NO. Considering
our present state of knowledge on EDRF, the above
metabolic pathway would suppose that vascular endothelial cells generate an unstable, lipophilic organic
nitrite or nitrate precursor that readily traverses membrane barriers and reacts with myoglobin to generate
EDRF or NO. The presently available information on
EDRF does not rule out the possibility of its formation
within smooth muscle cells after the release of an
endogenous nitrite or nitrate precursor from endothelial
cells. Alternatively, the NO may be generated within
the endothelial cells by presently unknown mechanisms. Regardless of its site of formation, if EDRF is
indeed NO, the metabolic pathways for its formation
in vascular tissue will have to be elucidated.
References
1. Furchgott RF, Zawadzki JV: The obligatory role of endothelial
cells in the relaxation of arterial smooth muscle by acetylcholine. Nature 1980;288:373-376
2. Chand N, Altura BM: Acetylcholine and bradykinin relax
intrapulmonary arteries by acting on endothelial cells: Role in
lung vascular disease. Science 1981;213:1376-1378
Circulation Research
878
Downloaded from http://circres.ahajournals.org/ by guest on June 15, 2017
3. De Mey JG, Vanhoutte PM: Role of the intima in cholinergic
and purinergic relaxation of isolated canine femoral arteries. J
Physiol (Lond) 1981;316:347-355
4. Gruetter CA, Lemke SM: Bradykinin-induced endotheliumdependent relaxation of bovine intrapulmonary artery and vein.
Eur J Pharmacol 1986;122:363-367
5. Ignarro LJ, Harbison RG, Wood KS, Kadowitz PJ: Activation
of purified soluble guanylate cyclase by endothelium-derived
relaxing factor from intrapulmonary artery and vein: Stimulation by acetylcholine, bradykinin and arachidonic acid. J
Pharmacol Exp Ther 1986;237:893-9OO
6. Furchgott RF: The role of endothelium in the responses of
vascular smooth muscle to drugs. Annu Rev Pharmacol Toxicol
1984;24:175-197
7. Ignarro LJ, Kadowitz PJ: Pharmacological and physiological
role of cyclic GMP in vascular smooth muscle relaxation. Annu
Rev Pharmacol Toxicol 1985;25:171-191
8. Gruetter CA, Gruetter DY, Lyon JE, Kadowitz PJ, Ignarro LJ:
Relationship between cyclic GMP formation and relaxation of
coronary arterial smooth muscle by glyceryl trinitrate, nitroprusside, nitrite and nitric oxide: Effects of methylene blue and
methemoglobin. J Pharmacol Exp Ther 1981;219:181-186
9. Ignarro LJ, Lippton H, Edwards JC, Baricos WH, Hyman AL,
Kadowitz PJ, Gruetter CA: Mechanism of vascular smooth
muscle relaxation by organic nitrates, nitrites, nitroprusside
and nitric oxide: Evidence for the involvement of S-nitrosothiols as active intermediates. J Pharmacol Exp Ther 1981;
218:739-749
10. Ignarro LJ, Burke TM, Wood KS, Wolin MS, Kadowitz PJ:
Association between cyclic GMP accumulation and
acetylcholine-elicited relaxation of bovine intrapulmonary
artery. J Pharmacol Exp Ther 1984;228:682-690
11. Holzmann S: Endothelium-induced relaxation by acetylcholine
associated with larger rises in cyclic GMP in coronary arterial
strips. J Cyclic Nucleotide Protein Phosphor Res 1982;8:
409-419
12. Rapoport RM, Murad F: Agonist-induced endotheliumdependent relaxation in rat thoracic aorta may be mediated
through cGMP. Circ Res 1983;52:352-357
13. Diamond J, Chu EB: Possible role for cyclic GMP in
endothelium-dependent relaxation of rabbit aorta by acetylcholine. Comparison with nitroglycerin. Res Commun Chem
Pathol Pharmacol 1983;41:369-381
14. Martin W, Villani GM, Jothianandan D, Furchgott RF:
Selective blockade of endothelium-dependent and glyceryl
trinitrate-induced relaxation by hemoglobin and by methylene
blue in the rabbit aorta. J Pharmacol Exp Ther 1985;232:
708-716
15. Griffith TM, Edwards DH, Lewis MJ, Henderson AH: Evidence that cyclic guanosine monophosphate (cGMP) mediates
endothelium-dependent relaxation. Eur J Pharmacol 1985;
112:195-202
16. Gruetter CA, Barry BK, McNamara DB, Gruetter DY,
Kadowitz PJ, Ignarro LJ: Relaxation of bovine coronary artery
and activation of coronary arterial guanylate cyclase by nitric
oxide, nitroprusside and a carcinogenic nitrosoamine. J Cyclic
Nucleotide Protein Phosphor Res 1979:5:211-224
17. Gruetter CA, Kadowitz PJ, Ignarro LJ: Methylene blue inhibits
coronary arterial relaxation and guanylate cyclase activation by
nitroglycerin, sodium nitrite and amyl nitrite. Can J Physiol
Pharmacol 1981 ;59:150-156
18. Gruetter CA, Barry BK, McNamara DB, Kadowitz PJ, Ignarro
LJ: Coronary arterial relaxation and guanylate cyclase activation by cigarette smoke, N'-nitrosonornicotine and nitric
oxide. J Pharmacol Exp Ther 1980;214:9-15
19. Martin W, Villani GM, Jothianandan D, Furchgott RF:
Blockade of endothelium-dependent and glyceryl trinitrateinduced relaxation of rabbit aorta by certain ferrous hemoproteins. J Pharmacol Exp Ther 1985;233:679-685
20. Katsuki S, Arnold W, Mittal C, Murad F: Stimulation of
guanylate cyclase by sodium nitroprusside, nitroglycerin and
nitric oxide in various tissue preparations and comparison to the
effects of sodium azide and hydroxylamine. J Cyclic Nucleotide Protein Phosphor Res 1977;3:23-35
21. Braughler JM, Mittal CK, Murad F: Effects of thiols, sugars,
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
Vol 61, No 6, December 1987
and proteins on nitric oxide activation of guanylate cyclase. J
BiolChem 1979;254:12450-12454
Ignarro LJ, Degnan JN, Baricos WH, Kadowitz PJ, Wolin MS:
Activation of purified guanylate cyclase by nitric oxide requires
heme: Comparison of heme-deficient, heme-reconstituted and
heme-containing forms of soluble enzyme from bovine lung.
Biochim Biophys Acta 1982;718:49-59
Murad F, Mittal CK, Arnold WP, Katsuki S, Kimura H:
Guanylate cyclase activation by azide, nitro compounds, nitric
oxide, and hydroxyl radical and inhibition by hemoglobin and
myoglobin. Adv Cyclic Nucleotide Protein Phosphorylation
Res 1978;9:145-158
Craven PA, DeRubertis FR: Restoration of the responsiveness
of purified guanylate cyclase to nitrosoguanidine, nitric oxide,
and related activators by heme and heme proteins: Evidence for
the involvement of the paramagnetic nitrosyl-heme complex in
enzyme activation. J Biol Chem 1978;253:8433-8443
Ignarro LJ, Wood KS, Ballot B, Wolin MS: Guanylate cyclase
from bovine lung: Evidence that enzyme activation by
phenylhydrazine is mediated by iron-phenyl hemoprotein
complexes. J Biol Chem 1984;259:5923-5931
Ignarro LJ, Adams JB, Horwitz PM, Wood KS: Activation of
soluble guanylate cyclase by NO-hemoproteins involves NOheme exchange: Comparison of heme-containing and hemedeficient enzyme forms. J Biol Chem 1986;261:4997-5002
Ohlstein EH, Wood KS, Ignarro LJ: Purification and properties
of heme-deficient hepatic soluble guanylate cyclase: Effects of
heme and other factors on enzyme activation by NO, NO-heme,
and protoporphyrin IX. Arch Biochem Biophys 1982;218:
187-198
Antonini E, Brunori M: in Neuberger A, Tatum EL (eds):
Frontiers of Biology. Amsterdam, North-Holland, 1971, vol
21, pp 13-33
Assendelft OWV: Spectrophotometry of Hemoglobin Derivatives. Springfield, 111., Charles C. Thomas, 1970, pp 47-73
Ignarro LJ, Byrns RE, Wood KS: Endothelium-dependent
modulation of cyclic GMP levels and intrinsic smooth muscle
tone in isolated bovine intrapulmonary artery and vein. Circ Res
1987;60:82-92
Edwards JC, Ignarro LJ, Hyman AL, Kadowitz PJ: Relaxation
of intrapulmonary artery and vein by nitrogen oxide-containing
vasodilators and cyclic GMP. J Pharmacol Exp Ther 1984;
228:33-42
Ignarro LJ, Harbison RG, Wood KS, Wolin MS, McNamara
DB, Hyman AL, Kadowitz PJ: Differences in responsiveness
of intrapulmonary artery and vein to arachidonic acid: Mechanism of arterial relaxation involves cyclic GMP and cyclic
AMP. J Pharmacol Exp Ther 1985;233:56O-569
Bell FK, O'Neill JJ, Burgison RM: Determination of the
oil/water distribution coefficients of glyceryl trinitrate and two
similar nitrate esters. J Pharm Sci 1963;52:637-639
Schwartz SM: Selection and characterization of bovine aortic
endothelial cells. In Vitro 1978;14:966-980
Berliner JA: Regulation of endothelial cell replication and
adhesiveness studied in defined medium. In Vitro 1981;
17:985-992
Winer BJ: Statistical Principles in Experimental Design, ed 2.
New York, McGraw-Hill, 1962
Harbison RG, Wood KS, Hyman AL, Kadowitz PJ, Ignarro LJ:
Endothelium-dependent and independent regulation of cyclic
GMP formation and relaxation of intrapulmonary artery by
peptides (abstract). Fed Proc 1985;44:1235
Marklund S, Marklund G: Involvement of the superoxide anion
radical in the autoxidation of pyrogallol and a convenient assay
for superoxide dismutase. Eur J Biochem 1974;47:469-474
MoncadaS, Palmer RMJ, Gryglewski RJ: Mechanism of action
of some inhibitors of endothelium-derived relaxing factor. Proc
NatlAcadSci USA 1986;83:9164-9168
Ignarro LJ, Gruetter CA, Hyman AL, Kadowitz PJ: Molecular
mechanisms of vasodilatation, in Poste G, Crooke ST (eds):
Dopamine Receptor Agonists. New York, Plenum Publishing
Corporation, 1984, pp 259-288
Gibson QH, Roughton FJW: The kinetics and equilibria of the
reactions of nitric oxide with sheep haemoglobin. J Physiol
(Lond) 1957; 136:507-526
Ignarro et al
Vascular EDRF and Nitric Oxide Radical
42. Martin W, Furchgott RF, Villani GM, Jothianandan D:
Phosphodiesterase inhibitors induce endothelium-dependent
relaxation of rat and rabbit aorta by potentiating the effects of
spontaneously released endothelium-derived relaxing factor. J
Pharmacol Exp Ther 1986;237:539-547
43. Wolin MS, Wood KS, Ignarro LJ: Guanylate cyclase from
bovine lung: A kinetic analysis of the regulation of the purified
soluble enzyme by protoporphyrin IX, heme and nitrosylheme. J Biol Chem 1982;257:13312-13320
44. Rubanyi GM, Vanhoutte PM: Superoxide anion and hyperoxia
inactivate endothelium-derived relaxing factor. Am J Physiol
1986;250:H822-H827
45. Gerzer R, Bohme E, Hofmann F, Schultz G: Soluble guanylate
cyclase purified from bovine lung contains heme and copper.
FEBSLett 1981 ;132:71-74
46. Green LC, Tannenbaum SR, Goldman P: Nitrate synthesis in
the germfree and conventional rat. Science 1981;212:56—58
47. Reith JF, Szakaly M: Formation and stability of nitric oxide
myoglobin. I. Studies with model systems. J Food Sci
1967;32:188-193
879
48. Reith JF, Szakaly M: Formation and stability of nitric oxide
myoglobin. II. Studies on meat. J Food Sci 1967;32:194-196
49. Walters CL, Casselden RJ, Taylor AM: Nitrite metabolism by
skeletal muscle mitochondria in relation to haem pigments.
Biovhim Biophys Ada 1967;143:310-318
50. Fox JB, Ackerman SA: Formation of nitric oxide myoglobin:
Mechanisms of the reaction with various reductants. J Food Sci
1968;33:364-370
51. Needleman P, Johnson EM: Mechanism of tolerance development to organic nitrates. J Pharmacol Exp Ther 1973;
184:709-715
52. Needleman P, Jakschik B, Johnson EM: Sulfhydryl requirement for relaxation of vascular smooth muscle. J Pharmacol
Exp Ther 1973; 187:324-331
KEY WORDS • endothelium • relaxation • cyclic GMP • nitrite <
vasodilation • EDRF • nitric oxide
Downloaded from http://circres.ahajournals.org/ by guest on June 15, 2017
Endothelium-derived relaxing factor from pulmonary artery and vein possesses
pharmacologic and chemical properties identical to those of nitric oxide radical.
L J Ignarro, R E Byrns, G M Buga and K S Wood
Downloaded from http://circres.ahajournals.org/ by guest on June 15, 2017
Circ Res. 1987;61:866-879
doi: 10.1161/01.RES.61.6.866
Circulation Research is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright © 1987 American Heart Association, Inc. All rights reserved.
Print ISSN: 0009-7330. Online ISSN: 1524-4571
The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://circres.ahajournals.org/content/61/6/866
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in
Circulation Research can be obtained via RightsLink, a service of the Copyright Clearance Center, not the
Editorial Office. Once the online version of the published article for which permission is being requested is
located, click Request Permissions in the middle column of the Web page under Services. Further information
about this process is available in the Permissions and Rights Question and Answer document.
Reprints: Information about reprints can be found online at:
http://www.lww.com/reprints
Subscriptions: Information about subscribing to Circulation Research is online at:
http://circres.ahajournals.org//subscriptions/