Skip main navigation

Carbon Monoxide-Releasing Molecules

Characterization of Biochemical and Vascular Activities
Originally published Research. 2002;90:e17–e24


Carbon monoxide (CO) is generated in living organisms during the degradation of heme by the enzyme heme oxygenase, which exists in constitutive (HO-2 and HO-3) and inducible (HO-1) isoforms. Carbon monoxide gas is known to dilate blood vessels in a manner similar to nitric oxide and has been recently shown to possess antiinflammatory and antiapoptotic properties. We report that a series of transition metal carbonyls, termed here carbon monoxide-releasing molecules (CO-RMs), liberate CO to elicit direct biological activities. Specifically, spectrophotometric and NMR analysis revealed that dimanganese decacarbonyl and tricarbonyldichlororuthenium (II) dimer release CO in a concentration-dependent manner. Moreover, CO-RMs caused sustained vasodilation in precontracted rat aortic rings, attenuated coronary vasoconstriction in hearts ex vivo, and significantly reduced acute hypertension in vivo. These vascular effects were mimicked by induction of HO-1 after treatment of animals with hemin, which increases endogenously generated CO. Thus, we have identified a novel class of compounds that are useful as prototypes for studying the bioactivity of CO. In the long term, transition metal carbonyls could be utilized for the therapeutic delivery of CO to alleviate vascular- and immuno-related dysfunctions. The full text of this article is available at

Although it has been known for a long time that carbon monoxide (CO) is generated in the human body,1 only in recent years have scientists begun to explore the possible biological activities of this gaseous molecule. The main endogenous source of CO is heme oxygenase, which exists in constitutive (HO-2 and HO-3) and inducible (HO-1) isoforms; heme serves as substrate for HO-1 and HO-2 in the formation of CO, free ferrous iron, and biliverdin, the latter being rapidly converted to bilirubin by biliverdin reductase.2 There is general consensus, supported by extensive published reports, that HO-1 represents a pivotal inducible defensive system against stressful stimuli, including UVA radiation, carcinogens, ischemia-reperfusion damage, endotoxic shock, and several other conditions characterized by production of oxygen-derived free radicals.2–4 As part of its physiological and cytoprotective actions, heme oxygenase-derived CO appears to play a major role as neurotransmitter,5–7 regulator of sinusoidal tone,8 inhibitor of platelet aggregation,9 and suppressor of acute hypertensive responses.10,11 In addition, exogenously applied CO has been shown to protect against lung injury in vivo,12,13 prevent both production of proinflammatory cytokines14 and endothelial cell apoptosis,15 and suppress graft rejection in mouse-to-rat cardiac transplants16; all these effects are simulated by transfection of the HO-1 gene. Thus, consistent findings reveal a series of important cellular functions that support a versatile and previously unidentified role for CO. It is interesting that many of the novel properties pertaining to CO have strong analogies with the well-established biological activities elicited by nitric oxide (NO), another gaseous molecule produced intracellularly by a family of ubiquitous enzymes known as NO synthases. Research in the field of NO has been largely facilitated by the development of a variety of organic compounds that spontaneously release NO and can reproduce a physiological or pathophysiological function of NO; however, no attempts have been made so far to identify compounds that are capable of carrying and delivering CO into biological systems.

Our laboratory has been working for the past few years on the characterization of transition metal carbonyls as potential CO-releasing molecules (CO-RMs). These complexes are compounds that contain a heavy metal such as nickel, cobalt, or iron surrounded by carbonyl (CO) groups as coordinated ligands.17 From inorganic chemical studies conducted on these substances in vitro, it is known that certain ligands in a metal complex can promote, either sterically or electronically, dissociation of CO.18–20 Furthermore, photodissociation and consequent elimination of the CO group(s) following exposure to light has been reported in the case of specific metal carbonyls.21,22 The discovery that carbonyl complexes possess such interesting and promising features in vitro prompted us to examine the ability of some of these compounds to promote a physiological response in biological systems. We report here that dimanganese decacarbonyl and tricarbonyldichlororuthenium (II) dimer release CO under appropriate conditions and can elicit specific vascular activities that are reminiscent of those mediated by the HO-1/CO pathway.

Materials and Methods


Iron pentacarbonyl ([Fe(CO)5]), dimanganese decacarbonyl ([Mn2(CO)10]), tricarbonyldichlororuthenium (II) dimer ([Ru(CO)3Cl2]2), and ruthenium (III) chloride hydrate (RuCl3) were purchased from Sigma-Aldrich Company, Ltd. Tri-dimethyl sulfoxide-dichlororuthenium (II) dimer [Ru(DMSO)3Cl2]2 was synthesized as previously described.23 Hemin (ferriprotoporphyrin IX chloride) and tin protoporphyrin IX (SnPPIX) were from Porphyrin Products, Inc. The guanylate cyclase inhibitor, [1H-1,2,4Oxadiazole[4,3-a]quinoxalin-1-one] (ODQ), was obtained from Alexis Corporation, and polyclonal rabbit anti-HO-1 antibodies were purchased from Stressgen. Horse heart myoglobin, NG-nitro-l-arginine methyl ester (L-NAME), and all other reagents were from Sigma, unless otherwise specified.

Detection of CO Release

The release of CO from metal carbonyl complexes was assessed spectrophotometrically by measuring the conversion of deoxymyoglobin (deoxy-Mb) to carbonmonoxy myoglobin (MbCO). The amount of MbCO formed was quantified by measuring the absorbance at 540 nm (extinction coefficient=15.4 mmol/L−1 · cm−1). Myoglobin solutions (66 μmol/L final concentration) were prepared fresh by dissolving the protein in 0.04 mol/L phosphate buffer (pH 6.8). Sodium dithionite (0.1%) was added to convert myoglobin to deoxy-Mb prior to each reading. Because CO loss from [Fe(CO)5] and [Mn2(CO)10] occurs only by photodissociation,24,25 the release of CO was induced by exposing these metal carbonyl complexes to a cold light source and allowing the gas to diffuse through a membrane before reacting with myoglobin (see Figure 1A). In contrast, CO released from [Ru(CO)3Cl2]2 was quantified by adding aliquots of stock solutions (10 μL) of the carbonyl complex in DMSO directly to the myoglobin solution. All the spectra were measured using a Helios α spectrophotometer.

Figure 1. Detection of CO released from carbonyl metal complexes by photodissociation. A, Schematic representation showing the device used to measure CO released from CO-RMs. Five hundred microliters of iron pentacarbonyl ([Fe(CO)5], 99.9%) or 1 mL of dimanganese decacarbonyl ([Mn2(CO)10], 0.013 mol/L in DMSO) (see also chemical structure) were placed in a plastic tube. A cell culture insert (Costar) was sealed on top in order to create 2 separate chambers with a 0.6 cm air space between the solution and the insert membrane (0.4 μm). One and a half milliliters of deoxy-Mb solution (66 μmol/L) was placed in the insert that was finally covered with parafilm. The carbonyl solution was then exposed to cold light to stimulate CO release, allowing the gas to diffuse through the membrane into the myoglobin solution. Aliquots of the myoglobin solution were taken at different times and the conversion of deoxy-Mb to MbCO measured spectrophotometrically. B, Typical spectrum of MbCO (A) and deoxy-Mb (B). Deoxy-Mb was converted to ≈100% MbCO after bubbling the myoglobin solution (66 μmol/L) for 45 seconds with 1% CO gas. C and D, Time course of the conversion of deoxy-Mb to MbCO during light-mediated CO dissociation from [Fe(CO)5] and [Mn2(CO)10], respectively. The concentration of MbCO was also calculated over time in the case of [Mn2(CO)10] (E) by measuring the change in absorbance at 540 nm (ε540=15.4 mmol/L−1 · cm−1).

Nuclear Magnetic Resonance Analysis

The release of CO from metal carbonyl complexes was also assessed by nuclear magnetic resonance (NMR) analysis. All the spectra were recorded at room temperature and measured using a Bruker AMX400 NMR spectrometer. [RuCl2(CO)3]2 (0.020 g) was dissolved in d6-DMSO (0.5 mL) and the 13C(1H) NMR spectrum run immediately. The 13CO signals were detected after 23 minutes of signal accumulation. The sample was then left for the 13C NMR spectrum to accumulate overnight. In order to compare chemical shifts with the literature23 and make the assignment of the signals clearer, [RuCl2(CO)3]2 was first dissolved in DMSO (50 μL), and the solution further diluted in 0.6 mL CDCl3 prior to NMR analysis. The signals were referenced to the central signals of d6-DMSO at δ 39.5 and CDCl3 (deuterated chloroform) at δ 77.0, respectively.

Cell Viability Assay

Bovine vascular smooth muscle cells were obtained from the Coriell Cell Repository (Camden, NJ) and grown in Dulbecco’s Minimal Essential Medium (MEM) as previously described.26 Confluent cells were treated with different concentrations of [Ru(CO)3Cl2]2 for various times and cell viability was assessed using a colorimetric assay kit from Promega as previously described.26

Immunohistochemistry and Northern Blot Analysis

For immunohistochemistry analysis, sections of heart muscles (5 μm thickness) were treated with 0.3% H2O2 in methanol to block endogenous peroxidase activity. Immunohistochemical staining was performed using rabbit polyclonal antibody against HO-1 (1:1000 dilution) as previously described.11 Total RNA was then extracted using a modification of the method described by Chomczynski and Sacchi.27 RNA was run on a 1.3% denaturing agarose gel containing 2.2 mol/L formaldehyde and transferred onto a nylon membrane overnight. The membrane was hybridized using [α-32P]dCTP-labeled cDNA probes to rat HO-1 and GAPDH genes, and bands were analyzed using a densitometer as previously described.11,28

Isolated Aortic Ring Preparation

Transverse ring sections of thoracic aorta were isolated from male Lewis rats (Harlan UK Ltd) and suspended under a 2-g tension in an organ bath containing oxygenated Krebs-Henseleit buffer at 37°C, as previously described by our group.11 The relaxation response to cumulative doses of [Ru(CO)3Cl2]2 was assessed in aortic rings precontracted with phenylephrine (3 μmol/L) in the presence of various agonists. Control rings were similarly treated by adding equal doses of DMSO (vehicle) to the organ bath. All animal studies were performed according to Home Office Regulations.

Isolated Rat Heart Preparation

Langendorff preparations were performed using male Lewis rat hearts (300 to 350 g) as previously described by our group.29 Coronary perfusion pressure (CPP), a parameter indicative of coronary vessel contractility, was continuously measured by a pressure transducer connected to the aortic cannula and data analyzed with an AcqKnowledge software (BIOPAC System Inc). Hearts removed either from control rats (vehicle-treated) or from animals that were pretreated with hemin (75 μmol/kg, IP) the day before were initially equilibrated for 20 minutes on the Langendorff apparatus and then perfused with L-NAME (25 μmol/L final concentration) to elicit vasoconstriction. The extent of CPP increase by L-NAME was also monitored over time in hemin-treated animals that received a heme oxygenase inhibitor (SnPPIX, 40 μmol/kg IP) 1 hour prior to heart excision and in control hearts that were perfused with buffer supplemented with [Mn2(CO)10] (13 μmol/L final concentration). Because [Mn2(CO)10] releases CO only by photodissociation, Krebs buffer containing [Mn2(CO)10] was exposed to a cold light source immediately before entering the aortic cannula.

Animal Studies

Lewis rats (280 to 350 g) were anesthetized by intramuscular injection of 1 mL/kg Hypnorm. Specially designed femoral artery and venous catheters were then surgically implanted as previously described10 and blood pressure monitored continuously using a polygraph recorder. A dose-dependent response on blood pressure was initially assessed by intravenous injection of [Ru(CO)3Cl2]2 (5, 10 and 20 μmol/kg) or [Ru(DMSO)3Cl2]2 (negative control). Control rats (vehicle-treated) and animals that were pretreated with hemin (75 μmol/kg, IP) 24 hours prior to blood pressure monitoring were then administered with an intravenous injection of 30 μmol/kg L-NAME to elicit an increase in mean arterial pressure. The extent of blood pressure increase by L-NAME was also monitored over time in hemin-treated animals that received SnPPIX (40 μmol/kg, IP) and in control rats previously injected with [Ru(CO)3Cl2]2 (20 μmol/kg, IV). In these 2 groups, SnPPIX or [Ru(CO)3Cl2]2 were administered to animals 1 hour or 20 minutes prior to L-NAME injection, respectively.

Statistical Analysis

All data are expressed as mean±SEM. Differences between the groups analyzed were assessed by the Student’s 2-tailed t test, and an analysis of variance (ANOVA) was performed where more than 2 treatments were compared. Results were considered statistically significant at P<0.05.


Characterization of CO-Releasing Molecules (CO-RMs): Detection of CO Liberated From Transition Metal Carbonyl Complexes and NMR Studies

Direct addition of iron pentacarbonyl ([Fe(CO)5]) or dimanganese decacarbonyl ([Mn2(CO)10]) to myoglobin solutions did not result in any appreciable formation of carbonmonoxy myoglobin (MbCO) over time (data not shown). This is consistent with the notion that these two transition metal carbonyl complexes do not release CO unless stimulated by light.24,25 Therefore, a source of cold light was utilized to promote CO release from carbonyls by photodissociation. The schematic representation of the device used for measuring CO from [Fe(CO)5] and [Mn2(CO)10] is shown in Figure 1A. On illumination, [Fe(CO)5] and [Mn2(CO)10] produced a change in the absorbance spectrum of myoglobin; in both cases the distinctive identified spectra were the ones typical of MbCO, as confirmed by bubbling CO gas to a solution of deoxy-Mb (Figure 1B). A gradual increase in MbCO formation was observed over time with [Fe(CO)5] and [Mn2(CO)10], respectively (Figures 1C and 1D). Interestingly, and in contrast to the above studied compounds, a different metal carbonyl complex, tricarbonyldichlororuthenium (II) dimer ([Ru(CO)3Cl2]2 in DMSO), rapidly elicited MbCO formation when added directly to a deoxy-Mb solution (Figure 2B). The amount of MbCO formed was dependent on the concentration of [Ru(CO)3Cl2]2 used; a fitting analysis of the saturation curve of MbCO revealed that for each mole of [Ru(CO)3Cl2]2, approximately 0.7 moles of CO are liberated (Figure 2C). The 13C NMR spectrum showed that [Ru(CO)3Cl2]2 freshly dissolved in DMSO does not exist as a dimer; in fact, 2 distinct peaks corresponding to tri-carbonyl (1) and di-carbonyl (2) monomers could be identified (Figure 3). The NMR analysis reveals that, during the solubilization process, DMSO acts as a coordinated ligand to ruthenium, thereby promoting the formation of the monomers. Interestingly, the detection of di-carbonyl monomers demonstrates that CO is liberated; the 13C NMR spectrum also suggests that the ratio between 1 and 2 is 40:60.

Figure 2. Spontaneous release of CO from [Ru(CO)3Cl2]2. A, Chemical structure of tricarbonyldichlororuthenium (II) dimer ([Ru(CO)3Cl2]2). B, Absorption spectrum of deoxy-Mb and MbCO before and after interaction of myoglobin with [Ru(CO)3Cl2]2. The metal carbonyl complex was solubilized in DMSO (9.7 mmol/L stock solution), aliquots of 2 to 32 μL were added directly to 1 mL of deoxy-Mb solutions (66 μmol/L), and absorption spectrum determined immediately after mixing the samples by inversion. C, Determination of the amount of MbCO formed after interaction of deoxy-Mb with various concentrations of [Ru(CO)3Cl2]2 (see Figure 1E for details).

Figure 3. Identification of the reaction products between [RuCl2(CO)3]2 and DMSO by NMR analysis. a, 100.62 MHz 13C(1H) NMR spectrum taken during the first 23 minutes of the reaction between freshly prepared [RuCl2(CO)3]2 and d6-DMSO. The solution very slowly produced fine bubbles of a gas, presumably CO, implied by the formation of 2. When the experiment was repeated by dissolving initially the metal complex in DMSO and then diluting with CDCl3, the assignment of the signals coincided with the published 13C chemical shifts of fac-[RuCl2(CO)3(DMSO)] (1, δ 183.0, 186.8), cis, cis, trans-[RuCl2(CO)2(DMSO)2] (2, δ 185.0), and cis, cis, cis-[RuCl2(CO)2(DMSO)2] (3, δ 186.0, 191.9) (reference by Alessio et al23). b, 13C NMR spectrum taken after [RuCl2(CO)3]2 in d6-DMSO was warmed at 50°C for 5 minutes and left to accumulate overnight. In addition to the peaks that could be assigned to 1, 2, and 3, there are carbonyl signals at δ 187.9 and 190.5 due to unidentified species.

Effect of [Ru(CO)3Cl2]2 on Cell Viability

As there are no previous studies on the use of metal carbonyl complexes in biological systems, it was necessary to evaluate the potential cytotoxic effect of these compounds. Because we noticed that exposure of [Fe(CO)5] to light gradually resulted in deposition of a green-brown precipitate, we decided to abandon the studies on this metal carbonyl. Treatment of vascular smooth muscle cells for 3 hours with [Ru(CO)3Cl2]2 (0 to 420 μmol/L final concentration) did not promote any detectable cytotoxicity (Figure 4B). Similarly, cell viability was well-preserved after exposure to this metal carbonyl for 3 hours followed by an additional 21-hour incubation in complete medium. A pronounced cytotoxic effect (>50% loss in cell viability) was only apparent after prolonged exposure (24 hours) to very high concentrations (>400 μmol/L) of [Ru(CO)3Cl2]2. Treatment of cells with the same amounts of vehicle (DMSO) or equivalent molar concentrations of ruthenium (RuCl3) did not cause any appreciable decrease in cell viability over time (Figures 4A and 4C, respectively). In the case of [Mn2(CO)10] (0 to 100 μmol/L), no major cytotoxicity on smooth muscle cells was detected after exposure for 24 hours in complete medium (data not shown).

Figure 4. Viability of bovine vascular smooth muscle cells treated with [Ru(CO)3Cl2]2. Confluent cells were incubated with complete media in the presence of various concentrations of the vehicle DMSO (A), [Ru(CO)3Cl2]2 (B), or RuCl3 (C). Cell viability was assessed using a colorimetric assay kit after 3 (filled bars) or 24 hours (crosshatched bars) incubation, or after 3 hours exposure to the agents followed by 21-hour incubation in complete media (open bars). Results are expressed as the mean±SEM (n=6); *P<0.05.

Vasodilatory Effect of CO Released From [Ru(CO)3Cl2]2

To investigate whether CO released from CO-RMs evokes specific biological activities, we first assessed the effect of [Ru(CO)3Cl2]2 on vessel contractility using the isolated aortic ring model. Consecutive additions of [Ru(CO)3Cl2]2 to aortic rings precontracted with phenylephrine elicited a rapid and significant vasodilatation (Figure 5); the extent of relaxation was already pronounced after the first addition of the compound (45% more than control). Interestingly, after extensive washing, the phenylephrine-induced contraction was completely restored in control but not in [Ru(CO)3Cl2]2-treated vessels, indicating that CO-RMs produces long-lasting effects (Figure 5). The vasodilatory response mediated by CO-RMs was almost totally abolished when reduced Mb, which avidly binds CO, was added to the buffer. Because CO modulates signal transduction mechanisms also via increased production of cGMP, we investigated the effect of a selective inhibitor of guanylate cyclase (ODQ, 10 μmol/L) on vessel contractility. As expected, ODQ considerably reduced the vasodilatation observed after the first 2 additions of [Ru(CO)3Cl2]2; however, it is of interest that the third addition of CO-RM still elicited a substantial vasodilatory action despite the presence of ODQ.

Figure 5. Vasodilatory properties of [Ru(CO)3Cl2]2 and mechanism(s) of relaxation by CO-RMs. Isolated rat aortas precontracted with phenylephrine (Phe, 3 μmol/L) were treated with 3 cumulative additions of [Ru(CO)3Cl2]2 (CO-RM, 222 μmol/L) or DMSO (vehicle) and changes in tension continuously recorded. After washing, contractility was assessed again by exposing vessels to phenylephrine. The percentage of relaxation after each addition of CO-RM is reported for vessels preincubated for 15 minutes with myoglobin (Mb, 200 μmol/L) or the guanylate cyclase inhibitor, [1H-1,2,4Oxadiazole[4,3-a]quinoxalin-1-one] (ODQ, 10 μmol/L). Results are expressed as the mean±SEM (n=6); *P<0.05 vs vehicle; †P<0.05 vs CO-RM.

Attenuation of Vasoconstriction by CO-RMs In Vivo: Comparison With Endogenously Generated CO

Additional experiments were conducted to determine the biological activity of CO-RMs on vascular function in vivo and compare it with HO-1-derived CO. Changes in coronary perfusion pressure (CPP) of isolated rat hearts were monitored to examine the effect of [Mn2(CO)10] on coronary vessel contractility. Infusion with the NO synthase inhibitor, NG-nitro-l-arginine methyl ester (L-NAME), caused a time-dependent increase in CPP, which reached a maximum (3-fold) after 30 minutes of perfusion (Figure 6C). Notably, perfusion of hearts with light-stimulated [Mn2(CO)10] (13 μmol/L) significantly delayed vasoconstriction and maintained CPP at much lower levels at the end of perfusion. When the buffer containing [Mn2(CO)10] was not exposed to light, thus omitting the CO-induced release process, the extent of constriction mediated by L-NAME was unaffected (data not shown); in addition, perfusion with manganese chloride (negative control) had no effect on myocardial CPP (data not shown). The effect observed with [Mn2(CO)10] could be similarly reproduced by induction of HO-1 in heart tissue. We previously reported that treatment of rats with hemin results in increased production of cardiac bilirubin, which is equimolar to the endogenously generated CO.29Figure 6A and 6B shows that HO-1 protein and mRNA are highly expressed in hearts 24 hours after hemin treatment; interestingly, the immunostaining for HO-1 protein was primarily confined to the vessels of cardiac muscle (Figure 6A, right panel). As anticipated above, the rise in CPP mediated by L-NAME in hemin-treated hearts was markedly attenuated (P<0.05) to an extent similar to that produced by [Mn2(CO)10] (Figure 6C); predictably, the effect of hemin was completely reversed by tin protoporphyrin IX (SnPPIX), a heme oxygenase inhibitor. In vivo experiments were also conducted to evaluate the usefulness of CO-RMs in modulating mean arterial pressure (MAP) and attenuating the acute hypertensive response. As shown in the Table, intravenous administration of increasing concentrations of [Ru(CO)3Cl2]2 (5, 10, and 20 μmol/kg) produced a transient decrease in blood pressure, which was significant at the highest dose and returned to basal levels within 30 minutes of injection. Interestingly, treatment of animals with [Ru(DMSO)3Cl2]2, which does not contain carbonyl groups (negative control), was totally ineffective, indicating that CO is effectively released from CO-RM in vivo. Oxyhemoglobin saturation remained unchanged throughout the various treatments (data not shown). When rats were infused with L-NAME (30 μmol/kg IV) a rapid and significant increase in blood pressure was observed (P<0.05); this effect was markedly suppressed by pretreatment of animals with a single infusion of [Ru(CO)3Cl2]2 (20 μmol/kg) prior to L-NAME administration (Figure 6D). Moreover, and in analogy with the data on coronary vasoconstriction in isolated hearts, treatment of animals with hemin resulted in a significant suppression of the L-NAME-mediated hypertensive responses, which once again was totally reversed by blockade of the heme oxygenase pathway with SnPPIX (Figure 6D).

Figure 6. Vasoactive effects of CO-RMs in vivo are simulated by HO-1 induction. A, Immunohistochemistry for HO-1 in cardiac tissue from control (left panel) and hemin-treated (right panel) animals. Hearts were removed from Lewis rats 24 hours after treatment with vehicle (control) or hemin (75 μmol/kg, IP) and immunostaining for HO-1 was assessed by the development of a brown color. B, Time course of HO-1 mRNA expression in cardiac tissue from hemin-treated animals. Rats were treated with hemin and hearts removed at different time points to assess HO-1 mRNA levels (+ve, positive control). C, Mitigation of coronary vasoconstriction by CO-RMs and HO-1 induction. Changes in coronary perfusion pressure (CPP) of isolated perfused hearts were monitored after infusion of L-NAME (25 μmol/L). The increase in CPP over time was measured in hearts removed 24 hours after treatment of rats with vehicle (control, ○), hemin (75 μmol/kg, ○), tin protoporphyrin IX (SnPPIX, 40 μmol/kg, ⧫), or hemin plus SnPPIX (⋄). Changes in CPP were also assessed in control hearts perfused with [Mn2(CO)10] (13 μmol/L, □) prior to L-NAME infusion. Results are mean±SEM (n=6); *P<0.05 vs vehicle-treated group (control). D, Mitigation of acute hypertension by CO-RMs and HO-1 induction. An increase in mean arterial pressure (MAP) in anesthesized animals was elicited by injection of a single dose of L-NAME (30 μmol/Kg, IV). Changes in MAP were monitored in animals 24 hours after treatment with vehicle (control, ○), hemin (75 μmol/kg, i.p, ♦), or hemin plus SnPPIX (40 μmol/kg, ⋄). The extent of blood pressure increase mediated by L-NAME was also measured over time in control rats previously injected with [Ru(CO)3Cl2]2 (20 μmol/kg, IV, □). Results are the mean±SEM (n=5); *P<0.05 vs vehicle-treated group (control).

Table 1. Effect of [RuCl2(CO)3]2 on Hemodynamic Parameters

Mean Arterial Pressure, mm HgHeart Rate, beats/min
0 min2 min20 min30 min60 min0 min2 min20 min30 min60 min
Freshly prepared solutions of [RuCl2(CO)3]2 or RuCl2(DMSO)3 (negative control) were injected intravenously at the doses indicated. Oxyhemoglobin saturation remained unchanged throughout the various treatments (data not shown). Data represent the mean±SEM (n=4).
*P<0.05 vs 0 min;
†Recording ended after no change.
Vehicle (DMSO)78±680±579±676±474±6277±16274±21280±30275±16271±21
    5 μmol/kg75±472±476±676±574±3307±13295±17266±21306±21277±14
    10 μmol/kg82±377±380±475±578±2300±6298±8294±10291±6290±16
    20 μmol/kg80±471±2*77±279±280±2272±10268±10264±14283±10283±11
    20 μmol/kg81±481±380±5289±20285±2299±15


Because of the strong experimental evidence implicating the heme oxygenase pathway and its products, bilirubin and CO, in the modulation of physiological and pathological processes, we were attracted by the idea of developing substances that could mimic the pharmacological action of heme oxygenase-derived CO. We initially explored the literature to ascertain whether chemical compounds capable of carrying carbonyl (CO) groups had already been described. Interestingly, we found that transition metal carbonyl complexes, which are used as catalysts in organic synthesis, possessed such a property and thus appeared excellent candidates to be tested as CO-releasing molecules (CO-RMs) in biological systems. Our studies focused on iron pentacarbonyl ([Fe(CO)5]), dimanganese decacarbonyl ([Mn2(CO)10]), and tricarbonyldichlororuthenium (II) dimer ([Ru(CO)3Cl2]2). All 3 compounds had the ability to convert deoxy-Mb to MbCO, implying that CO was liberated from the metal complexes. Unlike [Fe(CO)5] and [Mn2(CO)10], which released CO on exposure to light (see Figure 1A for details), [Ru(CO)3Cl2]2 in DMSO spontaneously promoted MbCO formation after addition to the myoglobin solution. That CO is effectively dissociating from [Ru(CO)3Cl2]2 was confirmed by NMR analysis; indeed, the NMR spectrum revealed the presence of 2 major peaks, which were identified as fac-[RuCl2(CO)3(DMSO)] and cis, cis, trans-[RuCl2(CO)2(DMSO)2], respectively, the tri-carbonyl and di-carbonyl monomers of [Ru(CO)3Cl2]2. The existence of di-carbonyl species clearly indicates that CO can be liberated during the formation of monomers; at present, we cannot exclude the occurrence of a possible equilibrium between the metal and CO in solution, but future studies using ligands different from DMSO will give us more information about the mechanism by which CO is released. The encouraging findings on the biochemical properties of these carbonyl carriers enabled us to design specific experiments and address important questions on the possible biological activity of CO-RMs in the cardiovascular system. Because [Mn2(CO)10] and [Ru(CO)3Cl2]2 did not exhibit apparent cytotoxicity in a concentration range of 40 to 210 μmol/L, we further investigated whether these 2 metal complexes could modulate vessel contractility ex vivo and blood pressure in the whole animal and analyzed a possible contribution of the guanylate cyclase pathway to this effect. This was a plausible approach because we have previously demonstrated that increased endogenous CO as a result of HO-1 induction in rat aortas markedly attenuates vasoconstriction in vitro and in vivo by increasing vascular cGMP levels.10,11

We report here that [Ru(CO)3Cl2]2 elicits profound vasodilatation when added to precontracted aortic rings. The relaxation process was abolished in the presence of myoglobin and was significantly attenuated by a guanylate cyclase inhibitor (ODQ), confirming that the pharmacological effect mediated by CO liberated from CO-RMs is, at least in part, cGMP-dependent. Despite the presence of ODQ, substantial vasodilation was still elicited after the third addition of [Ru(CO)3Cl2]2 to aortas, suggesting that CO released from CO-RMs may have accumulated in sufficiently high levels to displace ODQ from guanylate cyclase. Alternatively, elevated CO amounts could circumvent guanylate cyclase and exert relaxation via other independent cellular pathways.30 Notably, the effect of CO appears to be long-lasting as aortic rings previously treated with [Ru(CO)3Cl2]2 failed to regain contraction on a second challenge with phenylephrine.

The experiments conducted using the isolated heart model confirmed the vasoactive properties of another CO-RM, [Mn2(CO)10]. This metal carbonyl markedly attenuated L-NAME-mediated increase in coronary perfusion pressure; notably, this effect could only be achieved when [Mn2(CO)10] was stimulated by light to release CO. Hearts expressing high HO-1 in the vasculature following treatment of animals with hemin also displayed reduced contractility when challenged with L-NAME, and inhibition of heme oxygenase activity abolished the effect; this confirms the important role of endogenously produced CO in vascular control.10,11 Thus, augmented HO-1-derived CO can profoundly modulate cardiac vessel functions, and this effect can be mimicked by exogenously applied CO-RMs. Finally, our findings on the antihypertensive action of [Ru(CO)3Cl2]2 in animals support the possible utilization of CO-RMs in vivo. Consistent with a previous report from our group,10 we found that induction of the HO-1 system by hemin pretreatment considerably suppressed the increase in mean arterial pressure elicited by intravenous administration of L-NAME; as observed for the isolated aortic ring and heart preparations, SnPPIX restored the vasoconstrictor responses to L-NAME. Interestingly, infusion of [Ru(CO)3Cl2]2 strongly attenuated L-NAME-induced hypertension, indicating that CO-RMs do liberate CO once administered in vivo and can reproduce the effects of HO-1-derived CO.

The present study identifies a novel group of substances that are capable of carrying and delivering CO. In a similar fashion to endogenous HO-1-derived CO, CO-RMs also exert biological activities by eliciting vascular relaxation and mitigating both coronary vasoconstriction and acute hypertension. Recent evidence has highlighted the beneficial role of inhaled CO in rescuing HO-1-deficient mice from lethal ischemic lung injury.13 Moreover, potent antiinflammatory and antiapoptotic effects of exogenous CO gas have been described in models of endotoxin-induced inflammation and acute vascular rejection.14,16 The potential of using CO as therapeutic agent in a variety of disease states is, therefore, strongly emerging. In this context, the newly discovered CO-RMs could facilitate the progression toward pharmaceutical applications as they could be engineered to liberate CO with distinct kinetics and highly selective delivery to target tissues. The accessibility to and utilization of CO-RMs will greatly enhance our understanding on the chemistry of CO, its interaction with intracellular components, and its biological importance in physiology and disease.31

Presented in part at the 8th International Symposium on Mechanisms of Vasodilatation, Boston, Mass, May 31–June 3, 2001.

Original received September 25, 2001; resubmission received November 30, 2001; accepted December 20, 2001.

This study was supported by a Medical Research Council Hopping Award (to B.M and R.M.). The work of R.M. was supported by grants from the British Heart Foundation (PG/2001037) and the National Heart Research Fund (Leeds, UK).


Correspondence to R. Motterlini, Vascular Biology Unit, Dept of Surgical Research, Northwick Park Institute for Medical Research, Harrow, Middlesex HA1 3UJ, UK. E-mail


  • 1 Sjostrand T. Endogenous formation of carbon monoxide in man under normal and pathological conditions. Scan J Clin Lab Invest. 1949; 1: 201–214.CrossrefGoogle Scholar
  • 2 Maines MD. The heme oxygenase system: a regulator of second messenger gases. Annu Rev Pharmacol Toxicol. 1997; 37: 517–554.CrossrefMedlineGoogle Scholar
  • 3 Abraham NG, Drummond GS, Lutton JD, Kappas A. The biological significance and physiological role of heme oxygenase. Cell Physiol Biochem. 1996; 6: 129–168.CrossrefGoogle Scholar
  • 4 Foresti R, Motterlini R. The heme oxygenase pathway and its interaction with nitric oxide in the control of cellular homeostasis. Free Rad Res. 1999; 31: 459–475.CrossrefMedlineGoogle Scholar
  • 5 Verma A, Hirsch DJ, Glatt CE, Ronnett GV, Snyder SH. Carbon monoxide: a putative neural messenger. Science. 1993; 259: 381–384.CrossrefMedlineGoogle Scholar
  • 6 Maines MD. Carbon monoxide: an emerging regulator of cGMP in the brain. Mol Cell Neurosci. 1993; 4: 389–397.CrossrefMedlineGoogle Scholar
  • 7 Ingi T, Cheng J, Ronnett GV. Carbon-monoxide: an endogenous modulator of the nitric oxide-cyclic GMP signaling system. Neuron. 1996; 16: 835–842.CrossrefMedlineGoogle Scholar
  • 8 Suematsu M, Goda N, Sano T, Kashiwagi S, Egawa T, Shinoda Y, Ishimura Y. Carbon monoxide: an endogenous modulator of sinusoidal tone in the perfused rat liver. J Clin Invest. 1995; 96: 2431–2437.CrossrefMedlineGoogle Scholar
  • 9 Wagner CT, Durante W, Christodoulides N, Hellums JD, Schafer AI. Hemodynamic forces induce the expression of heme oxygenase in cultured vascular smooth muscle cells. J Clin Invest. 1997; 100: 589–596.CrossrefMedlineGoogle Scholar
  • 10 Motterlini R, Gonzales A, Foresti R, Clark JE, Green CJ, Winslow RM. Heme oxygenase-1-derived carbon monoxide contributes to the suppression of acute hypertensive responses in vivo. Circ Res. 1998; 83: 568–577.CrossrefMedlineGoogle Scholar
  • 11 Sammut IA, Foresti R, Clark JE, Exon DJ, Vesely MJJ, Sarathchandra P, Green CJ, Motterlini R. Carbon monoxide is a major contributor to the regulation of vascular tone in aortas expressing high levels of haeme oxygenase-1. Br J Pharmacol. 1998; 125: 1437–1444.CrossrefMedlineGoogle Scholar
  • 12 Otterbein LE, Mantell LL, Choi AMK. Carbon monoxide provides protection against hyperoxic lung injury. Am J Physiol. 1999; 276: L688–L694.MedlineGoogle Scholar
  • 13 Fujita T, Toda K, Karimova A, Yan SF, Naka Y, Yet SF, Pinsky DJ. Paradoxical rescue from ischemic lung injury by inhaled carbon monoxide driven by derepression of fibrinolysis. Nat Med. 2001; 7: 598–604.CrossrefMedlineGoogle Scholar
  • 14 Otterbein LE, Bach FH, Alam J, Soares M, Tao Lu H, Wysk M, Davis RJ, Flavell RA, Choi AM. Carbon monoxide has anti-inflammatory effects involving the mitogen-activated protein kinase pathway. Nat Med. 2000; 6: 422–428.CrossrefMedlineGoogle Scholar
  • 15 Brouard S, Otterbein LE, Anrather J, Tobiasch E, Bach FH, Choi AM, Soares MP. Carbon monoxide generated by heme oxygenase 1 suppresses endothelial cell apoptosis. J Exp Med. 2000; 192: 1015–1026.CrossrefMedlineGoogle Scholar
  • 16 Sato K, Balla J, Otterbein L, Smith RN, Brouard S, Lin Y, Csizmadia E, Sevigny J, Robson SC, Vercellotti G, Choi AM, Bach FH, Soares MP. Carbon monoxide generated by heme oxygenase-1 suppresses the rejection of mouse-to-rat cardiac transplants. J Immunol. 2001; 166: 4185–4194.CrossrefMedlineGoogle Scholar
  • 17 Herrmann WA. 100 Years of metal carbonyls: a serendipitous chemical discovery of major scientific and industrial impact. J Organomet Chem. 1990; 383: 21–44.CrossrefGoogle Scholar
  • 18 Darensbourg DJ, Klausmeyer KK, Reibenspies JH. CO-labilizing ability of the fluoride ligand in tungsten(0) carbonyl fluorides: x-ray structure of [Et4N]3[W2(CO)6F3]. Inorg Chem. 1995; 34: 4933–4934.CrossrefGoogle Scholar
  • 19 Darensbourg DJ, Klausmeyer KK, Reibenspies JH. Coordinatively unsaturated derivatives of group 6 metal carbonyls containing the π-donating ligand 3,5-di-tert-butylcatecholate. Inorg Chem. 1996; 35: 1529–1534.CrossrefMedlineGoogle Scholar
  • 20 Pearson J, Cooke J, Takats J, Jordan RB. Alkyne ligand enhancement of the substitution lability of mononuclear osmium, ruthenium, and iron carbonyls. J Am Chem Soc. 1998; 120: 1434–1440.CrossrefGoogle Scholar
  • 21 Wrighton MS, Ginley DS. Photochemistry of metal-metal bonded complexes: II, the photochemistry of rhenium and manganese carbonyl complexes containing a metal-metal bond. J Am Chem Soc. 1975; 97: 2065–2072.CrossrefGoogle Scholar
  • 22 Hughey JL, Bock CR, Meyer TJ. Flash photolysis evidence for metal-metal bond cleavage and loss of CO in the photochemistry of [n5-C5H5)Mo(CO)3]2. J Am Chem Soc. 1975; 97: 4440–4441.CrossrefGoogle Scholar
  • 23 Alessio E, Milani B, Bolle M, Mestroni G, Falechini P, Todone F, Geremia S, Calligaris M. Carbonyl derivatives of chloride-dimethyl sulfoxide-ruthenium(II) complexes: synthesis, structural characterization, and reactivity of Ru(CO)x(DMSO)4−xCl2 complexes (x=1−3). Inorg Chem. 1995; 34: 4722–4734.CrossrefGoogle Scholar
  • 24 Engelking PC, Lineberger WC. Laser photoelectron spectrometry of the negative ions of iron and iron carbonyls: electron affinity determination for the series Fe(CO)n, n=0,1,2,3,4. J Am Chem Soc. 1979; 101: 5569–5573.CrossrefGoogle Scholar
  • 25 Herrick RS, Brown TL. Flash photolytic investigation of photoinduced carbon monoxide dissociation from dinuclear manganese carbonyl compounds. Inorg Chem. 1984; 23: 4550–4553.CrossrefGoogle Scholar
  • 26 Clark JE, Foresti R, Green CJ, Motterlini R. Dynamics of haem oxygenase-1 expression and bilirubin production in cellular protection against oxidative stress. Biochem J. 2000; 348: 615–619.CrossrefMedlineGoogle Scholar
  • 27 Chomczynski P, Sacchi N. Single step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal Biochem. 1987; 162: 156–159.CrossrefMedlineGoogle Scholar
  • 28 Motterlini R, Foresti R, Bassi R, Calabrese V, Clark JE, Green CJ. Endothelial heme oxygenase-1 induction by hypoxia: modulation by inducible nitric oxide synthase (iNOS) and S-nitrosothiols. J Biol Chem. 2000; 275: 13613–13620.CrossrefMedlineGoogle Scholar
  • 29 Clark JE, Foresti R, Sarathchandra P, Kaur H, Green CJ, Motterlini R. Heme oxygenase-1-derived bilirubin ameliorates post-ischemic myocardial dysfunction. Am J Physiol Heart Circ Physiol. 2000; 278: H643–H651.CrossrefMedlineGoogle Scholar
  • 30 Wang R, Wang ZZ, Wu LY. Carbon monoxide-induced vasorelaxation and the underlying mechanisms. Br J Pharmacol. 1997; 121: 927–934.CrossrefMedlineGoogle Scholar
  • 31 Motterlini R, Foresti R, Green CJ. Studies on the development of carbon monoxide-releasing molecules: potential applications for the treatment of cardiovascular dysfunction.In: Wang R, ed. Carbon Monoxide and Cardiovascular Function. Boca Raton, Fla: CRC Press; 2002:249–271.Google Scholar