Correlation of local changes in extracellular oxygen and pH that accompany dopaminergic terminal activity in the rat caudate–putamen

Authors


Address correspondence and reprint requests to R. Mark Wightman, CB 3290 Venable Hall, Chapel Hill, NC 27599–3290, USA. E-mail: rmw@unc.edu

Abstract

Terminal activity causes an increase in local cerebral blood flow that can be quantified by measuring the accompanying increase in tissue oxygen. Alkaline pH changes can also follow neuronal activation. The purpose of these studies was to determine whether these changes in extracellular oxygen and pH correlate. Fast-scan cyclic voltammetry was used to detect changes in dopamine, pH and oxygen levels simultaneously in the caudate–putamen after electrical stimulation of the substantia nigra in anesthetized rats. The biphasic increases in oxygen and pH followed similar time courses, and were delayed a few seconds from the immediate release and uptake of dopamine. The changes following administration of neurotransmitter receptor antagonists as well as agents that modulate blood flow were identical for oxygen and pH. Two distinct mechanisms were identified that give rise to the oxygen and pH changes: blood vessel dilatation caused by nitric oxide synthesis after muscarinic receptor activation and adenosine receptor activation. We conclude that changes in blood flow accompanying terminal activity cause alkaline pH shifts by the rapid removal of carbon dioxide, a component of the extracellular brain buffering system.

Abbreviations used
BOLD

blood oxygen level dependent

E

potential

fMRI

functional magnetic resonance imaging

L-NAME

N-nitro-l-arginine methyl ester

NO

nitric oxide

SN/VTA

substantia nigra/ventral tegmental area

The brain depends on regulation of the circulation to maintain an adequate energy supply. Control of local blood flow by a variety of endogenous substances has been demonstrated. For example, the neurotransmitters GABA and glutamate are vasodilators in hippocampal slices (Fergus and Lee 1997; Lovick et al. 1999), and dopamine neurons directly innervate intraparenchymal vessels in the cortex causing vasoconstriction (Krimer et al. 1998). Blood flow is also regulated by metabolic byproducts such as adenosine, K+ or H+ (Sandor 1999). Classically, cerebral blood flow was determined by measuring the rate of hydrogen clearance at a platinum microelectrode after hydrogen formation in the brain (Young 1980). More recently, Lowry et al. (1997) have shown that regional increases in tissue oxygen, measured electrochemically, parallel the increases in cerebral blood flow measured by hydrogen clearance during behavioral activation. An increase in tissue oxygen occurs because oxygen utilization rates, although accelerated, are lower than oxygen delivery rates owing to vasodilatation (Fox and Raichle 1986). This increase in oxygen after terminal activation is the basis for the brain imaging techniques positron emission tomography, which measures blood flow, and blood oxygen level-dependent (BOLD) functional magnetic resonance imaging (fMRI), which measures blood oxygenation (Raichle 1998).

Cerebral blood flow is important not only in the delivery of metabolic nutrients but also in the removal of carbon dioxide, a byproduct of oxidative metabolism and component of the brain buffering system. The enzyme carbonic anhydrase catalyzes the hydration of carbon dioxide to produce carbonic acid, which can dissociate to bicarbonate and H+, allowing for rapid buffering of pH transients. Indeed, it has been postulated that an increase in blood flow could cause alkaline pH shifts by removal of carbon dioxide (Urbanics et al. 1978). We have observed pH changes in rats after naturally occuring dopamine transients (Robinson et al. 2001), evoked dopamine terminal activity (Budygin et al. 2001) and during training for intercranial self-stimulation (Kilpatrick et al. 2000). In addition, pH changes of similar magnitude have been observed in brain slice preparations, where no blood flow is present, and mechanisms for these changes due to GABA or glutamate neurotransmission have been elucidated (Kraig et al. 1983; Chen and Chesler 1992a).

The purpose of this study was to determine whether there is a correlation between oxygen and pH changes in the caudate–putamen after terminal activation. Using cyclic voltammetry, dopamine, pH changes and oxygen levels were monitored simultaneously after electrical stimulation of the dopamine cell bodies (Zimmerman and Wightman 1991). Identification of pH signals was verified by ion-selective microelectrodes. A biphasic increase in both oxygen and pH was observed that was delayed a few seconds from the immediate increase in dopamine. Administration of pharmacological agents known to inhibit either changes in blood flow or pH revealed that the pH signal always changed by the same percentage as the oxygen signal. Thus, the alkaline pH changes observed after terminal activation are due to increases in blood flow.

Experimental procedures

Cyclic voltammetry

Cylindrical carbon-fiber microelectrodes were prepared from 5-µm diameter carbon fibers (T-650; Amoco Corp., Greenville, SC, USA) and sealed in pulled capillary glass. The fiber was cut so that about 50 µm protruded from the end of the glass seal. The electrodes were epoxied (Miller-Stephenson, Danbury, CT, USA) and allowed to cure overnight at 100°C and then at 150°C. Electrodes were soaked in purified isopropanol before use (Bath et al. 2000). For dopamine fast-scan cyclic voltammetry, the electrode was held at − 0.4 V, and ramped to 1.0 V and back at 300 V/s every 100 ms. To detect oxygen, the electrode was held at 0 V, scanned to + 0.8 V, then −1.4 V and back to 0 V at 450 V/s every 100 ms. The computer-generated waveform was sent to an EI-400 potentiostat (Cyprus Systems, Lawrence, KS, USA) and data were collected through an acquisition board (National Instruments, Austin, TX, USA) interfaced with a computer. All color plots and cyclic voltammograms were background subtracted, to remove the large electrode charging current, by subtracting the average of a few cyclic voltammograms before the stimulation from each voltammogram. A flow injection system was used to calibrate the electrode response to dopamine, pH changes and oxygen after each experiment. For the oxygen calibrations, oxygen-, nitrogen- and air-saturated buffers were used after each experiment, and the concentration of oxygen in each buffer was calculated as described previously (Zimmerman and Wightman 1991). Using the extended waveform for oxygen, the electrode sensitivity at + 0.6 V for dopamine is 2.5 nA/µm and for pH is 2.0 nA per pH unit. At 0 V, the sensitivity for pH is 6 nA per pH unit and for oxygen is 0.001 nA/µm. The electrode sensitivity at − 1.4 V is 0.05 nA/µm for oxygen and 6 nA per pH unit for an alkaline pH shift.

Ion-selective microelectrodes

Ion-selective microelectrodes were prepared as described by Chen and Chesler (1992b). Two thin-wall capillaries (A & M Systems, Everett, WA, USA) were bound together with shrink wrap tubing (Alpha wire Corp., Elizabeth, NJ, USA), heated until soft, twisted together and then pulled to a point. The tips were broken back to a final combined diameter of 10 µm. One of the capillaries was back-filled with 150 mm NaCl (working barrel) and the other was filled with 150 mm NaCl, 150 mm phosphate buffer, pH 7.4 (reference barrel). The tip was placed in a mixed xylene, trimethylchlorosilane solution (3 : 1 by volume) and a small bolus of solution was repeatedly pulled into and ejected from the working barrel tip using a needle and syringe attached to the top of the electrode via Teflon tubing. The tip was then placed in the hydrogen ionophore cocktail (Hydrogen Ionophore Cocktail A; Fluka, Ronkonkoma, NY, USA) and a slight suction was applied. Once ionophore was present, the open end was sealed with dental wax.

Animals and surgery

Male Sprague–Dawley rats (250–350 g) (Charles River, Wilmington, MA, USA) were anesthetized with urethane (1.5 g/kg, i.p.). The core body temperature was maintained at 37°C using an isothermal pad (Delta Phase Pad; Braintree Scientific, Braintree, MA, USA). The rat was placed in a stereotaxic frame and holes were drilled precisely for the placement of the stimulating and working electrodes according to the atlas of Paxinos and Watson (1986). The carbon-fiber working electrode was lowered into the caudate–putamen (+ 1.2 mm anterior-posterior [AP], + 2.1 mm medial-lateral [ML], − 4.5 mm dorsal-ventral [DV]) and the bipolar stimulating electrode (Plastics One, Roanoke, VA, USA) was lowered to the substantia nigra/ventral tegmental area (SN/VTA) (− 5.6 mm AP, + 1.0 mm ML, − 7.5 mm DV). The dorsoventral coordinate of the electrodes was adjusted for maximal stimulated release. Sixty biphasic stimulation pulses at 60 Hz (2 ms, 125 µA per phase) were delivered every 5 min for pharmacological studies. A silver/silver chloride wire reference electrode was inserted on the contralateral side of the brain. All animal protocols and care were approved by the Institutional Animal Care and Use Committee of the University of North Carolina at Chapel Hill.

Drugs

Theophylline (60 mg/kg), atropine (5 mg/kg), acetazolamide (25 mg/kg), picrotoxin (2 mg/kg), kynurenic acid (300 mg/kg) and N-nitro-l-arginine methyl esther (L-NAME, 100 mg/kg) were used as received from Sigma–Aldrich (St Louis, MO, USA). Each dose was dissolved in 1 mL heated saline and administered i.p.

Statistical analysis

Statistical analyses except cross-correlations were performed using Graph Pad Prism (GraphPad Software Inc., San Diego, CA, USA) with significance tested at the 95% confidence level. Cross-correlations were performed using LabView (National Instruments). To determine the effect of drugs, three traces produced before the drug was given were averaged and compared with the average of three responses collected starting 40 min after drug administration. All data are presented as mean ± SEM.

pH subtraction

Changes in pH can interfere with detection of dopamine because their voltammetric waves overlap. Fig. 1(c) shows the characteristic cyclic voltammogram for dopamine, collected 1 s after stimulation, and Fig. 1(d) that for a pH change, collected 5 s after stimulation. To subtract out the change in pH, two points were chosen on the voltammogram, a reference point where there is no current from dopamine (point 1) and the peak oxidation current for dopamine (point 2) where there could be both dopamine and pH signal. The signal at point 1 (A1, in nanoamperes) is given by

image(1)
Figure 1.

Detection of dopamine and pH changes in vivo. (a) Concentration versus time plots for dopamine in the caudate–putamen. The measured trace, with pH interference, is the dotted line and the pH subtracted trace is the solid line. The line underneath the traces marks the duration of the 60-pulse, 60-Hz stimulation of the dopamine cell bodies. (b) Color plot shows the three-dimensional data collected by cyclic voltammetry. The x-axis represents time, the y-axis represents the applied potential (E), and the current is shown in pseudocolor. The purple and green circle at 5 s is the dopamine oxidation peak and the light colored yellow circle at the same time is due to dopamine reduction. The broad blue and green changes from 7 to 25 s are due to pH changes. The dashed white line shows the potential at which the traces in (a) were taken. (c) Cyclic voltammogram (CV) taken at 1 s after stimulation is characteristic of dopamine. (d) Cyclic voltammogram collected at 5 s after stimulation is characteristic of an alkaline pH shift.

where S1,H+ is the pH calibration factor (in nanoamperes per pH unit) at point 1 determined in vitro after the experiment and pH is the pH change. The signal at point 2 (A2) is

image(2)

where S2,DA and S2,H+ are the calibration factors at point 2 for dopamine and pH respectively. The system of two equations can then be solved for dopamine concentration ([DA])

image(3)

and the effects of the pH signal are subtracted.

Results

Simultaneous measurement of multiple analytes

Multiple analytes can be monitored simultaneously using cyclic voltammetry if each species has oxidation and reduction peaks at unique potentials. The three-dimensional data obtained from cyclic voltammetry experiments are most easily visualized with a color plot (Fig. 1b) (Michael et al. 1998). Time is plotted on the x-axis and scanned potential, starting at the bottom with the oxidative scan, on the y-axis, and current is plotted in pseudocolor. Figure 1 shows data collected in the caudate–putamen using a typical waveform for detecting dopamine, where the electrode was scanned from − 0.4 to 1.0 V and back at 300 V/s every 100 ms. A 60-pulse, 60-Hz stimulation was applied to the SN/VTA region at 5 s. In Fig. 1(b), the dopamine oxidation peak is clearly seen as a purple and green circle centered at 0.6 V starting at 5 s. The reduction peak is more difficult to distinguish, but occurs at the same time as a lighter yellow spot, on the reverse scan at around − 0.2 V. After the dopamine peaks, there are long-lasting current changes that occur over a broad range of potentials due to shifts in the electrode background current. These signals are similar to those observed in vitro in response to an alkaline pH shift.

To confirm the cause of the background shifts in vivo as pH changes, an ion-selective pH-sensitive microelectrode was used. The pH microelectrode was placed adjacent to the carbon-fiber electrode and the pH response after a 5-s, 60-Hz stimulation is shown in Fig. 2. An alkaline pH shift was observed at both the carbon-fiber and pH microelectrodes in the caudate–putamen after stimulation of dopaminergic neurons. In general, the pH response at the carbon-fiber electrode (Fig. 2a) was larger and had a faster rise time than the response of the pH microelectrode (Fig. 2b), suggesting that carbon-fiber microelectrodes might be better sensors for the fast, biphasic pH changes in vivo.

Figure 2.

Alkaline pH signals in the caudate–putamen in response to a 5-s, 60-Hz stimulation of the SN/VTA. (a) Signal measured with a carbon-fiber microelectrode. (b) Signal measured with an adjacent ion-selective pH electrode.

Current versus time traces were obtained from the color plot (Fig. 1), and the currents were converted to concentrations using calibration values obtained after the experiment. The dashed white line on the color plot shows the potential at which the dopamine concentration versus time trace was taken. The dotted line in Fig. 1(a) is the measured current, which rises owing to a change in dopamine concentration and then falls below baseline owing to an alkaline pH change. The solid line is the actual change in dopamine, with the interference from the pH change subtracted out. The dopamine signal has a faster rise than the pH and is quickly cleared by uptake whereas the pH signal has a delayed rise and remains raised. The cyclic voltammograms in Figs 1(c) and (d) were taken 1 and 5 s after stimulation, and are characteristic of dopamine and an alkaline pH shift respectively.

To detect oxygen as well as dopamine and pH, it is necessary to scan to a more negative potential to reduce oxygen (Zimmerman and Wightman 1991). The electrode was scanned from 0 V to + 0.8 V to − 0.4 V and back to 0 V every 100 ms at 450 V/s (Fig. 3). In the example shown in Fig. 3, the purple circle at 0.6 V around 5 s on the oxidative scan and the lighter colored yellow spot on the reductive scan at − 0.2 V are due to dopamine. The broad yellow (on the oxidative scan) and purple (on the reductive scan) peaks between 10 and 25 s are due to pH, and the blue and black areas around the − 1.4 V switching potential are due to the reduction of oxygen. The white lines indicate where the concentration versus time plots for the three species were taken. The oxygen and pH signals are both biphasic and follow similar time courses. Figures 1 and 3 were acquired in the same rat using the same electrode placement and stimulation parameters. As expected, measurements with the two different waveforms produce similar concentration profiles for the measured analytes.

Figure 3.

Simultaneous detection of dopamine, pH and oxygen changes in the caudate–putamen. Concentration versus time traces were obtained from the color plot as marked by the white lines and labeled to the right of the color plot. The bar underneath each trace marks the duration of stimulation. The dopamine signal rises and falls quickly after stimulation, whereas the alkaline pH shift and oxygen signals have a delayed rise and longer-lasting, biphasic signal. The color plot allows visualization of all the changes in oxygen, pH and dopamine.

Correlation of oxygen and pH signals

To investigate whether there was a correlation between the pH and oxygen changes, the pH and oxygen concentrations were averaged for 20 stimulations collected at 5-min intervals in the same rat. The changes for each were quite similar, with two distinguishable peaks for both oxygen and pH (Fig. 4). The inset graph is a correlation plot for the average oxygen and pH signals in this rat, which gives a correlation coefficient of 0.9. When examined in several animals (n = 42), the first oxygen maxima occurred 3–4 s after stimulation. The size and location of the second oxygen peak varied greatly between animals and with respect to both the stimulating and carbon-fiber electrode placement. The average correlation coefficient for the pH and oxygen signals was 0.77 ± 0.03. To determine whether the pH peak was delayed from the oxygen peak, a cross-correlation analysis was performed; on average the pH peak delayed the oxygen peak by 1.3 ± 0.2 s (p < 0.001, one-sample t-test compared with a theoretical mean of 0; n = 42). Average values for the stimulated dopamine, pH and oxygen changes are given in Table 1.

Figure 4.

Correlation of pH and oxygen signals in vivo. Twenty successive responses to a 1-s, 60-Hz stimulation, applied every 5 min, were averaged. The error bars, shown every tenth point, are the SEM. The bars under the traces mark the duration of stimulation. (a) Alkaline pH shift. (b) Oxygen increase. The inset graph is a correlation plot of pH and oxygen signals. The correlation coefficient is 0.9.

Table 1.  Average concentrations of evoked dopamine, oxygen and pH changes in the caudate-putamen
PeakSignal
  1. Signals were evoked by a 60-pulse, 60-Hz stimulation of the SN/VTA and are given as mean ± SEM (n = 42).

Dopamine1.6 ± 0.3 µm
O2 peak 137 ± 3 µm
O2 peak 236 ± 6 µm
pH peak 10.047 ± 0.006
pH peak 20.050 ± 0.009

Pharmacological effects of neurotransmitter receptor antagonists

Several recent studies have suggested that pH and blood flow changes can be attributed to the direct action of neurotransmitters. To test the hypothesis that oxygen and pH changes were mediated directly by neurotransmitters, antagonists of glutamate, GABA or muscarinic receptors were administered and the peak amplitudes were compared before and after drug administration. Sample traces of oxygen signals before and after i.p. administration of saline (1 mL), kynurenic acid (300 mg/kg), picrotoxin (2 mg/kg), and atropine (5 mg/kg) are shown in Fig. 5. These doses were chosen to maximize and compare drug effects because it is not practical to perform a dose–response curve for all these drugs in vivo. The average changes after drug administration, given as percentage of predrug values, are summarized in Table 2; 100% represented no change in signal. For the control experiments, there were no significant changes after saline administration (n = 6). Neither kynurenic acid, an ionotropic glutamate receptor antagonist, nor picrotoxin, a GABAA receptor antagonist, significantly changed any of the oxygen or pH peaks from the predrug values (n = 6). Kynurenic acid significantly decreased the stimulated dopamine release (p < 0.05, n = 6). Atropine, a muscarinic receptor antagonist, decreased both the first oxygen and pH peak to about half of the predrug value (p < 0.01, n = 6), but had no effect on the dopamine signal or the second oxygen or pH peak.

Figure 5.

Effects of neurotransmitter receptor antagonists. Sample traces are given of evoked oxygen signals before (solid line) and after (dotted line) the administration of (a) saline (1 mL), (b) kynurenic acid, a glutamate antagonist (300 mg/kg, i.p.), (c) picrotoxin, a GABAA receptor antagonist (2 mg/kg, i.p.) and (d) atropine, a muscarinic receptor antagonist (5 mg/kg, i.p.).

Table 2.  Changes in dopamine, oxygen and pH signals after pharmacological agents
DrugDopamineO2 peak 1O2 peak 2pH peak 1pH peak 2
  1. Values are percentages of the predrug response given as mean ± SEM (n = 6–8). aAtropine (5 mg/kg) was administered 60 min after L-NAME (100 mg/kg). Value was significantly different (*p < 0.05) from predrug value using a paired t-test.

Control (saline)92 ± 797 ± 8108 ± 14106 ± 13102 ± 9
Theophylline (60 mg/kg)131 ± 4*105 ± 2119 ± 8*90 ± 2411 ± 7*
Acetazolamide (25 mg/kg)94 ± 1447 ± 15*55 ± 20*39 ± 6*51 ± 8*
L-NAME (100 mg/kg)91 ± 1065 ± 7*104 ± 1766 ± 6*90 ± 7
Atropine (5 mg/kg)102 ± 759 ± 6*93 ± 754 ± 7*91 ± 8
Kynurenic acid (300 mg/kg)80 ± 5*89 ± 7106 ± 9100 ± 1094 ± 14
Picrotoxin (2 mg/kg)103 ± 1489 ± 8102 ± 2095 ± 1296 ± 15
L-NAME + atropinea90 ± 1460 ± 13*96 ± 1859 ± 7*100 ± 8

Pharmacological effects of vasoactive agents

To test the relationship between blood flow and pH changes, theophylline, an adenosine receptor antagonist (60 mg/kg, i.p.), acetazolamide, a carbonic anhydrase inhibitor (25 mg/kg, i.p.), and L-NAME, a nitric oxide (NO) synthesis inhibitor (100 mg/kg, i.p.) were administered. Figure 6 shows sample oxygen traces before and after administration of each drug and average percentages of the predrug response are summarized in Table 2. Theophylline, a demonstrated vasoconstrictor, significantly decreased both the second oxygen and pH peak (p < 0.01, n = 6) but did not affect the first oxygen or pH peak. Theophylline was the only vasoactive agent tested that affected dopamine signals, significantly increasing the stimulated dopamine levels (p < 0.005, n = 6). Acetazolamide disrupts brain buffering and is a known vasodilator. After acetazolamide, the dopamine signal was not significantly different but both oxygen and pH peaks were significantly decreased from their predrug values (p < 0.05, n = 8). NO is also a potent vasodilator, whose effect can be inhibited by NO synthase inhibitors, such as L-NAME. L-NAME significantly decreased the first pH and oxygen peaks (p < 0.05, n = 7) but did not affect the dopamine signal or the second oxygen or pH phase. Atropine (5 mg/kg) was also administered 60 min after L-NAME (100 mg/kg) and no further changes in oxygen or pH levels were observed (n = 4).

Figure 6.

Effects of vasoactive compounds. Sample traces are shown of evoked oxygen signals before (solid line) and after (dotted line) i.p. administration of (a) theophylline, an adenosine receptor antagonist (60 mg/kg), (b) acetazolamide, a carbonic anhydrase inhibitor (25 mg/kg) and (c) L-NAME, a nitric oxide synthesis inhibitor (100 mg/kg).

Discussion

It is well known that an increase in local blood flow accompanies an increase in terminal activity (Sandor 1999). In this work, we explored the chemical changes that accompany evoked neural activity in the caudate–putamen. Electrical stimulation of dopamine cell bodies in the SN/VTA region evoked instantaneous dopamine release in the caudate–putamen that was rapidly cleared by uptake (Wightman and Zimmerman 1990). The high time resolution of cyclic voltammetry also allowed biphasic increases in oxygen and pH to be resolved. This increase in oxygen has been documented in previous work with fMRI (Raichle 1998) and electrochemical sensors (Lowry et al. 1997) to arise from increased blood flow. The changes following administration of neurotransmitter receptor antagonists as well as agents that modulate blood flow were the same for oxygen and pH. In contrast, dopamine responded quite differently. Taken together, these data indicate that terminal activity, but not dopamine itself, increased blood flow in the caudate–putamen and caused an increase in local oxygen and pH.

Correlation of oxygen and pH changes

The electrical stimulation used in this study is quite specific. It evokes dopamine release that can be either enhanced pharmacologically, as with theophylline, or diminished, as with kynurenic acid. These responses are entirely consistent with previously characterized adenosine-induced depression (Ferre et al. 1993) and glutamate-induced enhancement (Kulagina et al. 2001) of dopamine release. Like previous experiments in which dopaminergic fibers in the medial forebrain bundle were stimulated, the evoked oxygen and pH changes are less evident when the stimulating electrode is placed away from the dopamine cell bodies. Although long (1 s) stimulus trains were employed in this work, these changes can be evoked with more transient stimuli. For example, the alkaline pH shifts are clearly present in the nucleus accumbens of freely moving rats with electrical stimuli that an animal will self-administer (Kilpatrick et al. 2000). They also accompany naturally occuring dopamine transients, demonstrating that they occur under physiological conditions.

Fast-scan cyclic voltammetry provides sufficient information for changes in all three components to be discerned simultaneously. The method is best suited to measurement of transient changes; therefore, basal levels were not explored. The oxidation of dopamine and the reduction of oxygen occur at potentials sufficiently separated for the two substances to be resolved. The pH signal, verified by correlation with an ion-selective electrode, partially overlaps the dopamine signal and arises from protonation or deprotonation of functional groups on the carbon-fiber surface (Runnels et al. 1999). Because the voltammetric shape and time course of the dopamine and pH changes differ, they can be resolved mathematically. Changes in pH also contribute to the current at which oxygen is detected, but the pH contribution to the oxygen response was sufficiently small in this application and could be neglected.

The dopamine signal, when the pH contribution was removed, revealed rapid release and uptake of the neurotransmitter. In contrast, the oxygen and pH signals were biphasic, lasting many seconds, and their time courses were well correlated. The size and shape of the oxygen and pH signals varied between animals and with electrode placement (as evidenced by the individual oxygen traces in Figs 5 and 6), but were reproducible at one location. The similarity of the shapes suggests that local blood flow, which causes the oxygen transients, also affects local pH. Cross-correlation of the oxygen and pH signals reveals that the alkaline changes lag the oxygen increase by over 1 s. The delay indicates that a kinetic barrier intervenes between local blood flow and local pH change. This step could be conversion of carbonic acid to carbon dioxide, which is removed by the increased blood flow (Huang et al. 1995). Consistent with this, acetazolamide, an inhibitor of carbonic anhydrase (Chen and Chesler 1992b), lowered both alkaline pH peaks. However, acetazolamide is a vasodilator (Frankel et al. 1992), and also caused a decrease in stimulated oxygen because the vessels were already partially dilated before terminal activation.

Mechanisms of oxygen and pH changes

Local blood flow is controlled in part by the receptor-mediated actions of neurotransmitters on blood vessels. In addition, a variety of vasoactive substances such as H+, K+, and adenosine are known. The changes in blood vessel diameter are large (Frankel et al. 1992) and can dramatically affect the local concentration of gaseous substances, as shown in Table 1 for oxygen. Indeed, because the brain is well buffered, only a substance that strongly affects pH, such as carbon dioxide, could cause changes of the magnitude observed (Chesler 1990).

In the caudate–putamen the only neurotransmitter that showed evidence of vasoactivity is acetylcholine. Atropine, a muscarinic antagonist, attenuated both the first oxygen and pH peak with no effect on either second peak. However, acetylcholine must be acting indirectly because the direct effect of acetylcholine on muscarinic receptors on smooth arterial muscle is vasoconstriction (Sandor 1999). In the cerebral cortex, dopamine is a vasoconstrictor (Krimer et al. 1998). However, in the caudate–putamen endogenous dopamine does not affect the local blood flow because inhibition of dopamine synthesis with α-methyl-p-tyrosine did not alter the evoked oxygen signal (Zimmerman et al. 1992). Norepinephrine, a vasoactive agent in the cortex (Adachi et al. 1991), has no effect in the caudate–putamen because yohimbine, an α-2-receptor antagonist, did not alter the oxygen response (Zimmerman et al. 1992). Although activation of NMDA or GABAA receptors has been demonstrated to cause vasodilatation in the hippocampus (Fergus and Lee 1997; Lovick et al. 1999), no change in either oxygen peak was observed after administration of kynurenic acid, an ionotropic glutamate receptor antagonist, or picrotoxin, a GABAA receptor antagonist. Glutamate and GABA have also been shown to directly affect pH in hippocampal slices through receptor-mediated mechanisms (Chen and Chesler 1992a). However, the absence of changes induced by these antagonists indicates that a GABAergic- or glutamatergic-induced pH change disassociated from blood flow does not contribute to the observed changes.

Other vasoactive agents were tested to probe the relationship between blood flow and pH changes. Theophylline, a non-specific adenosine receptor antagonist, decreased the second oxygen peak after terminal activity as well the second pH change. These changes, which we attribute to vasodilatation after activation of A2 receptors on smooth muscles (Phillis 1989), are consistent with the hypothesis that metabolites such as adenosine help to maintain the vasodilatory response (Iadecola 1993). To investigate NO, a known vasoactive agent, endothelial NO synthase was inhibited with L-NAME (Iadecola et al. 1994). The first oxygen and pH peaks were attenuated, similar to the decrease observed after atropine. However, atropine given after L-NAME administration had no additional effect. This suggests an integrated mechanism of action, such as NO release after activation of muscarinic receptors by acetylcholine (Iadecola et al. 1994).

The linked temporal behavior and biphasic nature of the pH and oxygen increases that accompany dopamine neurotransmission indicate that local pH is regulated by local blood flow, as has been demonstrated for local oxygen levels (Lowry et al. 1997). In this study, action potentials were generated by the electrical stimulation of dopamine cell bodies, similar to experiments where parallel fibers are stimulated to cause action potentials in Purkinje cells in the cerebellar cortex (Lauritzen 2001). These Purkinje cell studies have revealed that blood flow changes are related to afferent input function and are independent of efferent spike rate. Our studies show similar results in the caudate–putamen because the blood flow increase is not dependent on dopamine release. At least two distinct mechanisms give rise to the oxygen and pH changes: blood vessel dilatation caused by NO production after muscarinic receptor activation (peak 1) as well as adenosine receptor activation (peak 2). However, none of the agents tested completely eliminated the first peak, indicating the presence of a further mechanism that has not yet been accounted for. In all cases, the relative magnitude of the effect on the oxygen and pH peaks was the same, and there was specificity among the drugs with respect to which peaks were attenuated. Thus, we conclude that increases in blood flow cause alkaline pH shifts by altering the brain buffering system through the rapid washout of carbon dioxide.

Relevance to neurotransmission and its measurement

The alkaline pH shifts that accompany increased local blood flow could have several consequences. Physiologically, large (1 pH unit) alkaline shifts can alter receptor affinities (Vyklicky et al. 1990; Huang and Dillon 1999) and influence ion channel kinetics (Tombaugh and Somjen 1996). The effects of smaller pH changes are not as well understood but, for example, the NMDA channel open probability is strongly modulated by extracellular H+ concentration around physiological pH (Tang et al. 1990). Thus, small changes in extracellular pH could affect neurotransmission. From a measurement perspective, because cyclic voltammetry and other electrochemical techniques, such as chronoamperometry (Gerhardt and Hoffman 2001), are affected by pH, care must taken not to misinterpret pH shifts accompanying terminal activity as dopamine changes.

The time changes of blood flow observed with cyclic voltammetry are similar to those observed in laser-Doppler flowmetry (Malonek et al. 1997), and BOLD fMRI (Logothetis et al. 2001). Whereas fMRI has the obvious advantage of being non-invasive, cyclic voltammetry has better temporal and spatial resolution, so mechanisms of blood flow changes elucidated with cyclic voltammetry might aid in the interpretation of fMRI signals. The measured BOLD signal, the difference in deoxyhemoglobin concentration after a stimulus, probably corresponds to both the oxygen peaks measured with cyclic voltammetry. Recent studies have shown that, following administration of the adenosine receptor antagonists theophylline or caffeine, the change in BOLD signal after neuronal activation is actually greater than without drug (Morton et al. 2002; Mulderink et al. 2002). This contrasting effect may be attributed to the drugs acting as vasoconstrictors, lowering the basal blood flow, so the difference in blood flow is greater after neuronal activation. However, for this explanation to be reasonable there must be an additional mechanism that can increase blood flow after neuronal activation. Indeed, we have demonstrated a separate mechanism, muscarinic receptor activation that leads to NO production, which could increase the blood flow after adenosine receptor antagonism. Thus, the mechanisms for blood flow increases elucidated by cyclic voltammetry help explain how vasoconstrictors can cause increased BOLD contrast.

Acknowledgements

This work was supported by a grant from NIH (NS 15841) to RMW. We thank Dr Mitchell Chesler for instruction in fabricating ion-selective microelectrodes.

Ancillary