INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: sodium-hydrogen antiporter; NHE 3; medulla oblongata; central chemosensor

The exact nature of the adequate signal to central respiratory chemosensitivity is still unknown. Winterstein's "reaction theory" (1, 2) proposed H⁺ as the common and unique stimulus driving pulmonary ventilation during respiratory and metabolic acid-base disturbances. The major objection against the unifying "reaction theory" remained the different magnitude of ventilatory responses observed to metabolic and respiratory acid-base changes in the blood.

In the past, several research groups perfusing the brain ventricles with mock cerebrospinal fluid (CSF) tried to substantiate a common but differently accessible H⁺ sensor beyond the blood-brain barrier mediating the ventilatory responses to both hypercapnia and nonvolatile acids (see 1, 3). When, however, techniques for direct measurement of brainstem extracellular fluid (ECF) pH became available, its unique role in mediating ventilatory responses to inhaled CO₂ and to intravenous acid infusion had to be questioned (4). Nevertheless, several hints for a unique role of extracellular H⁺ were given by recent in vitro studies, in thin layer cell cultures of the medulla oblongata (8) and close to the ventral surface of superfused brainstem-spinal cord preparations of neonatal rats (9).

With the possibility of determining whole-brain pH_i by nuclear magnetic resonance techniques in conscious humans, there is now growing evidence for a role of brainstem intracellular pH as an adequate signal for central chemosensors (10). Brainstem intracellular pH as the target variable may be changed more easily by CO₂ through barriers and membranes than by fixed acids. Differences in pH_i may thus be responsible for otherwise unexplained respiratory responses to the isohydric rise in CSF PCO₂ in vivo (3) and in vitro (11).

More recently, interaction of both intra- and extracellular pH has been proposed, as neurons within chemosensitive areas of brainstem slices or cell cultures were shown not to regulate CO₂-induced intracellular acidification (15, 16), preferably in an acid environment. One reason for this seems to be the inhibition of Na⁺/H⁺ exchange by extracellular acidification (17). Further studies have clearly shown that CO₂-sensitive neurons in organotypic cultures from the ventrolateral medulla of newborn rats enhanced their bioelectric activity in response to a decrease in intracellular pH, independent of extracellular pH (16), and that the CO₂ response could be mimicked by different pharmacological inhibitors of the type 3 Na⁺/H⁺ exchanger (NHE3) in vitro (18). Neurons that showed sustained activation during hypercapnia also responded to low concentrations of S8218 (0.5-5.0 µM) with increased bioelectric activity (19). The lipophilic substance S8218, a potent inhibitor of the Na⁺/H⁺ exchanger type 3, was designed to easily penetrate the blood-brain barrier (Aventis Pharma, Frankfurt/Main, Germany).

The aim of this study was, therefore, to explore whether NHE3 inhibition also affects the central respiratory response to hypercapnia in vivo. Effects of intravenously applied S8218 on integrated phrenic nerve activity were investigated in anesthetized, mechanically ventilated rabbits, because we could also demonstrate expression of NHE3 within the medulla oblongata in this species. If hypercapnia could be mimicked by NHE3 inhibition, this would further support the assumption that intracellular acidification is an adequate signal for respiratory central CO₂ chemosensitivity. Furthermore, this would open new approaches to the treatment of respiratory dysfunctions.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

RT-PCR Study for Rabbit NHE3

Expression of NHE3 was achieved using the standard reverse transcription polymerase chain reaction (RT-PCR). The medulla oblongata (ca. 5 mm rostral to 5 mm caudal from the obex) was removed from three adult rabbits and snap frozen in liquid nitrogen. An equal portion of tissue was taken from the kidney's cortex as a positive control, as this organ strongly expresses NHE3 (20). Frozen organs were homogenized in 4 M guanidium thiocyanate containing 0.1 M 2-mercaptoethanol (10 ml/g tissue) using a Digamax homogenizer (DIAX 900, Heidolph, Kehlheim, Germany). Total RNA was isolated by acid phenol-chloroform extraction (21) and redissolved in DEPC-treated water. RNA concentration was calculated from optical density at 260 nm.

Expression of NHE3 in both organs was demonstrated by standard RT-PCR methods. In brief, 5 µg of total RNA from rabbit kidney and medulla oblongata was reverse transcribed into cDNA using oligo(dT₁₅) as primer for Moloney murine leukemia virus (M-MLV) reverse transcriptase (Promega, Heidelberg, Germany). Primers specific for rabbit NHE3 were used to obtain a 645-bp DNA fragment (5'-aag ccg ctg gtg cag tgg ctg aag g and 3'-gcc cag ctt ggc cga ctt gaa gga ctc c according to Genebank Accession number M87007). After an initial denaturation step at 94° C for 4 min, PCR was run 40 cycles with denaturation at 94° C for 60 s, annealing at 57° C for 90 s, and elongation at 72° C for 3 min. PCR products were separated on a 2% agarose gel, stained with ethidium bromide (0.5 µg/ml), and visualized under ultraviolet (UV) light.

Immunocytochemistry

The medulla oblongata of three further adult rabbits was removed, trimmed by transverse sectioning with a razor blade (ca. 5 mm rostral to 5 mm caudal from the obex), and immersed for 2-6 h in 4% paraformaldehyde in 0.1 M phosphate buffer (PB), pH 7.4. Meninges were removed with forceps. Tissue blocks were separated along the sagittal axis to obtain corresponding contralateral halfs for antibody staining and control, respectively. One brainstem was further cut in a frontal direction ca. 1 mm dorsal to the pyramids (see Figure 1). Endogenous peroxidase activity was blocked with a mixture of 0.3% H₂O₂ in cold methanol followed by cooling to 20° C for 1 h. Tissue was rehydrated to PB via 75%, 50%, and 25% methanol (30 min each). All the following steps were carried out on ice under gentle agitation. Tissue was immersed in 10% fetal calf serum (FCS), 0.1% NaN₃ in PB to block nonspecific binding for at least 1 h. Then a monoclonal anti-NHE3 antibody (Chemicon, MAB3134, dilution 1:100, incubation time: 2 d), an biotinylated anti-mouse immunoglobulin G (IgG) antibody (Sigma B-8250, dilution 1:300, incubation time: 12 h), and the ABC reagent (Vectorstain Elite, Vector Laboratories, incubation time: 6 h) were consecutively applied. After each incubation step tissue blocks were washed extensively with PB (five to eight changes over 12 h) to remove nonbound compounds. Bound ABC reagent was visualized with 3,3'-diaminobenzidine/NiCl₂ (Peroxidase Subtrate Kit, Vector Laboratories) according to the manufacturer's instruction. Control tissue was processed equally but not exposed to the anti-NHE3 antibody. Tissue blocks were mounted on silicon rubber with needles, submersed in PB, and viewed with an upright microscope (Olympus Bx50Wi).

View larger version (168K): [in this window] [in a new window]   Figure 1.   Detection of NHE3 immunoreactive (NHE3-ir) cells and NHE3 mRNA in the brainstem of adult rabbits. (A) NHE3-ir cells occur at sites dorsal from the pyramidal tract (A1,2) and also form a group near to the ventrolateral rim of the section (A4, arrows, micrograph taken ca. 1 mm rostral to the obex). Remote from this group only a few NHE3-ir cells were found (arrowhead). Immunostaining was absent in the contralateral control section (A3). All micrographs were taken from the surfaces of enbloc stained tissue. Directions of sectioning are shown in the diagram (upper right corner). Calibration bars: 50 µm (A1), 500 µm (A2-A4). (B) Amplification of an NHE3 coding DNA fragment using primers specific for rabbit NHE3 mRNA. Identical amounts (5 µg) of total RNA from adult rabbit medulla oblongata and kidney (positive control) were subjected to RT-PCR; distilled water (right lane) served as control. The arrow points to the 600-bp band of a hundred base-pair ladder (left lane). Note that PCR fragments from kidney (expected size 650 bp) and medulla oblongata are identical.

In Vivo Experiments

These experiments were performed in seven male rabbits with an average weight (± SEM) of 3.63 ± 0.14 kg. The animals were anesthetized by sodium pentobarbital. An initial injection of 60.1 ± 4.0 mg kg¹ was given prior to tracheotomy and dissection of a branch of the external jugular vein for continuous infusion of the anesthetic. Infusion was started 3-5 h before measurements and maintained at 7.5 ± 0.2 mg kg¹ h¹ throughout the experiment. Blood coagulation was prevented by 50-100 I.E. Sandoz-Heparin (Sandoz), usually given after each blood sampling.

The trachea was cannulated and connected to an open excess gas flow of oxygen-enriched air (fraction of inspired oxygen [FI_O2] = 0.37). For artificial ventilation, the animals were paralyzed by Alloferin (Hoffmann-La Roche AG, Grenzach-Wyhlen, Germany) and connected to a small animal respiration pump (Harvard Apparatus Co., Millis, MA). The femoral arteries were cannulated for continuously recording blood pressure and for arterial blood sampling, and the left femoral vein for intravenous injections.

During initial spontaneous breathing, tidal volume (VT) and inspiratory and expiratory durations (TI, TE) were evaluated from the respiratory flow signal, obtained by a Fleisch tube (size 0) connected with a Gould-Godart pneumotachograph (Type 17212, SensorMedics, Bilthoven, The Netherlands). Phrenic nerve activity was recorded from a distally cut C1-C5 root of the right phrenic nerve, desheathed, and placed on a bipolar silver electrode in close connection with a preamplifier system for extracellular recordings (DAM 50, World Precision Instruments, Sarasota, FL). The amplified nerve signals of the compound potential were rectified and integrated by a self-constructed leakage integrator with a time constant of 140 ms. During artificial ventilation, values for TI and TE were evaluated from the integrated phrenic nerve activity (IPNA). Airway CO₂ was continuously sampled and analyzed by infrared absorption (Binos 1, Leybold-Heraeus, Hanau, Germany). Blood pressure was measured by means of a catheter filled with heparin-saline, connecting one femoral artery to a pressure transducer (Statham P 23Gb) in combination with a AWP4 DC bridge amplifier (Astro-Med, Inc., W. Warwick, RI). Respiratory and cardiovascular variables were continuously recorded by a chart recorder type MT95K2 (Astro-Med, Inc., W. Warwick, RI).

Blood Analysis

From arterial blood samples, the O₂ partial pressure (Pa_O2) was measured directly (Clark-type electrode E5046, Radiometer, Copenhagen, Denmark), and the CO₂-partial pressure (Pa_CO2) was determined indirectly from pH measurements after equilibration with two CO₂-O₂ mixtures (BMS2 Mk2 blood microsystem and PHM 84 Research pH meter, Radiometer, Copenhagen, Denmark; precision gas mixing pump, Wösthoff, Bochum, Germany). Besides pH_a and actual bicarbonate (HCO₃a), this equilibration method directly provides the Pa_CO2/pH_a relationship of the actual blood sample, as well as standard bicarbonate (HCO₃st) and base excess (BE) concentrations. Blood analysis was performed at 38° C, rectal temperature being controlled at 38-38.5° C by a thermistor and heating pad. Concentrations of hemoglobin (Hb) and lactate were measured spectrophotometrically (Hitachi-100-10, Japan), using test solutions Merckotest (E. Merck, Darmstadt, Germany) and LACT MPR 3 (Boehringer, Mannheim, Germany), respectively.

Experimental Protocol

The animals inhaled O₂-enriched air (FI_O2 = 0.37) to maintain arterial PO₂ values in the range of 20 kPa. During spontaneous breathing, a first steady-state period of at least five breaths was awaited to obtain respiratory, cardiovascular, and (from arterial samples) blood gas and acid-base data. After bilateral vagotomy, animals were paralyzed (0.3 mg/kg Alloferin intravenously) and mechanically ventilated by adjusting tidal volume and pump rate to about 6.5 ml kg¹ and 50 min¹, respectively, to achieve an average arterial P_CO2 of about 4.1 kPa (Table 1). The apneic threshold for Pa_CO2 was searched by hyperventilating the animals until phasic phrenic nerve activity ceased. The CO₂ response of the phrenic nerve activity was tested at four steady-state levels of Pa_CO2 between 1.0 and 6.0 kPa above the individual threshold value first by returning to normal ventilation and then by adding CO₂ to the inspiratory gas mixture, each level being maintained for 10-12 min. After removal of the inspiratory CO₂ load, the effects of tentative NHE3 inhibition on respiratory variables were studied, first under eucapnic conditions. Infusions of S8218 (Aventis Pharma) were started at higher infusion rates of 1 ml/min (1 mg/ml saline for 20 min) to reach average (±SEM) cumulative doses of 9.2 ± 1.1 mg kg¹ at maximum, yielding average plasma concentrations of 0.9 ± 0.2 µg/ml. Subsequently, infusions were continued at lower rates of 0.1 ml/min to maintain plasma levels in the range of 0.3 µg/ml (about 10⁶ M). Under these conditions, measurements at apneic threshold and at the different CO₂ levels above were repeated and compared with control measurements.

Evaluation and Data Processing

Tidal volume (VT), durations of inspiration and expiration (TI,TE), end-tidal P_CO2 (PET_CO2) and/or amplitude of the integrated phrenic nerve activity (IPNA) were evaluated as the mean of five breaths. Respiratory rate (fR) and pulmonary ventilation () were calculated as fR = 60/TI+TE and = VT · fR. The integrated phrenic nerve activity or tidal phrenic amplitude (IPNA) was normalized by setting the maximum arbitrary value that could be obtained in each animal to 100 units (22); phrenic minute activity was calculated as IPNA · fR.

Statistical Analysis

Group mean steady-state values and standard errors of the mean (SEM) were calculated for different measured variables. Normal distribution was tested by the one-sample Kolmogorov-Smirnov test, before using a Student's two-tailed paired t test to compare differences of corresponding means under control conditions and during NHE3 inhibition. At the level of p_D  0.05 differences were regarded as being significant. In some cases, statistical analysis refers to n measurements repeated in all animals under comparable conditions. To explore the effect of S8218 on the respiratory pattern, linear regression analysis was performed for IPNA·fR as a function of IPNA in each animal, according to Hey and coworkers (23). Statistical analysis was in part carried out by SPSS 8.0 for Windows software (SPSS Inc. Chicago, IL).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Detection of NHE3 Immunoreactive (ir) Cells

Inspection of the outer ventral, ventrolateral, and dorsal surfaces of the fixed brainstem tissue revealed only a few large NHE3-ir cells close to the ventral (alongside the pyramidal tracts) or ventrolateral but not dorsal surface. NHE3-ir cells were, however, found at the transversally cut surfaces (inspected ca. 4-5 mm rostral to the obex) dorsal to the pyramidal tract (Figure 1A1,2) and were also grouped in ventrolateral areas (Figure 1A4). Only a few NHE3-ir cells were scattered in the mediolateral (see, e.g., Figure 1A4) and dorsal parts of the section. Somata of NHE3-ir cells were large (diameter 20-30 µm) and had one to five dendrite-like processes (see Figure 1A). Due to their morphological appearance NHE3-ir cells were regarded as neurons. No NHE-ir cells or structures were found in negative controls, where the specific antibody was omitted (Figure 1A3). We conclude that NHE3-ir neurons occurat least in partin areas that have a prevalence for central chemosensitivity and/or contribute to the generation of breathing rhythm.

RT-PCR for Rabbit NHE3

Expression of mRNA for the NHE3 was demonstrated in the medulla oblongata of three adult rabbits by means of RT-PCR. A single band of ca. 650 bp was obtained, being weaker but nevertheless identical to that obtained from the rabbit kidney cortex (Figure 1B). The size of this amplified fragment was very close to what could be expected from the sequence of the rabbit NHE3 gene.

Baseline Values in Vivo

To avoid pump-triggered mechanical feedback from the lungs to the respiratory centers, the vagus nerves were cut before artificial ventilation.

Table 1 shows baseline data during spontaneous breathing and during constant artificial ventilation under normocapnic conditions. Normocapnia was adjusted to somewhat higher levels than during spontaneous breathing, to approximately 1 kPa above the apneic threshold Pa_CO2 (see also Figure 5). Blood gases, acid-base values, and cardiovascular variables remained rather constant throughout the experiment.

View larger version (18K): [in this window] [in a new window]   Figure 5.   Effect of NHE3 inhibition on average central respiratory CO2 responses. Mean CO2 responses ± SEM from seven rabbits (same as in Figure 4), of integrated phrenic minute activity (IPNA·fR), integrated tidal phrenic nerve activity (IPNA), and respiratory rate (fR) before (closed circles) and after (open circles) cumulative application of S8218. Mean absolute values underlying the ratios shown by Figure 2 before and at the end of infusion are also included. Significant difference compared with the control condition at corresponding levels of PaCO2; *p < 0.05 (paired t test).

View larger version (13K): [in this window] [in a new window]   Figure 4.   Effect of NHE3 inhibition on single CO2 responses of central respiratory drive. Phrenic minute activity (IPNA · fR) as a function of arterial PCO2 (PaCO2) in seven artificially ventilated vagotomized rabbits, under control conditions (left panel  ) and after cumulative application of about 9 mg/kg S8218 (right panel ). One animal was additionally chemodenervated (broken line).

View larger version (15K): [in this window] [in a new window]   Figure 2.   Dose-response of central respiratory drive to S8218. Relative mean change ± SEM in six rabbits of phrenic minute activity (IPNA · fR) and arterial PCO2 (PaCO2) in response to an increasing cumulative intravenous dose (up to 9.2 ± 1.1 mg/kg) of the brain permeant NHE3 inhibitor S8218. Significant change compared with the control condition; *p < 0.05 and **p < 0.01 (paired t test).

Dose-Response Curves

Average dose-response curves of central respiratory minute drive (IPNA · fR) and Pa_CO2 during constant artificial ventilation were estimated in six rabbits with intact carotid chemoreflexes (Figure 2). During cumulative application up to about 9 mg/kg of the NHE3 inhibitor S8218, an overall average plasma concentration of 0.9 ± 0.2 µg/ml (about 2.9 µM) was reached, based on n = 27 measurements. This significantly enhanced IPNA · fR by an average of +51.0 ± 6.4%, p < 0.0001, due to a rise in both respiratory rate (+25.7 ± 4.0%, p < 0.0001) and tidal phrenic activity (+19.8 ± 3.1%, p < 0.001). The rise in fR was caused by about equivalent shortening in TI and TE, the ratio TI/TE being unchanged. No transient change in Pa_CO2 was also induced by infusion of S8218; blood acid-base status (BE) and mean arterial pressure () likewise remaining unchanged. There was, however, a small but significant and dose-dependent decrease in heart rate (average: 8.9 ± 1.1%, p < 0.0001).

Effect of S8218 on CO₂ Response Curves

CO₂ responses were investigated in seven rabbits for the range of Pa_CO2 between that for apneic threshold and up to about 6 kPa above.

In the experiment shown in Figure 3, the apneic threshold was reached at a Pa_CO2 of 3.0 kPa under control conditions. After application of the substance, even much stronger hyperventilation typically did not abolish rhythmic phrenic nerve discharge but elicited irregular small amplitude burst discharges. Because this behavior was observed in four of seven animals, a high average "near-threshold" value of respiratory frequency with a broad range of deviation resulted for the whole group (see Figure 5). As can be seen from Table 2, treatment with S8218 lowered the average threshold Pa_CO2 and PET_CO2 by more than 0.4 kPa (3 mm Hg) at unchanged metabolic acid-base status and oxygenation.

View larger version (56K): [in this window] [in a new window]   Figure 3.   Effect of NHE3 inhibition in a rabbit hyperventilated to apnea. From top to bottom: endtidal PCO2(PETCO2), integrated phrenic nerve activity (IPNA), and phrenic nerve compound potential of a male rabbit (3.11 kg). With NHE3 inhibition (compound potential attenuated by factor 0.5), stronger hyperventilation than under control conditions, causing much lower values of PaCO2 and PETCO2, does not completely abolish rhythmic nerve discharge, but rather elicits irregular oscillations (to be observed in four of seven animals).

The above-threshold CO₂ responses of central respiratory drive are shown in Figures 4 and 5. Additionally, Figure 5 includes the mean phrenic nerve activity during spontaneous breathing, estimated from each rabbit before onset of mechanical ventilation, the spontaneous Pa_CO2 being about 0.5 kPa (4 mm Hg) higher than that at apneic threshold (Tables 1 and 2). Furthermore, Figure 5 includes the mean absolute values underlying the ratios shown in Figure 2 before and at the end of infusion.

NHE3 inhibition (plasma level of S8218 kept at 1.1 µM by slow maintenance infusion throughout the CO₂ test) did significantly augment the central respiratory response in the whole range of Pa_CO2 tested above threshold. Both tidal phrenic amplitude (IPNA) and respiratory rate (fR) were up to 20-25% higher than under control conditions, yielding a considerable average rise of phrenic minute activity by 38.0 ± 8.5% (n = 35, p < 0.0001). In contrast, values of were not different under control conditions and during treatment, and the CO₂-induced diminution of heart rate was abolished by S8218.

Breathing Pattern in Response to CO₂ and upon NHE3 Inhibition

Respiratory pattern analysis was performed for hypercapnia and NHE3 inhibition by plotting IPNA · fR versus IPNA (Figure 6). The majority of widely overlapping data covers the frequency range between 35 and 65 min¹, regardless of whether achieved by elevated PCO₂ or by infusion of S8218. Individual regression analysis revealed highly linear correlations, r ranging between 0.88 and 0.99. The average slopes resulting for CO₂ inhalation (m = 55.3 ± 3.1 min¹) and for S8218 infusion (m = 60.3 ± 3.5 min¹) were not significantly different (n = 7, p > 0.20).

View larger version (19K): [in this window] [in a new window]   Figure 6.   Respiratory pattern analysis for hypercapnia and NHE3 inhibition. Phrenic minute activity (IPNA · fR) as a function of tidal phrenic activity (IPNA), according to Hey and coworkers (23), for control CO2 responses (closed circles) and for responses to cumulative application of S8218 (open circles). Isofrequency lines: a = 65 min1, b = 50 min1, c = 35 min1. Note that individual responses to NHE3 inhibition show patterns similar to CO2 responses, with no tendency to enhance activity in favor of tidal amplitude.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Our main finding was that intravenous application of S8218, a novel brain permeant NHE3 inhibitor, increased respiratory phrenic nerve activity at different levels of Pa_CO2 and decreased the threshold PCO₂ for hypocapnic apnea in anesthetized, vagotomized, paralyzed, and artificially ventilated rabbits.

Constancy of Experimental Conditions

The present experimental approach has the advantage that paired comparisons between control and treatment can be performed for each animal to eliminate interindividual scattering. However, this approach requires stable conditions throughout the experiment to verify the drug effects. For that reason, constancy of anesthesia was attempted by adjusting the appropriate continuous infusion of the anesthetic long before starting the measurements (see METHODS). Likewise, Table 1 may serve to demonstrate stability of preparation and anesthesia, showing no time-dependent changes of characteristic cardiorespiratory variables, neither in the sense of enhanced sympathetic tone or vigilance nor in the sense of progressive metabolic acidosis, which type of influences could have contributed to the lowered apneic threshold and hence left-ward shift of the respiratory CO₂ response upon treatment with S8218.

Furthermore, separate control experiments, albeit conducted for other purposes and with inactive peripheral chemoreflexes, clearly substantiate that CO₂ responses of the same duration as in the present study can be reproducibly elicited. Figure 7 shows repeated CO₂ responses of phrenic minute activity in nine otherwise untreated rabbits, ventilated with O₂-enriched air. No significant change in the apneic threshold PCO₂ (0.07 ± 0.14 kPa, n = 9, p > 0.60) or in phrenic minute activity at above threshold levels of PCO₂ (2.0 ± 5.9%, n = 44, p > 0.70) could be discerned. Based on these different supports, it is very unlikely that the drug effects of S8218 were the result of inconstant depth of anesthesia or other experimental inaccuracies.

View larger version (12K): [in this window] [in a new window]   Figure 7.   Check for time-dependent effects on repeated CO2 challenges. Two successive estimates (controls I and II) of phrenic minute activity (IPNA · fR) as a function of arterial PCO2 (PaCO2) in nine vagotomized rabbits ventilated with O2-enriched air. Mean body weight (3.6 ± 0.1 kg) and anesthesia (initial: 55 ± 3 mg kg1; infusion: 6.8 ± 0.2 mg kg1 min1) are similar as in the study group (Figure 4). Note constancy of PaCO2 at apneic threshold and of central respiratory CO2 drive.

Prevalence of NHE3 in the Brainstem of Adult Rabbits

NHE3 is a major sodium/proton exchanger in intestine and kidney, but shows very little abundance in the brain. Up to now, NHE3 had only been demonstrated within the medulla oblongata of newborn rats, first on the mRNA level (24) and recently also as an immunoreactive protein (18), using monoclonal antibodies raised against a rat NHE3 sequence from the kidney (25). Now the presence of NHE3 mRNA also has been shown in the adult rabbit brainstem, using RT-PCR (Figure 1B). Also in this species NHE3 immunoreactive neurons occur in brainstem areas, which have a prevalence for central chemosensitivity and/or rhythm generation (Figure 1A). Based on these findings, we felt encouraged to use the rabbit for testing effects of NHE3 inhibition by S8218 on the respiratory system in vivo.

Role of NHE3 versus Other NHE Subtypes in Central Chemosensitivity

Our experimental approach is the first to study the NHE3 blocker S8218 in vivo. The results clearly demonstrate an activating effect of systemically applied S8218 on phrenic nerve activity at different levels of Pa_CO2 in mechanically ventilated rabbits. The observed effects appear analogous to the enhanced bioelectric activity induced by NHE3 inhibition in cultured chemosensitive neurons from the ventrolateral medulla of newborn rats (18), and are compatible with a putative role for the NHE3 in central respiratory chemosensitivity of adult rabbits. Bulbospinal neurons related to the vascular system most likely are not involved, as we did not observe activating effects of S8218 on the baseline level and CO₂ response of arterial blood pressure and heart rate.

Experiments in rats, reaching similar mean plasma concentrations as in the rabbit study (0.8 ± 0.1 µg/ml, n = 5) have shown that S8218 can easily cross the blood-brain barrier and is enriched in the brain tissue by factors between 8 and 15 compared with blood. Based on these considerations, in our experiments a concentration up to 15 µM is to be expected in the central nervous system (CNS), sufficient to block completely human or rat NHE3 (IC₅₀ 1.6 and 0.8 µM, respectively). Due to this excellent penetration through the blood- brain barrier, the concentrations of the NHE3 inhibitor we used leave no doubt about its primary action within the CNS. In the blood compartment, no effect on (renal) metabolic acid-base regulation could be seen. Likewise, a diuretic effect, which could have been expected from NHE3 inhibition in the proximal tubule, was not observed in rats even at 50 mg/kg, most likely due to the lipophilicity of the compound. A leading role for NHE3 in central chemosensitivity is supported by our in vitro studies on organotypic cultures from the ventrolateral medulla oblongata of neonatal rats. By immunoreactivity, NHE3 was shown to be localized at the outer surface of neurons, pointing to a functional role for this protein as a membrane transporter. Selective inhibition of the NHE3 with substances S1611 or S3226 activated only CO₂-sensitive neurons (18), whereas NHE1 inhibition widely failed to induce hypercapnia-like responses. With regard to the order of magnitude and time course, intracellular acidification by NHE3 inhibition closely mimicked that induced by CO₂.

In much the same way as in vitro, NHE1 inhibition also failed to induce hypercapnic-like cardiorespiratory effects in vivo. Microinjections of 100 µM ethyl-isopropyl-amiloride (EIPA) into the ventrolateral medulla did not add to preganglionic sympathetic nerve activation elicited by hypercapnic brainstem perfusion in the cat (26). Likewise, cisternal infusion of amiloride between 10⁵ and 10² M failed to change the CO₂ sensitivity of pulmonary ventilation in anesthetized rabbits (27, 28). Although no effect of amiloride could be found on the ventilatory response to hypercapnia, this NHE1 blocker was able to lower the bicarbonate concentration within the CSF, implying a small rise in baseline ventilation, mainly due to an augmented tidal volume (27). The constancy of acid-base variables upon NHE3 inhibition excludes metabolic acidosis as being the cause for the observed leftward shift of the CO₂ response and for the respiratory pattern response elicited by S8218 (see below).

Although we cannot completely exclude possible side effects on other NHE subtypes due to high drug concentrations in brain tissue, the involvement at least of NHE1 appears rather unlikely. This is also supported by in vitro studies showing NHE1 and NHE2 inhibition with amiloride and HOE 642 to reduce neuronal activity, at least in the hippocampus, where all types of NHE (except NHE3) are present (29).

A possible interference of S8218 with NHE5 can likewise not completely be excluded, as recently Erlichman and Leiter observed a rise in pulmonary ventilation upon NHE5 inhibition in mice (personal communication). However, NHE3 inhibition by S8218 had no effects in guinea pig hippocampal slices, where the NHE5 inhibitor harmaline clearly suppressed bioelectrical activity (30).

Role of Peripheral Chemoreceptors

Unintentional contributions of peripheral chemoreceptors were primarily prevented by maintaining high arterial oxygen levels throughout the experiment (Tables 1 and 2). Likewise, no metabolic acidosis did develop, which could have been able to lower the apneic threshold PCO₂ via carotid chemoreflex activation (31). As far as the influence of NHE inhibition on peripheral chemoreceptors is concerned, Na⁺/H⁺ exchange sensitive to 150 µM EIPA has been demonstrated in type I glomus cells (32). However, at plasma levels ranging about 10⁶ M of the blocking agent S8218, unspecific inhibition of NHE1 (IC₅₀ of S8218 for human NHE1 about three times higher) is rather unlikely. Up to now, there is no direct evidence for NHE3 being involved in peripheral chemoreception. Previous studies have shown that in nonhypoxic rabbits, carotid chemoreceptor activation by CO₂ is minimal, and to a negligible amount contributes to the ventilatory CO₂ response (33). Quite in agreement with these observations, the one chemodenervated animal in the present study did not differ from the intact group with respect to lower Pa_CO2 at apneic threshold and higher CO₂ responses upon NHE3 inhibition (Figure 4, broken lines). Although verification is needed by more experiments with peripherally chemodenervated animals, much evidence up to now favors a predominant effect of NHE3 inhibition on central respiratory chemosensitivity.

Intracellular [H⁺] as the Adequate Stimulus to Central Chemosensitivity?

Our results suggest that the central respiratory system is activated by NHE3 inhibition. Assuming that NHE3 at medullary chemosensitive neurons is blocked by S8218, inhibited acid extrusion and hence augmented intracellular acidification could be the proximate stimulus for the induced respiratory drive. In agreement with this assumption, intracellular pH measurements in brainstem slices (15, 17) and organotypic cell cultures of medullary neurons (16, 18) point to pH_i as an adequate stimulus for central respiratory chemosensitivity. The extracellular pH also seems to be involved, although more indirectly, as among numerous effects, including ion channel gating, transmitter release, or metabolic changes (34), pH_e modulates the degree of acid extrusion in chemosensitive sites (17) and changes the affinity of receptors for neuronal transmitters (29, 30). In principle, all these alterations may also modify or even orchestrate the response of the respiratory network to hypercapnia.

Up to now, quantitative relationships have been established only between neuronal activity and extracellular pH. Thereby, recordings of neuronal activity in brainstem slices revealed that most cells did not respond uniquely to a reduction in extracellular pH (by various CO₂/HCO₃ combinations), but were also activated by high CO₂ at normal pH_e (13). Likewise, in explant tissue cultures of the medulla oblongata neuronal firing frequency was enhanced only by hypercapnic but not by isocapnic acidosis (12).

Similarly, reports on isolated brainstem spinal cord preparations of the neonatal rat are not unequivocal with respect to the role of extracellular pH. Whereas superfusion with acid solutions at constant PCO₂ generally led to increases in respiratory (phrenic or hypoglossal) motor drive (9, 11, 35), isohydric hypercapnia was effective in only some preparations (11, 35), or completely failed to drive respiratory motor output (9). Surface acidification with CO₂/HCO₃-free HEPES buffer was nearly ineffective (11) or even depressive on respiration (9), comparable to in vivo studies lacking ventilatory responses to cerebral ventricular perfusion with acid nonbicarbonate buffers (3). These contradictory results in reduced systems still imply, although they do not prove, intracellular acidification as a major factor mediating the central respiratory CO₂ drive.

There are also hints for a role of brainstem intracellular pH in respiratory control, at least during hypercapnia, from whole animal studies on carbonic anhydrase inhibition. Respiratory responses to acetazolamide were not correlated with medullary surface pH and small focal applications of the inhibitor reduced and slowed the respiratory response to CO₂ steps, as to be expected from less CO₂ being converted to intracellular H⁺ (36). On the other hand, there is a lack of ventilatory reaction to endogenous brainstem lactacidosis (2, 7), not compatible with pH_i as the only adequate signal under all circumstances. This leaves the possibility that intracellular acidification by sources other than CO₂ is not uniquely converted into a signal for central chemosensitivity.

CO₂ Mimetic Breathing Pattern upon NHE3 Inhibition

Effects of NHE3 inhibition on breathing pattern are visualized as a plot of IPNA · fR versus IPNA (Figure 6). It can be seen that breathing pattern responses to hypercapnia and to infusion of S8218 are qualitatively similar, suggesting a CO₂ mimetic action of NHE3 inhibition. Moreover, statistical analysis revealed no differences in slopes of individual regression lines (23) under both conditions, indicating equivalent frequency contributions to the total respiratory response. Compatible with this observation, respiratory responses to hypercapnia and to NHE3 inhibition seem to be mediated by intracellular acidification of the same subgroup of chemosensitive brainstem neurons, whereby NHE3 is a substantial transporter, widely expressed in sensors for CO₂.

However, regarding the breathing pattern, it is doubtful that respiratory responses to metabolic acidosis are mediated by the same neuronal subsystem. In contrast to NHE3 inhibition, ventilatory drives induced by acid CSF, either low bicarbonate isocapnic or isobicarbonate hypercapnic, exhibit completely different breathing patterns (2). Unequivocally, low HCO₃ ventricular perfusion elicited predominant tidal volume responses (1, 14, 37). Likewise, the eucapnic response of ventilation to cisternal amiloride infusion, leading to CSF acidification, was in favor of VT (27).

Analogous observations on breathing pattern responses in isolated brainstem-spinal cord preparations as well led to the assumption of different subsystems of central chemosensitivity (11, 35). Thus, the possible role of NHE3 in respiratory compensation of metabolic acidosis remains to be elucidated.

Physiological and Pathophysiological Implications of NHE3 Inhibition

The CO₂ mimetic effect of NHE3 inhibition we have shown led to an additional respiratory drive at any level of Pa_CO2 in anesthetized rabbits. Also in healthy humans, the general activity of NHE in body cells was assumed to determine the ventilatory CO₂ response. The maximal NHE activity during intracellular acidification was genetically determined and different between subjects, but rather similar for all body cells in the same individual (38). Compatible with the suggested role for NHE in central chemosensitivity, subjects with strong (thrombocyte) NHE activity only weakly reacted to inhaled CO₂, whereas those with weak NHE activity distinctly enhanced pulmonary ventilation during experimental rebreathing maneuvers (39). Thus, high overall NHE activity was assumed to be linked with CO₂ retainment manifest in "nonresponders" during professional or accidental CO₂ exposure. By analogy, an increased sodium-proton antiporter activity predisposed patients to obstructive sleep apnea (40).

Besides the CO₂ mimetic effect of NHE3 inhibition, a strong interference with respiratory rhythmogenesis could be demonstrated in the present study. In all animals, experimentally induced hypocapnic apnea could not be reached at CO₂ levels as high as under control conditions but required stronger hyperventilation. In four of the seven rabbits studied some respiratory oscillations persisted in spite of the strongest hyperventilation, characterized by irregular high-frequency low-amplitude phrenic discharges. Because hypercapnia in rabbits normally attenuates tonic phrenic nerve activity, in this respect S8218 did not simply act CO₂ mimetic, but may interfere with reciprocal (inspiratory/postinspiratory) network properties in a more complex manner.

Nevertheless, a growing susceptibility to respiratory arrest during sleep and/or to periodic breathing was evident in healthy subjects when the apneic threshold was close to the spontaneous awake level of PCO₂ (41). Therefore, the maintenance of respiratory rhythm by lowering the CO₂ threshold has also been clinically attempted, for example, using acetazolamide for medical treatment of patients with chronic obstructive pulmonary disease (COPD) and/or sleep-disordered breathing (42, 43).

In the present experiments, maintenance of normal respiratory rhythm at levels of PCO₂ lower than under control conditions was achieved by NHE3 inhibition, tentatively in chemosensitive brainstem neurons.

General Conclusion

Systemic application of the brain permeant NHE3 inhibitor S8218 acts on the central respiratory drive and breathing pattern predominantly as a CO₂ mimetic in anesthetized rabbits, most likely through inhibition of acid extrusion from brainstem chemosensitive neurons, which have been shown to express NHE3. Additionally, rhythmogenesis is supported by S8218, due to a lowering of the threshold PCO₂ for central apnea.