Journal of Applied Physiology

Hyperbaric oxygen and chemical oxidants stimulate CO2/H+-sensitive neurons in rat brain stem slices

Daniel K. Mulkey, Richard A. Henderson III, Robert W. Putnam, Jay B. Dean


Hyperoxia, a model of oxidative stress, can disrupt brain stem function, presumably by an increase in O2 free radicals. Breathing hyperbaric oxygen (HBO2) initially causes hyperoxic hyperventilation, whereas extended exposure to HBO2 disrupts cardiorespiratory control. Presently, it is unknown how hyperoxia affects brain stem neurons. We have tested the hypothesis that hyperoxia increases excitability of neurons of the solitary complex neurons, which is an important region for cardiorespiratory control and central CO2/H+ chemoreception. Intracellular recordings were made in rat medullary slices during exposure to 2-3 atm of HBO2, HBO2 plus antioxidant (Trolox C), and chemical oxidants (N-chlorosuccinimide, chloramine-T). HBO2 increased input resistance and stimulated firing rate in 38% of neurons; both effects of HBO2 were blocked by antioxidant and mimicked by chemical oxidants. Hypercapnia stimulated 32 of 60 (53%) neurons. Remarkably, these CO2/H+-chemosensitive neurons were preferentially sensitive to HBO2; 90% of neurons sensitive to HBO2 and/or chemical oxidants were also CO2/H+ chemosensitive. Conversely, only 19% of HBO2-insensitive neurons were CO2/H+ chemosensitive. We conclude that hyperoxia decreases membrane conductance and stimulates firing of putative central CO2/H+-chemoreceptor neurons by an O2 free radical mechanism. These findings may explain why hyperoxia, paradoxically, stimulates ventilation.

  • central chemoreception
  • reactive oxygen species
  • cardiorespiratory control
  • intracellular recording
  • hyperoxia

the central nervous system (CNS) is especially sensitive to oxidative stress. For example, hyperoxia, which is a popular model of oxidative stress, induced by breathing high levels of oxygen at hyperbaric pressure [i.e., hyperbaric oxygen (HBO2)], can rapidly disrupt neural function and result in CNS O2 toxicity (16). Neurological responses to hyperoxia vary, depending on the oxygen tension in the brain and the duration of exposure. For example, the CNS response to hyperoxia can range from moderate, but reversible, changes in neural activity (7, 49), to violent and reversible seizures at higher levels of oxygen (16), to irreversible motor deficits and ultimately death at the highest dosages of hyperoxia (16). In each of these instances, the effects of hyperoxia on the CNS are thought to result from increased production and accumulation of O2 free radicals and subsequent oxidation of cellular components vital to maintaining normal mechanisms of neuronal excitability (16).

The cardiorespiratory centers of the brain stem, similarly, are sensitive to a broad range of inspired oxygen. Hypoxia increases alveolar ventilation, primarily by stimulation of the peripheral chemoreceptors (18), although a central stimulatory effect on ventilation has also been reported (47, 64). Paradoxically, hyperoxia, which decreases peripheral chemoreceptor activity (18), also increases ventilation, thus suggesting that hyperoxia has a central effect (17, 32, 48). For example, breathing high levels of oxygen at normobaric pressure [∼1 atmosphere absolute (ATA)] or at HBO2 early on stimulates breathing in intact mammals (4, 46) and in those lacking peripheral chemoreceptor input (17, 32, 34, 48). When CO2 levels are maintained during hyperoxic hyperventilation (i.e., isocapnic hyperoxia), the increased ventilatory response to hyperoxia is graded and quite large, with a significant stimulation occurring while breathing just 30% O2 (compared with air, 21% O2) at normobaric pressure (4). Similarly, the ventilatory response to hyperoxia continues to increase as high levels of O2 are breathed at hyperbaric pressure up to 8 ATA (17). This paradoxical increase in ventilation is known as “hyperoxic hyperventilation” (48). With continued exposure to higher levels of HBO2, the production of reactive O2 species (24, 27) and oxidative damage rapidly increases (42) to the extent that respiration becomes unstable (17, 46). Thus these effects of hyperoxia on respiration apparently serve no adaptive value and are analogous to the dose-dependent effects of a drug; at low concentrations and/or short exposures, hyperoxia stimulates ventilation in a graded, reversible fashion, whereas, at higher concentrations and/or longer exposures, hyperoxia causes dyspnea and abnormal respiration.

Because the hyperoxic hyperventilatory response was retained after carotid deafferentation (17, 32, 48) and abolished by anesthesia (32), it was proposed that hyperoxia stimulated respiratory neurons and/or central CO2/H+-chemoreceptor neurons of the respiratory control system (17, 32, 48). The cellular mechanism by which hyperoxia stimulates breathing is unknown; however, several possibilities have been proposed, including the Haldane effect with increased arterial Pco2 (45) and direct stimulation of respiratory neurons and/or the central CO2/H+ chemoreceptors (17, 48). In this study, we have tested three hypotheses. The first hypothesis was that hyperoxia stimulates neurons in a cardiorespiratory control area of the brain stem. To test this hypothesis, we studied the effects of HBO2 on neurons from a region of the dorsal caudal medulla oblongata that contributes to cardiorespiratory control and central CO2/H+ chemoreception; specifically, the nucleus tractus solitarius (NTS), which, along with the dorsal motor nucleus of vagus, comprises the solitary complex (SC) (20, 36). If neurons in the SC were sensitive to acute exposure to hyperoxia, then our next hypothesis was to determine whether changes in neuronal excitability during hyperoxia were due to increased O2 free radicals and redox modulation by comparing neuronal activity during hyperoxia to that measured in the presence of an antioxidant or a chemical oxidant. Finally, if neurons in the SC were sensitive to hyperoxia, then our third hypothesis was that hyperoxia stimulates putative central CO2/H+ chemoreceptors in the SC (17, 48). Thus we tested neuronal responses in the same cell to normobaric hypercapnia, HBO2, and HBO2 with hypercapnia. Our findings show that HBO2 decreases membrane conductance and preferentially stimulates putative CO2/H+-chemoreceptor neurons in the SC by a mechanism that involves redox modulation and presumably O2 free radicals. Preliminary reports of these data were previously published (50, 52).


Brain Slices

Slices were prepared from juvenile and adult Sprague-Dawley rats (body weight >75 g), as previously described (22). Anesthesia was not used because these agents have a depressant action on neurons (54) and because these agents antagonize neuronal sensitivity to hyperbaric pressure (68). The animal was killed by rapid decapitation. The brain stem was removed and blocked and then submerged in ice-cold (4-6°C) artificial cerebrospinal fluid (aCSF) of the following composition (in mM): 125 NaCl, 5.0 KCl, 1.3 MgSO4, 26 NaHCO3, 1.24 KH2PO4, 2.4 CaCl2, and 10 glucose at 300 mosM. Transverse slices (300 μm thick) were cut, starting at obex and proceeding rostrally through the medulla oblongata for ∼900 μm. Slices from the caudal medulla contain the dorsal chemosensitive area and cardiorespiratory neurons in the region of the SC (20, 36).

Control Conditions

Under control conditions, aCSF was equilibrated with 95% O2-5% CO2 gas mixture at barometric pressure (Pb) of ∼1 ATA (i.e., normobaric pressure) and 37°C to produce Po2 of ∼720 Torr and Pco2 of ∼40 Torr (36, 51). This control level of Po2, which is typically used for brain slice studies (1, 23), is actually very hyperoxic at normobaric pressure compared with the intact CNS (Refs. 8, 28, 51; and see Table 1). However, because a more physiological control level of tissue Po2 (PtiO2) for the brain slice preparation has not yet been defined (1, 23, 51), we equilibrated control medium with 95% O2 at Pb of 1 ATA to enable comparison of our data with the results of previous studies. Therefore, to study the effects of further hyperoxia on brain stem slices, as in the present study, requires the use of a hyperbaric chamber, where aCSF Po2 values in excess of 760 Torr (1 ATA) can be achieved (22, 44, 51).

View this table:
Table 1.

Intact CNS and brain slice terminology for describing levels of neural tissue PO2

Hyperoxia, Hypercapnia, and the Rationale for Gas Levels Used

Hyperoxia administered at hyperbaric pressure is known as “hyperbaric oxygen, HBO2.” As shown in Table 1, HBO2 is defined in the intact CNS as a PtiO2 in excess of 200 Torr measured when breathing 100% O2 at >1 ATA. However, as stated above, the typical control PtiO2 in the brain slice at normobaric pressure is very hyperoxic, compared with the intact animal breathing normobaric air, and mimics neural PtiO2 during HBO2 (in vivo), only minus the physical pressure component (23, 51, 53). Therefore, in this study, HBO2 in the brain stem slice was defined as an aCSF Po2 >720 Torr at Pb >1 ATA. Brain stem slices were exposed to two different levels of HBO2, depending on Pb (∼1,680 and ∼2,470 Torr of O2) (53). Acute exposure to HBO2 raised Po2 in the core of the brain stem slice to ∼1,000-∼1,500 Torr (51), which is comparable to PtiO2 measured in the intact CNS of rats breathing 95% O2-5% CO2 at 4-5 ATA, or 100% O2 at 5-6 ATA (23, 40). In rats, O2-induced seizures (i.e., CNS O2 toxicity) occur after 3-5 h at 3 ATA O2 and more rapidly at 5-6 ATA O2 (63). Blood-gas measurements in humans show that arterial Po2 increases from ∼90 Torr while normobaric air is breathed to a maximum of 1,900-2,000 Torr while 100% O2 is breathed at 3.5 ATA (45). Hyperbaric oxygen therapy typically uses >1 to 3 ATA O2 (65). Consequently, we felt that the levels of HBO2 used in the present study provided a good model for studying early changes in neuronal excitability during exposure to hyperoxia.

“Normocapnia” in the intact animal refers to an arterial Pco2 ranging from ∼35 to 45 Torr (arterial pH ∼7.35-7.45). Typically, a 25-26 mM bicarbonate-buffered aCSF solution, equilibrated with 5% CO2 at normobaric pressure, produces a comparable aCSF Pco2 of ∼40 Torr and an extracellular pH (pHo) of 7.45 (59). In the brain stem slice, control intracellular pH (pHi) is ∼7.24 under these conditions, and hypercapnic acidosis (going from 5 to 10 or 15% CO2) decreases pHi by ≤0.2 to ∼0.3 pH units (20, 30, 59). “Hypercapnia” (i.e., hypercapnic acidosis) describes aCSF equilibrated with increased levels of CO2 at Pb of 1 ATA. For this study, we increased CO2 from 5 to 15% (Pco2 ∼114 Torr), which produces a pHo of ∼6.9 and a pHi of ∼6.9 in brain stem neurons at 37°C (30), to identify CO2/H+-chemosensitive neurons. By comparison, in the intact animal, breathing 10% CO2 increases Pco2 in the CNS from 42 ± 2 to 82 ± 3 Torr (2), which decreases pHo in the CNS from 7.31 to 7.06 and decreases pHi from 7.05 to 6.93 (2). Similarly, 10% CO2 (and these levels of hypercapnic acidosis and intracellular acidosis) stimulates ventilation (2). Consequently, the level of hypercapnic acidosis that we elected to use (increase from 5 to 15% CO2) represents an appropriate test stimulus. Our laboratory's previous studies indicate that CO2/H+-chemosensitive neurons in the SC have a graded increased firing rate response to 5-15% CO2 (19). Moreover, we have determined that the pHi and firing rate responses to hypercapnic acidosis of putative chemoreceptor neurons in the brain stem are graded, fairly linear responses that are maintained over a very broad range of CO2, including 5-50% CO2 (20). Finally, “hypercapnic HBO2” describes aCSF that was both hypercapnic and hyperoxic, as defined above, and is administered to the brain stem slice at 3 ATA. In the intact animal, the addition of hypercapnia to HBO2 also increases cerebral blood flow and thus increases PtiO2 for any given level of inspired Po2 (45, 46). In the brain slice, which is devoid of blood flow, this will not occur.

Hyperbaric Chamber

Brain stem slices were incubated in control medium (95% O2-5% CO2) at ∼25°C for at least 1 h. Brain slices were selected individually and transferred to a tissue chamber inside a hyperbaric chamber customized for brain slice electrophysiology, the details of which are described elsewhere (22). The tissue slice was supported on a nylon mesh and submerged ∼1 mm deep in a bath of aCSF (∼5-ml volume) that was delivered at a constant flow rate of 2 ml/min by using a high-pressure liquid chromatography pump. This hyperbaric chamber enables us to independently change Pb, Po2, and Pco2 while maintaining an intracellular recording of membrane potential (Vm) in a single neuron (22). The utility of this novel in vitro model lies in its ability to generate a range of PtiO2 values in brain slices that are known to alter CNS function in vivo (3, 17, 46, 48, 63), but at lower levels of total ambient pressure (Pb) (23).


Intracellular recordings of Vm were made by using the Axoclamp 2A microelectrode clamp (head stage gain = 0.1×; Axon Instruments), which, as previously reported (22), is insensitive to the Pb range used in this study (≤4 ATA; i.e., 1 ATA room pressure plus 3 atm gauge pressure). The recording electrode was made from borosilicate glass (1B100F-4, World Precision Instruments) by using a onestage Flaming/Brown micropipette puller (P87, Sutter Instrument). The electrode, filled with 3 M potassium acetate, was connected to the head stage by a Ag-AgCl wire (Medwire) and had a tip resistance that ranged from 100 to 150 MΩ (19, 21). A low-resistance (< 1 MΩ) combination Ag-AgCl and potassium-gluconate agar bridge reference electrode was placed in the tissue bath to complete the recording circuit (22). All voltage recordings were performed in current-clamp (“bridge”) mode. Input resistance (Rin) was measured by using current-pulse protocols created by AxoScope 6.0.3 software and a Digidata 1200A data-acquisition system (Axon Instruments). During each experiment, Vm, Pb (helium atmosphere), and integrated firing rate signals (10-s bins) were displayed on a chart recorder and stored on videotape (Vetter PCM recorder model 400).

Test Conditions

Compression medium: helium. “Normobaric pressure” refers to ambient pressure measured in our laboratory with a mercury barometer, which typically ranged from 738 to 752 Torr (0.97-0.99 ATA, which we have rounded off here as 1 ATA in all of our normobaric experiments). “Hyperbaric pressure” refers to ambient pressure inside the pressurized hyperbaric chamber (i.e., room pressure plus gauge pressure). In our studies, ambient pressure was raised above room pressure by using 100% helium (22, 23). As in previous studies (22, 29, 53), helium was used as the compression medium for testing neuronal barosensitivity. Helium is inert and of low solubility in water and lipid membranes at Pb < 245 ATA (5, 23); therefore, the effects of helium over the range of Pb values used in this study (≤4 ATA) are attributed to the effects of pressure per se rather than the narcotic actions of increased partial pressure of helium (5, 23, 53). Before compression, room air was purged from the chamber atmosphere and replaced with 100% helium. During the ensuing helium compression (2-3 ATA), the Po2 and Pco2 in aCSF, which were established at normobaric pressure, did not increase further because there were no additional sources of O2 and CO2 in the overlying atmosphere to be driven into the slice (22, 23, 49, 51).1 In all experiments, the rate of compression and decompression was controlled at ∼2 ATA/min. Brain slice temperature was maintained at 37 ± 0.5°C during compression and decompression (22).

HBO2 and hypercapnic perfusates. Hyperbaric oxygenated perfusate was made by equilibrating aCSF with 98.0 or 98.3% O2 at 2.2 or 3.3 ATA in separate high-pressure sample cylinders (1-liter volume) to produce corresponding Po2 values of ∼1,640 or ∼2,470 Torr in the perfusate (22, 51). It is important to note that Pco2 of the hyperoxic medium was maintained relative to control (pH ∼7.45) by decreasing the CO2 to 2 and 1.65% at 2.2 and 3.3 ATA, respectively. The high-pressure sample cylinders were maintained 0.2-0.3 atm above hyperbaric chamber pressure so that HBO2 perfusate would flow to the brain slice, which was pressurized to 2.0 or 3.0 ATA helium. Despite this slight pressure differential between the sample cylinders and the hyperbaric chamber, and the large O2 gradient from the perfusate to the chamber atmosphere (see footnote), gas bubbles were rarely observed at the brain slice. Normobaric hypercapnic perfusate was made by equilibrating aCSF with 85% O2-15% CO2 at room pressure and 37°C to produce a perfusate Po2 of ∼650 Torr and a Pco2 of ∼115 Torr. Hypercapnic HBO2 perfusate (i.e., CO2 + HBO2) was made by equilibrating aCSF with 95% O2-5% CO2 at 3.3 ATA to produce a perfusate Po2 of ∼2,380 Torr and Pco2 of ∼125 Torr.

Oxidants and antioxidants. Chemical oxidants used were chloramine-T (CT; Sigma-Aldrich) and N-chlorosuccinimide (NCS; Sigma-Aldrich), both of which are water soluble and fairly specific oxidizers of the amino acids cysteine and methionine (56, 62). However, neither CT nor NCS are thought to produce free radicals. The concentrations of CT (500 μM) and NCS (1.0 mM) were chosen based on their ability to decrease synaptic efficacy in hippocampal slices during 30-min exposures in a manner similar to H2O2 (56). The antioxidant used was Trolox C (Sigma-Aldrich). Trolox C is an analog of vitamin E made water soluble by replacing the hydrophobic alkyl group at carbon 1 with a hydrophilic carboxyl group. The antioxidant properties of vitamin E are conferred by the hydroxyl group at carbon 6, which does not differ between the two compounds. Therefore, like vitamin E, Trolox C can function as an antioxidant by scavenging O2 free radicals as well as repairing some types of oxidative damage (9, 25). The concentration of Trolox C used in this study ranged from 100 to 200 μM, which has been shown to block the electrophysiological effects of H2O2 on hippocampal neurons (56).

Analysis and Data Presentation

Data were analyzed by using pCLAMP 6.0.3, origin 5.0, and Sigmastat software packages. Spontaneous firing rate was integrated into 10-s bins by using a window discriminator and rate meter (model RAD-II-A, Winston Electronics). Additional software packages, AxoScope 7.0 and CorelDraw 8.0, were used for data presentation. Neurons were considered HBO2 sensitive if HBO2 increased firing rate by ≥20%. Paired sample t-tests (P ≤ 0.05) were used to determine when the mean population difference (firing rate, Rin, after-hyperpolarizing potential) differed significantly from zero. Statistical difference between multiple groups was determined by one-way ANOVA and Tukey's multiple comparison test. Contingency tables were used to compare the incidence of HBO2 responsiveness to CO2/H+ chemosensitivity and barosensitivity (53). The significance of associations between these tabulated parameters was determined by using the Fisher's exact test with the Yates continuity correction when appropriate (P ≤ 0.05). All data are presented as means ± SE. Gas tension in the perfusate and tissue are reported in Torr, where 760 Torr is equal to 1 ATA.


The data presented were obtained during intracellular recordings made from 113 SC neurons. A healthy recording had a Vm, as measured by the amplifier, more negative than -40 mV and action potential amplitude >50 mV. These cells had Vm values of -53 ± 6.0 mV (range -41 to -68 mV) and Rin values of 110 ± 4 MΩ (range 36-222 MΩ).

Response to HBO2

In a typical experiment, an intracellular recording was established under control conditions at normobaric pressure. After 10-30 min, the hyperbaric chamber was compressed to 2.0 or 3.0 ATA with 100% helium. This level of helium compression, which was used to mimic the effects of pressure per se (23), was not expected a priori to affect neuronal activity because, in previous studies, compression up to ≤10 ATA had no measurable effects on cellular activity (reviewed in Refs. 23, 43). However, we observed in some SC neurons that Pb ≤ 4 ATA of helium increased firing rate and decreased Rin (i.e., these neurons were barosensitive), but that there was not a strong relationship between neuronal sensitivity to either HBO2 or CO2/H+ and hyperbaric helium. Details regarding barosensitivity of SC neurons to this range of hyperbaric helium and its potential physiological significance are reported in the companion paper (53).

While maintaining a stable recording at hyperbaric pressure for 5-10 min, the perfusate Po2 was increased from control level (∼720 Torr) to HBO2 (∼1,680 or 2,470 Torr) without any additional increase in ambient pressure (23, 51). This level of hyperoxia caused a small membrane depolarization (≤3 mV) and a significant increase in firing rate by 216 ± 39% in 43 of 113 (38%) neurons (Fig. 1, A and B). During HBO2-induced excitation, Rin increased from 110 MΩ by 8 ± 2 MΩ (n = 31) (Figs. 1C and 2C). These responses to HBO2 were usually reversible within 10-15 min of switching back to control medium. The average duration of exposure was 10 min, and neuronal responses to either level of HBO2 [bath Po2 values of ∼1,680 Torr (n = 7) and ∼2,470 Torr (n = 106)] were similar; therefore, electrophysiological measurements made at either level of HBO2 were pooled for further analyses. The remaining 70 of 113 (62%) of SC neurons tested did not respond to this level and duration of HBO2 and were termed “HBO2 insensitive” (e.g., see Fig. 4C, left). In addition, amplitude of the afterhyperpolarization was not significantly affected by HBO2 (t-test, not shown). These results indicate that HBO2 increases Rin (i.e., decreases membrane conductance) and increases excitability of a subpopulation of SC neurons (i.e., not all neurons tested in the SC were sensitive to acute exposure to hyperoxia).

Fig. 1.

The excitatory firing rate (FR) and input resistance (Rin) responses of a hyperbaric oxygen (HBO2)-sensitive neuron to multiple bouts of HBO2 was reversible and repeatable. A: traces of barometric pressure (Pb) and integrated FR (∫FR; impulses/s) and calculated bath Po2 show that compression to 3 atmospheres absolute (ATA) with 100% helium did not affect FR in this neuron. At 3 ATA, repeated exposures to HBO2 (<10 min) reversibly increased FR. The FR response to HBO2 was unchanged over the course of the 3-h recording (all data not shown). B: representative membrane potential (Vm) traces shown on an expanded time scale (resting Vm -58 mV) during the HBO2-induced stimulation. C: expanded average Vm traces (n = 5) during -0.2-nA current injections to measure Rin (magnitude of the Vm change is proportional to Rin and inversely related to membrane conductance). Note that Rin reversibly increased during HBO2. B and C: traces 1-3 were taken at the times indicated in A.

Fig. 2.

Antioxidant blocked HBO2 sensitivity. A: the trace of ∫FR (impulses/s) shows the FR response to 3 bouts of HBO2 (Pb = 3 ATA helium). An initial exposure to 3.3 ATA HBO2 increased FR. After 90 min of incubation in medium containing the antioxidant Trolox C, a second exposure to 3.3 ATA HBO2 did not affect FR. Washing out Trolox C for ∼30 min restored the excitatory FR response to a third exposure to HBO2. B: average Vm traces (n = 5) during -0.2-nA current injections show that the HBO2-induced increase in Rin also was blocked by Trolox C. C: bar graph showing average increase (Δ) in Rin (means ± SE) of HBO2-sensitive (n = 31), HBO2-sensitive plus Trolox C (n = 4), and HBO2-insensitive (n = 43) neurons. **Values significantly different from zero (t-test) at P < 0.001.

Fig. 4.

CO2/H+-sensitive neurons are stimulated by HBO2 and chemical oxidants. A: Pb, bath Po2, and ∫FR (impulses/s) were measured in an SC neuron during exposure to HBO2, hypercapnia, and CT. Helium compression to 3 ATA did not affect Po2 and FR, but, while at 3 ATA, exposure to HBO2 increased bath Po2 and, shortly thereafter, FR. At Pb of 1 ATA, the neuron showed a typical CO2/H+ excitatory response to normobaric hypercapnia for putative CO2 chemoreceptors of the respiratory control system (19). Exposure to the oxidant CT (500 μM) also increased FR. B: the trace of FR shows the response of a CO2/H+-chemosensitive neuron to repeated bouts of 15% CO2. Subsequent exposure to the oxidant NCS (1 mM) also increased FR. Notice that FR responses to chemical oxidants CT and NCS usually were not fully reversible (arrow). C: traces of Pb, segments of Vm on an expanded time scale, and ∫FR (impulses/s) for a CO2/H+-insensitive neuron show that 20 min of 3.3 ATA HBO2 did not affect FR or Rin. Twenty minutes later, after decompression to 1 ATA, exposure to 15% CO2 and CT (500 μM) also did not affect FR.

Antioxidant, Chemical Oxidants, and HBO2

To determine whether HBO2 increases Rin and stimulates firing rate by redox modulation (26) involving free radicals (16, 38), two types of experiments were conducted. First, the antioxidant Trolox C was used to determine whether it could block neuronal sensitivity to HBO2. Second, the chemical oxidants CT and NCS were administered at normobaric pressure and control PtiO2 to determine whether they would mimic neuronal sensitivity to HBO2.

In the first experiment, as demonstrated in Fig. 2A, we found that the antioxidant Trolox C significantly reduced the HBO2-induced increase in firing rate from 256 ± 80% to only 14 ± 4% in the four cells tested. Similarly, the increase in Rin during HBO2 was significantly blocked by Trolox C (0.5 ± 4 MΩ) (Fig. 2, B and C).

In the second experiment, we found that exposure to the oxidants CT (Fig. 3, A-D) and NCS (Fig. 3D), for 5-15 min (14 ± 1 min), at normobaric pressure and control Po2, significantly increased firing rate (240 ± 47% increase) in 12 of 19 (63%) neurons and caused a corresponding significant increase in RinRin = 9.8 ± 4.5 MΩ) (Fig. 3, C and E). Thus CT and NCS had similar effects on neuronal activity, as did HBO2.

Fig. 3.

HBO2 response was mimicked by chemical oxidants. A: ∫FR (impulses/s) and calculated bath Po2 show that 3.3 ATA HBO2 stimulated FR. Similarly, exposure to the oxidizing agent chloramine-T (CT; 500 μM) 45 min later or to 1 mM N-chlorosuccinimide (NCS) (data not shown) also stimulated FR. B: representative Vm traces show that the recording was healthy and stable during exposure to HBO2 and CT. C: average Vm traces (n = 5) during -0.2-nA current injection. The oxidants CT and NCS (data not shown) mimicked HBO2-induced increased Rin. D: bar graph showing the average FR response to HBO2 and chemical oxidants. **Oxidative stress, imposed by either 10 ± 0.5 min of HBO2 or 14 ± 1 min of CT or NCS, significantly increased FR (T-test, P < 0.001) compared with control. *However, the FR responses to chemical oxidants were less reversible after 10-20 min washout than acute HBO2 and remained significantly different from control (ANOVA and Tukey's multiple-comparison test, P < 0.05). Values are means ± SE. E: bar graph showing average ΔRin (means ± SE) of solitary complex (SC) neurons sensitive to HBO2 (n = 31) (redrawn from Fig. 2C) and oxidizing agents (CT, n = 2 or NCS, n = 10). Values significantly different from zero (t-test) at **P < 0.001 or *P < 0.05.

Response of HBO2-Sensitive Neurons to Hypercapnic Acidosis

The above findings indicate that some, but not all, neurons in the SC are sensitive to acute exposure to oxidative stress. To test the hypothesis that central CO2/H+-chemoreceptor neurons are selectively sensitive to oxidative stress (17, 48), we exposed 60 neurons to elevated CO2 at room pressure (i.e., normobaric hypercapnic acidosis), HBO2, and chemical oxidants, in random order. We found that 32 of 60 (53%) SC neurons tested were CO2/H+ sensitive, that is, they exhibited a significant (≥20%) increase in firing rate (Fig. 4), usually with an increase in Rin (data not shown). We attribute this response to hypercapnic acidosis to decreased K+ conductance caused by decreased pHi (21, 30). In addition, 29 of 60 (48%) of the SC neurons were stimulated by HBO2 and/or chemical oxidants. Of the 29 neurons stimulated by HBO2 and/or chemical oxidants, 26 (90%) were also stimulated by hypercapnic acidosis (Fig. 4, A and B). Conversely, only 6 of 31 (19%) HBO2-insensitive neurons were also stimulated by hypercapnic acidosis; most HBO2-insensitive neurons (81%) were insensitive to hypercapnia (Fig. 4C). Therefore, these results indicate that there is a very strong relationship in the SC between neuronal sensitivity to hypercapnic acidosis and to hyperoxia (Fisher's exact test, P < 0.001).

Both CO2 and HBO2 have similar effects on firing rate (increase) and Rin (increase) in SC neurons, which raises the possibility that CO2 and HBO2 might be sensed by a similar mechanism involving increased O2 free radicals. This unusual possibility seems plausible, given that hypercapnia and intracellular acidosis can increase production of reactive oxygen species (38). Moreover, SC neurons undergo a sustained intracellular acidification (i.e., no apparent pHi regulation) during hypercapnic acidosis (59). To test this possibility, we attempted to block the CO2/H+-chemosensitive response with an antioxidant. If an antioxidant could block the CO2/H+ response, it would suggest that CO2/H+-chemosensitive neurons respond to CO2 and HBO2 by a common signaling pathway that involves O2 free radicals. We found that the antioxidant (Trolox C, 100-200 μM) decreased the HBO2-induced increase in firing rate, as shown previously (Fig. 2), but it did not significantly reduce the firing rate response to hypercapnic acidosis in the four CO2-chemosensitive neurons tested (Fig. 5).

Fig. 5.

Antioxidant did not block CO2 sensitivity. A: the trace of ∫FR (impulses/s) and calculated Pco2 and Po2 show the FR response of a CO2/H+- and HBO2-sensitive neuron to 15% CO2 (Pco2 of ∼114 Torr) and HBO2 (Po2 of ∼2,470 Torr), respectively (30 min between exposures). While at Pb of 3 ATA helium, the slice was perfused for 30 min with Trolox C. In the presence of Trolox C (200 μM), a second bout of HBO2 (Po2 of ∼2,470 Torr) had little effect on FR, whereas a second exposure to 15% CO2 (Pco2 of ∼114 Torr) still increased FR. The results from both experiments suggest that hypercapnic acidosis modulates neuronal excitability by a mechanism independent of free radicals. B: bar graph shows %ΔFR for HBO2- and CO2/H+-sensitive neurons in control medium and medium containing the antioxidant Trolox C (100-200 μM). The FR response to HBO2 (256 ± 80%, n = 4) was significantly reduced (14 ± 4%, n = 4) by Trolox C; however, CO2/H+ chemosensitivity (226 ± 52%, n = 4) was not affected by Trolox C (185 ± 53%, n = 4). Values are means ± SE. *Statistical difference was determine by T-test at P < 0.05.

To further test the possibility that CO2/H+ and HBO2 stimulate SC neurons by separate mechanisms, we exposed a group of CO2/H+-sensitive neurons, while in HBO2 alone, to hypercapnic acidosis plus HBO2 (i.e., hypercapnic HBO2). If hypercapnic HBO2 does not change firing rate by an additional amount, it would suggest that CO2/H+ and HBO2 share a common signaling component. Alternatively, if the response to hypercapnic HBO2 were equal to or greater than the summed firing rate change evoked by hypercapnia alone and HBO2 alone, it would suggest that CO2/H+ and HBO2 affect SC neurons by separate signaling mechanisms. We found that hypercapnic HBO2 increased firing rate of all six CO2/H+- and HBO2-sensitive neurons tested (t-test, P < 0.05). In one-half of the cases (n = 3), the firing rate during hypercapnic HBO2 was greater than the summed change in firing rate to each stimulus alone (Fig. 6A). In the other one-half of the cases (n = 3), the firing rate during hypercapnic HBO2 was equal to the sum change in firing rate to each stimulus alone (Fig. 6B). Together, the results from these two sets of experiments (Figs. 5 and 6) indicate that CO2/H+ chemosensitivity is not dependent per se on pHi-induced production of O2 free radicals and suggest that SC neurons are stimulated by both HBO2 and CO2/H+ through two separate signaling pathways. Figure 7 shows our working model for the CO2/H+ and HBO2 sensitivity of SC neurons.

Fig. 6.

SC neurons were stimulated by hypercapnia during HBO2. A: traces of ∫FR (impulses/s), Vm, and calculated Pco2 and Po2 show a typical CO2/H+-sensitive and HBO2-sensitive neuron (30 min between exposures). During HBO2, the slice was exposed to hypercapnic HBO2 (5% CO2-95% O2 at 3.3 ATA), which increased Po2 and Pco2 from ∼720 and 40 Torr to 2,380 and 125 Torr, respectively. As illustrated in A, the FR response of a CO2- and HBO2-sensitive neuron to hypercapnic HBO2 was greater than the sum change in FR caused by hypercapnia or HBO2 alone (representative of 3 of 6 neurons tested). B: trace of FR shows an example of a FR response to hypercapnic HBO2 that was equal to the summation of the individual CO2 and HBO2 responses (representative of the other 3 of 6 neurons tested).

Fig. 7.

Working model of CO2/H+ and HBO2 sensitivity in SC neurons. We hypothesize that hypercapnia and HBO2 stimulate CO2/H+-chemosensitive neurons by separate mechanisms that involve decreased conductance of different K+ channels (i.e., increased Rin) and, in doing so, cause an increase in respiratory output (i.e., increased ∫FR). We also propose that hypercapnic acidosis may increase free radical production (a; Ref. 38) and/or that O2 free radicals produced during HBO2 cause an intracellular acidification (b; Refs. 67, 69). The result of either interaction would conceivably increase the neuronal response to the stimulus. See text at end of discussion for further details. C, control response.


We have described, for the first time, the effects of HBO2, chemical oxidants, and an antioxidant on the electrical activity of individual neurons in the mammalian CNS, specifically, in a cardiorespiratory control area of the brain stem. Our results support the hypothesis that CO2/H+-chemosensitive SC neurons are stimulated by HBO2, presumably by an O2 free radical-dependent mechanism, which results in an increased Rin. Therefore, these results may explain, in part, previous observations that hyperoxia paradoxically stimulates breathing in vivo (hyperoxic hyperventilation) in normal humans (4, 46) and animals (17), as well as in those lacking peripheral chemoreceptor input (32, 34, 48).

Hyperoxia and Oxidative Stress

Hyperoxia has been shown to increase the production of O2 free radicals before the onset of signs of CNS O2 toxicity (27). O2 free radicals have been shown to modulate neuronal excitability (26). For these reasons, hyperoxia is thought to affect CNS function by an O2 free radical-dependent mechanism (16). In this study, we present evidence suggesting that HBO2 increased firing rate and Rin of SC neurons by an O2 free radical-dependent mechanism, and, in doing so, our results provide support for the long-held free radical theory of CNS O2 toxicity (16). This conclusion is supported by our observation that Trolox C blocked the increase in firing rate and Rin during HBO2.

Also consistent with the general hypothesis that HBO2 produces O2 free radicals, which in turn oxidize cellular proteins, was our observation that, like HBO2, chemical oxidants increased firing rate and Rin. The oxidants tested are reportedly “fairly” specific oxidizers of cysteine and/or methionine residues (62). Because they mimic the effects of HBO2, our data suggest that HBO2 sensitivity of SC neurons may result from the oxidation of cysteine and/or methionine residues located in critical cellular proteins. However, the similarities in sensitivity to HBO2 vs. chemical oxidants were more qualitative than quantitative in some aspects. For example, exposures to chemical oxidants tended to be less reversible [Figs. 3D and 4, A and B (arrows)] and stimulated more neurons (63%) than exposures to HBO2 (38%). The direct comparison between the effects of oxidants and HBO2 is complicated because 1) the type and amount of free radical produced during HBO2 (e.g., superoxide, nitric oxide, etc.) will depend on many factors, including metabolic activity, nitric oxide synthase activity, and free iron (16, 38); 2) as demonstrated in vitro, the effects of oxidative stress vary, depending on the specific oxidizing and reducing agents involved (35); 3) unlike CT and NCS, which presumably are stable until they encounter cysteine or methionine, highly reactive O2 free radicals have short life spans and equally short diffusion distances; and 4) in our experiments, chemical oxidants were applied, on average, for a longer duration than HBO2, 14 ± 0.5 min vs. 10.1 ± 0.5 min, respectively. Regardless, these results support our untested hypothesis that more severe or prolonged oxidative stress will cause irreversible damage and, in doing so, have a more disruptive effect on neuronal activity.

Together, given that an antioxidant blocked the effects of HBO2 on firing rate and Rin, whereas chemical oxidants mimicked these effects of HBO2, this suggests that HBO2 may stimulate neuronal activity in the SC by a free radical-dependent mechanism, possibly involving the redox modulation of an outward K+ channel and/or an inward Cl- channel. This possibility is consistent with previous findings that activities of several types of K+ channels are affected by chemical oxidants (35). Our results are the most direct evidence supporting a role of redox modulation in the hyperoxia response of CNS neurons.

Hyperoxia and Respiratory Control: Central CO2/H+ Chemoreception

CO2/H+-chemosensitive neurons, located in several areas of the brain stem (15), are thought to function as central chemoreceptors for the respiratory control system and to provide the primary afferent input to the respiratory rhythm generator (15, 61). Previous in vivo studies in rats have shown that normobaric (32, 48) and hyperbaric hyperoxia (17) stimulate breathing in the absence of input from peripheral chemoreceptors. This so-called hyperoxic hyperventilation was also demonstrated in humans (4), as well as in those lacking peripheral chemoreceptor function (34). These results suggest that hyperoxia has a direct effect on central respiratory control, possibly by stimulating respiratory neurons, such as CO2/H+ chemoreceptors of the respiratory control system (17, 32, 49). Our results show that HBO2 sensitivity is not equivalent among SC neurons and that it is indeed the putative central CO2 chemoreceptors that are the first neurons affected by acute exposure to HBO2.

Although our results may help explain how hyperoxia stimulates breathing, they also beg the question: why are neurons that provide the primary stimulus for breathing (i.e., central CO2/H+ chemoreceptors) highly sensitive to hyperoxia? In the intact organism, hyperoxic hyperventilation will further increase neural PtiO2 after escape from hyperoxic cerebral vasoconstriction (24) and thus exacerbate the effects of oxidative stress. Therefore, hyperoxic hyperventilation is not likely to be an adaptive response for the organism. It is more likely that hyperoxic hyperventilation is an early manifestation of CNS O2 toxicity in which the stimulatory effects of hyperoxia on respiration rapidly deteriorate to respiratory failure with prolonged HBO2 exposures (17, 63).

The levels of hyperoxia used in this study during HBO2 are similar to those that cause CNS O2 toxicity in vivo (23, 51). CNS O2 toxicity is ultimately manifest as grand mal convulsions (16); however, other autonomic and motor symptoms may also occur. For example, bradycardia, dyspnea, and altered respiratory neural reflexes have been reported (17, 33, 63, 66). We found that only 38% of neurons responded to Po2 values as high as ∼2,500 Torr, thus suggesting that the level of hyperoxia used in this study was not excessively large or unphysiological. We suspect that longer HBO2 exposures and/or more potent oxidative stressors will disrupt more SC neurons and have greater effects on neuronal activity. For example, longer exposure to hyperoxia may cause abnormal changes in neuronal excitability that may or may not be reversible (16).

Oxidative stress disrupts respiratory control. For example, chronic oxidative stresses in the form of high levels of iron or alterations in CNS antioxidant levels are thought to contribute to sudden infant death syndrome (37, 58). Furthermore, hyperoxia at normobaric pressure, a commonly used model of oxidative stress (23), has been shown to destabilize breathing in infants presenting with recurrent apnea and cyanosis (6) and to increase the occurrence of sleep-related apneic episodes in rats (12). These results suggest that oxidative stress affects neurons in brain stem respiratory control centers, possibly CO2/H+-chemosensitive neurons, which are thought to provide the primary stimulus for breathing (61). Indeed, our results show that central CO2/H+ chemoreceptors in the SC are highly sensitive (increased firing rate and Rin) to various forms of oxidative stress. In addition, a recent study has shown that microinjection of S-nitrosothiols (which are a byproduct of protein oxidation mediated by reactive nitrogen species) into the NTS stimulated breathing in intact rats (47). Furthermore, ozone, a form of oxidative stress that disrupts breathing, particularly in children, increased Rin and depolarized monkey NTS neurons (11). Together, these observations suggest that the effects of oxidative stress on respiration could be, in part, mediated by CO2/H+-chemosensitive SC neurons. However, it will be important in future studies to determine whether central chemoreceptors located in other regions of the brain stem (15, 58) are likewise sensitive to oxidative environments.

CO2-O2 Interactions

We also tested the possibility that HBO2 and CO2/H+ detection involves a common mechanism involving O2 free radicals. We found that the CO2 signaling pathway in the SC is functionally separate from that of HBO2. This conclusion was supported by the finding that Trolox C blocked the increase in firing rate and Rin during HBO2, but it did not significantly affect the firing rate response to hypercapnia. In addition, we determined that exposure to CO2 during HBO2 significantly increased firing rate, thus further suggesting that CO2/H+ and HBO2 are sensed by separate signaling mechanisms.

It is possible, however, that, despite these separate mechanisms, high CO2 and HBO2 interact in a way that increases sensitivity to each stimulus when applied in combination (i.e., hypercapnic HBO2). There are several ways by which CO2 and HBO2 signaling pathways might exhibit a positive interaction. Hypercapnic acidification may increase free radical production by at least three known mechanisms: 1) acidification can dissociate iron from transferrin (38) and possibly ferritin (57) to facilitate iron catalysis of superoxide and hydrogen peroxide to the very reactive hydroxyl radical (16, 38); 2) hypercapnic acidification could increase the formation of reactive nitrogen species (38); and 3) acidification could affect reduction-oxidation reactions, many of which are pH dependent (60). Conversely, hyperoxia may increase neuronal responsiveness to hypercapnia. For example, previous studies found that hyperoxia caused intracellular acidosis in astrocytes, C6 glioma cells (67), and cardiac myoblasts (69) by reducing glycolysis and increasing ATP hydrolysis. Moreover, previous work in glomus cells found that hyperoxia can affect sensitivity to subsequent exposure to low O2 (hypoxia) and hypercapnia (33).

Possible interactions between CO2 and hyperoxia are of interest because CO2 retention during hyperoxia can occur when hyperbaric gases are breathed, as during underwater diving or as the result of certain medical conditions (e.g., chronic obstructive pulmonary disease) and subsequent O2 therapy (23). Carbon dioxide retention during hyperoxia will exacerbate any changes in respiration that are dependent on central chemoreceptors. For example, previous in vivo studies have shown that acute hypercapnia worsens CNS O2 toxicity by lowering the threshold Po2 at which seizures occur, increasing the severity of seizures and decreasing survival time (3, 13). Rats acutely exposed to 8.3% CO2 (arterial Pco2 = 70.8 Torr), while at 3-4 ATA HBO2, manifest symptoms of CNS oxygen toxicity much earlier than animals at the same Po2 not exposed to CO2 (13). Traditionally, the effects of hypercapnia on O2 toxicity have been attributed to CO2-induced vasodilation of cerebral vessels and increased delivery of hyperoxic blood to the CNS (45). The brain slice is devoid of blood flow; hence, this preparation is well suited to determine whether there is an interaction between CO2 and HBO2 at the cellular level. We found that, of the six neurons stimulated by hypercapnic HBO2, one-half showed a firing rate response that was greater than the sum change in firing rate to each stimulus alone. This observation neither supports nor disproves the possibility that there is a synergistic interaction between CO2 and HBO2. An interaction between the signaling pathways is supported by one CO2/H+- and HBO2-insensitive neuron that was stimulated by hypercapnic HBO2 (data not shown). Clearly, more studies are needed to determine whether CO2/H+ and HBO2 signaling pathways interact at the level of the single neuron.

Hyperoxic Control Perfusate and Control Activity

Our findings also raise an important issue with respect to the use of the rodent brain slice model, which is widely used for the study of neural function. In the vast majority of these studies, the brain slice is perfused with a “control” solution equilibrated with 95% O2 at room pressure. As shown in Table 1, this will result in slice PtiO2 values between 150 and 436 Torr, depending on the thickness of the slice and other factors (23, 51). These values are strikingly hyperoxic compared with normal values of Po2 (normobaric normoxia) in the intact CNS of an animal breathing air: i.e., <10-34 Torr (40, 51) or even as low as 1-3 Torr (28). In fact, PtiO2 in the control brain slice is similar to the PtiO2 measured in the brain of intact rats breathing 100% O2 at 2.2-2.9 ATA (HBO2) (23, 40, 51). By comparison, the level of PtiO2 that occurs in the brain slice during the typical test used for “hypoxia” is actually closer to PtiO2 in the CNS under normoxic air conditions in vivo (Table 1) (40, 41). Thus it will be important in future studies to determine a physiological level of control Po2 for the brain slice preparation. This will be especially important for neurons in the SC, which we show here to be highly sensitive to hyperoxia (23).


Our results may explain, in part, the phenomenon of hyperoxic hyperventilation (4, 17, 32, 45, 48), as well as the disruptive effects of hyperoxia on cardiorespiratory control (63). These results also raise the question as to which other populations of neurons in the CNS, including neurons in other CO2/H+-chemoreceptor areas of the brain stem (15, 61), are selectively sensitive to free radicals and how this phenomenon may contribute to the onset of CNS dysfunction in general (16, 26, 63). Furthermore, the observations reported here raise a technical question regarding the brain slice preparation: that is, what effect does a hyperoxic control perfusate (95% O2 at a Pb of 1 ATA) have on neuronal function in the isolated brain slice preparation (23)? This is an important question, given that hyperoxia can modulate specific groups of neurons (14, 28), as reported here. Therefore, it will be necessary in the future to determine a more nearly physiological control Po2 in the brain slice preparation to fully understand the effects of changes in O2, both hypoxic and hyperoxic, on CNS neuron function.


The research was supported, in part, by National Heart, Lung, and Blood Institute Grant R01-HL-56683 (to R. W. Putnam and J. B. Dean), Office of Naval Research Grant N000140110179 (to J. B. Dean and R. A. Henderson III), and Department of Defense/Office of Naval Research-Defense University Research Instrumentation Program Grant N000140210643 (to J. B. Dean, R. A. Henderson III, and R. W. Putnam). The Wright State University Biomedical Sciences PhD program supported D. K. Mulkey.


The authors thank Phyllis Douglas for technical assistance.


  • 1 We have previously measured an O2 diffusion gradient that extends from the O2-rich perfusate, at a maximum depth of ∼450 μm from the gas-liquid interface, into the low-O2 helium atmosphere above the tissue bath (51).

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract