Changes in arterial Po2, Pco2, and pH are the strongest stimuli sensed by peripheral and central chemoreceptors to adjust ventilation to the metabolic demand. Erythropoietin (Epo), the main regulator of red blood cell production, increases the hypoxic ventilatory response, an effect attributed to the presence of Epo receptors in both carotid bodies and key brainstem structures involved in integration of peripheral inputs and control of breathing. However, it is not known whether Epo also has an effect on the hypercapnic chemoreflex. In a first attempt to answer this question, we tested the hypothesis that Epo alters the ventilatory response to increased CO2 levels. Basal ventilation and hypercapnic ventilatory response (HCVR) were recorded from control mice and from two transgenic mouse lines constitutively expressing high levels of human Epo in brain only (Tg21) or in brain and plasma (Tg6), the latter leading to polycythemia. To tease apart the potential effects of polycythemia and levels of plasma Epo in the HCVR, control animals were injected with an Epo analog (Aranesp), and Tg6 mice were treated with the hemolytic agent phenylhydrazine after splenectomy. Ventilatory parameters measured by plethysmography in conscious mice were consistent with data from electrophysiological recordings in anesthetized animals and revealed a blunted HCVR in Tg6 mice. Polycythemia alone and increased levels of plasma Epo blunt the HCVR. In addition, Tg21 mice with an augmented level of cerebral Epo also had a decreased HCVR. We discuss the potential implications of these findings in several physiopathological conditions.
- carbon dioxide
- ventilatory control
- carotid bodies
- respiratory network
the glycoprotein hormone erythropoietin (Epo) plays an essential role in erythropoiesis (30, 31, 46). It is also a cytokine characterized by low expression and activity under normal conditions and rapid upregulation during hypoxia (21, 50), tumor growth (11), and trauma (25). Epo is produced by the peritubular capillary endothelial cells in the kidney and liver, and its basal level is upregulated when the arterial Po2 (PaO2) decreases below normal levels. It is used for the treatment of patients after chemotherapy (27) or in need of repeated blood transfusions (28, 36). Epo and its receptor (EpoR) have been detected in many tissues, suggesting that Epo has a much broader field of action than previously recognized. In particular, Epo in brain plays an important role in stroke (42), multiple sclerosis (57), schizophrenia (10), epilepsy (32), and retinopathy (55).
Several studies characterized the effects of Epo on breathing. In adult rats, central Epo injections increase minute ventilation (73) and induce a phrenic long-term facilitation (7). Epo increases the ventilatory response to hypoxia in both mice (60) and humans (17). The soluble EpoR, an endogenous Epo antagonist expressed in the mouse brain, regulates the ventilatory acclimatization to hypoxia (59, 60) and decreases hypoxic respiratory activity in vitro (29). Although Epo contributes significantly to oxygen homeostasis, its potential role in modulating the ventilatory response to changes in PaCO2 has been little explored (1). Assessing this role may be relevant because hypoxia-mediated hyperventilation increases the elimination of CO2 that exerts a powerful drive to breathe (13, 20). Changes in CO2 levels are detected by peripheral chemoreceptors in the carotid bodies and central chemoreceptors in multiple sites in the brainstem, including those in the retrotrapezoid nucleus (18–20, 49, 53, 54, 64, 69). Because EpoR is present in these structures, increased Epo levels may increase or decrease the hypercapnic ventilatory response (HCVR). Here, we studied two transgenic mouse lines overexpressing Epo in brain and plasma (Tg6) or in brain only (Tg21) and compared their baseline ventilation and responses to CO2 inhalation to those of wild-type (WT) mice. Plethysmographic measurements in conscious mice and electrophysiological recordings in anesthetized animals showed normal basal ventilation but a decreased HCVR in mutant mice. Results from chronic Epo injections in WT mice and pharmacological restoration of normal hematocrit values in Tg6 animals indicated that both polycythemia and high plasmatic Epo levels reduced the HCVR. We conclude that Epo decreased the hypercapnic chemoreflex via hematological (polycythemia), plasmatic (carotid bodies), and central effects.
MATERIALS AND METHODS
Experiments were performed in accordance with French national legislation (JO 87–848) and European Communities Council Directive (22 September 2010, 2010/63/EU, 74). Permit numbers A13-505 and 13-227 for C. Menuet and C. Gestreau, respectively, were delivered by Direction Départementale de la Protection des Populations, Préfecture des Bouches du Rhône, France.
Transgenic mouse lines overexpressing Epo were generated by microinjection of human Epo cDNA driven by the human platelet-derived growth factor (PDGF) β-chain promoter into the pronuclei of fertilized oocytes derived from B6C3 hybrid mice (56). One resulting transgenic mouse line TgN(PDGFBEPO)321ZbZ (Tg6) showed increased Epo levels in plasma (12-fold vs. WT control animals) and brain (26-fold vs. WT), accompanied by a doubled hematocrit (23, 68) and increased [25.3 (0.7) g/dl] hemoglobin concentration. This erythrocytic line was backcrossed to C57Bl/6 mice for more than 12 generations by mating hemizygous males to WT C57Bl/6 females. Half of the offspring were hemizygous for the transgene, whereas the other half were WT and thus used as controls (WT-Tg6). The second transgenic mouse line termed TgN(PDGFBEPO)322ZbZ (Tg21) was bred to homozygosity in a C57Bl/6 background and showed increased Epo levels in brain only (4-fold vs. WT) (71). Both WT (WT-Tg21) and Tg21 animals presented normal hematocrit and hemoglobin concentration [15.4 (0.5) g/dl]. Because aged Tg6 mice display organ degeneration (23), we used exclusively transgenic (23 Tg6 and 13 Tg21) and WT (23 WT-Tg6 and 14 WT-Tg21) mice of 2–3 mo of age in our experiments.
Determination of Ventilation by Plethysmography
Ventilation of conscious mice was recorded using noninvasive whole-body plethysmography, as previously described (47). Before recordings, the mice were habituated to the plethysmograph chamber to reduce stress. Only periods of breathing without body movements were studied. We used a constant flow plethysmograph (EMKA Technologies, Paris, France) with 200-ml animal chamber ventilated with air (600 ml/min). Spirograms were stored and analyzed (Spike 2; Cambridge Electronic Design, Cambridge, UK) to calculate the mean respiratory frequency (Rf, expressed in cycles/min), the tidal volume normalized by the body weight (VT, μl/g), and the minute ventilation normalized by the body weight (VE, ml·g−1·min−1, with VE = Rf × VT/1,000). For VT measurements, calibrations were performed by injecting a small volume (100 μl) of air into the recording chamber. We analyzed the effect of imposed hypercapnic challenge by first recording ventilation during inhalation of room air for 20 min and then during inhalation of a hypercapnic mixture (4% CO2, 21% O2, balance N2) for 5 min (47). Because hypercapnia usually induces transient agitation, only recordings during the last 3 min of the hypercapnic challenge were analyzed.
Analysis of Respiratory Parameters by In Vivo Electrophysiology
To further assess the HCVR of WT and transgenic animals, electrophysiological recordings were obtained from anesthetized animals.
Mice were put in an induction chamber flushed with 3% isoflurane in air. Volatile anesthesia was maintained via a face mask by reducing the inspired isoflurane concentration to 1.5–2.0%. Body temperature was maintained at 37°C using a temperature-controlled heating pad.
Surgery and electrophysiological recordings.
A pair of copper electrodes was implanted in an external intercostal muscle to record electromyographic inspiratory activity (EIC). These electrodes also recorded the electrocardiogram (ECG). After a tracheostomy, the cervical vagus nerve was dissected free from surrounding tissue and mounted on coiled bipolar silver electrodes to record vagal nerve activity (VNA). Although VNA displays a biphasic (inspiratory and postinspiratory) pattern in unanesthetized preparations such as the in situ preparation (8), VNA respiratory bursts recorded in our conditions mostly consisted in inspiratory discharge, probably attributable to a greater depression of postinspiration by anesthesia. The raw signals (EIC and VNA) were digitized (sampling frequency 5 kHz), amplified (gain 5,000 to 20,000), and filtered (band pass 0.1–3.0 kHz), and moving averages were obtained (leaky integrator with 100-ms time constant). Both raw and integrated signals were stored on the computer for offline analyses (Spike 2; Cambridge Electronic Design). In some experiments (4 WT-Tg6 and 3 Tg6 mice), end-tidal CO2 was also measured using a CO2 analyzer (Microcapstar; CWE, Ardmore, PA).
Fifteen minutes after completion of surgery, all mice had stable heart rate (HR), Rf, VNA and EIC. Signals were then recorded for 3 min under basal condition (control period). Subsequently, mice were exposed for 3 min to a normoxic hypercapnic gas mixture (7% CO2-21% O2, balance N2). Signals were recorded during the 3-min hypercapnic challenge as well as for 5 min immediately after (recovery period). This protocol was chosen based on preliminary data showing 1) no obvious change in ventilation when anesthetized mice were exposed to 4% CO2, and 2) a clear increase in ventilation in response to 7% CO2. Of note, the ventilatory response to 7% CO2 was not saturated, as further increases in respiratory signal amplitude could be recorded when mice were exposed to 12% CO2 (data not shown).
Duration (expressed in seconds), peak amplitude, and area under the curve (AUC) of respiratory bursts (expressed in arbitrary units) were measured from the integrated EIC and VNA signals. For each animal, the values were first measured during control periods and compared with data obtained during and after the hypercapnic challenge. Changes in peak amplitude and AUC of integrated EIC and VNA were expressed as percentages of control values. The raw and the integrated EIC signals were used to compute HR (expressed in beats/min) and Rf, respectively. Peaks corresponding to QRS complex of the ECG were first detected and then transformed into marker events for HR using Spike2 software (Cambridge Electronic Design). For each animal, mean HR and Rf values were computed in control and hypercapnic conditions. All values (EIC, VNA, HR, and Rf) were then averaged using all animals selected for analyses (n = 7–9 per group) and expressed as group mean (standard deviation). At the end of the experiments, blood samples were collected from the heart to measure Epo levels in plasma, and animals were euthanized by intracardiac injection of T61 (Intervet S.A., Beaucouzé, France).
We used pharmacological treatments in WT and transgenic mice to assess the different factors contributing to the reduction of the HCVR. We followed the same protocol as previously described by Schuler and colleagues (58). Each pharmacological protocol was followed by plethysmographic assessment of the HCVR of the mice.
Aranesp treatment of WT-Tg6 mice.
To examine the effect of polycythemia on the HCVR, the hematocrit (Htc) of WT-Tg6 mice was increased using subcutaneous injections of Aranesp (NESP) (darbopoietin-α, Amgen; 12.5 μg/kg) twice a week. At least 10 days of treatment were necessary to reach a high level of Htc, as observed from blood samples taken from the tail vein of the animals and Htc measurements after centrifugation.
Splenectomy and phenylhydrazine treatment of Tg6 mice.
To determine the effect of Epo on the HCVR, a splenectomy was first performed using a left-side abdominal laparotomy (58, 66) on 6-wk-old Tg6 mice because the massive erythropoiesis occurring in the spleen would compensate for any attempt to decrease Htc (15, 58, 66). Once Tg6 mice were 10 wk old, their Htc was reduced to normal-range levels (40–45%) by subcutaneous administration of freshly prepared phenylhydrazine hydrochloride (PHZ, P6926; Sigma, St. Louis, MO) that causes chemical hemolysis (65). Animals received one PHZ injection (0.12 mg/g body wt) every 2 days for 6 days. Blood samples were taken from the tail veins of the animals, and Htc was measured by centrifugation. A group of sham-operated Tg6 animals (Tg6-SHAM) was used as controls.
Quantification of Plasma Epo
Blood samples were drawn by cardiac puncture into heparinized Eppendorf tubes, and plasma was collected after centrifugation at 18,000 g. Protein concentrations of plasma were determined by the Bradford protein assay (Bio-Rad, Hercules, CA). Subsequently, Epo levels were quantified using an 125I-labeled Epo-based radioimmunoassay (RIA) (Amersham, Zurich, Switzerland), according to previously published protocols (30). The lower detection limit of our RIA was 4 U/l, and the intra-assay/interassay variances were <2% and <6%, respectively.
Analysis was performed using StatView software (Abacus Concepts, Berkeley, CA), and values were expressed as means (standard deviation). Repeated-measures ANOVA (2-way) followed by a post hoc protected least significant difference of Fisher (PLSD) test were used to analyze data from plethysmography and electrophysiology. One-way ANOVA followed by a PLSD post hoc test was used to analyze data from Htc and plasma Epo measurements. Unpaired Student's t-tests were used to analyze data from end-tidal CO2 values. Regression analyses were also performed to analyze the relation between HCVR and Htc values or plasma Epo concentration. Differences were considered significant when P < 0.05.
Decreased HCVR in Awake Tg6 Mice
Unrestrained conscious WT-Tg6 (n = 8) and Tg6 mice (n = 9), the latter showing constitutive overexpression of Epo (in brain and plasma) and polycythemia, were used to measure VE, VT, and Rf under basal and hypercapnic conditions. The Tg6 group was composed of both Tg6-SHAM (n = 4) and Tg6 (n = 5) mice because their ventilatory parameters were similar under both normocapnic (VT, Rf, and VE, P = 0.73, 0.36, and 0.49, respectively) and hypercapnic (VT, Rf, and VE, P = 0.3, 0.24, and 0.75, respectively) conditions. Table 1 summarizes the results obtained for each parameter, group, and condition and reports the statistical values of the two-way repeated-measures ANOVA. Post hoc tests revealed that neither VE nor ventilatory pattern (Rf and VT) differed between WT-Tg6 and Tg6 mice exposed to normoxic normocapnic (basal) condition (Fig. 1, A, B, D–F). Exposure to normoxic hypercapnia (21% O2-4% CO2 for 5 min) triggered a robust ventilatory response in both strains (Fig. 1, A and C). However, VE was lower (P < 0.01) in Tg6 than in WT-Tg6 mice (Fig. 1D; Table 1). The HCVR of Tg6 mice was blunted by ∼60% due to significantly lower Rf (P < 0.05; Fig. 1E) and VT (P < 0.05; Fig. 1F).
Decreased Intercostal and Vagal Respiratory Activities in Anesthetized Tg6 Mice Exposed to Hypercapnia
Under basal (room-air) conditions, similar end-tidal CO2 levels (P = 0.97) were found in some of the isoflurane-anesthetized WT-Tg6 [4.5 (0.3) %, n = 4/9] and Tg6 [4.6 (0.2) %, n = 3/7] mice. Also, WT-Tg6 and Tg6 mice had similar HR and Rf values in both the control and hypercapnic periods (Table 2). Respiratory drives to thoracic and laryngeal muscles were also evaluated by measuring mean duration, peak amplitude, and AUC of the integrated EIC and VNA in both mice groups (Figs. 2 and 3; Table 3). In basal and hypercapnic conditions, there were no differences between strains in duration of EIC or VNA (Fig. 3). However, the AUC of both respiratory signals measured during the HCVR was significantly lower in Tg6 than in WT-Tg6 mice (P < 0.05 and P < 0.01 for EIC and VNA, respectively). This difference was mainly attributable to a significant decrease (∼50%; P < 0.01 for both parameters) in peak amplitude of the respiratory signals (Figs. 2 and 3; Table 3). The mean amplitude and AUC of VNA measured during recovery in WT-Tg6 animals remained significantly elevated compared with Tg6 mice (P < 0.05 for both parameters) (Fig. 3B; Table 3).
Polycythemia Alone Decreases the HCVR
The erythropoiesis-stimulating agent NESP and treatment with the hemolytic agent PHZ were used to manipulate Htc values in WT and Tg6 mice. NESP significantly increased the Htc of WT-Tg6 mice (WT-NESP, n = 6), and PHZ effectively reduced the Htc of Tg6 mice (Tg6-PHZ, n = 7) to normal values (Fig. 4A; Table 4). Subsequent evaluation of ventilatory parameters in these mice and comparison between all groups showed changes in basal ventilation and in the magnitude of the HCVR (Table 1). In control condition, WT-NESP mice had a lower VE (P < 0.05) than WT-Tg6 mice, and Tg6-PHZ had a higher VE (P < 0.05) than Tg6 animals (Fig. 4B). During hypercapnia, VE varied in the following order: WT-Tg6 > WT-NESP > Tg6-PHZ > Tg6 mice (Fig. 4B; Table 1). Compared with values in hypercapnia obtained in control (WT-Tg6) mice, Rf was significantly lower (P < 0.05 for both comparisons) in Tg6 mice as well as in Tg6-PHZ animals with a normal Htc level (Fig. 4B; Table 1). Furthermore, a linear regression analysis showed a significant relation (F = 9.226, P = 0.0074, R2=0.386) between HCVR and Htc values measured in the same WT-Tg6 (n = 8), WT-NESP (n = 6), and Tg6 mice (n = 6), suggesting that an increase in Htc (polycythemia) is associated with a decrease in HCVR (Fig. 4C). Of note, values from Tg6-PHZ mice were not included in the regression analysis attributable to a confounding factor (increased plasma Epo levels) discussed below.
Increased Plasma Epo Blunts the HCVR
Treatments with either NESP or PHZ altered, not only the Htc, but also the levels of plasma Epo. Indeed, both WT-Tg6 animals injected with NESP (WT-NESP) and Tg6 mice injected with PHZ (Tg6-PHZ) had significant increases in plasma Epo concentrations (Fig. 5A; Table 4). We interpret the increased level of plasma Epo in Tg6-PHZ mice as a probable consequence of hypoxemia attributable to the hemolytic action of PHZ. Thus we used data from WT-Tg6 (n = 7), WT-NESP (n = 5), and Tg6 mice (n = 6) to analyze the relation between the HCVR and plasma Epo concentration by a linear regression. This analysis showed a significant relation (F = 14.938, P = 0.0014, R2 = 0.483), suggesting that an increase in plasma Epo concentration is associated with a decrease in HCVR (Fig. 5B).
Reduced HCVR in Awake Tg21 Mice
In this study, a second strain of transgenic mice (Tg21) was used to analyze the effect of cerebral Epo on the HCVR. Tg21 mice overexpress Epo in brain only and show normal levels of plasma Epo and Htc (71). Respiratory parameters in Tg21 (n = 7) and their corresponding control group (WT-Tg21, n = 8) were first recorded by plethysmography (Fig. 6). Results from statistical comparisons of parameters by two-way repeated-measures ANOVA are illustrated in Table 1. Post hoc tests revealed that WT-Tg21 and Tg21 mice had similar basal ventilatory parameters (VE, Rf, and VT; P > 0.05 for all comparisons) (Fig. 6). In response to CO2, however, Tg21 mice had a significantly lower VE than WT-Tg21 mice (P < 0.05; Fig. 6A). This lowering in VE by ∼34% in Tg21 animals was mainly due to a lower Rf (P < 0.05; Fig. 6B) because VT values did not differ significantly (Fig. 6C).
Decreased Intercostal and Vagal Activities in Anesthetized Tg21 Mice Exposed to Hypercapnia
Electrophysiological measurements in anesthetized WT-Tg21 (n = 6) and Tg21 (n = 6) mice revealed similar HR and Rf under both control and hypercapnic conditions (Table 5). By contrast, there were significant differences in respiratory drive to the pump (EIC) and valve (VNA) muscles between the two groups of mice (Table 6). Compared with WT-Tg21 mice, Tg21 mice exposed to 7% CO2 had lower intercostal and vagal respiratory activities (Figs. 7 and 8). The blunted HCVR of Tg21 mice was mainly due to significantly lower peak amplitude and AUC of both intercostal (Figs. 7 and 8A) and vagal discharges (Figs. 7 and 8B). The amplitude and AUC of VNA did not recover similarly between the two groups of animals (Fig. 8; Table 6), probably attributable to changes in magnitude of the HCVR.
The present study examined the effects of high Epo levels and polycythemia on the respiratory responses to hypercapnia. We took advantage of a transgenic mouse line (Tg6) showing constitutive high levels of Epo in brain (26-fold vs. WT) and plasma (12-fold vs. WT) and high Htc (71). Results in Tg6 mice showed for the first time that overexpression of Epo decreased the HCVR. In addition, data obtained after treatment of control (WT-Tg6) mice with NESP as well as injection of PHZ in transgenic Tg6 mice with splenectomy clearly showed that both polycythemia and increased Epo concentration in plasma reduced the HCVR. Finally, a decrease in the HCVR was also observed in a second strain of transgenic mice (Tg21) that overexpress Epo only in brain (4-fold vs. WT). Overall, these data suggest that Epo decreases the HCVR via both peripheral and central effects and could thus exert a deleterious effect on breathing in response to hypercapnia.
Decreased HCVR Attributable to Polycythemia
Our data showed a strong, inverse linear relation between Htc levels and the amplitude of HCVR in untreated control (WT-Tg6), control treated with Epo (WT-NESP), and Tg6 mice. Increasing the number of red blood cells increases the level of carbonic anhydrase, the enzyme catalyzing the reversible hydration of CO2 into bicarbonate and protons (39). Because inhibition of carbonic anhydrase, for instance with acetazolamide, induces a mild renal metabolic acidosis and increases ventilation (6, 40), we can assume that an elevation of carbonic anhydrase increases the buffering capacity of blood, thus lowering the CO2/pH chemical drive on respiratory centers. In addition, a higher Htc would confer a higher capacity to form carboxyhemoglobin. Such a high level of carboxyhemoglobin likely helps to increase CO2 elimination by pulmonary gas exchange and lower plasma CO2. Thus polycythemia may exert indirect effects contributing to a decrease in the stimulus presentation at both peripheral and central CO2 chemoreceptors. This may explain why basal ventilation in conscious WT-NESP mice was decreased compared with WT-Tg6 animals and why it was increased in Tg6-PHZ compared with Tg6 mice. Of note, splenectomy and phenylhydrazine treatment in Tg6 mice (Tg6-PHZ) restored normal Htc level but not a normal HCVR. Data from these mice did not fit to the linear relation between Htc levels and amplitude of the HCVR found in the other groups of animals, suggesting that further parameters contributed to decrease the ventilatory response.
Decreased HCVR Attributable to Increased Plasma Epo Concentration
Our data revealed a strong, inverse linear relation between plasma Epo levels and the amplitude of HCVR in control untreated (WT-Tg6), control treated with Epo (WT-NESP), and Tg6 mice. These results strongly suggest that a high plasma Epo concentration blunts the HCVR although the underlying mechanisms remain speculative. Under physiological conditions, an intact blood-brain barrier excludes large glycosylated molecules such as Epo (4, 43). Moreover, when large doses of Epo (5,000 U/l or higher) are administered systemically, only about 1–2% of the Epo crosses the blood-brain barrier (2, 41, 63, 72). Thus high plasma Epo concentration may not be able to affect central chemoreceptors. By contrast, the impact of plasma Epo concentration on carotid bodies cannot be discounted. Indeed, Epo receptors are expressed in glomus cells (34, 60), and plasma Epo increases the ventilatory response to hypoxia by interacting with peripheral chemoreceptors (16, 60, 61). Further studies are needed to determine how Epo in carotid bodies interacts with O2- and CO2-sensitive channels and/or neurotransmitters.
Decreased HCVR Attributable to Increased Brain Epo Concentration
Centrally produced Epo is suggested to act directly on brainstem respiratory centers to increase the hypoxic ventilatory response via Epo receptors located in medullary structures controlling respiratory activity (16, 60–62). These data suggest that, in addition to the peripheral effects of Epo discussed above, Epo may also blunt the HCVR via a central action.
Comparisons of data from Tg6 and Tg21 mice.
Tg21 mice overexpress Epo only in brain and show normal chemical blood parameters and Htc (71). The present data from plethysmography and electrophysiology revealed a lower HCVR in these mice consistent with the hypothesis that Epo directly decreases the central respiratory drive during hypercapnia. The attenuation of ventilatory response to hypercapnia in Tg21 mice was less pronounced than that observed in Tg6 mice. This difference may be explained by additive peripheral and central effects of Epo in Tg6 mice or by the fact that Tg6 mice show a higher overexpression of central Epo (26-fold vs. WT) compared with Tg21 animals (4-fold vs. WT) (71).
In our experiments, two different in vivo strategies were used to assess the HCVR in Tg6 and Tg21 mice. Plethysmography enables recording of ventilation in unrestrained, conscious mice. This useful technique, however, provides only a global measure of ventilation and sometimes detects artefactual increases in VT when mice struggle against partly obstructed upper airways (9, 67). Effective ventilation requires the coordinated activity of upper airway muscles (regulation of resistance) and pump muscles (flow generation) (3). Therefore, we further assessed the ventilatory response of mice using electrophysiological approaches under anesthesia to describe the effects of Epo on cranial (VNA) and spinal (EIC) respiratory motor outputs. Of note, although Tg6 mice had increased plasmatic and brain Epo concentrations as well as polycythemia, our data failed to detect significant changes in end-tidal CO2, Rf, and HR in basal condition between Tg6 and control mice, suggesting comparable levels of respiratory drive and cardiac activity. This is in good agreement with a study showing no alteration in blood pressure, HR, and cardiac output in Tg6 mice despite excessive hematocrit values (66). Vasodilatation and enhanced erythrocyte flexibility likely explain the regulated adaptation to high blood viscosity in these mice (66). In contrast, our results showed that the decreased HCVR of Tg6 and Tg21 mice was associated with a lower drive to both ventilatory pump (EIC) and valve (VNA) muscles. A simple explanation of these results could be that brain Pco2 and pH do not change to the same extent in control vs. transgenic mice upon exposure to the same FiCO2. This could be the case, for example, if cerebral blood flow was decreased by excessive Epo concentration, thereby reducing the stimulus presentation to the CNS. However, opposite changes in cerebral blood flow have been found in the two strains of mice used in the present study, with a 77% decrease in Tg6 and a 22% increase in Tg21 (14). Furthermore, the latter study showed similar oxygen delivery in transgenic and control mice. Thus we speculate that similar changes in central CO2/pH levels occurred in both mutant and control mice during hypercapnia.
Putative Mechanism of Epo-Induced Central Desensitization to Hypercapnia
Our results suggest that Epo acts directly on brainstem respiratory centers to decrease the HCVR. Therefore, Epo may target one or more central chemosensitive structures that are widely distributed throughout the brainstem (20, 53, 54). Among them, the retrotrapezoid nucleus (RTN) is a key component of CO2/pH chemosensitivity (20, 49). Indeed, both GPR4 receptors and TASK-2 channels located on the cell membrane of RTN neurons detect changes in CO2/pH and contribute to the normal respiratory chemoreflex (18, 33, 52, 69). Interestingly, mutant Task-2−/− mice have an increased ventilatory response to hypoxia as well as decreased sensitivity to changes in CO2/pH (18, 69). The similarity between respiratory phenotypes of Tg6/Tg21 and Task-2−/− mice suggests that Epo may target RTN neurons and increase the efflux of potassium through TASK-2 channels. Further studies are needed to know whether Epo can modulate chemosensitive currents in RTN neurons. In addition, enhancement of the ventilatory response to hypoxia is associated with a decrease in brainstem catecholamine turnover (60). As catecholaminergic groups have been also recognized as CO2/pH chemosensitive areas (22, 26, 38), they may be other putative mediators of the Epo-induced hypercapnic desensitization.
Astrocytes also mediate a CO2/pH chemosensitive drive to respiration (19). Astrocytes are the main sources of Epo in the brain (44, 45, 70) and express Epo receptors (5, 51). Central Epo may thus act in an autocrine/paracrine manner in chemosensitive areas on both astrocytes and neurons. In addition, high numbers of Epo receptors are found in astrocytic endfeet around brain capillaries, and these receptors could be bound by circulating Epo (41). The latter mechanisms may explain the greater reduction of the HCVR in Tg6 compared with Tg21 mice.
Perspectives and Significance
Our results show that Epo decreases the chemoreflex upon exposure to elevated FiCO2 via both peripheral and central effects. This may have implications in several conditions. The first is during physiological acclimatization to hypobaric hypoxia in which the initial phase of hyperventilation results in secondary hypocapnia and alkalosis and thus hypoxic depression or roll off. If Epo also decreases the chemoreflex in response to low CO2 levels, it may thus counteract the secondary depression of breathing, maintain higher ventilation, and contribute to a better acclimatization to hypoxia. Such an effect could act in concert with or contribute to the Epo-mediated enhancement of the ventilatory response to hypoxia (16, 17, 60). Further experiments, such as measurements of the apneic threshold (69) in Tg6 and/or Tg21 mice or in control mice chronically treated with Epo, are needed to test this hypothesis. Our observations may also be relevant for chronic mountain sickness (CMS), a condition in which high-altitude residents affected by this disease develop a blunted ventilatory response to hypoxia, inducing severe and persistent hypoxemia, leading to increased Epo production, polycythemia, pulmonary hypertension, right heart failure, and death (37, 48). A greater end-tidal Pco2 and a lower pH are found in patients with CMS compared with control high-altitude dwellers (12, 37), and hypoventilation in CMS patients is associated with reduced central chemosensitivity to CO2 (12, 24, 35). Epo may play a crucial role in the development of CMS, not only by inducing a pathological level of Htc, but also by reducing the respiratory drive attributable to blunted CO2/pH chemoreflex, contributing to hypoventilation and thus hypoxemia. Alternatively, the reduced drive may help retain CO2 and consequently maintain perfusion to CO2-sensitive vascular beds.
This work was supported by grants from the Conseil Régional Provence Alpes Côte d'Azur (PACA), Ville de Marseille, and French National Agency for Research (ANR Respitask) to C. Gestreau.
No conflicts of interest, financial or otherwise, are declared by the authors.
Author contributions: C.M. and C.G. conception and design of research; C.M., H.K., T.D.l.P.d., and C.G. performed experiments; C.M., H.K., T.D.l.P.d., and C.G. analyzed data; C.M., H.K., and C.G. interpreted results of experiments; C.M., H.K., T.D.l.P.d., and C.G. prepared figures; C.M., H.K., and C.G. drafted manuscript; C.M. and C.G. edited and revised manuscript; C.M., H.K., T.D.l.P.d., and C.G. approved final version of manuscript.
We express our gratitude to Dr. Jorge Soliz for contributing to the experimental design, participating in some of the experiments, interpreting the data, and editing the manuscript. The authors thank Dr. Steve Iscoe for appraisal of the manuscript and editorial assistance. We are also indebted to Professor Max Gassmann, who is responsible for the source of animals.
Present address of Clement Menuet: Central Cardiovascular Regulation, Department of Physiology, University of Melbourne, Victoria 3010, Australia.
- Copyright © 2016 the American Physiological Society