Roles of nitric oxide, carbon monoxide, and hydrogen sulfide in the regulation of the hypothalamic–pituitary–adrenal axis


Address correspondence and reprint requests to Cesare Mancuso, Institute of Pharmacology, Catholic University School of Medicine, Largo Francesco Vito, 1 – 00168 Rome, Italy. E-mail:


J. Neurochem. (2010) 113, 563–575.


The importance of stress in modifying human behavior and lifestyle is no longer a matter of debate. Although mild stress enhances the immune response and prevents infections, prolonged stress seems to play pathogenic roles in depression and neurodegenerative disorders. The body has developed an adaptive stress response consisting of cardiovascular, metabolic, and psychological changes, which act in concert to eliminate stressors. One of the major components of this response is the hypothalamic–pituitary–adrenal axis, also known as the stress axis. Over the last 30 years, many studies have documented the integrated stress-axis regulation by neurotransmitters. They have also demonstrated that gaseous neuromodulators, such as NO, CO, and H2S, regulate the hypothalamic release of neuropeptides. The specific effects (stimulatory vs. inhibitory) of these gases on the stress axis varies, depending on the type of stress (neurogenic or immuno-inflammatory), its intensity (low or high), and the species studied (rodents or humans). This review examines the complex roles of NO, CO, and H2S in modulation of stress-axis activity, with particular emphasis on the regulatory effects they exert at the hypothalamic level.

Abbreviations used:

adrenocorticotropin hormone


arginin vasopressin






corticotrophin-releasing hormone






endothelium-derived relaxing factor


endothelial nitric oxide synthase


γ-glutamylcysteine synthetase (EC


gonadotropin-releasing hormone


glutathione synthetase (EC


heme oxygenase






inducible nitric oxide synthase


NG-nitro-l-arginine methyl ester




neuronal nitric oxide synthase




prostaglandin E2


pyridoxal 5-phosphate


paraventricular nucleus


soluble guanylyl cyclase


supraoptic nucleus

Maintaining good health is a major goal of all humans. Unfortunately, despite our efforts to preserve wellness, every day intrinsic and/or extrinsic forces disturb this complex dynamic state of equilibrium, also known as homeostasis (Tsigos and Chrousos 2002). These forces are called stressors, and they can have various origins, the most important of which are infectious diseases and psychosocial events. The body has developed a ‘general adaptation response’ consisting of cardiovascular, metabolic, and psychological changes aimed at counteracting the effects of these stressors (Pedersen et al. 2001; Tsigos and Chrousos 2002). This response is under the control of the hypothalamic–pituitary–adrenal (HPA) axis, which is responsible for the release of glucocorticoid hormones from the adrenal gland. The metabolic and anti-inflammatory activities of these steroid hormones are key components of the adaptive stress response (Pedersen et al. 2001; Tsigos and Chrousos 2002).

Over the past 2–3 decades, many studies have addressed the mechanisms by which the ‘classic’ neurotransmitters influence the HPA axis, especially at the central level (Calogero et al. 1988, 1989; Grossman et al. 1993; Jessop 1999; Leonard 2005). Compelling evidence has also emerged on the stress-regulating roles of gaseous neuromodulators, such as NO, CO, and H2S, in particular on the effects they produce at the hypothalamic and/or pituitary levels (Mancuso et al. 1997a; Navarra et al. 2000; Rivier 2002; Wu and Wang 2005; Calabrese et al. 2007). Whether these gases stimulate or inhibit the stress axis is still a matter of debate: their effects vary, depending on the type and intensity/duration of the stress and the animal species being studied. The aim of this paper is to review the literature on the roles of NO, CO, and H2S in modulation of stress-axis activity with particular emphasis on its central regulation.

The HPA axis or stress axis

The hypothalamus plays key roles in many brain functions, including the control of the autonomic nervous system and regulation of body temperature, food intake, and hormone secretion. Located within the diencephalon, it is delimited anteriorly by the optic chiasm, laterally by the optic tracts and temporal lobes, and posteriorly by the mammillary bodies (Toni et al. 2004). The hypothalamus is divided into the pre-optic, supraoptic, tuberal, and mammillary regions, each containing different groups of cells – or nuclei – with distinct characteristics and regulatory mechanisms. The pre-optic, supraoptic (SON), and paraventricular (PVN) nuclei are particularly important because they are involved in the release of hypothalamic neuropeptides (Toni et al. 2004). The pituitary, which is regarded as the master endocrine gland, is located in the sella turcica, a small cavity within the sphenoid bone. It is divided into two regions: the anterior pituitary, or adenohypophysis, and the posterior pituitary, or neurohypohysis (Toni et al. 2004).

The hypothalamus and the pituitary are closely related, both anatomically and functionally. In fact, the stalk of the pituitary is directly connected to the median eminence, a small swelling on the lower aspect of the hypothalamus that is one of the few brain areas not protected by the blood–brain barrier (Toni et al. 2004). A complex system of portal vessels conveys hypothalamic neuropeptides to both the anterior and posterior regions of the pituitary. The neurons located in the PVN and SON synthesize corticotrophin-releasing hormone (CRH) and arginine vasopressin (AVP), the neuropeptides responsible for activation of the stress axis at the hypothalamic level (Toni et al. 2004). CRH is released into the median eminence by axons of the parvicellular population of PVN neurons. From here, the peptide enters the primary portal plexus of the hypothalamus and is transported through the long vessels to the anterior pituitary, where it stimulates the release of adrenocorticotropin hormone (ACTH) by corticotroph cells. Direct communication between some of these axons and the anterior pituitary has also been demonstrated (Toni et al. 2004). As for AVP, axons of neurons located in the magnocellular portions of the PVN and SON release this neuropeptide into the median eminence. The short portal vessels of the sinusoidal plexus then transport the AVP to the anterior pituitary, where it stimulates ACTH release. Other axons transport AVP directly to the posterior pituitary, where it is released into the systemic circulation (Toni et al. 2004).

During non-stressful situations, the release of both CRH and AVP is characterized by a pulsatile circadian rhythm, with higher-amplitude pulses during the morning hours. Under conditions of stress, both the amplitude and synchronization of CRH and AVP pulses increase, producing significantly higher blood levels of ACTH (Tsigos and Chrousos 2002). Pituitary ACTH then stimulates the adrenal cortex to release glucocorticoid hormones, which promote energy metabolism (gluconeogenesis, lipolysis, and proteolysis) and anti-inflammatory activity. The increases in plasma glucocorticoid levels then reduce the release of ACTH and CRH through a negative feedback mechanism, thereby preventing excessive GC stimulation that can lead to depression and also neurodegenerative disorders (Pedersen et al. 2001).

Nitric oxide

Until recently, NO was regarded mainly as a toxic air pollutant produced by automobile engines and power plants. Its production by living cells was discovered only in the 1980s. In early studies performed at that time, rodent macrophages were found to release increased amounts of nitrate and nitrite (two products of NO metabolism) after exposure to bacterial lipopolysaccharide (LPS) (Stuehr and Marletta 1985). These in vitro findings were later confirmed by the in vivo demonstration that LPS administration stimulates urinary nitrite and nitrate excretion in mice (Billiar et al. 1990). Another important landmark in the characterization of NO was the study by Furchgott and Zawadzki (1980). They reported that acetylcholine or carbachol stimulation of pre-contracted vascular smooth muscle preparations was followed by endothelial production of a short-lived vasodilator that was different from prostacyclin, and they called this substance endothelium-derived relaxing factor (EDRF) (Furchgott and Zawadzki 1980). By 1987, various pieces of evidence had been put together, and EDRF turned out to be none other than NO (Hutchinson et al. 1987; Ignarro et al. 1987; Palmer et al. 1987), and shortly thereafter, l-arginine was identified as its precursor (Palmer et al. 1988; Stuehr et al. 1989). These seminal discoveries were followed by reports demonstrating the important roles played by NO in the regulation of inflammation, smooth muscle relaxation, and the immune response (Bogdan 2001; Coleman 2001; Rattan 2005).

The first evidence of NO’s activity as a neurotransmitter was reported by Garthwaite et al. (1988) They demonstrated that glutamate stimulation of cerebellar NMDA receptors caused neurons to release a diffusible molecule with features very similar to those of EDRF. In the years that followed, NO’s multiple functions in the CNS were documented, including the roles it played in cognitive function (induction and maintenance of synaptic plasticity) and the control of sleep, appetite, body temperature, and neurosecretion (McCann 1997; Rivier 2001, 2002, 2003; Guix et al. 2005; Calabrese et al. 2007). And in the peripheral nervous system, NO was shown to regulate the non-adrenergic non-cholinergic relaxation of smooth muscle in the corpora cavernosa and the gastrointestinal tract (Van Geldre and Lefebvre 2004; Toda et al. 2005).

Nitric oxide synthesis is catalyzed by the nitric oxide synthase (NOS, EC family of enzymes. In the presence of oxygen and NADPH, these enzymes convert l-arginine to into l-citrulline and NO (Fig. 1). Three NOS isoforms have been identified in the CNS and in peripheral structures: (i) neuronal NOS (nNOS, type I); (ii) endothelial NOS (eNOS; type III); and (iii) inducible NOS (iNOS, type II) (Bredt 1999; Guix et al. 2005). Activation of the different NOS isoforms requires various factors and cofactors. Formation of a calcium/calmodulin complex is necessary for the dimerization of both nNOS and eNOS, and their enzymatic activity also requires cofactors such as tetrahydrobiopterin (BH4), flavins, and thiols (Bredt 1999). Phosphorylation/dephosphorylation processes also play important roles in regulating the activity of these two NOS isoforms. Several kinases and phosphatases are known to increase or decrease NOS activity, and specific phosphorylation sites have also been identified in the NOS sequences (Table 1; Boo and Jo 2003; Zhou and Zhu 2009). Endogenous factors and disease states can also stimulate eNOS and nNOS activity and/or expression, thereby increasing NO production (Table 1) (Chatterjee et al. 2008; Zhou and Zhu 2009). Unlike eNOS and nNOS, iNOS can bind to calmodulin even in the presence of very low intracellular concentrations of calcium, so its ability to generate NO is calcium-independent (Schulz et al. 1992; Xie et al. 1992). In the brain, nNOS has been found in neuronal populations of the cerebral cortex, striatum, hippocampus (CA1 region and dentate gyrus), the lateral dorsal and pedunculopontine tegmental nuclei, the cerebellum (granule cells), and the hypothalamic PVN and SON (Dawson and Snyder 1994). Neuronal NOS immunoreactivity has also been detected in astrocytes, cerebral blood vessels, and the posterior pituitary (Dawson and Snyder 1994; Guix et al. 2005). Endothelial NOS is expressed by vascular endothelial cells in the brain, and it regulates cerebral blood flow, but eNOS immunoreactivity has also been detected in the cerebellum, olfactory bulb, cerebellar cortex, dentate gyrus of the hippocampus, and the bed nucleus of the stria terminalis (Bredt et al. 1990). Inducible NOS levels in the CNS are generally fairly low. However, expression in astrocytes and microglia increases following viral infection or trauma (Dawson and Dawson 1996). Activation of iNOS requires gene transcription, and its induction can be influenced by amyloid-β-peptide, LPS, and proinflammatory cytokines. Activation can also be blocked by glucocorticoids, inhibitory cytokines, prostaglandins (PG), tissue growth factors, and protein synthesis inhibitors (Sharma 1998; Guix et al. 2005) (see also Table 1).

Figure 1.

 Synthesis of NO. In the presence of oxygen, NADPH, and cofactors, such as heme, flavins (FMN, FAD), tetrahydrobiopterin and thiols, nitric oxide synthase (NOS) catalyzes the 5-electron oxidation of the terminal guanidino nitrogen of l-arginine, which results in the formation of l-citrulline and nitric oxide free radical (•NO).

Table 1.   Main factors regulating nitric oxide production through effects exerted on nitric oxide synthase (NOS) isoforms
  1. aUp-regulation (↑)/down-regulation (↓) of NOS gene expression or NOS activity.

  2. bStudies performed in rat cortical and hippocampal neurons.

  3. cThe amino acid position number refers to the sequence of human endothelial NOS.

  4. Aβ, amyloid-β-peptide; CaMK, calcium/calmodulin-dependent protein kinase; CHX, cycloheximide; DEX, dexamethasone; Hsp90, heat shock protein 90; LPS, bacterial lipopolysaccharide; NOSIP, nitric oxide synthase-interactive protein; PG, prostaglandin; PKC, protein kinase C; PP, protein phosphatase; PSD 95, post-synaptic density protein 95; S, serine residue; T, threonine residue; TNF-α, tumor necrosis factor-α.

Bilirubin↑ Expression↑ Expression↑ Expression
Calcium/calmodulin↑ ActivityCalcium/calmodulin↑ ActivityCHX↓ Expression
Dephosphorylation Dephosphorylation DEX↓ Expression
PP1(S847b)↑ ActivityPP2A (S1177c)↓ ActivityIFN-γ↑ Expression
Hsp90↑ ActivityPP1 (T495c)↑ ActivityIL-1↑ Expression
NMDA↑ ActivityHypoxia↑ ExpressionIL-2↑ Expression
NOSIP↓ ActivityLPS↑ ExpressionIL-4↓ Expression
Phosphorylation Phosphorylation IL-10↓ Expression
CaMKII (S847b)↓ ActivityAkt (S615c, S1177c)↑ ActivityPGA2↓ Expression
CaMKI (S741b)↓ ActivityCaMKII (S1177c)↑ ActivityTNF-α↑ Expression
Parkinson’s disease↑ ExpressionPKC (S114c, T495c)↓ Activity  
PSD95↑ ActivityPhysical exercise↑ Expression  
  TNF-α↑ Expression  

Nitric oxide and the stress axis

Immunohistochemical studies have revealed increased levels of nNOS mRNA and protein and enhanced nNOS activity in the PVNs of rats exposed to stressful conditions, such as forced swimming, immobilization, and endotoxin administration (Lee et al. 1995; Kishimoto et al. 1996; Tsuchiya et al. 1997; Kostoglou-Athanassiou et al. 1998; Sanchez et al. 1999). Interestingly, simultaneous exposure to neurogenic stress (e.g., restraint) and immuno-inflammatory stress (e.g., administration of LPS) stimulated the NO system in the rat PVN to a greater extent than either stress alone (Yang et al. 1999). Significant increases in nNOS mRNA and activity have also been described in the anterior pituitary and adrenal gland (Kishimoto et al. 1996) of rats subjected to neurogenic stress.

Despite the strength of these morphological data, NO’s functional contribution to the regulation of stress-axis activity is still questioned. It has, in fact, been found to exert both stimulatory and inhibitory effects on the HPA axis (Fig. 2). Early studies showed that l-arginine, the precursor of NO, and NO donors had no effect on basal release of CRH or AVP in rat hypothalamic explants, but they significantly reduced the secretion of these neuropeptides stimulated by K+ and IL-1β (Costa et al. 1993; Yasin et al. 1993). In contrast, Karanth et al. (1993) found that l-arginine enhanced both the basal and IL-2-stimulated release of CRH from rat mediobasal hypothalami, and this effect was reduced by the non-selective NOS inhibitor l-NG-monomethyl arginine. These contradictory in vitro findings were followed by in vivo studies, which showed that NO’s role in the regulation of the stress axis depends on the brain area involved and the type of stress.

Figure 2.

 Differential regulation of the stress axis by nitric oxide. NO can exert permissive (solid arrows) or preventive (dotted arrows) effects on the hypothalamic–pituitary–adrenal axis, depending on the type of stressor. NO blunts stress-axis activation triggered by blood-borne signals, such as IL-1β or arginine vasopressin (AVP), by reducing the release of corticotrophin-releasing hormone (CRH), adrenocorticotropin hormone (ACTH), and corticosterone. (A stimulatory role for NO on the IL-2-mediated release of CRH from rat mediobasal hypothalamus has been documented in only one study; Karanth et al. 1993). As for the stress-axis activity caused by bacterial lipolysaccharide (LPS) stimulation, NO reduces ACTH release elicited by low-dose LPS exposure and increases that produced by exposure to high-dose LPS. When the stress is neurogenic in origin, NO enhances stress-axis activation by increasing the release of CRH and AVP from the hypothalamus and ACTH from the pituitary gland. NO blocks the release of pituitary ACTH and adrenal corticosterone in response to cholinergic agonists, but it exerts permissive effects on the α2-adrenoceptor-mediated increase of ACTH release. In the paraventricular and supraoptic nuclei of the hypothalamus, NO increases the GABA-mediated post-synaptic currents, thus reducing neuronal excitability and exerting a negative control on the stress axis.

It is well established that blood-borne signals, including circulating IL-1β, LPS, or AVP, act preferably at the pituitary-median eminence level. Rivier and Shen (1994) found that inhibition of NO synthesis in rats significantly increased the plasma levels of ACTH and corticosterone observed after peripheral administration of IL-1β. Similarly, the NOS inhibitor NG-nitro-l-arginine methyl ester (l-NAME) exacerbated the ACTH increase provoked by intravenous administration of AVP, but not CRH (Rivier and Shen 1994). Interestingly, however, the inhibition of NOS activity did not affect the HPA activation induced by the intracerebroventricular (i.c.v.) administration of IL-1β (Rivier and Shen 1994). The effect of bacterial LPS on the HPA axis, which is mediated at least in part by NO, is quite complex. Low-dose LPS administration (0.5 μg/kg) is associated with activation of the stress axis in rats, as reflected by significant increases in plasma ACTH levels, and this effect is potentiated by concomitant peripheral administration of l-NAME. It is worth noting that the effect of l-NAME in this experimental system was counteracted by immunoneutralization of endogenous AVP, thus demonstrating the importance of the NO–AVP interaction (Rivier 2003). High-dose LPS administration (> 50 μg/kg) also increased plasma ACTH, but this effect was inhibited by l-NAME and unrelated to endogenous AVP. Therefore, it is probably caused by marked neuronal activation at the PVN level (Rivier 2003).

Neurogenic stressors, such as sustained physical exercise, mild footshock, or restraint, can modulate HPA axis activity through effects exerted at either the hypothalamic or pituitary level. In rats and pigs, the inhibition of NOS activity decreases the ACTH response to restraint or shock and the up-regulation of CRH and AVP transcription in the hypothalamic PVN, whereas the i.c.v. administration of NO increases CRH and AVP transcript levels in this nucleus (Turnbull et al. 1998; Rivier 2001, 2003; Jankord et al. 2009). In contrast, l-NAME acts at the hypothalamic level to augment the increased ACTH levels caused by physical exercise (Jankord et al. 2009).

The HPA axis is under the control of several neurotransmitters. It has been clearly demonstrated that the stress axis is strongly stimulated by activation of the central cholinergic and adrenergic systems, whereas activation of the GABAergic system by stimulation of the GABAA receptor subtype exerts a tonic negative effect (Grossman et al. 1993; Jessop 1999; Bugajski et al. 2004) (Fig. 2). Increased hypothalamic release of CRH has been observed after cholinergic stimulation in the rat (Calogero et al. 1988, 1989; Grossman et al. 1993; Ohmori et al. 1995). The acetylcholine-receptor (muscarinic and nicotinic) agonists, carbachol and nicotine, increase plasma levels of ACTH and corticosterone in rats, and this effect is potentiated by the NOS inhibitors NG-nitro-l-arginine and l-NAME (Gadek-Michalska and Bugajski 2005; Bugajski et al. 2006). Interestingly, NOS inhibitors exacerbate the carbachol-induced HPA activation in rats concomitantly exposed to the neurogenic stress of crowding (Bugajski et al. 2006). As for the adrenergic system, the blockade of NO synthesis by l-NAME or l-N6-(1-iminoethyl)lysine, which is a selective inhibitor of iNOS reportedly reduces the secretion of ACTH and corticosterone elicited by stimulation of α1, α2, and β-adrenoreceptors in rats (Bugajski et al. 2004; Gadek-Michalska and Bugajski 2008). However, Seo et al. (2003) demonstrated that only the ACTH release stimulated by the α2 subtype is modulated by NO. Regarding the GABAergic system, NO, alone or combined with physicoemotional stress, such as forced swimming, has been found to increased GABA release in the rat SON (Engelmann et al. 2002). In the PVN and SON, this gas has also been shown to increase the frequency of miniature inhibitory GABAergic post-synaptic currents, thus reducing neuronal excitability and contributing to the negative control of HPA axis activity (Stern and Ludwig 2001; Li et al. 2003).

Nitric oxide has been demonstrated to exert most of its physiological functions, including smooth muscle cell relaxation and neurotransmission, via activation of the soluble guanylyl cyclase (sGC; EC GMP pathway (sGC/cGMP) and downstream regulation of protein kinase G or cyclic nucleotide-gated channels (Fig. 3; Calabrese et al. 2007). The importance of the NO/sGC/cGMP system in the modulation of the HPA axis has yet to be defined. NO/sGC/cGMP signaling has been shown to increase the release of gonadotropin-releasing hormone (GnRH) from the rat hypothalamus (Canteros et al. 1995), but activation of this pathway does not appear to be necessary for the hypothalamic release of CRH or AVP (Terrell et al. 2003; Kadekaro 2004). The latter conclusion was recently challenged by the results of a double-blind, placebo-controlled, cross-over clinical trial, in which healthy male athletes were treated with tadalafil, a long-acting type-V phosphodiesterase inhibitor that increases cGMP concentrations. Exercise-induced activation of the HPA axis, assessed by measurement of salivary cortisol concentrations, was significantly potentiated by tadalafil (Di Luigi et al. 2008). Induction of cyclooxygenase (COX, EC and the increased PG levels it produces play major roles in HPA axis activation (Fig. 3). Inhibition of PG synthesis has been shown to inhibit the stimulatory effect of IL-1β on ACTH release in rats and its potentiation by l-NAME (Rivier 1998). Prostaglandin blockade also counteracts the l-NAME- and NG-nitro-L-arginine-induced increases in ACTH and corticosterone release triggered by nicotine administration in rats (Bugajski et al. 2004).

Figure 3.

 The main signal transduction pathways involved in the biological activities of NO, CO, and H2S. NO and CO – formed by nitric oxide synthase (NOS) and heme oxygenase (HO) activities, respectively (see Fig. 4) – interact primarily with hemoproteins and activate soluble guanylyl cyclase (sGC) and cyclooxygenase (COX) (solid arrow). By modulating the activity of these enzymes, NO and CO regulate neurotransmission, smooth muscle relaxation, cell survival, and neuropeptide release. H2S, which is formed by cystathionine-β-synthase (CBS) and cystathionine-γ-lyase (CSE) activities (see Fig. 6), stimulates adenylyl cyclase (AC), mitogen-activated protein kinases (MAPK), ATP-sensitive potassium channels, and the transcription factor NF-E2-related factor-2 (Nrf2), thereby controlling neuronal activity, vascular smooth muscle relaxation, cell proliferation, and the oxidant/antioxidant balance (solid arrow). The three gases can also interact with one another in numerous and complex ways. Theoretically speaking, both NO and CO could inhibit CBS activity, thus reducing H2S formation through this pathway, but a careful analysis of the binding constants suggests that only CO plays a role in such regulatory loop (for further information see text). Moreover, H2S scavenges NO, transforming it into an inactive S-nitrosothiol (dotted arrow). H2S also stimulates HO activity and increases CO formation by activating Nrf2 (solid arrow), a well-known activator of HO-1. In this case, the CO-mediated inhibition of CBS mentioned above represents a negative feed-back mechanism designed to avoid excessive stimulation of CO formation by H2S.

Figure 6.

 The cysteine metabolism pathways leading to H2S formation. Cystathionine-β-synthase (CBS) and cystathionine-γ-lyase (CSE) synthesize H2S from l-cysteine. The two enzymes have to act in concert to generate H2S in many tissues (right side of the diagram); in others only one of the two is necessary (left side of the diagram). It is noteworthy to mention that formation of H2S is an alternative pathway for CBS, the main one leading to the formation of glutathione.

Carbon monoxide

Carbon monoxide is one of the most toxic molecules in nature. It is formed during the combustion of organic material, including wood or coal, under low oxygen tension, and it binds irreversibly to oxyhemoglobin, thereby preventing oxygen delivery to tissues (Wu and Wang 2005). As a result of its lethal nature, CO has been extensively studied over the last 50 years, mainly from a toxicological point of view. However, in 1963, endogenous production of CO in man was documented (Coburn et al. 1963) although the biochemical pathway leading to its formation and its physiologic importance remained obscure. Later, Tenhunen et al. (1969) described heme oxygenase (HO, EC, a microsomal enzyme that catalyzes the oxidation of the alpha-meso-carbon bridge of heme moieties, generating equimolar amounts of CO, ferrous iron, and biliverdin (Fig. 4). Heme oxygenase is functionally coupled to the cytosolic enzyme biliverdin reductase (EC, which reduces biliverdin to bilirubin, an endogenous molecule whose interactions with free radicals in the intracellular milieu have been studied extensively (Mancuso et al. 2003, 2006b,c, 2008; Stocker 2004; Barone et al. 2009) (Fig. 4). The physiologic importance of CO was finally revealed by Verma et al. (1993) who demonstrated the role played by this gas in synaptic plasticity. These seminal observations were followed by several papers demonstrating the importance of CO as an endogenous neuromodulator in both the central and peripheral nervous systems. In particular, CO (along with NO) was shown to be involved in the regulation of hippocampal long-term potentiation, non-adrenergic non-cholinergic smooth muscle relaxation, inflammation, and apoptosis (Wu and Wang 2005).

Figure 4.

 The heme degradation pathway. In the presence of oxygen and NADPH, heme moieties derived from hemoproteins are oxidatively cleaved by heme oxygenase (HO) at the α-meso-carbon bridge. The result is equimolar amounts of CO, FeII, and the linear tetrapyrrole biliverdin. Biliverdin then undergoes reductive modification at the C10 (γ bridge) by the NAD(P)H-requiring enzyme biliverdin reductase (BVR), which yields bilirubin as the final product.

Given the ubiquitous distribution of the inducible and constitutive isoforms of HO (HO-1 and HO-2, respectively), the generation of CO can reasonably be expected to occur in all organs. As far as the nervous system is concerned, neurons are characterized by very high HO activity under basal conditions. Most of this activity is accounted for by HO-2, which is expressed in neuronal populations of the forebrain, hippocampus, hypothalamus, midbrain, basal ganglia, thalamus, cerebellum, and brainstem (Maines 1997). Heme oxygenase-1 is present in much smaller amounts and is localized in scattered groups of neurons, including the ventromedial nucleus and PVN of the hypothalamus (Maines 1997). Therefore, it seems that activation of HO-1 and HO-2 resulting in the formation of CO can be induced by numerous noxious stimuli (Table 2) within nuclei that play primary roles in the central regulation of the stress response (Mancuso 2004). More recently, HO-1 has been found in GT1-7 cells, a hypothalamic neuronal cell line endowed with neuroendocrine activity in vitro (Mores et al. 2008). HO-1 has also been detected in glial cells, where its expression can be induced by oxidative stress (Dwyer et al. 1995).

Table 2.   Factors that increase carbon monoxide production by enhancing the expression or activity of heme oxygenase (HO) isoforms
  1. aUp-regulation (↑) of HO gene expression or HO activity.

  2. Aβ, amyloid-β-peptide; LPS, bacterial lipopolysaccharide; RNS, reactive nitrogen species; ROS, reactive oxygen species.

↑ ExpressionCalcium/calmodulin↑ Activity
Heat shock↑ ExpressionGlucocorticoids↑ Expression
Heme↑ ExpressionOpiates↑ Expression
LPS↑ Expression  
Nitric oxide↑ Expression  
ROS↑ Expression  
RNS↑ Expression  
Sodium arsenite↑ Expression  

Carbon monoxide and the stress axis

The first study to address the potential impact of CO on the HPA axis was conducted by Pozzoli et al. (1994). Hemin, the precursor of CO in the HO system, was found to have no effect on the basal release of CRH in rat hypothalamic explants, but it significantly reduced that stimulated by K+ or by IL-1β. This effect was attributed to CO, rather than to hemin itself, because it was significantly diminished by the HO inhibitor zinc-protoporphyrin-IX (Pozzoli et al. 1994) (Fig. 5). CO’s major role in the regulation of the stress axis was later corroborated by studies showing that (i) hemin is transformed into biliverdin and CO in the rat hypothalamus under the same experimental conditions that inhibit stimulated CRH release, and (ii) the increased CO formation is responsible for the inhibition of K+-stimulated AVP and oxytocin release in rat hypothalamic explants in vitro (Fig. 5) (Kostoglou-Athanassiou et al. 1996, 1998; Mancuso et al. 1997a). The latter effect was indeed mediated by CO formation because it was counteracted by HO inhibitors and reproduced by direct incubation of hypothalami in CO-saturated medium (Kostoglou-Athanassiou et al. 1996, 1998; Mancuso et al. 1997a). Importantly, neither biliverdin nor bilirubin had any effect on basal or stimulated release of CRH, AVP, or oxytocin (Pozzoli et al. 1994; Kostoglou-Athanassiou et al. 1996; Mancuso et al. 1997a). These in vitro findings were later confirmed by the results of in vivo studies, which are much more difficult to carry out owing to the relative inability of metalloporphyrins, including hemin, to cross the blood–brain barrier at pharmacological doses. To overcome this limitation, the metalloporphyrins were administered to rats by the i.c.v. route. Under these conditions, tin-protoporphyrin-IX, a well-known inhibitor of HO activity and CO synthesis, significantly potentiated the LPS-induced increase in circulating AVP levels while reducing the hypothalamic content of this neuropeptide (Mancuso et al. 1999). The results of these studies are consistent with the view that HO products (probably CO) exert inhibitory effects on the stimulated release of CRH, AVP, and oxytocin by specific hypothalamic neurons.

Figure 5.

 Role of CO in the regulation of stress axis. CO inhibits IL-1β- and K+-stimulated release of corticotrophin-releasing hormone (CRH), arginine vasopressin (AVP) and oxytocin (OXY) (dotted arrow) from the rat hypothalamus. It has also been proposed as an important mediator of the pyrogenic response to neurogenic and immuno-inflammatory stressors (solid arrow).

Although it is only indirectly related to the stress axis, CO’s effect on the release of GnRH is worth mentioning. In in vitro studies, hemin caused dose-dependent increases in basal GnRH release in both hypothalamic explants and immortalized cells, and these effects were specifically inhibited by zinc-protoporphyrin-IX (Brann et al. 1997) in the rat. Although the mechanism of CO’s action on GnRH release has not been fully investigated, it is reasonable to hypothesize that PGs are involved, since CO has been shown to increase prostaglandin (PG) E2 production in the rat hypothalamus (Mancuso et al. 1997b, 1998), and this prostanoid is known to be a major physiological stimulus to GnRH release (Rivest and Rivier 1995; Mancuso et al. 1997c). These results highlight CO’s opposite effects on GnRH and CRH release. It is well established that stress and reproductive function are inversely related. For example, IL-1β, a well-known mediator of inflammatory stress, increases CRH and neurohypophyseal hormone release while blunting GnRH secretion via a direct action at the hypothalamic level (Rivest and Rivier 1995). Within this framework, CO’s role appears to be physiologically consistent: it prevents excessive HPA axis activation and, at the same time, enhances reproductive processes.

A peculiar manifestation of HPA axis activation is an increase in body temperature. Rats exposed to a variety of neurogenic stress stimuli, such as restraint, handling, or cage switching, exhibit increases in body temperature referred to as stress fever (Oka et al. 2001). Papers by Luiz Branco’s group have clearly demonstrated that CO is a mediator of this pyrogenic response (Fig. 5). Additional support for this hypothesis emerged from studies by Sanchez et al. (1999) and Steiner and Branco (2000, 2001), who found that HO inhibition decreases LPS-induced fever in rats, whereas heme overload triggers increases in body temperature. More recently, these investigators studied the relationship between the HO–CO system and physicoemotional stress in rats (Steiner et al. 2003). They found that i.c.v. administration of the HO inhibitor zinc-deuteroporphyrin-bis-glycol had no effect on the body temperature of euthermic rats, but it markedly attenuated restraint-induced fever (Steiner et al. 2003). These results confirmed that the HO–CO pathway is not involved in the tonic regulation of body temperature, but when HO is activated by a specific stimulus CO becomes a pyrogenic neuromodulator.

Early studies showed that CO exerts its biological effects in the nervous system by activating cytosolic sGC and thereby increasing intracellular cGMP levels (Maines 1997; Steiner et al. 2002) (Fig. 3). This paradigm has been challenged by several reports, which demonstrated that CO is only a weak activator of sGC. Indeed, it causes a much less substantial increase (up to fourfold) in cGMP production (Burstyn et al. 1995; Kharitonov et al. 1995; Ma et al. 2007) than that elicited by NO (100-fold) (Kharitonov et al. 1995; Ma et al. 2007). These findings and those from previous studies on NO/COX interaction suggested that the neuroendocrine effects of CO might be ascribed to its regulation of COX, an enzyme that is abundantly expressed throughout the brain, including the hypothalamus (Breder et al. 1992, 1995). Studies by our group have demonstrated close correlation between CO levels, which influence the release of hypothalamic neuropeptides (CRH, AVP and oxytocin), and COX activity (Mancuso et al. 2006a, 2007). In fact, hemin increased COX activity in rat hypothalamic explants and in primary cultures of rat hypothalamic astrocytes. The increase in PGE2 formation was inhibited by the specific HO inhibitor tin-mesoporphyrin-IX and reversed by the CO scavenger hemoglobin, which shows that it was indeed the result of CO generation stemming from the HO-mediated degradation of hemin (Mancuso et al. 1997b). Rat hypothalamic explants incubated in CO-saturated medium also exhibit significant increases in PGE2 release (Mancuso et al. 1998). The role of PGs, in particular PGE2, as endogenous intermediates in the cytokine-stimulated activation of the HPA is well established (Rivest and Rivier 1995). Therefore, these experiments provided early evidence of the role of COX/PG pathway in the CO-mediated activation of the stress axis in the rat.

Hydrogen sulfide

Hydrogen sulfide (H2S) is the third member of the family of gaseous neurotransmitters. Nitric oxide and CO are odorless gases and can reach toxic concentrations in the environment without being detected. In contrast, H2S has a typical odor that is often compared with that of rotten eggs. The human nose can detect levels of H2S that are 400 times lower than those found to be toxic (Wang 2002), so H2S intoxication is highly unlikely.

The first evidence of endogenous H2S production in the brain emerged in 1989 from studies of rats (where the gas was found at concentrations close to 100 μM) and of normal human postmortem samples (Goodwin et al. 1989; Warenycia et al. 1989; Li and Moore 2008). However, these figures have recently been challenged by Furne et al. (2008), who reported endogenous H2S levels in the mouse brain and liver of around 15 nM. Later, endogenous production of H2S was also demonstrated in vascular tissue (Hosoki et al. 1997; Zhao et al. 2001).

For the most part, H2S is synthesized from l-cysteine by cystathionine-β-synthase (CBS, EC and/or cystathionine-γ-lyase (CSE, EC (Wang 2002). In many tissues, the two enzymes act in concert to generate H2S; in others only one of the two is necessary (Wang 2002). Like nNOS, eNOS, and HO-2 activities, CBS activity in the brain is calcium/calmodulin-dependent (Dominy and Stipanuk 2004; Li and Moore 2008) (Table 3). The presence of a heme prosthetic group makes CBS sensitive to redox changes, and this explains why it is inactivated in the presence of increased intracellular levels of hydrogen peroxide, NO, and CO (Table 3) (Taoka and Banerjee 2001; Maclean et al. 2002; Łowicka and Bełtowski 2007). A careful analysis of the binding constants reveals that CBS binds CO ∼200-fold more tightly than it does NO (Taoka and Banerjee 2001). Furthermore, the dissociation constant of CBS for NO is significantly higher than for sGC (∼300 μM vs. 0.25 μM) (Stone and Marletta 1996; Taoka and Banerjee 2001). In this light, binding of NO to CBS is unlikely to occur under patho-physiological circumstances in which NO levels are not high enough to affect this enzyme. Conversely, the more favorable kinetics makes CO the endogenous gas able to regulate H2S synthesis through CBS activity (Taoka and Banerjee 2001). The expression of CSE can be induced by inflammatory stimuli, such as LPS (Table 3) (Li et al. 2005; Li and Moore 2008). Interestingly, both CBS and CSE are active in the brain, although the former is considered to be responsible for most of the H2S production (Awata et al. 1995; Abe and Kimura 1996; Li and Moore 2008). The expression of CBS mRNA has been demonstrated in various areas of the rat brain, including the hippocampus, cerebellum, cerebral cortex, and brainstem (Abe and Kimura 1996; Li and Moore 2008). Hydrogen sulfide can be also formed non-enzymatically, but this alternative route, which involves the reduction of elemental sulfur by reducing equivalents formed during glucose oxidation, is far less important (Searcy and Lee 1998; Wang 2002).

Table 3.   Major regulators of hydrogen sulfide production through both cystathionine-β-synthase (CBS) and cystathionine-γ-lyase (CSE) expression/activity
  1. aUp-regulation (↑) or down-regulation (↓) of CBS/CSE gene expression or activity.

  2. bBecause of binding parameters, CO rather than NO is view as the endogenous gaseous compound involved in the pathophysiologic regulation of CBS activity (for further information see text).

  3. DM-2, type-2 diabetes mellitus; LPS, bacterial lipopolysaccharide; SAME, S-adenosyl methionine; TNF-α, tumor necrosis factor-α.

Calcium/calmodulin↑ ActivityDM-2↑ Expression
Carbon monoxideb↓ ActivityLPS↑ Expression
Glucagon↑ ExpressionPancreatitis↑ Expression
Glucocorticoids↑ Expression  
Hydrogen peroxide↓ Expression  
Insulin↓ Expression  
Nitric oxideb↓ Activity  
SAM↑ Activity  
TNF-α↑ Activity  

In the nervous systems of rodents and humans, H2S has been implicated in the regulation of many functions, including the induction of hippocampal long-term potentiation, nociception, neuronal hyperpolarization (particularly in the CA1 region of the hippocampus and the dorsal raphe nucleus), stimulation of enteric neurons and the related prosecretory effects, and smooth muscle relaxation (Reiffenstein et al. 1992; Wang 2002; Schicho et al. 2006; Kawabata et al. 2007; Li and Moore 2008). Numerous interactions between H2S and intracellular signaling pathways have been documented. First of all, the gas has been shown to be a direct activator of adenylyl cyclase, ATP-sensitive potassium channels, the mitogen-activated protein kinase system, and the transcription factor NF-E2-related factor-2, and, through these pathways, it regulates neuronal activity, vascular smooth muscle relaxation, cell proliferation, and the oxidant/antioxidant balance, respectively (Kimura 2000; Zhao and Wang 2002; Yang et al. 2004a,b; Tang et al. 2005; Li and Moore 2008; Calvert et al. 2009). Other biological actions of H2S depend on its interaction with other gaseous compounds. For example, H2S binding to NO results in the formation of an inactive nitrosothiol and thereby reduces NO-dependent signaling (Li and Moore 2007). It can also activate sGC or COX by up-regulating HO-1 activity and increasing CO formation (Fig. 3).

Hydrogen sulfide and the stress axis

Hydrogen sulfide’s interactions with the HPA are not as thoroughly documented as those of NO and CO. The incubation of rat hypothalamic explants with NaHS, which releases H2S in solution, is known to reduce the K+-stimulated release of CRH, and a similar effect can be obtained by treating hypothalamic tissue with S-adenosyl-methionine, a precursor of the H2S formed through CBS activity (Dello Russo et al. 2000; Navarra et al. 2000). This in vitro evidence has been further supported by the results of an in vivo study of rats treated with S-adenosyl-methionine and then exposed to physical stress. S-adenosyl-methionine significantly reduced the plasma corticosterone levels in rats exposed to cold (4°C) for 1 h, but it had no effect on basal glucocorticoid release (Navarra et al. 2000).


The importance of NO, CO, and H2S in the regulation of the stress axis is beyond dispute, but their specific roles include complex interactions with one another and effects that vary with the characteristics of the stress and type of cell/organisms involved. Nitric oxide, CO, and H2S do not seem to be involved in the tonic regulation of the stress axis, but they can reduce its excessive activation by different stimuli. Particularly important is the ability of both NO and CO to reduce the release of hypophysiotropic CRH and AVP triggered by exposure to immuno-inflammatory stressors, such as LPS or IL-1β. This highlights the major role played by these gases in preventing the potentially detrimental effects of excess GC production during prolonged inflammatory states or infectious diseases (Costa et al. 1996). However, in the presence of physicoemotional stress, NO and CO may act in concert to stimulate the stress axis, by influencing PVN activity (Turnbull et al. 1998). By reducing the HPA activation triggered by other forms of neurogenic stress, such as hypothermia, H2S can balance the stimulatory effects of both NO and CO. Finally, although it has been demonstrated only in vascular tissues (Zhao et al. 2001), NO’s ability to induce endogenous H2S production could also be important in the nervous system, and this may represent an additional means by which the gaseous neurotransmitters modulate the activity of the HPA axis.


The authors gratefully acknowledge the contributions of Drs Eugenio Barone and Raffaella Siciliano in creating the figures. The manuscript was edited by Marian Everett Kent, and the study was supported by grants from the Catholic University (Fondi Ateneo 2008 and 2009) to CM. The authors report no conflict of interest in the presented work.