Review article: carbon monoxide in gastrointestinal physiology and its potential in therapeutics


  • This commissioned review article was subject to full peer-review and the authors received an honorarium from Wiley, on behalf of AP&T.

Correspondence to:

Dr G. Farrugia, Enteric NeuroScience Program, Mayo Clinic, 200 First Street SW, Rochester, MN 55905, USA.




While carbon monoxide (CO) is a known toxin, it is now recognised that CO is also an important signalling molecule involved in physiology and pathophysiology.


To summarise our current understanding of the role of endogenous CO in the regulation of gastrointestinal physiology and pathophysiology, and to potential therapeutic applications of modulating CO.


This review is based on a comprehensive search of the Ovid Medline comprehensive database and supplemented by our ongoing studies evaluating the role of CO in gastrointestinal physiology and pathophysiology.


Carbon monoxide derived from haem oxygenase (HO)-2 is predominantly involved in neuromodulation and in setting the smooth muscle membrane potential, while CO derived from HO-1 has anti-inflammatory and antioxidative properties, which protect gastrointestinal smooth muscle from damage caused by injury or inflammation. Exogenous CO is being explored as a therapeutic agent in a variety of gastrointestinal disorders, including diabetic gastroparesis, post-operative ileus, organ transplantation, inflammatory bowel disease and sepsis. However, identifying the appropriate mechanism for safely delivering CO in humans is a major challenge.


Carbon monoxide is an important regulator of gastrointestinal function and protects the gastrointestinal tract against noxious injury. CO is a promising therapeutic target in conditions associated with gastrointestinal injury and inflammation. Elucidating the mechanisms by which CO works and developing safe CO delivery mechanisms are necessary to refine therapeutic strategies.


In the late 19th century, carbon monoxide (CO) was already known to be present in the blood, but its exact origin was still unknown. The correlation between haem degradation and endogenous CO production was shown later, first in vitro[1] and later, also in vivo.[2] The responsible enzyme, haem oxygenase (HO; encoded by HMOX genes), was identified much later in liver microsomes.[3, 4]

As illustrated in Figure 1, HO has two main isoforms; HO-1 and 2 derived from different genes.[5] HO-2 is constitutively expressed,[5] while HO-1 can be rapidly induced in response to several chemical and physical stressors.[6-8] A third isoform was previously identified as HO-3 in rats, but is now regarded as a pseudo-gene and not thought to generate a functional protein.[9] HO-1 provides cells with an inducible protective mechanism against oxidative stress, injury and inflammation via several mechanisms. Despite the different molecular structure of HO-1 and 2, both enzymes catalyse the same nicotinamide adenine dinucleotide phosphate (NADPH)-dependent reaction, responsible for the ring-opening and degradation of haem, converting it into equimolar amounts of free iron (Fe2+), biliverdin and CO (Figure 1). The products of this reaction have been proven to be responsible for many, if not all, of the physiological functions of HO. Importantly, HO results in the removal of free haem, which is cytotoxic and catalyses the production of hydroxyl radicals. Haem degradation also results in the extraction of the reactive iron atom from haem, to neutralise it to Fe3+ or for reutilisation in other reactions. The formed biliverdin is further reduced to bilirubin by biliverdin reductase and both have been reported to possess anti-inflammatory and antioxidative properties.[10, 11] In addition, the produced CO has been shown to possess many antioxidative and anti-inflammatory properties.[12] The relative importance of each of these factors is not entirely elucidated, but CO clearly exerts a wide array of beneficial and protective effects seen upon HO-1 stimulation. This review will focus on the physiological role of CO in the gastrointestinal tract.

Figure 1.

Mechanisms of CO generation. Carbon monoxide (CO) is generated when haem is degraded by either haem oxygenase-1 (HO-1) or haem oxygenase-2 (HO-2) using NADPH as a co-factor. Regulation of HO-2 is not well defined, but occurs in response to nerve activity, activation of Ca2+ calmodulin or glucocorticoids. HO-1 activity is increased by increased protein expression in inflammation and injury and by substrate (haem) availability. Free iron and biliverdin are also generated by this process. Biliverdin is degraded to bilirubin by biliverdin reductase.


The content of this manuscript was obtained from the literature that has informed our work on this field going back many years. We have included a summary of our own work, information obtained from discussion with colleagues and data from broad literature searches. The searches were carried out using the Ovid Medline 1946 to June Week 3 2013 database. We searched for English language reviews on HO and searched all English language literature containing the terms: ‘gastr$, intest$, stomach OR colon’ AND ‘heme oxygenase, haem oxygenase OR carbon monoxide’.


Haem oxygenase

Under basal conditions, inducible HO-1 is found mainly in the spleen and other tissues where senescent red blood cells are degraded. In nearly all other tissues, including the gastrointestinal tract, HO-1 expression is only detectable after stimulation by a wide variety of stimuli not limited to oxidative stress,[13] hypoxia[14-16] or inflammation.[17] A higher expression in the gastrointestinal tract was seen during colitis,[18, 19] inflammatory bowel disease (IBD)[19] and gastric ulcers.[20] Conversely, HO-2 is constitutively expressed, but can be relatively weakly regulated by glucocorticoids,[21, 22] indicating a different physiological function from HO-1. HO-2 is also more widely expressed in basal conditions and was originally identified and purified from rat testes.[23] Other tissues with high HO-2 expression include neurons, endothelial cells, liver, kidneys and spleen.[24-26] HO-2 immunoreactivity is present throughout the gastrointestinal tract in subsets of enteric neuronal cell bodies and fibres in the myenteric plexus and deep muscular plexus in humans[27] and commonly used animal models.[28-31] Particularly high levels of HO-2 are expressed in neurons of the pyloric and ileocaecal sphincters.[32] HO-2 is also present in nonneuronal cells of the mucosal epithelium,[33] smooth muscle cells and endothelium of blood vessels[34] and interstitial cells of Cajal (ICC).[34]

Intracellular targets of CO

The molecular targets and cellular effects of CO, generated by HO, are summarised in Figure 2. The synthesis and release of CO by cells are directly determined by the availability of haem and the expression and activity of basal or inducible HO. It appears that most, if not all, cells can synthesise haem. Exogenous application of CO can often replicate the effects seen under HO-1 stimulation, although there is still uncertainty about the effective local and systemic concentrations reached to induce physiological effects. The ability of CO to diffuse through membranes bypassing cell surface receptors makes it an ideal signalling molecule for fast modulation of physiological processes. CO exerts its effects through a limited number of signal transduction pathways. The specific signalling pathways vary, depending on the cell type and circumstances (Figure 2). The most common pathways described are as follows.

Figure 2.

Targets and effects of carbon monoxide (CO). CO regulates the activity of multiple protein targets to mediate its effects as an intracellular signalling molecule, hyperpolarising factor, anti-inflammatory agent and cytoprotectant (dep, dependent; act, activated).

Binding of haem proteins

Carbon monoxide possesses strong metal-binding properties, making all metallo-proteins containing haem plausible binding targets. CO is capable of regulating the activity of the following proteins: haemoglobin, myoglobin, cytochrome c,[35] cytochrome P450,[36] nitric oxide synthase (NOS), catalase, prostaglandin H synthase, NADPH oxidase (Nox), and the transcription factors NPAS2,[37] Bach-1 and Bach-2. By binding haem, CO inhibits binding of oxygen to haemoglobin with consequent toxic effects. CO can also bind to mitochondrial cyclooxygenase (COX), which inhibits oxidative phosphorylation and adenosine triphosphate (ATP) production and increases production of reactive oxygen species (ROS). ROS, although considered toxic for the cells, are important modulators of signal transduction, regulating biological processes, such as mitochondrial biogenesis.[38, 39]

Carbon monoxide[40, 41] binds to iron of the haem moiety of soluble guanylate cyclase (sGC), albeit less avidly than nitric oxide (NO), thereby stimulating its activity and increasing production of cGMP. This pathway appears to be involved in smooth muscle relaxation and vasodilatation.[42] Indeed, CO has been shown to cause relaxation throughout the gastrointestinal tract by a cGMP-dependent mechanism in canine jejunum,[43] ileal smooth muscle strips from guinea pigs,[44] anal sphincter smooth muscle strips from opossum,[45] feline lower oesophageal sphincter,[46] and in the fundus[47] and oesophagogastric junction from pigs.[48] CO (300 μM) and the CO-releasing molecule, (CORM)-2, increased cGMP levels and relaxed gastric fundus,[49] jejunal smooth muscle strips[50] and distal colon[51] in mice, effects that could be abolished by the sGC inhibitor ODQ. Accordingly, basal levels of ileal cGMP are lower in HO-2 knockout mice compared with their wild type counterparts. Nonadrenergic noncholinergic (NANC) relaxations and cGMP elevations after electrical field stimulation were markedly lower in HO-2 knockout mice, an effect not observed in muscle strips of the inner anal sphincter.[31, 52] YC-1, a superoxide-sensitive stimulator of sGC, strongly enhanced the amplitude of the CO-induced relaxation of fundic smooth muscle strips from pigs.[47] cGMP-induced smooth muscle relaxation is probably mediated mainly by activation of cGMP-dependent protein kinase I, which ultimately reduces intracellular Ca2+ concentrations through voltage-gated ion channels[53] CO can also increase the levels of cAMP by regulating phosphodiesterases.[54] A recent paper identified a unique haem domain as a regulatory unit for soluble adenylate cyclase that can bind CO and NO.[55] This may represent a new pathway of cAMP stimulation.

Signalling through the mitogen-activated protein kinase pathway

Carbon monoxide downregulates the production of the pro-inflammatory cytokines, tumour necrosis factor-α, interleukin-1, and macrophage inflammatory protein-1β production and increases the anti-inflammatory cytokine interleukin-10 (IL-10), thereby protecting cells against inflammation-induced apoptosis. These anti-inflammatory and tissue-protective effects were shown to be mediated through the mitogen-activated protein kinase kinase (MKK-3)/p38 mitogen-activated protein kinase pathway, a pathway known to be involved in the protection against environmental stress.[12, 56] In another study, it was shown that activation of p38 mitogen-activated protein kinase (MAPK) by CO promotes nuclear translocation of heat shock factor-1, a transcription factor that stimulates the expression of heat shock protein-70.[57] CO also offers protection against LPS-induced sepsis in mice through the JNK pathway and the transcription factor AP-1.[58] Further investigation in LPS-induced macrophages showed that the first step in the anti-inflammatory pathway of CO involves an increase in ROS through uncoupling protein-2, ultimately resulting in p38 MAPK activation and PPARγ SUMOylation.[59] Other signalling pathways include STAT,[60, 61] PI3K-Akt[60] and stabilisation of HIF1α.[62] A more recent study found that CO inhibits TNF-α induced NADPH oxidase 4 activity in brain endothelial cells via a pathway involving extracellular signal-related kinase (ERK)1/2, p38 MAPK and Akt, ultimately resulting in a decreased ROS concentration and offering protection against apoptosis.[63] This suggests that the pathways may cross-talk.

Activation of ion channels

In recent years, it has become clear that many physiological effects of CO are mediated through activation of ion channels, either through direct binding or by binding to haem-containing proteins associated with these channels. The earliest reports described the ability of CO to activate delayed rectifier-like K+ currents in human jejunal smooth muscle cells and rabbit corneal epithelial cells resulting in membrane hyperpolarisation, possibly mediated by cGMP.[43, 64, 65] In cardiomyocytes, the inhibition of L-type Ca2+ currents is through an indirect mechanism that is due to binding of CO to complex III of the mitochondrial electron transfer chain and increased ROS production. Thus, a paradox exists; even though CO is increasing generation of ROS, the inhibition of Ca2+ influx pathway provides another protective mechanism against damage.[66] Ion channels are involved in the protective effects of CO against oxidative stress-induced apoptosis of neurons, a process implicated in cerebral ischaemia and neurodegenerative diseases. In primary cultures of hippocampal neurons, it was shown that CO selectively inhibits the delayed rectifier-like K+ channel Kv2.1, thereby preventing intracellular K+ loss, an early step in apoptosis. This process was also mediated through ROS and protein kinase G activity.[67] In H441 cells, a human bronchiolar epithelial cell line, CO inhibits Na+ channels independently from cGMP.[68]

Other ion channels modulated by CO include, but are not limited to, the large-conductance voltage and Ca2+-activated K+ channel BK(Ca), which contribute to oxygen sensing and vascular tone[69] and ATP-gated P2X receptors that are involved in purinergic neurotransmission.[70] In human intestinal smooth muscle cells, exogenous CO can activate L-type Ca2+ channels through a pathway that involves increased NO and cGMP levels, and also PKA, but not PKG.[71]

Carbon monoxide also modulates colonic ion transport. In a colonic epithelial cell line (Caco2), stimulation of endogenous CO production by haemin (a ferrous haem molecule – Fe2+ protoporphyrin IX) or administration of the CORM-2 induced active Cl secretion. This effect was inhibited with 1H-[1,2,4]oxadiazolo[4,3-a]quinoxalin-1-one (ODQ), an inhibitor of the sGC, suggesting a role for the cGMP-dependent pathway.[72] In the rat distal colon, CORM-2 increases short-circuit current, carried by changes in transepithelial transport of Cl and HCO3 in a concentration-dependent manner.[73]

Role of CO in normal physiology of the gastrointestinal tract

Carbon monoxide as a nerve-derived gaseous neuromodulator involved in gastrointestinal contractility

Given its hyperpolarising properties and its involvement in NANC neurotransmission (Figure 2), CO fulfils some of the criteria for a neurotransmitter. However, there are limitations to this concept in that neurotransmitters typically bind on membrane-bound receptors on effector cells, while gases can diffuse through the membrane. Moreover, in the stomach and small intestine, CO originates from HO-2 expressed in ICC as well as from neurons. In addition, gases like CO and NO are not stored in synaptic vesicles and there is only limited evidence of rapid regulated release of CO from neurons. A pathway in which Ca2+ influx activates protein kinase C, regulating CK2 and finally HO-2 activity, has been proposed for mouse anal sphincter relaxation,[74] but the time course of these events is somewhat slow compared with true synaptic transmitter release, although HO-2 activity in colonic submucosal neurons is rapidly blocked by tetrodotoxin (TTX, see below). Another example of regulated release of CO has been shown in vascular smooth muscle from pig cerebral microvessels in response to glutamate.[75] What is clear is that CO is critical to inhibitory neurotransmission in some parts of the gastrointestinal tract, particularly in the interplay between CO and NO (see below) in inhibitory neurotransmission,[76] with CO appearing to be required for the full effect of NO.

Interplay between NO and CO, differences and similarities

Nitric oxide was the first gas to be identified as an intercellular messenger, stimulating investigators to look into messenger roles for other gases like H2S and CO. When generated by neuronal nitric oxide synthase (nNOS), NO is an authentic fast neurotransmitter in all respects, except that it is not stored in vesicles and does not have a cell surface receptor. Based on its molecular similarity to NO, CO was one of the next candidates to be considered.[77] Besides the structural resemblance, the enzyme systems that produce them endogenously are often expressed in the same cells, also in the gastrointestinal tract[27] and both gases are known to activate guanylate cyclase, albeit with differing affinities.[78] NO and CO work together, for example, to set the resting membrane potential in enteric smooth muscle cells. Both NO and CO also work as co-transmitters during NANC neurotransmission in the mouse small intestine. NANC neurotransmission is markedly reduced in both nNOS knockout and HO-2 knockout mice and even more pronounced in the double knockout, suggesting that both CO and NO are necessary for normal function.[76]

Both gases can also strengthen or decrease one another's action. In mouse jejunal circular smooth muscle strips, the NOS inhibitor L-NAME significantly reduced the CO- and CORM-2-induced relaxations, suggesting an NO-mediated amplification of CO signalling.[50] In cerebellar granule cell cultures, however, exogenous CO administration reduced the NO-mediated cGMP increase.[40] Other than a role in fast signalling from neurons to the target, NO appears to be a key signalling molecule during nervous system development[79] and CO has been reported to act antagonistically to NO in enteric neuron migration in the locust.[80]

Although CO and NO can bind to the same molecules, they often interact through different binding sites. Wu et al. found that, in rat vascular smooth muscle cells, both NO and CO can stimulate large-conductance calcium-activated K+ channels, but the effects of CO are primarily mediated on the α subunit, while NO's site of action is the β subunit.[81]

Carbon monoxide as a gastrointestinal smooth muscle hyperpolarising factor

Carbon monoxide is a hyperpolarising factor in gastrointestinal smooth muscle (Figure 2). A gradient in resting membrane potential exists along the long axis of the stomach and across the thickness of circular smooth muscle layers of the stomach and intestines.[82-84] This gradient allows a graded contractile response to a stimulus. CO production was shown to mirror this gradient, which, together with the lack of membrane potential gradient in the gut of HO-2 knockout mice, confirmed that CO is required for the generation of the RMP gradient.[84] Moreover, the gradient was absent in the small intestines of W/Wv knockout mice, suggesting that CO formed by HO-2 present in ICC in the myenteric region is responsible for the generation and maintenance of the gradient.[85] In colonic circular smooth muscle layers, however, the transwall gradient is reversed compared with the stomach and small intestine, and the major source of CO responsible for the gradient appears to be from submucosal neurons.[86] This hyperpolarising effect of CO appears to be due to dynamic regulation of HO-2 activity in colonic neurons as inhibition of neuronal firing using low concentrations of the Na+ channel blocker, TTX, does quickly remove the membrane potential gradient (J.H. Szurszewski and L. Sha – personal communication).

The role of CO and the potential for CO as a therapy in gastrointestinal disease

A role for CO in disease and injury has been demonstrated in many organ systems, with the effects reflecting the highly inducible nature of HO1 and the potent physiological effects of CO and biliverdin (Figure 2, Table 1).[87, 88] The elements that link these diseases are the anti-inflammatory effects of CO and biliverdin, increased perfusion of the tissues following smooth muscle relaxation, clearance of cytotoxic free haem and reduction in injury due to the antioxidant properties of CO and biliverdin. The elevation of HO-1 levels following injury or inflammation appears to have a protective effect against further injury, but this response is varied in duration and magnitude. Genetic changes can contribute to this variable response[89] and treatment with CO-based therapies may sustain, replace or restore the activity of HO-1.[90]

Table 1. Summary of CO-based therapies tested in clinical gastrointestinal diseases and animal models of the diseases
DiseaseSpeciesDelivery mechanisms testedImpact of therapyRefs
  1. CO, carbon monoxide; HO, haemeoxygenase.

Diabetic gastroparesisMouseHaeminReversal of delayed gastric emptying [117-119]
  CO inhalationAlternative-activation of macrophages 
 HumanHaemin – clinical trialUnder investigation [96]
Post-operative ileusMouseCO inhalationPrevented symptoms [125]
  CO-releasing moleculesPrevented inflammation [105]
  Interleukin-10Prevented symptoms [127]
 PigCO inhalationPrevented inflammation [126]
Organ transplantationRatCO or CO-releasing molecules in transportation mediumImproved outcomes [129, 131, 132]
  CO inhalation  [130]
 HumanAccidental CO exposureNo detrimental effect [133]
Inflammatory bowel diseaseHumanInhalation in cigarette smokeReduced symptoms in ulcerative colitis [133]
 MouseCO inhalation, HO induction by cobalt protoporphyrin, CO-releasing molecules, biliverdinImproved outcomes in Th1- and Th2-mediated colitis and TNBS and DDS colitis models. Suppression of microbiome–host interactions [111, 134-138]
SepsisRatBiliverdinProtects against microbial sepsis [141]
  Ischaemic preconditioningProtects against LPS organ failure [143]
 MouseCO-releasing molecules, cobalt protoporphyrinIncreased bacterial clearance in microbial sepsis [138, 140]

In experimental animals, approaches that increase CO levels in the target tissue improve a variety of gastrointestinal disorders, for example, diabetic gastroparesis, IBD, sepsis and post-operative ileus and outcomes following intestinal transplant. Most of these diseases are directly or indirectly associated with inflammation or increased oxidative stress in the affected tissues and, even in the absence of inflammation, there is good reason to propose that modulation of immune cells contributes to the beneficial response to CO delivery.

In the absence of HO-1, or when HO-1 is expressed at low levels, significant tissue injury and diseases are observed. Mice homozygous for knockout of the HO-1 allele often die in utero and do not survive long after birth[91] and patients with congenitally low HO-1 expression suffer serious consequences.[92, 93] There are at least three identified polymorphisms in the human Hmox1 gene that associate with reduced HO-1 expression and worse outcomes in diseases including pulmonary, cardiovascular and neurological diseases as well as outcomes of renal transplantation and haematological disorders.[89] These polymorphisms can affect the susceptibility of an individual to disease as well as the response to treatments that are targeted to increase HO-1 activity and CO production. For gastrointestinal disorders, there is one report of increased risk of gastric cancer in individuals with polymorphisms in the Hmox1 gene.[94] For mice, our work has identified a (TCTCT)n repeat polymorphism in the mouse hmox1 promoter that is different from the (GT)n repeat polymorphism in humans and appears to predispose diabetic animals to development of delayed gastric emptying.[95] Identifying and understanding the impact of these polymorphisms are an important step towards choosing CO-based therapies that will be most effective in any specific individual.

Delivery of CO-based therapies

Increased levels of CO may be achieved by direct delivery of the gas or increased local production. Due to the strong antioxidant effects of bilirubin, the other product of HO-1 activity, it is possible that increased HO-1 activity could promote a more effective therapy. As a gaseous molecule with a high affinity for a number of biological targets, CO is a difficult substance to quantify in complex systems. Thus, the free CO concentration reaching a cell and diffusing towards an intracellular target, or in a bodily fluid such as blood, is not known. The impact of CO delivery on the organism can be assessed by using a fairly simple spectrophotometric method for measuring COHb levels due to the high affinity of haemoglobin for CO and the rapid equilibration of the binding process. The %COHb is usually reported as an indication of the CO burden in an organism. Free CO as delivered by inhalation can be monitored by sampling the gas and measuring the levels using electrochemical or spectroscopic techniques. Typical doses are in the 10 to 100s parts per million (ppm).

The major challenge to the use of CO-based therapies for gastrointestinal disease and injury is the delivery mechanism. Direct inhalation of CO is convenient with the effective doses in animals being less than 250 ppm.[90] Ongoing clinical trials are evaluating doses of 150 ppm administered for 3 h up to twice weekly.[96] Nevertheless, there remains considerable resistance to the use of this poisonous substance for therapeutic purposes. Even though the levels of haemoglobin saturation with therapeutic CO administration remain significantly below the COHb levels of 5–13% found in moderate smokers, it has been observed that neurological and cardiac injury might still be a problem at quite low doses of CO.[97] World Health Organization (WHO) guidelines recommended that blood COHb levels be limited to below 2.5% or delivery of CO at 87.1 ppm for 15 min[98] based on achieving the lowest reasonable levels in an occupational setting. Clearly, unnecessary exposure to an environmental toxin should be limited as much as possible, but there is no evidence that such low limits need to be applied to short-term therapeutic CO administration. Existing studies on the toxicology of CO used either very high doses or prolonged exposure. In the heart, acute exposure to CO can cause arrhythmias; however, only when administered at >1500 ppm.[99] Exposure of rats for 4 weeks at continuous levels of CO in the 30–100 ppm range resulted in cardiac remodelling.[97] The nervous system, like the heart, is an organ that is dependent on aerobic respiration and, as such, is particularly sensitive to disruption by CO. Neurological symptoms that are diagnostic for CO poisoning include headaches, dizziness, confusion and loss of consciousness. However, these symptoms are described in cases where the COHb levels were higher than 10%.[100, 101] Other neurological symptoms, including visual perception, and other functional impairments appear to be detectable when COHb exceeds 5%, as reviewed in detail in the WHO report.[98] However, it is unclear if these effects are transient and fully reversible and so tolerable in an effective therapy for other disease. We conclude that an abundance of care is necessary in assessing the degree and duration of cardiac and neurological changes when trying to establish inhaled CO as therapy. However, there is good reason to believe that a therapeutic dose can be found and side effects can be avoided by limiting treatments to short, low doses with adequate recovery in between. The therapeutic window for CO dosing does not appear to be much narrower than that for several other drugs, including other inhaled gases, such as NO, that have been brought into limited clinical use.[102]

The potential concerns regarding systemic delivery of CO have resulted in an effort to develop targeted delivery. Focused delivery of CO to affected tissues appears preferable to systemic administration and this may be achieved by localised upregulation of HO-1 or localised delivery of a CO-releasing agent. The molecules are metal carbonyls that are stable as solids, but have been created to release CO in solution. They are based on a couple of different chemical backbones, are variably water soluble and release CO at varying rates.[103, 104] They are clearly effective in vivo when tested in animal models of a variety of diseases, including post-operative ileus in mice.[90, 104-106] With respect to localised delivery, the study on post-operative ileus demonstrated that CORM-3, but not CORM-A1, could be administered by intraperitoneal injection without significantly altering blood COHb levels.[105] However, there are some concerns about the toxicity of these compounds that need to be resolved before applying them to clinical practice[104], and, for each compound, the effective concentration of CO and duration of CO release need to be carefully characterised for each application. Similarly, induction of HO-1 is not without problems; free haem is toxic, induces oxidative stress and causes inflammation, and it is by inducing HO-1 that this toxicity is ameliorated.[107] This has not prevented approval of haemin for treating porphyria in studies that found that haem-based compounds were safe.[108] Haemin increased plasma HO-1 protein concentration four to fivefold and HO-1 activity more than 15-fold relative to baseline in healthy people.[109] Other alternatives may be to use some of the other compounds that induce HO-1 and that are approved for use in humans, which include IL-10 (as discussed elsewhere) and other drugs.[110-112]

Another way to develop better CO-based therapies for gastrointestinal disease and injury may be to exploit our increasing understanding of the downstream signalling pathways for CO. The complexity of the response to treatment is not fully elucidated and there are overlapping effects, but those effects are frequently synergistic. For example, the induction of HO-1 by CO, which in turn generates more CO or IL-10, is another example. IL-10 can be generated in response to biliverdin or CO,[10, 12] it has direct cytoprotective effects[113] and induces HO-1.[114] Successful use of IL-10 in treating gastroparesis in mice is outlined below.

CO as a therapeutic agent

Diabetic gastroparesis

Gastroparesis is a syndrome characterised by delayed gastric emptying in the absence of mechanical obstruction of the stomach. The symptoms that lead to diagnosis include postprandial fullness (early satiety), nausea, vomiting and bloating and often abdominal pain.[115] Gastroparesis in patients with diabetes is a significant cause of morbidity not only due to the symptoms of gastroparesis but also due to the effects on the timing of nutrient delivery due to impaired gastrointestinal function and the consequences of this on effective glycaemic control.[116]

Diabetic mice can also develop gastroparesis as detected by delayed gastric emptying, a symptom that develops in animals that fail to sustain HO-1 expression in the gastric smooth muscle. In the non-obese diabetic mice, the slowing of gastric emptying can be reversed by inducing HO-1 using haemin,[117] by administering CO or by administering IL-10.[118, 119] Simultaneous inhibition of HO-1 with chromium mesoporphyrin did not block the improvement in gastric emptying observed in CO-treated mice, indicating that CO is sufficient for successful therapy. IL-10 also induced HO-1 expression and reversed the slowing of gastric emptying, but it is not clear if the upregulation of HO-1 expression is necessary for the favourable response to IL-10. In mice, the HO-1 that appears to be necessary for protecting diabetic animals from development of delayed gastric emptying is expressed in a population of alternatively activated macrophages. These macrophages have the characteristics of the M2c phenotype of mouse macrophage as they express HO-1 and CD163 and release IL-10.[120] IL-10 is an autocrine factor that supports survival of M2c macrophages[121] and also suppresses release of inflammatory cytokines from conventionally activated M1 macrophages.[122] Thus, there appear to be reciprocal interactions between IL-10 and HO-1[123] and, although it is difficult to separate these interactions, they should be investigated to establish optimal targets for CO/HO-1 based therapies. One therapeutic option that merits further consideration is a combination of IL-10 and CO treatment, which might act in a synergistic fashion and thus significantly reduce the necessary dose of either compound.

Post-operative ileus

The development of prolonged hypomotility or ileus following abdominal surgery and intestinal manipulation prolongs hospital stays and can contribute to complications.[124] The cellular processes that cause ileus are complex and include both local processes that include macrophage activation and release of cytokines as well as neuronally mediated pathways. CO-based therapies in animal models of ileus reduce the severity and duration of ileus. In mice, exposure to 250 ppm of CO by inhalation for 1 h, prior to intestinal manipulation, induced HO-1 in leucocytes and macrophages and prevented development of ileus.[125] In a subsequent study, pre-treatment by inhalation of 75 ppm of CO for 3 h was also shown to be effective in preventing ileus in pigs as well as mice.[126] Another group demonstrated that pre-treatment with two different CO-releasing molecules, CORM-3 and CORM-A1, also reversed ileus following surgical manipulation in mice.[105] The effect was associated with increased levels of anti-inflammatory cytokines, including IL-10, and reduced levels of pro-inflammatory cytokines. Thus, IL-10 has potential as a preventive therapy for postileus as demonstrated in a study that reported faster recovery in intestinal transit and smooth muscle contractility in mice treated with IL-10 prior to surgical manipulation to induce ileus.[127] In mice pre-treated for ileus with CORM-3, p38-MAPK signalling was upregulated, whereas ERK MAPK signalling was downregulated.[105] Thus CO-based therapies hold promise for the treatment of post-operative ileus because the treatment can be administered on an acute basis and potentially delivered close to the site of action by intraperitoneal injection.[128] However, this promise has not yet been translated into practice, perhaps reflecting the caution associated with use of gases as therapeutic agents.

Organ transplantation

Several studies in experimental animals have identified the potential for CO-based therapies in organ transplantation, including gastrointestinal transplants.[129-132] The basis for these studies is the beneficial effect of CO on ischaemia/reperfusion injury, suppression of inflammation and overall reductions in oxidative stress. The objectives are to preserve the tissue after collection and prior to transplantation and to protect the tissue from reperfusion injury and rejection. Addition of CO to the storage medium during transfer of the organ can be achieved either by bubbling the solution or addition of CO-releasing molecules and both these methods improve outcomes after transplantation in animals for several different organs,[132] including intestinal transplantation in rats.[131] Exposure of recipient animals to CO at 250 ppm for 1–3 h significantly improved the outcomes of intestinal transplantation in rats[129] by a mechanism that was associated with amelioration of inflammatory injury to the transplanted organ. There is no evidence that exposure of the donor organ to CO during storage and transfer has deleterious effects on the success of the transplant in animals. Furthermore, multiple studies on the consequences of transplanting organs from donors who died from CO exposure have determined that graft failure is not more likely for those organs.[132] In fact, animal studies indicate that any procedure that exposes the donor organ to CO prior to transplantation may improve the chances of graft survival.

Inflammatory bowel disease

Several studies have investigated the possible utility of CO in treating IBD and determining the nature of changes in HO-1 activity in colonic inflammation in humans and animal models. Clinical observation has indicated that CO might alleviate the symptoms of ulcerative colitis, specifically in the context of CO inhalation by smokers.[133] Of course, CO is one of the many components in cigarette smoke with significant biological effects, not least nicotine. However, this is not an effect that is true for all IBDs; for example, patients with Crohn's disease do not report similar benefits from smoking,[133] which probably reflects the complex interaction of multiple factors that contribute to various manifestations of intestinal inflammation. In focused investigations using animal models of colitis to study the specific effect of CO on intestinal inflammation, changes to HO-1 expression and activity are central to both the development of inflammation and therapies that prevent or reduce the damage. HO-1 expression is suppressed in macrophages by interferon-γ and other factors in mice with Th1-mediated intestinal inflammation and CO protects the colon from injury by blocking the inhibitory effects of interferon-γ on HO-1.[134] In colitis mediated by Th2 cytokines, namely the TCRα (−/−) mouse, CO reduced the injury in active colitis through IL-10- and HO-1-dependent pathways.[135] The other product of HO-1 activity, biliverdin, decreases damage in colitis caused by dextran sodium sulphate administration to mice. This effect is notable as it is not replicated by CO administration.[136] In 2,4,6-trinitrobenzine sulphonic acid-induced colitis in mice, CO inhalation reduces the injury by a mechanism that involved suppression of TNF-α expression.[137] More recently, the CO-releasing molecule, ALF-186, has been demonstrated to alter the interaction between intestinal microbiota and host mice.[138] Thus, CO-based therapies can have beneficial effects in intestinal inflammation by affecting host responses and environmental factors. Indeed, it appears that a long-standing therapy for IBDs, 5-aminosalicylic acid, acts, at least in part, by induction of HO-1.[111]


The gastrointestinal tract represents the major interface between the pathogens that cause sepsis and the body. Impairment of gastrointestinal barrier function is a key step to the development of sepsis and reduced motility is a rapid and sustained consequence of sepsis. Bacterial invasion and the toxins released activate immune responses, cytokine production and widespread inflammation.[139] In this situation, a robust inflammatory response against the pathogen is necessary and at least one pathogen, enterohaemorrhagic Escherichia coli, suppresses the response by upregulating HO-1 and reducing inducible NO synthase levels in enterocytes.[140] On the other hand, biliverdin appears to protect against polymicrobial sepsis in rats,[141] CO from HO-1 increases the host defence response to microbial sepsis in mice[142] and increased intestinal HO-1 expression attenuates lipopolysaccharide-induced multi-organ failure in rats.[143] On balance, it appears that upregulating HO-1 will be beneficial to a host in its interactions with commensal and pathogenic microbiota. In addition to reducing inflammation in the IL-10 (−/−)model of colitis, induction of HO-1 by cobalt (III) protoporphyrin IX chloride (CoPP) or administration of ALF-186 also increased bacterial clearance of pathogenic Salmonella typhimurium in wild-type mice.[138] Furthermore, knockdown of HO-1 in mouse macrophages reduced their bacteriocidal activity.[138] These observations create a rationale for CO-based therapies for sepsis in as much as the treatments can reverse motility and barrier changes, repair tissue injury and enhance pathogen clearance, while allowing appropriate antimicrobial responses.


In summary, CO is an important physiological regulator of gastrointestinal function and generation of CO is necessary to sustain a healthy gastrointestinal tract. CO derived from HO-2 appears to be predominantly involved in neuromodulation and setting the smooth muscle membrane potential, while CO derived from HO-1 plays an important role in protection of the gastrointestinal tract from damage from injury or inflammation. Targeting CO as a therapy is a promising approach, but refinements to the therapeutic approach depend upon understanding the mechanism by which these different delivery methods achieve their beneficial effects. Convenience of delivery can be enhanced and unwanted side effects can be avoided by targeting the specific downstream effectors of the administered drug. In addition, the mechanisms by which CO works are not fully elucidated and understanding these pathways better is necessary to further refine the therapeutic approach.


Guarantor of the article: Gianrico Farrugia.

Author contributions: All the authors contributed to the conception, research, writing and review of this article. All authors approved the final version of the manuscript.


Declaration of personal interests: None.

Declaration of funding interests: We thank Kristy Zodrow for excellent secretarial assistance with this manuscript and members of the Enteric NeuroScience Program at Mayo Clinic for helpful discussions. This work was supported by National Institutes of Health grants DK57061, DK68055 and DK74008.