Inflammation and host defense are necessary and intrinsic processes that serve to protect organisms from a series of pathological challenges. The mechanisms that accompany the inflammatory response involve multiple cell types, signalling pathways and transcriptional factors and inflammation appears to be relevant for the vast majority of chronic diseases as well as in acute conditions 68. That HMOX1 is a key player in mitigating inflammation was first reported in a model of carrageenin-induced pleurisy in rats, in which the evolution of inflammation was accompanied by a dramatic increase in HMOX1 levels and inhibition of heme oxygenase activity enhanced inflammatory markers 69. In addition, HMOX1 deficiency in human subjects exhibited high levels of vascular inflammation and oxidative stress 70, a finding which is highly reproducible in mice lacking this stress protein 71. Although bilirubin and biliverdin, endowed with potent antioxidant properties, may be important contributors that fight inflammation 72 73, CO gas applied exogenously is often found to recapitulate many of the anti-inflammatory actions elicited by HMOX1 74. Our work on the discovery and characterization of CO-RMs was carried out while novel findings by Otterbein et al. described the powerful effect of CO gas in inhibiting the production of pro-inflammatory cytokines (TNF-α, IL-1β) stimulated by lypopolysaccharide (LPS) in vitro and in vivo, showing at the same time that CO induced the expression of the anti-inflammatory cytokine IL-10 and that mitogen activated protein kinases (MAPKs) mediated this phenomenon 8. This and other exciting work stimulated our efforts in the development of CO-RMs and in trying to understand their efficacy in disease models. From an anti-inflammatory perspective, CO-RMs can affect multiple cell types and pathways that coordinate the inflammatory cascade (see Table  1 for a summary of the anti-inflammatory activities of CO-RMs in various in vitro and in vivo models). For example, Urquhart et al found that CORM-3 strongly reduced neutrophil extravasation in the peritoneum of zymosan-treated mice and inhibited expression of adhesion molecules in human polymorphonuclear neutrophils (PMNs) 54. Still focusing on PMNs, Sun and co-workers showed that CORM-2 attenuated the leukocyte sequestration, Nfkβ activation and endothelial protein expression of ICAM-1 in the lung of thermally injured mice 75. The multiple effects of CO-RMs were particularly well dissected in a study by Masini et al. where human PMNs primed to elicit an inflammatory response were co-incubated with rat endothelial cells or perivascular mast cells 50. Here the authors clearly showed that CORM-3 down-regulated the oxidative burst in PMNs, the over-expression of adhesion molecules in PMNs and endothelial cells and the release of histamine and up-regulation of an activation marker by mast cells. These results indicate how CORM-3 modulates acute inflammation by reducing the activation of PMNs, the first responders in host defense, but also by inhibiting expression of molecules and inflammatory factors that perpetuate the inflammatory process. In RAW macrophages and BV-2 microglia we have also shown concentration-dependent decreases in nitrite and TNF-α production by CORM-2 and CORM-3 following challenge with LPS 55 56 57.

The in vivo anti-inflammatory action of CO-RMs has also been consistently described. The group of Alcaraz has performed a series of detailed investigations in arthritis models 21 33 49 and demonstrated that daily treatment with CORM-2 or CORM-3 can effectively suppress the clinical and histopathological manifestations of disease. Levels of PGE-2 and many other inflammatory mediators were reduced in the joint and this resulted in a better preservation of cartilage tissue and bone structures 33. However, modulation of inflammatory molecules levels surely is not the only mechanism contributing to the CO-RMs mediated protection against inflammation and the data by Lancel and colleagues point to mitochondria as very important cellular organelles that are affected by CO-RMs. In a model of sepsis induced by cecal ligation, CORM-3 administration conserved cardiac mitochondrial function by preventing sepsis-mediated damage to mitochondria thus preserving membrane potential and respiration and inducing mitochondrial biogenesis 36. In the heart of mice fed a high fat diet to mimic a metabolic syndrome-like disorder CORM-3 also stimulated mitochondrial biogenesis 76. The mode of action and efficacy of CO-RMs may also depend on the timing of administration in relation to the pathology studied, as exemplified recently by our investigation in a model of hemorrhagic stroke in rats 19. Indeed, we observed that CORM-3 pre-treatment (5 min) or post-treatment (3 days) of rats after the onset of the hemorrhage elicited protective effects while administering the compound 3 hours after the stroke, in correspondence to the acute phase of the disease process, resulted in exacerbation of damage. The striking observation of this study is that one single dose of CORM-3 could modify the long-term inflammatory scenario that followed the hemorrhagic stroke by redirecting and limiting the infiltration of peripheral leukocytes and neutrophils in the brain and reducing the local activation of brain microglia and astrocytes induced by the stroke. Importantly, CORM-3 appeared to finely tune the levels of TNF-α, by allowing its positive action in reparatory processes but inhibiting its detrimental effects. Thus, a growing body of literature supports a beneficial role of CO-RMs in inflammatory models but future investigations are required to better establish their therapeutic applications (see Figures  2 and 3 for the proposed mechanism of action of CO-RMs in vitro and in vivo).

It is well accepted that inflammatory stimuli promote a variety of responses that participate to exacerbate damage in cells and tissues but also promote the resolution of inflammation. Oxidative stress, derived from excessive and persistent production of reactive oxygen species (ROS) and a possible decrease in antioxidant defences, accompanies or precedes the increased amounts of inflammatory mediators upon inflammatory challenge. Because CO has a high affinity for different heme-containing proteins – cytochromes in mitochondria and NADPH oxidase in the cell 77 - that contribute to regulate the levels of ROS, it is intriguing that part of the anti-inflammatory activities of CO-RMs may derive directly from inhibiting the generation of these damaging (or signalling) species. CO-RMs have shown a tendency to modulate pathways that produce ROS and the chemical nature of transition metal carbonyls might favour this reaction by allowing a selective transfer of CO from the CO-RMs to the target 78 79. In RAW macrophages treated with LPS or PMA-stimulated neutrophils CORM-2 inhibited NADPH activity and overproduction of superoxide anion (O₂⁻) 80. Similarly, CORM-A1 reduced the accumulation of ROS induced by TNF-α in pig cerebral microvascular endothelial cells, possibly by acting on a specific subunit of NADPH oxidase (Nox4) which is highly expressed in these cells 40 81. Notably, a reduction in oxidative stress was reported also in chondrocytes from cartilage specimens of patients suffering from osteoarthritis, emphasizing both the relevance of these findings in primary human tissue and the idea that pathological processes occurring in diseased tissue can be modified by application of CO-RMs 51. Oxidative stress levels were also significantly reduced by CORM-3 in intestinal tissue in a clinically relevant model of postoperative ileus and this was accompanied by partial restoration of antioxidant capacity levels 41. Increased production of ROS following TNF-α/cycloheximide exposure was also diminished by CORM-A1 in a mouse intestinal epithelial cell line 82. In summary, different CO-RMs can inhibit ROS/oxidative stress that results from inflammation, thus affecting an early and crucial mechanism that modulates subsequent inflammatory processes (see Figure  2).

Inflammation is a complex phenomenon; thus, it is anticipated that any anti-inflammatory properties of CO-RMs would involve a number of metabolic pathway. Overproduction of NO following up-regulation of inducible NO synthase (iNOS) is a critical step in the initiation and propagation of the inflammatory response 83 and diverse actions of CO-RMs in relation to this system have been described although with quite conflicting findings. We have observed that CORM-2 and CORM-3 decrease NO levels produced by macrophages stimulated with LPS without affecting iNOS protein expression 57 and because of these results we have postulated that CO from CO-RMs inhibited the activity of iNOS, a heme-containing protein already shown in purified form to be blocked by CO gas 84. Similar results were obtained in microglia by Min KJ et al. 85 while Megias and colleagues actually demonstrated that iNOS expression was reduced by CORM-2 in Caco-2 cells challenged with a combination of IL-1β, TNF-α and IFN-γ 52. Other authors have reported the same observation in the gut 41 and in the spinal cord 23 following inflammatory damaging conditions, strengthening the idea that indeed CO-RMs exert an inhibitory effect on iNOS induction and activity. This would not perhaps be surprising if we consider that CO-RMs seem to affect the activation of Nfkβ 24 25 52, which controls the expression iNOS and is a master regulator of major pathways in inflammation. However, until detailed studies designed to dissect the effect of CO-RMs on activity versus induction of iNOS are performed, it will not be clear whether CO-RMs can inhibit NO generation or iNOS expression. It may well be that inhibition of both can occur simultaneously or that one effect or the other will prevail depending on the inflammatory condition, tissue analyzed and the type of CO-RM investigated.

The threat of bacterial infection is omnipresent in surgical settings, wounds and contaminated food, any of which may lead to fatal consequences. Interestingly, CO-RMs have been shown to possess anti-bacterial properties that may be among the important therapeutic applications envisioned for this class of compounds. Lack of HMOX1 in mice resulted in an exaggerated lethality after cecal ligation and puncture (CLP), which caused polymicrobial sepsis 86. However, the administration of CORM-2 was able to increase phagocytosis, decrease circulating bacterial counts, and rescue HMOX1^–/– mice from the exaggerated mortality of CLP-induced sepsis, even when applied 6 hours after the initiation of infection. This is a remarkable result that emphasizes how these molecules can exert pleiotropic actions in such a complicated and severe pathological scenario. Desmard et al. also demonstrated that CORM-3, CORM-2 and, to a lesser extent, CORM-371, exert anti-bacterial actions against P. Aeruginosa in vitro and in vivo 17 87. The ruthenium-based CO-RMs appeared more effective and CORM-A1 only exerted a transient bacteriostatic action, highlighting again the importance of the metal in mediating some activities of CO-RMs and perhaps guiding CO to the appropriate cellular target. More detailed work has been performed to investigate the direct effect of CO-RMs on different bacteria and the results have been reviewed elsewhere 88 89.

Using more biochemical-oriented approaches, it has been possible to determine that terminal oxidases are targeted by CORM-3 when inhibiting bacterial growth 17 44, thus impairing bacterial respiration. Moreover, oxidative stress caused by CO-RMs is another factor explaining some mechanistic actions of these compounds. According to Tavares et al., exposure of E. Coli to CORM-2 or a molybdenum-containing CO-RM increased the levels of intracellular ROS as well as causing DNA damage and disruption of Fe-S clusters 45. The authors also showed that thiol-based antioxidants prevented the anti-microbial properties of CO-RMs, a finding we corroborated in studies using P. Aeruginosa 17 87. However, in our work no effect of CORM-2 or CORM-3 on ROS production was detected as assessed by the use of a fluorescent probe 17, while in another collaborative investigation it was observed that thiols reduced the ROS production stimulated by CORM-2 in P. Aeruginosa biofilms but that this reduction was not accompanied by inhibition of bacterial growth 46. Thus, the collective evidence suggests that CO-RMs interacts with metal-based proteins present in bacteria to exert various effects that are related to their bactericidal or bacteriostatic properties. However, it is possible that other pathways, susceptible to changes following application of CO-RMs, contribute to CO-RMs-mediated anti-microbial effects. In this regard, data obtained from microarray analysis of bacterial genes affected by exposing E. Coli to CORM-2 and CORM-3 have provided fascinating information about the pathways responding to CO-RMs 44 90. It should be noted that one study looked at aerobically and anaerobically grown E Coli and CORM-2 while the other investigated CORM-3 in anaerobically grown E Coli, thus already indicating that different results should be expected from this analysis. Saraiva and colleagues have nicely summarized the diverse, and perhaps still incomplete, information collected in these two transcriptomic approaches showing that some pathways are typically changed in anaerobic conditions, some only in the aerobic state and some are instead commonly altered in E Coli grown either in aerobic or anaerobic conditions 88. Of interest, genes involved in zinc homeostasis and the bacterial response to oxidative stress (SoxRS and OxyR) were increased in both conditions, perhaps stressing that, irrespective of the growth medium, the presence of ruthenium in CO-RMs and their propensity to cause oxidative stress/inhibit the respiratory complexes will consistently occur. The fact that genes modulating biofilm pathways are affected is also a clear signal that E. Coli is suffering from exposure to CO-RMs and thus attempts to boost its resistance to these agents by promoting biofilm formation. In addition, increased methionine metabolism is triggered by CO-RMs, which may still be linked to an oxidative stress response.

In summary, few, but well designed and informative reports support the idea that CO-RMs are useful compounds to be employed alone or in combination with other antibiotics 46 to fight bacterial infection, taking into account the important notion that the bactericidal actions of CO-RM-s are elicited at concentrations that do not harm mammalian cells 17.

Assessing CO liberation from CO-RMs has been a priority since our discovery of these compounds. Initially we developed a myoglobin assay for the detection of carbon monoxy myoglobin (MbCO) and employed an amperometric CO electrode to determine the rate and quantity of CO released 12 16. Gas chromatography techiques have also been used by others to assess the spontaneous liberation of CO from CO-RMs in solution. In parallel, we used bioassays such as relaxation of aortic vessels and inhibition of the inflammatory response in macrophages to assess the CO-mediated pharmacological effects of CO-RMs 12 57 91. In most cases we found a very good correlation between the rate and mode of CO release by CO-RMs and their effect on aortic ring relaxation. The results from several studies have also enabled us to propose that the chemical structure of metal carbonyls CO-RMs might facilitate the direct transfer of CO from CO-RMs to the intracellular target(s) as it appears that the release of CO from certain metal carbonyls (i.e. CORM-2 and CORM-3) requires an acceptor 17 44 92. This may enhance selectivity for the action of CO from metal carbonyls and the difference with CO gas applied exogenously would be that its diffusion into cells might be limited or hampered by the encounter of many proteins potentially able to bind CO, including the prototypic intracellular target(s) 17 77. Although this concept needs to be substantiated, results on the bactericidal effects of CO-RMs (see above) and an interesting article published recently reports data in this direction. Wang and colleagues have developed a genetically encoded fluorescent probe, which is capable of selectively detecting CO inside living cell 93. The probe, named COSer for CO sensor, consists of a permuted yellow fluorescent protein inserted into the regulatory domain of the bacterial protein CooA, a heme-dependent transcription factor known to bind CO with high affinity and selectivity. It was found that the fluorescent intensity of HeLa cells transfected with COSer increased after addition of 5 μM CO gas and a higher response was obtained with 10 μM. Interestingly, the fluorescence intensity was even stronger in cells treated with CORM-2 as a very significant response was obtained with only 1 μM CORM-2 and to obtain a given fluorescence intensity, more CO gas was needed with COSer-transfected cells than with the purified probe. These findings led the authors to state that CORM-2 provided an alternative and more controllable method for CO delivery to cells and could have possibly reduced the difficulty they encountered in getting CO into cells by using simple CO solutions. Similar findings were recently obtained by Michel and co-workers, who have synthesized a palladium-based fluorescent probe that is capable of detecting CO with high selectivity both in aqueous solutions and in living cells. Notably, CORM-3 was used in their experiments as the source of CO revealing that, unlike CO gas, concentrations as low as 1 μM CORM-3 were sufficient to trigger fluorescence in cells loaded with the palladium probe 94.

The use of the MbCO assay to assess the rate and amount of CO liberated by CO-RMs has been recently questioned 95. In our experiments we showed that while MbCO is promptly formed after addition of CORM-2 or CORM-3 to a solution containing reduced Mb, a sensitive CO electrode failed to detect any CO upon addition of these two CO-RMs 17. However, CO release from CORM-A1, a boranocarbonate, is detected by the Mb assay and by the electrode with comparable results, indicating the spontaneous liberation of CO from the compound. McLean and colleagues have shown that in the case of CORM-3 and CORM-2 liberation of CO and the consequent formation of MbCO is facilitated by dithionite, which is usually added in excess to the assay for keeping Mb in a reduced state 95. The authors concluded that the MbCO assay should be abandoned and propose the use of hemoglobin (Hb) since it binds CO with a much greater affinity than oxygen and does not require deoxygenation by dithionite. We believe these data indicate that dithionite and other sulphite can accelerate the release of CO from CO-RMs and that rates of CO release obtained with the MbCO assay should be interpreted cautiously, but we would like to add few important considerations still in favour of the MbCO assay. First, the results by McLean and colleagues seem to imply that liberation (or transfer) of CO from CORM-2 and CORM-3 to a prototypic target (i.e. Mb) cannot occur in the presence of a deoxygenated reduced heme but is triggered only by interaction with anions such as sulphites. That this is not the case is elegantly described by Obirai and colleagues in an interesting report published few years ago revealing quite the opposite as CORM-2 was demonstrated to directly transfer CO to a heme(FeII)/heme(FeIII) redox couple 92. Using a cyclic voltammetry method, the authors proved that when CORM-2 is added to an argon deaerated phosphate buffer solution containing an electrode coated with the heme redox couple but in complete absence of dithionite, a heme(FeII)-CO complex is formed. Secondly, the determination of the rate of CO liberation from CO-RMs by using an in vitro biochemical assay is rather approximate and we always judged it best to interpret our data on CO liberation using a combination of approaches as these compounds are designed for their possible therapeutic use in vivo. That is the reason why in our studies on the characterization of CO-RMs we always coupled the quantification of CO liberation in vitro with data obtained using bioassays that reflect more closely the behaviour of these compounds in complex biological systems. For instance, despite the fact that CORM-2 and CORM-3 are stable compounds in solution and may not liberate CO spontaneously, they still cause a rapid relaxation in isolated vessels and hypotension in animals suggesting that these compounds are fast CO releasers in vivo in line with the MbCO assay data. The bioactive effects mediated by the rapid CO release from these two CO-RMs have been corroborated by using pharmacological tools (i.e. inactive CO-RMs or CO-RMs depleted of CO) 15 57 91 or by comparison with compounds that release CO much slower in vitro and in vivo (i.e. CORM-A1 and CORM-371) 16 17. As a further example, we have recently employed the MbCO assay to determine that CORM-401, a manganese-containing CO-RM, liberates 3-4 CO per molecule 60. We found that the relaxation exerted by this molecule in aortic rings is approximately 3-fold more pronounced than that elicited by the same concentration of CORM-A1, which has a half-life similar to CORM-401 but liberates only 1 CO (unpublished results). Third, the use of oxygenated Hb instead of reduced Mb poses other relevant issues, such as the presence of 4 hemes and the co-operative effect of CO binding to the hemes, which will make it more difficult to quantify the amount and kinetics of CO released. These considerations, together with the results of the MbCO assay and the recent findings with the fluorescent probes reported above, strongly indicate that the release of CO from CORM-2 and CORM-3 occurs when the metal carbonyl is in the vicinity of a reduced iron acceptor (MbFe(II) or heme(II)). The results have also important implications on the efficacy of metal carbonyl CO-RMs in delivering CO to prototypic intracellular targets.