Abstract
The carotid body (CB) is a neural crest-derived organ whose major function is to sense changes in arterial oxygen tension to elicit hyperventilation in hypoxia. The CB is composed of clusters of neuron-like glomus, or type-I, cells enveloped by glia-like sustentacular, or type-II, cells. Responsiveness of CB to acute hypoxia relies on the inhibition of O2-sensitive K+ channels in glomus cells, which leads to cell depolarisation, Ca2+ entry and release of transmitters that activate afferent nerve fibres. Although this model of O2 sensing is generally accepted, the molecular mechanisms underlying K+ channel modulation by O2 tension are unknown. Among the putative hypoxia-sensing mechanisms there are: the production of oxygen radicals, either in mitochondria or reduced nicotinamide adenine dinucleotide phosphate oxidases; metabolic mitochondrial inhibition and decrease of intracellular ATP; disruption of the prolylhydroxylase/hypoxia inducible factor pathway; or decrease of carbon monoxide production by haemoxygenase-2. In chronic hypoxia, the CB grows with increasing glomus cell number. The current authors have identified, in the CB, neural stem cells, which can differentiate into glomus cells. Cell fate experiments suggest that the CB progenitors are the glia-like sustentacular cells. The CB appears to be involved in the pathophysiology of several prevalent human diseases.
SERIES “HYPOXIA: ERS LUNG SCIENCE CONFERENCE”
Edited by N. Weissmann
Number 5 in this Series
The carotid body (CB), a small neural crest-derived paired organ located at the carotid bifurcation (fig. 1a⇓), is a principal component of the homeostatic acute oxygen-sensing system required to activate the brainstem respiratory centre to produce hyperventilation during hypoxaemia (e.g. in high-altitude residents or in patients with chronic obstructive pulmonary diseases) 2–4. The CB is one of the most irrigated organs in the body and receives blood through a branch arising from the external carotid artery. The CB parenchyma is organised into glomeruli: clusters of cells, in close contact with a profuse network of capillaries, and afferent sensory fibres joining the glossopharyngeal nerve (fig. 1a⇓ and b). The most abundant cell types in the CB glomeruli are the neuron-like glomus, or type-I, cells, which are enveloped by processes of glia-like, sustentacular type-II cells (fig. 1c⇓). The CB also contains some autonomic neurons and fibres, which seem to have an efferent regulatory action on glomus cells 5.
Anatomical organisation of the carotid body (CB). a) Histological section of the mouse carotid bifurcation immunostained with an antibody against tyrosine hydroxylase (TH; brown colour). The clusters (glomeruli) of TH+ cells forming the CB can be easily identified. b) Schematic representation of a CB glomerulus with indication of type-I and type-II cells, blood vessels (V) and afferent sensory nerve fibres (NF). c) Appearance of a rat TH-positive type-I glomus cell. d) Appearance of a rat glial fibrillary acidic protein (GFAP)-positive type-II cell with processes surrounding GFAP-negative cells. Cell nuclei are stained with 4′,6-diamidino-2-phenylindole. ICA: internal carotid artery. Scale bars = 100 μm (a) and 5 μm (c and d). Modified from 1.
Glomus cells are physiologically complex, as they express a broad variety of voltage- and ligand-gated ion channels, as well as transient receptor potential and background K+ channels. They contain secretory vesicles packed with neurotransmitters, notably ATP, dopamine and acetylcholine, among others 6. Voltage-gated ion channels have been studied in detail in patch-clamped glomus cells from several species 2. Macroscopic ionic currents recorded from these cells are composed of outward (mediated by several classes of K+ channels) and inward (mediated by Na+ and/or Ca2+ channels) components (as discussed hereunder). Quantitatively, the proportion of the different subtypes of K+, Na+ and Ca2+ channels expressed in glomus cells greatly varies among the mammalian species studied. Owing to the presence of voltage-gated membrane channels, glomus cells are electrically excitable and can repetitively generate action potentials. This property is particularly evident in rabbit glomus cells, with relatively large voltage-dependent Na+ currents 7. Glomus cell membrane depolarisation induces a reversible neurosecretory response, dependent on extracellular Ca2+ influx, which can be easily monitored by amperometry 8, 9. Thus, glomus cells behave as presynaptic-like elements that establish contact with the postsynaptic sensory nerve fibres.
The precise functional significance of the numerous neurotransmitters that exist in the CB is still under debate. The CB is among the most dopaminergic structures in the body and, as extracellular dopamine inhibits the Ca2+ channels in glomus cells, it has been suggested that this transmitter has an autocrine role 10. In contrast, ATP and, possibly, acetylcholine appear to be the major active neurotransmitters at the glomus cell-afferent fibre synapse 6, 11. There are other amines and several neuropeptides in the CB whose functional significance is, as yet, not well known. Glomus cells also have high levels of neurotrophic factors, which seem to exert a local autocrine and paracrine action 12. Among these factors, the glia cell line-derived neurotrophic factor (GDNF) has attracted particular attention because it is highly expressed in adult glomus cells 13–15. As GDNF can promote the survival of dopaminergic neurons, CB transplants have been used for intrastriatal delivery of dopamine and GDNF in parkinsonian animal models and in some pilot clinical studies on Parkinson's disease patients (see section Carotid body function and mechanisms of disease).
Of the cells in the CB parenchyma, ∼15–20% are type-II cells, which in vivo exhibit long processes surrounding type-I cells (fig. 1b⇑ and d). Type-II cells are nonexcitable and lack most of the voltage-gated channels characteristic of type-I cells 7, 16. The molecular interactions between type-I and type-II cells, possibly critical for the physiology of the organ, are basically unknown. Classically, type-II cells were considered to belong to the peripheral glia with a supportive role. However, recent experimental data have shown that the adult CB is a functionally active germinal niche. In this regard, it has been strongly suggested that type-II cells are indeed dormant stem cells that in response to physiological hypoxia can proliferate and differentiate into new glomus cells (see section Carotid body plasticity in chronic hypoxia: adult carotid body stem cells) 1.
RESPONSES OF GLOMUS CELLS TO ACUTE HYPOXIA: MODEL OF CAROTID BODY O2 SENSING
Glomus cells are polymodal arterial chemoreceptors, activated not only by hypoxia but also by other stimuli, most notably hypercapnia, extracellular acidosis and hypoglycaemia 2, 17. It is, however, the sensitivity to acute changes of O2 tension what makes the CB essential for the classical adaptive hyperventilatory reflex in response to hypoxaemia. Although the function of the CB as an acute oxygen sensor has been known since the first half of the 20th century, it was during the past 15–20 yrs that the basic cellular events underlying this physiological process were unveiled. It is now broadly accepted that glomus cells are the chemoreceptive elements in the CB and that they contain several classes of O2-sensitive K+ channels whose open probability decrease during hypoxia 2–4. Voltage-dependent K+ channels were initially reported to be O2 sensitive in rabbit CB cells 18, but other K+ channel types (particularly Ca2+-dependent maxi-K+ and twin pore acid-stimulated K+ channel-like background channels) have been also found to be modulated by O2 in several CB preparations 19, 20. Inhibition of the K+ channels leads to glomus cell membrane depolarisation and increase in the firing frequency of the cells, thus resulting in Ca2+ channel opening, transmembrane Ca2+ influx and transmitter release. The major steps in the chemotransduction process are: hypoxic inhibition of the K+ currents (fig. 2a⇓) and inhibition of single K+ channel activity (fig. 2b⇓) 18, 21, external Ca2+-dependent increase of cytosolic Ca2+ in hypoxia (fig. 2c⇓) 8, 23, and catecholamine release from hypoxic glomus cells (fig. 2d⇓) 8, 9. The dose-dependent cellular responses to hypoxia (increase of cytosolic Ca2+ concentration and catecholamine release) almost perfectly match the characteristic hyperbolic correlation between arterial O2 tension and the afferent discharges of the CB sinus nerve or the increase in ventilation seen in vivo. Within the context of this broadly accepted “membrane model” of CB O2 sensing, schematically summarised in figure 3⇓, it is worth remarking that, although the participation of mitochondria in CB O2 sensing is under debate (see section Mechanisms of carotid body acute O2 sensing), the experimental data available unequivocally indicate that they do not contribute to the hypoxia-induced rise of cytosolic Ca2+ concentration necessary to trigger glomus cell secretion.
Responses of glomus cells to hypoxia. a) Ionic currents recorded from a patch-clamped rabbit glomus cell during a depolarising pulse from -80 mV to +20 mV. Inward calcium (ICa) and outward potassium (IK) currents are indicated. Voltage-gated sodium channels were blocked with tetrodotoxin. Note the selective reversible inhibition of the K+ current by exposure to hypoxia (switching from an O2 tension 150 to ∼15 mmHg). Modified from 9. b) Single K+ channel activity recorded in an excised membrane patch exposed to normoxic and hypoxic solutions. Note that lowering O2 tension reduces the single channel open probability without affecting the single channel current amplitude. Modified from 21. c) Changes in cytosolic calcium concentration, [Ca2+], in rabbit glomus cells as a function of oxygen tension. The inset shows that hypoxia (H) induces a rise of cytosolic Ca2+ concentration, which is strictly dependent on extracellular Ca2+ influx. Modified from 8, 9. d) Catecholamine secretion from rabbit glomus cells as a function of oxygen tension. The inset illustrates the secretory response of a single cell to hypoxia as monitored by amperometry. Modified from 22.
Membrane model of glomus cell oxygen sensing. The steps in chemosensory transduction are as follows. 1) Decrease of O2 tension (PO2), 2) O2 sensing, 3) closure of potassium channels, 4) cell depolarisation, 5) opening of calcium channels, 6) increase of cytosolic calcium concentration, [Ca2+], 7) transmitter release and 8) activation of afferent fibres, which send the information to the central nervous system (CNS). Although these steps in chemosensory transduction have broad experimental support, the nature of the O2 sensor and the mechanisms by which changes in O2 tension regulate K+ channel activity are still unknown (question mark). ΔVm: change in membrane voltage. Modified from 2, 24.
The membrane model of CB oxygen sensing discussed in the preceding paragraph has been generalised to other neurosecretory or contractile cells acutely responding to hypoxia. Among those are neonatal adrenomedullary chromaffin cells 25–27, cells in the neuroepithelial bodies of the lung 28 or PC12 cells 29, as well as pulmonary arterial myocytes 30, 31. These cells belong to the homeostatic acute O2-sensing system that allows mammalian fast adaptation to hypoxic environments 4.
MECHANISMS OF CAROTID BODY ACUTE O2 SENSING
Despite progress in CB cellular physiology, the molecular mechanisms underlying glomus cell O2 sensing, i.e. how the change in O2 tension is translated into decrease of K+ conductance, remain essentially unknown. Several possible O2-sensing mechanisms, including the direct interaction of O2 with the ion channels or their indirect modulation through O2-sensing molecules, have been postulated 2, 24. Since numerous K+ channel types have been reported to be O2 sensitive, it is assumed that there could be several O2 sensors coexisting in the same cell or distributed among the different O2-sensitive cell types, even in closely related animal species. Understanding CB O2 sensing at the molecular level is, however, challenged by the small size of the organ, which precludes elaborated biochemical and molecular biology experiments, and the gaseous nature of the detected molecule, which is easily diffusible across cell membranes and difficult to keep under strict control in the open chambers normally used for in vitro studies. In addition, O2 responsiveness of isolated CB glomus cells is often lost, because of damage during the vigorous enzymatic and mechanical treatment needed for their dispersion. Finally, some possible O2-sensing mechanisms have been inferred from pharmacological experiments using compounds (as, for example, mitochondrial inhibitors) that might have nonspecific effects 32, 33, hence providing misleading conclusions. The mechanisms of CB O2 sensing are summarised in the following sections, emphasising the proposals that are currently under debate and the knowledge generated by the current authors’ experimental work.
Redox-based O2 sensor: reduced nicotinamide adenine dinucleotide phosphate oxidase
A plausible form of CB O2 sensing is the conversion of O2 into reactive oxygen species (ROS), which would in turn alter the redox status of signalling molecules and the function of membrane ion channels. The two ROS-producing sites postulated as O2 sensors are the reduced nicotinamide adenine dinucleotide phosphate (NADPH) oxidase and mitochondria systems.
NADPH oxidase is found in neutrophils and histochemically localised in the CB, although its presence in the chemosensitive glomus cells is not well documented. This enzyme has been proposed to transduce O2 levels by changing the rate of superoxide anion production, which, after conversion to hydrogen peroxide, oxidises ion channels and other molecules. The neutrophil oxidase is an oligomer composed of the membrane-bound catalytic complex (formed by gp91phox and p22phox), a cytochrome, and several cytosolic regulatory subunits (p47phox and others). Although impaired O2 sensitivity of airway chemoreceptor cells has been reported in gp91phox-null mutant mice 34, hypoxia responsiveness of CB and other cells appears to be unaltered 35, 36. Moreover, in patch-clamped glomus cells from these animals the modulation of the O2-sensitive K+ current by O2 tension is unchanged 37. Surprisingly, genetic suppression of another component of the neutrophil's oxidase (p47phox) results in mice with increased basal activity in the carotid sinus nerve and exacerbated ventilatory response to hypoxia 38. This phenotype suggests nonspecific modifications in the p47phox knockout mouse rather than the selective alteration of the O2-sensing machinery in the CB cells. Altogether, these studies indicate that the phagocytic NADPH oxidase is not directly involved in CB O2 sensing, although it is conceivable that other isoforms, existing in numerous tissues 39, could contribute to the hypoxia responsiveness of CB cells. The entire concept of redox-based O2 sensing in glomus cells is, however, challenged by the finding that the reduced/oxidised glutathione ratio in CBs remains unchanged during exposure to hypoxia, despite the fact that this quotient increases after incubation of CBs with N-acetylcysteine, a precursor to reduced glutathione and an ROS scavenger 40.
Mitochondrial dysfunction
Several investigators have traditionally considered mitochondria to be the site of O2 sensing in glomus cells because, similar to hypoxia, inhibitors of the electron transport chain (ETC) or mitochondrial uncouplers increase the afferent activity of the CB sinus nerve 41. This proposal was complemented by reports that hypoxia and cyanide (an inhibitor of mitochondrial complex IV) lead to Ca2+ release from mitochondria in dispersed glomus cells 42. As indicated in the previous section, the mitochondrial hypothesis of CB O2 sensing has lost much support after the discovery of O2-regulated K+ channels and experimental demonstration that the Ca2+ ions needed for glomus cell secretion in hypoxia enter the cell via plasmalemmal voltage-gated Ca2+ channels 8, 9, 23. The interest in mitochondria has, however, resurged more recently because mitochondria uncouplers raise cytosolic Ca2+ and reduce background K+ permeability in glomus cells 43, 44. So, it could be that in hypoxia, mitochondria generate signals that alter membrane ionic conductances (e.g. through modification of the cell redox status or via reduction of cytosolic ATP). In fact, it has been proposed that the redox modulation of membrane K+ channels is the reason for the O2 sensitivity of other acutely responding cells 45. Conversely, decrease of intracellular ATP in hypoxia could result in either the direct closure of ATP-regulated background K+ channels 46 or the increase in AMP/ATP ratio leading to AMP kinase activation. AMP kinase could, in turn, modulate membrane ion channels thus eliciting cell depolarisation 47. In favour of this hypothesis is the existence of O2-sensitive background K+ channels, which appear to be modulated by mitochondrial uncouplers and ATP. In addition, mRNA of AMP kinase is detected in glomus cells (unpublished data of the current authors). In contrast with these observations, the current authors have shown that in the presence of saturating concentrations of mitochondria ETC inhibitors acting at different complexes (I, II, III and IV), hypoxia can still activate transmitter release from glomus cells, thus suggesting that mitochondrial inhibition and hypoxia might activate glomus cells through separate pathways 32. Moreover, patch-clamped glomus cells loaded with a high concentration of Mg-ATP (3–5 mM) still respond to hypoxia 8, 18. Interestingly, the current authors have also observed that rotenone, but no other agents inhibiting complex I at different sites, can block hypoxia responsiveness of glomus cells, thus suggesting that a rotenone-binding site participates in O2 sensing 32. This effect of rotenone seems to be quite specific, as glomus cell responsiveness to hypoglucaemia is unaffected by rotenone 48. However, the pharmacological data must be interpreted with caution, as it is highly likely that at the concentrations used, ETC inhibitors have nonspecific effects on the voltage-gated ion channels 33. It is known, for example, that rotenone can reversibly inhibit K+ currents in cells devoid of mitochondria 49.
Mitochondria have also been associated with CB O2 sensing because mutations in the mitochondrial complex II (particularly in the small membrane-anchoring subunit of succinate dehydrogenase (SDHD)) are the main cause of familiar hereditary paraganglioma (PGL), a highly vascularised and often catecholamine-secreting CB tumour 50. As PGLs display cellular hyperplasia/anaplasia similar to the CB of individuals exposed to chronic hypoxaemia 51, 52, it has been proposed that the ultimate cause of tumorigenesis is a defect in sensing environmental O2 levels 50, 53–55. The current authors have tested this hypothesis by generating a knockout mouse model lacking SDHD. Whereas null animals die early during embryonic life, heterozygous SDHD+/- mice develop normally without apparent signs of respiratory distress. SDHD+/- animals show, however, a 40–50% decrease of mitochondrial complex II activity in all the tissues tested (brain, liver, heart and kidney) and a small (∼15%) increase in the number and size of glomus cells 56. Despite these structural changes, the response to hypoxia of glomus cells in SDHD+/- mice was unaltered, or even augmented, in comparison with SDHD+/+ litter mates, indicating that partial deficiency of complex II activity does not seem to alter glomus cell responsiveness to hypoxia.
In summary, although the exact role of mitochondria in CB function is not fully clarified, the data available thus far suggest that these organelles do not directly contribute to the primary steps in CB O2 sensing. However, mitochondrial dysfunction (e.g. in extreme hypoxia or after addition of ETC inhibitors) might result in metabolic alterations leading to changes in membrane ion channels that could modulate glomus cell activity. In accord with this idea, glomus cells in partially deficient SDHD mice (SDHD+/-), although with normal O2 sensing, exhibit an abnormally high resting secretory activity and a constitutive ∼50% reduction in total K+ current density 56.
Prolyl/asparagyl hydroxylases and hypoxia-inducible factor pathway
The best-studied O2 sensors are probably the prolyl/asparagyl hydroxylases, enzymes that utilise molecular O2 (together with Fe2+ and α-ketoglutarate as cosubstrate) to hydroxylate specific proline/asparagine residues, respectively, of hypoxia-inducible transcription factors (HIF)-1α, and its isoforms, as well as other molecules which, in turn, regulate the expression of numerous hypoxia-sensitive genes. In the absence of O2, the lack of hydroxyl groups in specific proline and asparagine residues of the HIF molecule prevents its degradation by the proteasome and facilitates its stabilisation, dimerisation with HIF-1β, translocation to the nucleus, and transcriptional activity 57, 58. Hydroxylation of HIF in the presence of O2 occurs in a few minutes, hence it is conceivable that O2-dependent hydroxylases could also modulate ion channels and thus participate in the acute responses to hypoxia. The current authors have tested this plausible hypothesis using CB slices incubated with saturating concentrations of dimethyloxalylglycine (DMOG), a membrane-permeant competitive inhibitor of α-ketoglutarate that completely and nonselectively inhibits hydroxylases 59. It is well known that DMOG mimics hypoxia and induces the expression of HIF-dependent genes 59, 60. However, after incubation of CB slices with DMOG, glomus cells retain their normal responsiveness to acute hypoxia. Preliminary experiments on prolyl hydroxylase 3-null mice performed in the current authors’ laboratory have also shown that their CB sensitivity to acute hypoxia is unaltered.
It has also been suggested that HIF-1α could directly participate in the acute responsiveness to hypoxia, since the plastic changes in the chemosensory activity (augmented ventilatory response and long-term facilitation) induced by sustained and intermittent chronic hypoxia are altered in HIF-1α+/- mice 61, 62. CB cells in slices from HIF-1α+/- mice show, however, a marked secretory response to hypoxia indistinguishable from that measured in homozygous HIF-1α+/+ wild-type mice 63. These data suggest that, although HIF-1α may contribute to CB functional plasticity, partial deficiency of the transcription factor does not significantly alter the intrinsic acute O2 sensitivity of CB glomus cells.
Haemoxygenase-2
Haemoxygenase (HO)-2 is an antioxidant enzyme constitutively expressed in most cells, including CB cells 64–66. This enzyme uses O2 to convert haem into biliverdin, iron and carbon monoxide 67. The possible involvement of HO-2 in CB acute O2 sensing has been suggested because it co-immunoprecipitates with heterologously expressed maxi-K+ channels and its inhibition with small interfering RNA abolishes the O2 modulation of recombinant channels 65. HO-2 is expressed in rat CB glomus cells and, in addition, native maxi-K+ channels recorded in patches excised from these cells are activated by HO-2 substrates (haem and NADPH). Based on these data it has been proposed that HO-2 could act as an O2 sensor through the production of CO, which is by itself a maxi-K+ channel activator 67, 68.
Although the proposal that HO-2 participates in O2 sensing is quite attractive 69 it has been challenged by experiments performed on the HO-2 knockout mouse, which develop normally, without alteration in haematocrit or signs of respiratory distress during the first postnatal 2–3 months, although they manifest cardiorespiratory alterations at advanced age 67, 70. The current authors have studied in detail the secretory responses to acute hypoxia of glomus cells from HO-2+/+, HO-2-/- and HO-2+/- animals using CB slices 63, 22. In all cases, secretion rate increased drastically upon exposure to low O2 tension. The dose–response curves obtained from glomus cells exposed to different O2 tensions were indistinguishable in HO-2-deficient and wild-type mice, suggesting that partial or complete HO-2 deficiency do not alter glomus cell O2 sensitivity. It can be also disregarded that the embryonic absence of HO-2 is compensated by upregulation of HO-1, an inducible HO, since the mRNA expression of this enzyme in CB tissue from HO-2-null animals is not significantly increased. Moreover, HO-1 does not seem to compensate for HO-2 deficiency, since within the CB it is expressed predominantly in blood vessels and, even in HO-2-/- animals, it is absent from the clusters (glomeruli) of tyrosine hydroxylase (TH)+ glomus cells 22.
Although glomus cell responsiveness to hypoxia is normal in HO-2-null animals, it seems that HO-2 deficiency causes CB phenotypic alterations secondary to redox dysregulation 65. HO-2-null young adults (<3 months) showed a marked upregulation of cyclophilin and TH, the rate-limiting enzyme for catecholamine synthesis highly expressed in CB glomus cells. In contrast, CB Slo1 mRNA (the maxi-K+ channel α-subunit gene) was significantly downregulated in HO-2-null mice in comparison with controls. These alterations in the CB gene expression profile, although unrelated to the mechanisms of CB O2 sensing, are compatible with a subclinical cellular oxidative stress, which could also be responsible for a small, but significant, CB growth observed in HO-2-null animals 22.
In summary, no definitive conclusion can be drawn to date regarding the molecular mechanisms of CB O2 sensing. There are numerous hypotheses and interesting proposals under debate but clarification of this important physiological process must await future experimental work.
CAROTID BODY PLASTICITY IN CHRONIC HYPOXIA: ADULT CAROTID BODY STEM CELLS
In addition to its role as an acutely responding arterial chemoreceptor, the CB is special among the adult neural and paraneural organs because it grows several-fold upon exposure to chronic hypoxia. In humans, this adaptive response occurs during acclimation to high altitude 52, 71, 72 or in patients suffering cardiopulmonary diseases presenting hypoxaemia 51, 73 (see section Carotid body function and mechanisms of disease). The current authors have recently studied in detail the morphological changes induced by chronic hypoxia in mouse and rat CB, with the aim of identifying the progenitors that could be used for in vitro expansion of CB dopaminergic glomus cells 1. Mouse CBs from animals kept in normoxia (21% O2 atmosphere) show the typical histological organisation of the organ with clusters of TH+ glomus cells (fig. 4a⇓). Exposure of the animals to hypoxia (10% O2 atmosphere) induces a marked CB enlargement caused by dilation and multiplication of blood vessels, as well as expansion of the parenchyma, with increased number of TH+ glomus cell clusters (fig. 4b⇓). To analyse the origin and formation of new glomus cells, mice were treated with BrdU (a marker selectively incorporated in replicating DNA) and, subsequently, maintained several days in a hypoxic environment. After a few days in hypoxia, even before the CB growth became macroscopically obvious, numerous BrdU+ TH+ cells were observed, indicating the appearance of new glomus cells (fig. 4c⇓). Brief (2-h) BrdU pulses were also applied to animals that had been kept in hypoxia for several days, in order to test whether some TH+ cells could be captured in the process of division. In some of these experiments, TH+ BrdU+ cells (fig. 4d⇓) were observed, suggesting that, as reported previously 12, glomus cells might undergo mitosis upon activation by hypoxia. TH+ glomus cells cannot produce clonal neurospheres in vitro (as discussed hereunder), so it is likely that their mitogenic potential is limited and possibly depends on the level of hypoxia and animal age 1. The time course of CB structural changes induced by hypoxia is shown in figure 4e⇓. Although BrdU incorporation into the CB tissue is observed immediately after exposure to hypoxia, the newly formed glomus cells (BrdU+ TH+) were predominantly seen after few days. This time course also suggested the existence in the CB of precursors, whose proliferation in hypoxia precedes their differentiation into glomus cells.
Carotid body (CB) growth in chronic hypoxia. Increase of CB size in a mouse exposed to hypoxia (10% O2) for 21 days (b), compared with normoxia (a). c) Tyrosine hydroxylase (TH)+ and BrdU+ cells (arrowheads) in a mouse CB exposed to hypoxia for 7 days. BrdU was administered each day from the beginning of exposure to hypoxia. d) TH+ and BrdU+ cells (arrows) in a mouse CB exposed to hypoxia for 7 days. A pulse of BrdU (in d) was administered 2 h before animal sacrifice. Cell nuclei are stained with 4′,6-diamidino-2-phenylindole. TH+ cells are shown as red, and BrdU+ as green. Scale bars = 50 μm. e) Progressive changes of mouse CB cell number and volume during exposure to hypoxia. ▪: volume; •: BrdU+ cells; ▴: TH+ BrdU+ cells. Reproduced and modified from 1 with permission from the publisher.
The precursors giving rise to glomus cells have been identified using enzymatically dispersed CB cells plated in floating conditions. To stimulate clonogenic proliferation, cells are cultured under moderate hypoxia (3% O2), a condition that mimics the hypoxic stimulation of CB growth in vivo. Under these conditions, ∼1% of the plated CB cells give rise to neurospheres, typical colony-like structures formed by growing neural stem cells (fig. 5a⇓ and b). In contrast to the typical spherical shape of neurospheres formed by stem cells isolated from other neural (central or peripheral) areas 74, 75, most of the CB-derived neurospheres have characteristically one or two large blebs budding out of the main core (fig. 5a⇓ and b). Immunocytochemical analysis of thin-section neurospheres have revealed the presence of nestin (a typical neural stem cell marker)-positive cells within the main core, and clusters of differentiated TH+ and nestin- cells within the blebs attached through a hilus (fig. 5c⇓ and d). The TH+ blebs resembled in shape the glomeruli characteristic of the in situ CB and grew to a large size after several weeks in culture (fig. 5e⇓). This morphological and immunological pattern (core of nestin+ cells preceding the blebs with TH+ cells) is consistently observed in most of the CB neurospheres studied (fig. 5f–h⇓). The clonal origin of CB neurospheres has been confirmed by single cell deposition experiments (fig. 5i–k⇓).
Carotid body (CB) stem cells. a) Neurospheres formed by dispersed CB cells after 10 days in culture; b) examples of the typical blebs (arrows) emerging from the neurospheres. Immunohistochemical analysis of a neurosphere thin section with bright field representation (c), and illustrating the presence of nestin+ progenitors (green) within the neurosphere core, and tyrosine hydroxylase (TH)+ glomus cells (red) within the bleb (d). e) Grown neurosphere (20 days in culture) with large blebs containing differentiated TH+ cells. Time course of rat CB neurosphere formation: f) 5 days; g) 7 days; h) 10 days. Organisation of the neurosphere core containing nestin+ progenitors precedes the appearance of TH+ glomus cells. Sequential photographs of a clonal colony illustrating the formation of a typical CB neurosphere from a single CB stem cell: i) 0 days; j) 5 days; k) 10 days. Scale bars = 100 μm (a and b) and 50 μm (c–k). Modified and reproduced from 1 with permission from the publisher.
The data described in the previous section indicate that the CB contains stem cells from which TH+ cells (resembling glomus cells) can be differentiated in vitro. The current authors have studied the physiology of stem cell-derived TH+ cells in order to test whether they behave as true matured glomus cells. TH+ cells within the neurosphere buds generated in vitro were subjected to voltage clamp using the whole-cell configuration of the patch-clamp technique. The recording in figure 6a⇓ illustrates that the newly formed cells have small inward Ca2+ currents (ICa) followed by larger outward K+ currents (IK). The amplitude, time course and voltage dependence of the outward current were similar to those recorded from cells in rat CB slices or after enzymatic dispersion 17, 76. As in normal CB glomus cells, blockade of the K+ outward current with internal Cs+ revealed the presence of typical inward, non- (or slowly) inactivating Ca2+ currents (fig. 6a⇓). The newly formed glomus cells responded to hypoxia with an acute surge of catecholamine secretion indistinguishable from that evoked in the CB in vivo (fig. 6b⇓) and they also expressed GDNF mRNA, a trophic factor characteristic of adult glomus cells (fig. 6c⇓) 15, 76. These data indicate that TH+ cells derived in vitro from CB progenitors exhibit the characteristic complex functional properties of mature glomus cells 1.
Physiological properties of in vitro differentiated glomus cells. a) Recording of calcium (ICa) and potassium (IK) voltage-dependent currents obtained from a patch-clamped glomus cell in a neurosphere bleb: depolarisation from -80 mV to +20 mV (top). Patch-clamp recording of voltage-gated calcium currents after blockade of potassium channels with intracellular caesium ions (bottom). b) Glomus cell catecholamine secretory response to hypoxia (switching from a solution with O2 tension of 145 mmHg (19.3 kPa) to another with ∼15 mmHg (2.0 kPa)). Each spike represents a single exocytotic event. c) RT-PCR analysis of rat carotid body (CB), superior cervical ganglion (SCG) and CB-derived neurospheres (NS) to show the selective expression of glia cell line-derived neurotrophic factor mRNA. M: marker. d) Hypothetical sequence of cellular events occurring within the carotid body during exposure to hypoxia. Glial fibrillary acidic protein (GFAP)+ type-II cells are considered to be the progenitors activated by hypoxia to produce nestin+ cells, which give rise to tyrosine hydroxylase (TH)+ glomus cells. Modified and reproduced from 1 with permission from the publisher.
Altogether, the data summarised in the present section indicate that the adult CB is a neurogenic niche where new neuron-like glomus cells can derive from progenitors. In fact, this is the first example of neural crest-derived stem cells with a recognisable function identified in the adult peripheral nervous system. Based on numerous cell fate experiments both in vivo and in vitro 1, the current authors have proposed the model for neurogenesis depicted in figure 6d⇑. Rat glial fibrillary acidic protein (GFAP)-positive type-II cells are viewed as quiescent (or slowly dividing) CB stem cells that can be reversibly converted to nestin+ intermediate progenitors. Upon exposure to hypoxia, the equilibrium is displaced towards the nestin+ population, giving rise to TH+ glomus cells. Therefore, the adult CB is a well-identified neurogenic centre that can be used for research on the molecular mechanisms of neurogenesis. Knowledge on CB stem cell physiology could also facilitate the expansion of human CBs for use in cell therapy (see section Carotid body function and mechanisms of disease).
CAROTID BODY FUNCTION AND MECHANISMS OF DISEASE
The CB is mainly known for its role in the control of respiration; nevertheless, it also has increasing clinical significance, as there is mounting evidence that CB dysfunction is involved in the pathophysiology of several human diseases, some of them of high prevalence.
Pathologies associated with primary alterations of carotid body O2 sensing
CB sensitivity to hypoxia develops during the early postnatal period and this correlates with an enhanced Ca2+ rise in response to hypoxia and increase in K+ current amplitude. Maturation of CB chemosensitivity is particularly important in the newborn since, in addition to increasing ventilation and sympathetic tone, activation of the CBs facilitates arousal from sleep and switch from nasal to oral breathing. Loss of chemosensitivity due to CB denervation around the time of birth produces severe respiratory disturbances in rats, piglets and lambs, exposing the newborn to respiratory instability and unexpected death 77. In several animal species, however, the hyperventilatory response to hypoxia, abolished by CB denervation, is re-established totally or partially several months after the surgery (possibly due to activation of other chemoreceptors). In contrast, glomectomy due to tumour surgery in humans results in complete and sustained lack of hypoxia responsiveness 78. Asthmatic humans treated by bilateral CB ablation have blunted responses to hypoxia, mainly during sleep, and some have died suddenly and unexpectedly 79.
It is believed that some respiratory disorders of the newborn, such as the sudden infant death syndrome (SIDS) could be due to primary alterations of the CB chemoreceptors 80, 81. Abnormalities in CB size or transmitter content have been reported in victims of SIDS. A common histological finding in CB from SIDS patients is the overgrowth of sustentacular/progenitor cells with decrease of glomus cell number 82, 83. Glomus cells from patients affected by SIDS contain a lesser number of dense core vesicles and appear to have higher CB dopamine content (released to the extracellular medium) than in normal children 84. This could be a cause of CB hypochemosensitivity, as dopamine is known to inhibit Ca2+ currents in glomus cells 10. Nicotine acting on peripheral chemoreceptors may delay CB resetting after birth and attenuate the protective chemoreflex response, thus increasing vulnerability to hypoxic episodes in the newborn. This could explain the association between maternal smoking and SIDS syndrome 85. It has been suggested that SIDS is probably not a sudden event but may be preceded by a relatively long period of hypoxia due to failure of reflex mechanisms 86. More recently, vascular endothelial growth factor, a gene induced by chronic hypoxia, was found increased in cerebrospinal fluid of infants who died of SIDS as compared with controls 87, thus further supporting the view that SIDS is caused by a decreased sensitivity of chemoreceptors.
The congenital central hypoventilation syndrome (CCHS) is a life-threatening disorder with impaired ventilatory response to hypoxia and hypercapnia that as SIDS appears to be also related to CB dysfunction. In some CCHS patients, necropsy showed >50% decrease in the number of TH+ CB glomus cells, increase in sustentacular cells and decrease in the number of dopaminergic vesicles 88. Numerous cases of CCHS are associated with genetic mutations. Mutations inherited from one of the parents have been found in the coding regions of endothelin-1, brain-derived neurotrophic factor and receptor tyrosine kinase (RET). All of these genes participate in development of neural crest-derived tissues 89. Mutations in the tyrosine kinase domain of RET are particularly interesting, since they also appear in Hirschsprung's disease, which is associated with ∼20% of the cases of CCHS. RET is part of the multicomponent receptor complex of GDNF, and both RET and GDNF are highly expressed in CB cells 14, 15. A recent study has shown heterozygous de novo mutations in PHOX2B, another gene necessary for the early development of neural crest-derived cells and for the formation of reflex circuits in the autonomic nervous system, in 18 out of 29 individuals with CCHS 90, 91. These data suggest that alteration of CB development and function is also associated with genetic CCHS. Hence, it could be that primary alterations of the CB germinal niche are a major cause of respiratory reflex dysfunction seen in CCHS and SIDS patients.
Hypoventilation in adults with chronic obstructive pulmonary disease results in “blue bloaters”, while those with normal ventilation are termed “pink puffers”. Offspring of the blue bloaters have a poorer ventilatory response to hypoxia than offspring of pink puffers, suggesting a familial component 4. The genetic influence on CB function is clear in two strains of rats, which had different CB calcium responses to acute hypoxia and different carotid sinus nerve traffic 92. The respiratory stimulant, doxapram, mimics the effect of hypoxia by inhibiting both voltage- and Ca2+-dependent K+ currents in glomus cells 93, providing further evidence of the importance of K+ channels in O2 sensing.
Carotid body and pathophysiology of chronic hypoxia
Exposure to chronic hypoxia, e.g. living at high altitude, produces a compensatory CB hypertrophy and cellular hyperplasia (see section Carotid body plasticity in chronic hypoxia: adult carotid body stem cells). The same occurs in situations in which alveolar gas exchange is compromised as, for example, in cystic fibrosis or cyanotic heart disease 74, 75. In these patients, stimulation of the respiratory centre by CB fibres is necessary to maintain the respiratory drive; thus, special precaution must be taken in the management of the patients to avoid excessive oxygenation and inhibition of CB activity. CB hypertrophia and cellular hyperplasia/anaplasia is also observed in CB tumours (chemodectomas or paragangliomas). These are relatively rare, mostly benign tumours in the neck that, besides the symptoms due to local compression, can also produce systemic hypertension 55. The most frequent cause of chemodectoma is the hereditary CB paraganglioma due to mutations in SDHD, a gene that encodes the small membrane anchoring subunit of SDHD in the mitochondrial complex II 50. The histological similarity between the CB growth in chronic hypoxia and paraganglioma has led to the suggestion that mitochondrial complex II participates in O2 sensing. As discussed previously (in the Mechanisms of carotid body acute oxygen sensing section), heterozygous SDHD knockout mice show a mild CB hypertrophy without alteration in acute responsiveness to lowering O2 tension 56. The grade of malignancy of CB paraganglioma is inversely associated with the number of GFAP+ type-II cells, a fact that could indicate that deregulation of CB progenitors (with type-II cell phenotype) participates in tumorigenesis 1.
An important health problem related to CB function is obstructive sleep apnoea syndrome (OSAS) 94. OSAS is a highly prevalent problem occurring at rates of 2–3% in children, 3–7% in middle-aged adults and 10–15% in healthy elderly subjects 95. It also has 30% prevalence among patients with so-called essential hypertension. CB seems to play a critical role in the development of hypertension associated with sleep apnoea. In rats exposed to 30 days of intermittent hypoxia (7 h per day), hypertension was observed but surgical denervation of peripheral chemoreceptors prevented the increase in arterial blood pressure. Adrenal demedullation and chemical destruction of the peripheral sympathetic nervous system by 6-OH dopamine also prevented hypertension 96, 97. In patients suffering from OSAS there is an increase in sympathetic activity, probably due to the recurrent arousal following the periods of apnoeas. However, it is believed that hypoxia per se also increases the sympathetic tone. Intermittent hypoxia in rats induces plastic changes in the CB, thus increasing its sensitivity and tonic sympathetic activation without obvious morphological alterations 62, 98. Similarly, OSAS patients have enhanced peripheral chemoreflex sensitivity and in those who experience repetitive hypoxaemia this increase might contribute to high levels of sympathetic activity even during normoxic daytime wakefulness 99, 100.
Carotid body and cell therapy
As the carotid body is a highly dopaminergic organ, it has been used in dopaminergic cell replacement for Parkinson's disease. Additional advantages of the carotid body for cell therapy rely on its survival in hypoxic environments, similar to that existing in the brain parenchyma after a tissue graft, and because it offers the possibility of autotransplantation in humans. Carotid body cell aggregates have been transplanted with excellent functional recovery in parkinsonian rats 101, 102 and monkeys 103. In a safety pilot study performed on PD patients, carotid body autotransplantation produced a clear amelioration in some cases 104. The beneficial effects of carotid body transplants are not only due to the local release of dopamine but also to a trophic action exerted on nigrostriatal dopaminergic neurons 14. The carotid body contains more glia cell line-derived neurotrophic factor than any other structure in adult mice 15. Therefore, glomus cells are ideal candidates to be used as biological pumps for the controlled endogenous release of glia cell line-derived neurotrophic factor and possibly other trophic factors with unique synergistic actions. In fact, carotid body grafting has also been shown to reduce neuronal death in an acute rat stroke model 105. The systematic clinical applicability of carotid body dopamine- and glia cell line-derived neurotrophic factor-producing cells is under investigation 1, 106.
Support statement
This research has been supported by the Spanish Ministry of Health (Instituto de Salud Carlos III), The Spanish Ministry of Science, and the Juan March and Marcelino Botín Foundations.
Statement of interest
None declared.
Footnotes
-
Previous articles in this series: No. 1: Wagner PD. The biology of oxygen. Eur Respir J 2008; 31: 887–890. No. 2: Zhou G, Dada LA, Sznajder JI. Regulation of alveolar epithelial function by hypoxia. Eur Respir J 2008; 31: 1107–1113. No. 3: Berchner-Pfannschmidt U, Frede S, Wotzlaw C, Fandrey J. Imaging of the hypoxia-inducible factor pathway: insights into oxygen sensing. Eur Respir J 2008; 32: 210–217. No. 4: Lévy P, Pépin J-L, Arnaud C, et al. Intermittent hypoxia and sleep-disordered breathing: curent concepts and perspectives. Eur Respir J 2008; 32: 1082–1095.
- Received April 14, 2008.
- Accepted April 22, 2008.
- © ERS Journals Ltd