The Human Carotid Body: Expression of Oxygen Sensing and Signaling Genes of Relevance for Anesthesia

• ❖ Although molecular mechanisms of the oxygen-sensing apparatus in the carotid body have been partly probed in animals, the mechanisms in humans are unknown.

• ❖ Six carotid bodies were removed during radical neck dissection and subjected to high-resolution gene analysis and immunohistochemistry.

• ❖ There are similarities but also key differences in genetic expression of putative components of oxygen sensing in human carotid body compared with that in other species.

OXYGEN is an essential substrate for all mammalian cells, and early detection of hypoxia is critical to protect organs and the whole body against cell damage and death. The carotid body (CB) is the global oxygen sensor that promptly regulates ventilation and cardiorespiratory reflexes during acute hypoxia.1Many drugs used in anesthesia depress regulation of breathing during acute hypoxia (e.g. , propofol, halogenated inhaled general anesthetics, and neuromuscular blocking agents [NMBAs]).2–5Although such agents clearly blunt the acute hypoxic ventilatory response (HVR) in humans, the mechanism(s) by which they interfere with oxygen sensing and signaling remains an enigma.

In brief, the acute HVR is triggered by stimulation of peripheral chemoreceptor cells (type 1 cells) in the CB.1The oxygen-sensing mechanism(s) in the CB is not fully understood but has been proposed to involve metabolic mitochondrial inhibition, production of oxygen radicals, inhibition of the hypoxia-induced factor pathway, or decreased carbon monoxide production by heme oxygenase 2 (HO-2) associated with the large-conductance potassium (BK) K⁺channel.6Early events in oxygen sensing involve closure of oxygen-sensitive K⁺channels in the CB type 1 cell, a subsequent membrane depolarization, and opening of voltage-dependent Ca²⁺channels. The resultant Ca²⁺influx triggers release of neurotransmitters, with a subsequent increase of activity in the afferent carotid sinus nerve. This activates areas in the brainstem responsible for regulation of breathing.1Among the multitude of transmitters released, acetylcholine and adenosine triphosphate (ATP) seem to be important excitatory transmitters, whereas dopamine and γ-aminobutyric acid (GABA) have modulatory and inhibitory functions, respectively.6,7General anesthetics and nondepolarizing NMBAs target ion channels such as GABA^Areceptors, nicotinic acetylcholine receptors (nAChRs), and leak K⁺channels.8–10In addition, propofol and nondepolarizing NMBAs blunt oxygen signaling in CB of many animal species, and the halogenated inhaled anesthetics halothane and sevoflurane have a distinct action on K⁺channels in the CB type 1 cells.11–14Thus, drugs used in anesthesia have the potential to target critical steps in global oxygen-sensing and signaling pathways of the CB. Although components of the oxygen sensing and signaling pathways in experimental animals have been partially characterized, the pathways of the human CB are at present unknown.1,6 

Consequently, as a first step toward a better understanding of these fundamental mechanisms, we here characterize components of the human oxygen-sensing and signaling pathways with a focus on GABAergic, cholinergic, purinergic, and dopaminergic receptor systems and K⁺channels of relevance in anesthesia.

The study was approved by the ethics committee on human research at Karolinska Institutet (Stockholm, Sweden). Six patients (male, American Society of Anesthesiologists classification I-II, aged 36–68 yr) scheduled for unilateral neck dissection to treat head and neck cancer were included in the study after informed written consent. None of the patients had received radiation or chemotherapy before surgery, none had a tumor involving the CB, and none was using nicotine. Exclusion criteria were severe cardiopulmonary (New York Heart Association Functional Classification more than 2), neurologic, or metabolic disease and smoking. Demographic data are presented in table 1.

During sevoflurane-narcotic based anesthesia with normoxic and normocapnic mechanical ventilation (Aisys; GE Healthcare, Madison, WI), the CB was removed by the head-and-neck surgeon after unilateral neck dissection. The CBs (weight range, 50–80 mg) were divided into pieces of equal size and immediately either frozen in liquid nitrogen for RNA isolation or fixed in 4% paraformaldehyde at 4°C for immunohistochemistry.

Snap frozen pieces of CBs weighing ∼20–35 mg were stored at −80°C until RNA was prepared. For isolation of total RNA, samples were homogenized using a Potter-Elvehjem Teflon pestle homogenizer in 1 ml TRIzol® (Invitrogen, Carlsbad, CA) and allowed to stand for 5 min at room temperature. The homogenate was then mixed with 0.2 ml chloroform, vigorously shaken for 15 s, and incubated at room temperature for 2–3 min. The mixture was centrifuged for 5 min at 12,000g  at 4°C, the aqueous phase was transferred to a fresh microcentrifuge tube, and 1 volume of 70% ethanol was added. Thereafter, RNA was isolated using an RNeasy kit (Invitrogen) according to the manufacturer's recommendations. The yields of RNA were from 1 to 2 μg. All RNA samples showed A  260/280 ratios between 1.7 and 2.0. RNA integrity was assessed either by running samples on a denaturing agarose gel or by using an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA).

Labeling and hybridization were performed according to standard Affymetrix (Santa Clara, CA) procedure at the Karolinska Institute Bioinformatics and Expression core facility**using Affymetrix Human Genome U133 2.0 Plus oligonucleotide microarrays. Affymetrix microarrays were analyzed with the use of Affymetrix Microarray Suite Software (MAS) to estimate expression values for each probe. Using the statistical detection algorithms in MAS version 5.0, we have separated transcripts of particular probe sets that are reliably detected (present or marginally present calls) from transcripts below the threshold of detection and therefore considered absent. Transcripts for which an absent call was determined were eliminated from further analysis.

Isolated total RNA (0.5 μg) was used in a complementary DNA synthesis reaction using SuperScript II First-Strand Synthesis System for reverse transcription-polymerase chain reaction (RT-PCR; Invitrogen) and oligo(dT) primers. The resulting complementary DNA was amplified using a 7500 Real-time PCR System (Applied Biosystems, Foster City, CA). Twenty-five microliters of amplification reaction included 200 ng complementary DNA template, TaqMan® Universal Master Mix, and a TaqMan® Gene Expression Assay for relevant genes (Applied Biosystems, Foster City, CA) ( appendix 1). All samples were amplified in triplicate and normalized to three housekeeping genes (β-actin, TATA box binding protein, and 18S ribosomal RNA).

After fixation in 4% paraformaldehyde for 2–4 h, CB pieces were cryoprotected in 30% sucrose in phosphate-buffered saline at 4°C for 3–4 days. Subsequently, the CB pieces were rapidly frozen in TissueTek O.C.T. compound (Sakura, Zueterwoude, Netherlands) and stored at −80°C. Fourteen-micrometer sections were cut on a cryostat and mounted onto microscope slides.15For determination of morphology, some slides were stained with hematoxylin and eosin (Histolab, Gothenburg, Sweden), and the cellular structures of the CB were identified with light microscopy using a Zeiss Axioskop 2 plus equipped with an AxioCam digital camera (Zeiss, Munich, Germany).15Corresponding slides were chosen for immunohistochemistry. After incubation in phosphate-buffered saline and blocking in phosphate-buffered saline with bovine serum albumin and Triton X-100, primary and secondary antibodies were applied as described previously.15Primary antibodies against tyrosine hydroxylase (TH) and β-III-tubulin were used to identify type 1 cells and antibodies against glial fibrillary acidic protein were used to identify type 2 cells ( appendix 2). Primary antibodies against the potential oxygen sensor HO-2, the TWIK-related acid sensitive K⁺channel-1 (TASK-1) and BK potassium channels, the α2, β3, γ2, and δ GABA^Areceptor subunits, the α3, α7, and β2 nAChR subunits, the ATP-receptor P2X², the adenosine A^2Areceptor, and the dopamine D²-receptor are described in  appendix 2.

Double immunofluorescent staining with anti-TH or anti-β-III-tubulin and antibodies against relevant receptors was performed to demonstrate colocalization in type 1 cells.15The secondary antibodies used were Alexa Fluor 488, 555, and 647 antibodies (Invitrogen), selected to match the animal specificity of the primary antibodies and used at a concentration of 1:500. Slides without primary antibodies were used as negative controls. The slides were mounted using Vectashield, and 4′,6-diamidino-2-phenylindole (Vector Laboratories, Burlingame, CA) was used to counterstain the nuclei. Colocalization was confirmed with double immunofluorescent staining. The CB was then visualized and photographed using a laser-scanning microscope (LSM 710 and ZEN 2009 software; Zeiss Inc., Jena, Germany).15 

PCR data are normalized to the reference genes according to the common ΔC^tmethod using the formula 2^{ΔCttarget gene (40 − Cttarget)/2ΔCtreference gene (40 − Ctreference gene)}, related to the expression of TH, and expressed as mean ± SEM (Prism 4.0; GraphPad Software, San Diego, CA).

CBs from the six patients included in the study were analyzed to describe gene expression and protein localization. Because of technical limitations, we were not able to apply all methods used in the study of all CBs: the CB from patient 1 was used only for hematoxylin-eosin and immunohistochemical staining, and the quality of the RNA obtained from patients 2 and 5 did not match the specific requirements of the microarray analysis. Moreover, the fixed CB piece from patient 6 was not suitable for immunohistochemistry. We used CBs from patients 3, 4, and 6 for microarray, those from patients 2–6 for real-time PCR, and those from patients 1–5 for immunohistochemical staining.

Hematoxylin-eosin staining revealed a typical CB appearance,16with dark type 1 cells arranged in an irregular cluster-like pattern (fig. 1). The size and arrangement of the type 1 cell clusters was heterogeneous, both between individuals and in different parts of a single CB. There was no infiltration of tumor tissue in the hematoxylin-eosin preparations.

The transcriptome of human CB was studied using Affymetrix Human Genome U133 2.0 Plus oligonucleotide microarray. MAS 5.0 analysis revealed approximately 25,000 probe sets with the assigned present or marginally present calls. After removal of redundant probe sets, the total number of genes expressed was approximately 13–14,000 (14,064, 13,253, and 13,658 genes expressed in patients 3, 4, and 6, respectively). This is consistent with the latest estimations of the average number of expressed genes in different human tissues, which varies from approximately 10,000 to 15 000.17Notably, the mouse CB transcriptome identified earlier is slightly smaller (approximately 10,000 genes),18,19which most likely reflects the higher complexity of the human CB.

Next, we analyzed the expressed gene lists obtained from the microarray assay for the presence of key genes that define the CB's chemosensory function and that are potential targets for drugs used in anesthesia, such as GABA^A, nicotinic, purinergic, dopaminergic receptors, K⁺channels, and potential oxygen sensors (table 2). To validate the microarray data, we performed real-time PCR analysis for all relevant genes using highly specific TaqMan® Gene Expression Assays (see Materials and Methods ). The two methods yielded similar patterns of expression for all selected genes with the exception of α3, α7, and β2 nAChR subunits, for which microarray analysis did not assign present calls in all patients. The real-time PCR of these genes showed, however, that at least α3 and α7 had detectable levels of expression in all five patients tested. β2 expression was found in three of five cases. Similar trends were observed for P2X²and TH; the microarray data showed expression to be silent, whereas the real-time PCR identified both transcripts (albeit in rather low amounts) in all five patients (table 2). The expression of GABA^Aγ2 and δ receptor subunits was detected by PCR in only two patients, whereas microarray data were negative. Because neither analytical method used would allow estimation of absolute expression values, we have related the PCR gene expression data to the expression of TH, which is considered to be very low at normoxia (fig. 2).19Such analysis demonstrates low abundance of GABA^Areceptor subunits as well as P2X²transcripts, which were detected only by real-time PCR. In general, such discrepancies between microarray and PCR data might be explained by the known tendency of microarray analysis to underestimate the extent of gene expression compared with real-time PCR, which is undoubtedly a more sensitive and specific analytical tool.20 

Immunohistochemistry was used to confirm results of real-time PCR and microarray and to provide information on specific cell localization of proteins. As described previously, TH and β-III-tubulin were used as type 1 cell markers and glial fibrillary acidic protein as type 2 cell marker15,21(fig. 3A). We detected sparsely distributed TH-positive cells in type 1 cell clusters, in line with previous findings in the human CB (fig. 3).22,23In some clusters, however, all cells were positive for TH (fig. 3B). β-III-tubulin, on the other hand, was positive in the majority of cells in all type 1 cell clusters, and we consequently used this marker to assess colocalization of relevant proteins. Positive immunostaining in the human CB was seen for GABA^Areceptor subunits α2, β3, and γ2, nAChR subunits α3, α7 and β2, ATP-receptor P2X², adenosine receptor A^2A, dopamine D²receptor, K⁺channels TASK-1 and BK, and HO-2 (figs. 4 and 5, table 2). The immunostaining for all proteins was positive in all patients tested with the exception of BK and the α7 subunit of the nAChR, where most of the CBs were positive (table 2). The immunostaining for the GABA^Areceptor δ subunit was negative (table 2). Based on morphology and colocalization with β-III-tubulin, we conclude that these proteins are localized in type 1 cells or in adjacent nerve endings but not in type 2 cells. The pattern for the α3 subunit of nAChR and D²is inconsistent with the rest; cells appear to be densely filled with the protein, suggesting a more intracellular localization than for the other proteins (fig. 5). A summary of all results is presented in table 2.

Hypoxia is a common cause of serious anesthesia-related adverse events in the early postoperative period, the most frequent underlying mechanisms being respiratory depression and airway obstruction.24–27The residual effects of general anesthetics and NMBAs on regulation of breathing and airway protection seem to have a key role in postanesthetic mortality and morbidity.28–30Several studies on regulation of breathing in humans have indirectly pointed toward a specific interaction of anesthetics with peripheral chemoreceptors, yet there are no studies to show mechanisms through which anesthetics might perturb chemoreception in human CB.2–5This study is the first to demonstrate human oxygen-sensing and signaling proteins that are also potential targets for drugs used in anesthesia. The major findings are that key sensing and signaling proteins previously found in animal studies and known to be of importance in the context of anesthesia are present in the human CB. On both the messenger RNA (mRNA) and protein levels, we demonstrate presence of relevant GABAergic, nicotinergic, purinergic, and dopaminergic receptors, as well as the potential oxygen sensor HO-2 and the TASK-1 and BK K⁺channels.

It is essential to assess multiple signaling proteins because studies on animal CBs during the last few decades have demonstrated a signaling protein cascade between the oxygen-sensitive CB type 1 cell and the postsynaptic carotid sinus nerve in response to hypoxia.1,6This cascade involves closure of oxygen-sensitive K⁺channels in response to hypoxia followed by depolarization of the membrane of the CB type 1 cell and a Ca²⁺-dependent neurotransmitter release. There are clear expressional and functional species differences regarding the identity of the oxygen-sensitive K⁺channels; various voltage-dependent (Kv) channels are important in mouse and rabbit CB, whereas the leak TASK-1 and BK K⁺channels are important in rat.6,31The neurotransmitters released during hypoxia, expression of the corresponding receptors, and especially their functions are also species dependent.15,32The most important excitatory transmitters at present seem to be acetylcholine acting on postsynaptic nAChRs, adenosine activating presynaptic A^2Areceptors, and ATP acting on both presynaptic and postsynaptic P2X²and P2׳ATP-receptors. Dopamine, on the other hand, has an inhibitory or modulatory role via  postsynaptic dopamine D²receptors.1,6Activation of postsynaptic GABA^Areceptors has recently been demonstrated to have an important inhibitory role in CB chemotransmission.7Thus, both oxygen-sensitive K⁺channels and important receptors display large species variability and, with the exception of the dopamine D²receptor, they have not been characterized in human CB. So far, the only receptors reported to be present in the human CB are histamine H1 and H3, dopamine D², NTR1, TH, and serotonin.22,23,33,34Those immunohistochemical studies were also done on tissue taken up to 48 h postmortem from a mixed population of humans in terms of age (e.g. , age from 4 months to 91 yr), which raises questions concerning how representative they are.

Several physiologic studies have demonstrated the importance of CB function for the HVR in humans. Studies of patients undergoing carotid endarterectomy or bilateral neck dissection show that the hypoxic response is blunted after a unilateral CB denervation and, furthermore, markedly reduced or even abolished after a bilateral denervation. It is noteworthy that the human HVR does not recover over time after CB denervation or resection, which contrasts to animals, in which the aortic bodies gradually take over and compensate for the loss of CB function.35Moreover, there is a large interindividual variability in the normal ventilatory response to acute hypoxia; in some persons, the HVR is even absent.36The reason for this interindividual variance in humans is poorly understood, and molecular studies on human CB chemoreception are therefore needed. Improved knowledge in this field might also have implications for postoperative care of patients in whom HVR is reduced or absent.

In this unique material, we chose to investigate each CB individually rather than pooling them together, as is commonly done in animal research because of the small size of the organ.18,19Because characterization of gene expression on the transcriptional level is often insufficient for prediction of protein translation, we characterized each CB on both mRNA and protein levels. We chose an immunohistochemical method to detect target proteins, which gave the added benefit of showing the cell type in which the expression occurs. This is of paramount importance because the CB tissue contains various cell types, including glomus type 1 cells and sustentacular type 2 cells; thus, we could make some differentiation between cell types. In addition, relying only on immunohistochemistry would have increased the risk of false positive results.37 

In all patients, the α2 and β3 GABA^Areceptor subunits, the α3 nAChR subunit, the adenosine A^2Aand dopamine D²receptors, as well as the TASK-1 and BK K⁺channels and HO-2 were expressed as both mRNAs and proteins. The α7 and β2 nAChR subunits were present in most patients, whereas the α1 and α3 GABA^Areceptor subunits, the α4 nAChR subunit, and the P2X³receptor were not detected. It is noteworthy that microarray data suggested that TH and P2X²were absent, but both PCR and protein analyses showed them to be present in all patients. This might reflect low levels of expression of these genes during normoxia, which, in combination with the above-mentioned issue of the sensitivity of microarray technique, might lead to such apparent contradictions in the experimental data. In addition, very few CB type 1 cells were immunopositive for TH, and the mRNA levels were also barely detectable, which is in line with previous findings that TH is much more sparse in the human CB than in those of experimental animals.22,23Interestingly, the gene expression of TH almost doubled during hypoxia in mouse,19but whether this occurs also in humans remains to be investigated. The GABA^Areceptor subunits δ, ρ1, and ρ2 were present only in single patients, and at very low levels in one modality, and were therefore considered not to be present in the human CB. Comparison of our data from humans with data from previous animal studies reveals large similarities between species but also interesting and potentially crucial differences. Here we demonstrate expression of the α2 subunit of the GABA^Areceptor, which is also expressed in rat CB but not in cat CB.38,39The α4 nAChR subunit was not found in the human material but has been detected at both the mRNA and protein levels in mouse and cat CB, and a pharmacological study indicates the presence of the α4β2 nAChR subtype.15,40,41Furthermore, in contrast to previous animal studies, we found no expression of the ATP-receptor P2X³in the human CB.42,43Thus, there are distinct differences in the expression of various receptors and receptor subunits in animal and human CB.

Whether this is of functional importance remains to be evaluated, but this study suggests that general conclusions about human CB oxygen-sensing and signaling function cannot be drawn solely on the basis of data from animals. Based on our findings in human CB, we speculate about potential target receptors/ion channels underlying the effects of drugs used in anesthesia on acute HVR and CB chemoreception. Although potentiation and activation of the GABA^Areceptor channel are the main mechanisms of anesthetic action of propofol, it also potently depresses CB chemoreception and HVR.3,4,9,12,14The most common GABA^Areceptors in the central nervous system consist of two α, two β, and one γ subunit, and in human CB, the α2, β3, and γ2 receptor subunits are expressed. Thus, it might be argued that propofol reduces CB chemoreception as a result of an interaction with an α2β3γ2 GABA^Areceptor, which is also one of the most common GABA^Areceptor subtypes expressed in the central nervous system.9Inhaled halogenated anesthetics target both K⁺channels and nAChRs, which also are present in the human CB.9Finally, nondepolarizing NMBAs have the potential to inhibit neuronal nAChRs present in human CB.13,44It must be emphasized that these are hypotheses based on a combination of our data from the study of human CB and previous results from studies on human and animal CB and can as such be seen only as speculations. To achieve a more comprehensive understanding of human CB oxygen sensing, signaling, and interactions with drugs used in anesthesia, functional studies using human material will be needed. A better understanding of the human CB is important not only in the context of anesthesia but also for a larger patient population, such as those with chronic obstructive pulmonary disease or obstructive sleep apnea syndrome and in patients in critical condition after trauma.

In conclusion, in unique samples of human CB, we here demonstrate presence of critical proteins in the oxygen-sensing and signaling cascade. Our findings demonstrate similarities to but also important differences from established animal models. In addition, our work establishes an essential platform for studying the interaction between anesthetic drugs and human CB chemoreception.

The authors thank Jan Lundgren, M.D., Ph.D. (Associate Professor, Department of Oto-rhino-laryngology, Karolinska University Hospital, Solna, Sweden), for help in conducting this project. We also thank the head-and-neck surgeons at the Department of Oto-rhino-laryngology, Karolinska University Hospital, who were involved in the surgical care of the patients.