• 2-Deoxyglucose 6-phosphate;
  • Microdialysis;
  • Intracellular pH;
  • Interstitial pH;
  • Hypercapnia;
  • Ischemia


  1. Top of page
  2. Abstract
  4. NMR measurements
  7. Acknowledgements

Abstract : A unique method for simultaneously measuring interstitial (pHe) as well as intracellular (pHi) pH in the brains of lightly anesthetized rats is described. A 4-mm microdialysis probe was inserted acutely into the right frontal lobe in the center of the area sampled by a surface coil tuned for the collection of 31P-NMR spectra. 2-Deoxyglucose 6-phosphate (2-DG-6-P) was microdialyzed into the rat until a single NMR peak was detected in the phosphomonoester region of the 31P spectrum. pHe and pHi values were calculated from the chemical shift of 2-DG-6-P and inorganic phosphate, respectively, relative to the phosphocreatine peak. The average in vivo pHe was 7.24 ± 0.01, whereas the average pHi was 7.05 ± 0.01 (n = 7). The average pHe value and the average CSF bicarbonate value (23.5 ± 0.1 mEq/L) were used to calculate an interstitial Pco2 of 55 mm Hg. Rats were then subjected to a 15-min period of either hypercapnia, by addition of CO2 (2.5, 5, or 10%) to the ventilator gases, or hypocapnia (Pco2 < 30 mm Hg), by increasing the ventilation rate and volume. pHe responded inversely to arterial Pco2 and was well described (r2 = 0.91) by the Henderson-Hassel-balch equation, assuming a pKa for the bicarbonate buffer system of 6.1 and a solubility coefficient for CO2 of 0.031. This confirms the view that the bicarbonate buffer system is dominant in the interstitial space. pHi responded inversely and linearly to arterial Pco2. The intracellular effect was muted as compared with pHe (slope = -0.0025, r2 = 0.60). pHe and pHi values were also monitored during the first 12 min of ischemia produced by cardiac arrest. pHe decreases more rapidly than pHi during the first 5 min of ischemia. After 12 min of ischemia, pHe and pHi values were not significantly different (6.44 ± 0.02 and 6.44 ± 0.03, respectively). The limitations, advantages, and future uses of the combined microdialysis/31P-NMR method for measurement of pHe and pHi are discussed.

Brain interstitial space is known to undergo significant changes in pH when normal metabolic processes are disturbed. Determination of interstitial pH (pHe) is key for understanding the process of brain cell pH regulation during systemic metabolic acid—base disturbances and characterization of the role of tissue acidosis in cellular injury during cerebral hypoxia or ischemia. Early attempts to measure pHe employed large glass electrodes or surface electrodes (Javaheri et al., 1983b), but the results were limited by technical difficulties. The double-barreled H+ liquid ion exchanger microelectrodes introduced by Ammann et al. (1981) were a major advance in the measurement of interstitial hydrogen ion concentrations in brain. Subsequently, microelectrodes were successfully employed to measure changes in pHe during seizures (Siesjö et al., 1985), metabolic acidosis and alkalosis (Javaheri et al., 1983a), focal ischemia (Nedergaard et al., 1991), global ischemia (Katsura et al., 1992 ; Harris and Symon, 1984 ; Mutch and Hansen, 1984), and hypercapnia (Katsura et al., 1994). Unfortunately, microelectrodes have a number of limitations ranging from sampling a relatively small volume of the interstitial space to errors resulting from possible diffusion of CO2 from exposed brain and problems with accurate and stable calibration. Values for pHe in normocapnic brain have been reported to cover a wide range, from a low of 7.07 (Javaheri et al., 1983a) to a high of 7.49 (Harris and Symon, 1984). More commonly, values in the range of 7.24 (Nedergaard et al., 1991) to 7.32 (Katsura et al., 1994) have been reported. Given the importance of accurate monitoring of pHe changes to the understanding of brain pathologies and the uncertainty in the measured brain pHe via microelectrodes, it seemed reasonable to explore alternative methods for its measurement.

In vivo phosphorus NMR has been used extensively to determine average intracellular pH (pHi) in brain, by taking advantage of the chemical shift of Pi relative to phosphocreatine (PCr) (for review, see Kauppinen et al., 1993). The signal arising from the Pi present in interstitial space, however, is small and difficult to deconvolute from the much larger intracellular Pi signal. The only NMR-based in vivo determinations of pHe are in the report of Portman et al. (1991), who measured a pHe of 7.32 ± 0.02 at a Paco2 of 41.8 ± 2.4 mm Hg in sheep, and a recent report from our lab (Fitzpatrick et al., 1996), in which we identified an extracellular peak at pH 7.25 in isolated canine brains that we believe to represent brain interstitial space.

Microdialysis probes have recently been introduced as a tool for the investigation of the brain microenvironment (Bungay et al., 1990). When implanted in the brains of animals, they can be used to sample various interstitial components including neurotransmitters (Benveniste et al., 1984), Pi (Scheller et al., 1992), malonaldehyde (Waterfall et al., 1995), and lactate (Taylor et al., 1996). Microdialysis probes have also been used to introduce various compounds into the brain, such as the neurotoxin fluoroacetate (Szerb and Redondo, 1993), hydrogen ion (Waterfall et al., 1996), and the lactate transport inhibitor probenecid (Taylor et al., 1996).

We have combined in vivo 31P-NMR and microdialysis to measure pHe in the brains of artificially ventilated, lightly halothane-anesthetized rats. A microdialysis probe was acutely inserted into the frontal lobe of the right hemisphere of rat brains near the center of the volume sampled by a surface coil tuned for the collection of 31P spectra. 2-Deoxyglucose 6-phosphate (2-DG-6-P) was microdialyzed until its peak was detected in the 31P spectrum. When placed in the interstitial space, 2-DG-6-P has recently been shown to be inert metabolically and to cross endothelial and cell membranes very slowly. 2-DG-6-P is localized to the interstitial space even when the interstitial-to-intracellular gradient is extraordinarily high (36 : 1) and during severe pathological conditions (Duong et al., 1998). The pHe was calculated from the chemical shift of 2-DG-6-P relative to PCr. In this report, we described the use of this technique to measure the response of pHe to hyper- and hypocapnia in the rat and to monitor the change of pHe during the first 7 min of ischemia produced by cardiac arrest. The limitations, advantages, and future uses of this method for measurement of pHe are also discussed.


  1. Top of page
  2. Abstract
  4. NMR measurements
  7. Acknowledgements

Animal preparation

Anesthesia was induced in 300- to 350-g Sprague-Dawley rats, using 2% halothane delivered through a mask. Cannulas were placed in both femoral artery and vein to permit monitoring of arterial pressure, glucose, gases, and pH and the infusion of fluids and medications. Body temperature was also monitored and maintained at 37°C with either a heating pad prior to NMR or warm air during NMR. The trachea was then exposed through a midline incision and a tracheotomy was performed. The animal was relaxed with tubocurarine chloride (1 mg/kg i.v., repeated as needed) and ventilated (Harvard Apparatus, Holliston, MA, U.S.A.) with 1% halothane in 1 : 1 NO2/O2 at a rate of 90 breaths/min with the stroke volume adjusted to maintain normocapnia. The skull was then exposed and the temporal muscles excised at the level of the zygomatic arch to minimize muscle contamination of the NMR signal. The insignificant contribution to the NMR signal of skeletal muscle is confirmed by the fact that after 6-7 min of ischemia, the PCr and ATP signals in our preparation are undetectable (data not shown). Typically, NMR signals from high-energy metabolites in ischemic skeletal muscle can persist for hours (Lagerwall et al., 1995). Previous work by Ackerman et al. (1984a) has shown that the NMR signal from blood 2,3-diphosphoglycerate does not significantly contaminate the brain Pi signal. Burr holes, 0.5 mm in diameter, were carefully drilled in the coronal suture 1.5 mm on either side of the bregma. Into the right burr hole, a custom-made brain microdialysis probe (Bioanalytical Systems, West Lafayette, IN, U.S.A.), with a cannula length of 7 mm and a membrane length of 2 or 4 mm, was placed into the cerebral cortex to a depth of ~2 or 4 mm. A nylon screw was placed in the other burr hole. Dental acrylic was used to secure the probe to the screw and skull.

In three rats, a cisternal puncture was performed and 0.1 ml of CSF was withdrawn. Care was taken to obtain the sampling atraumatically and anaerobically. The sample was analyzed for pH and Pco2 (i-STAT, Princeton, NJ, U.S.A.), and a value was calculated for the bicarbonate ion using the equation [HCO3] = 10 ^ (pH + log Pco2 - 7.61).

All animal procedures used in this study were conducted in strict compliance with the NIH Guide for the Care and Use of Laboratory Animals and approved by the University of Wisconsin Center for Health Sciences Research Animal Care Committee.

NMR measurements

  1. Top of page
  2. Abstract
  4. NMR measurements
  7. Acknowledgements

NMR data were collected on a DMX-400 wide-bore (9.4-T, 89-mm) spectrometer manufactured by Bruker Instruments (Bellerica, MA, U.S.A.). The animals were placed in a custom-made NMR probe and the skulls secured to a single-turn (2.3-cm-diameter) surface coil through which the microdialysis probe was placed. The circuit was a single-tuned (31P, 161.97-MHz) balanced-match radiofrequency circuit design (Murphy-Boesch and Koretsky, 1983). The animal was enclosed in air-bubble insulation (1/16 in thick), and a plastic tube delivered warm air into the probe to facilitate maintenance of body temperature. The animal was again immobilized, and an amnestic level of halothane (0.2%) was delivered in 1 : 1 N2O/O2 to minimize the metabolic effect of the anesthetic on the NMR spectra. Small-diameter (0.12-mm i.d.) fluorinated ethylene polypropylene Teflon tubing (1.5 m) connected the microdialysis probe to a syringe pump (Stoelting, Wood Dale, IL, U.S.A.) for the infusion of 2-DG-6-P. After placement in the magnet, the sample was shimmed until the linewidth of PCr was <80 Hz. Spectral data were collected by first applying a low-power (0.2-W) off-resonance (38.9 ppm relative to center frequency) square pulse for the suppression of bone phosphate signal followed by a 60-s (75-W) spectral excitation pulse on-resonance. The use of preacquisition saturation techniques for effective removal of baseline bone hump in rats has been validated by others (Ackerman et al., 1984b). For the hypercapnia studies, a series of 100 2K free-induction decays (FIDs) were summed using the following parameters : acquisition time 185 ms, time of repetition 3 s, and sweep width 34.2 ppm. For the study of ischemia, a series of 20 2K FIDs were collected. To improve signal-to-noise ratio, the data were truncated to 1K points, zero-filled to 4K, and multiplied by a 15-Hz exponential decay filter before the Fourier transform.

Peak fitting

Postacquisition processing to correct the baseline and fit the 2-DG-6-P and Pi peaks was accomplished with PeakFit (SPSS, Chicago, IL, U.S.A.). PeakFit uses the Marquardt-Levenberg algorithm for nonlinear fitting. It differs from similar software in that it makes possible adjustments of the preliminary estimates of the parameters based on the visualization of their effect on the actual fit before the start of each iteration. This procedure decreases the possibility of iterations converging to local minima. For a more complete discussion of our validation procedure, see Gilboe et al. (1993) and Fitzpatrick et al. (1996). The overlapping of the phosphomonoester (PME) resonances with the 2-DG-6-P signal can present a significant problem in fitting (Soto et al., 1996). This is minimized by collecting control spectra before 2-DG-6-P infusion and fitting the PME region (typically two peaks). This fit was subtracted from the spectra collected following 2-DG-6-P infusion to ensure an accurate measurement of the 2-DG-6-P chemical shift. Because pHe is more alkaline than pHi, the overlap is significantly less than when 2-DG-6-P is used to measure pHi. During ischemia, when the PME resonances shift significantly with intracellular acidification, the data were analyzed as a running summation of five 1-min files. The PME fit of each summation was compared with the previous PME fit to estimate the shift of the PME resonances during ischemia.

Calculation of pH from Pi and 2-DG-6-P spectra

The pH of intracellular and interstitial space was calculated from the chemical shift of Pi and 2-DG-6-P, respectively, with reference to PCr (Moon and Richards, 1973). PCr has a pK of 4.6. Because it does not ionize significantly in the range of these studies, it is a stable reference for measuring the chemical shift of Pi and 2-DG-6-P. The standard curve for Pi was constructed at 37°C by varying the pH and measuring the chemical shift of Pi in mock intracellular solutions containing 5.65 mM KH2PO4, 9.68 mM Na2HPO4, 119.6 mM KCl, 3.2 mM MgCl2, and 10 mM PCr (Fitzpatrick et al., 1996) and fitting the data to a form of the Henderson-Hasselbalch equation. The standard curve for 2-DG-6-P was also constructed at 37°C in the same manner by varying the pH of mock interstitial solutions containing 130 mM NaCl, 1.2 mM CaCl2, 0.8 mM MgCl2, 0.5 mM NaH2PO4, 3 mM KCl, 10 mM PCr, and 10 mM 2-DG-6-P (Katsura et al., 1992). The pHi was calculated from the standard curve for Pi using the equation pHi = log[(δs - 3.26)/(5.70 - δs)] + 6.75, where δs is the chemical shift of Pi (ppm) with reference to PCr. The pHe was calculated from the standard curve for 2-DG-6-P using the equation pHe = log[(δs - 4.18)/(7.56 - δs)] + 6.25, where δs is the chemical shift of 2-DG-6-P (ppm) with respect to PCr [see Ackerman et al. (1996) for a discussion of the limitation of NMR chemical shift pH measurement].

Experimental protocol

Fifteen minutes of control spectra was acquired in each of five animals. Then, 600 mM (pH 7.41) 2-DG-6-P (Sigma Chemicals, St. Louis, MO, U.S.A.) was microdialyzed at 0.1 ml/h until a prominent peak appeared at a slightly lower frequency than the PME region in the rat brain 31P spectra (~60 min). Hypercapnia was then induced in eight rats by adding 2.5, 5, or 10% CO2 to the intake gases of the ventilator. After a 5-min delay to allow for stabilization, arterial blood was sampled for measurement of blood pH, gases (i-STAT), and glucose (Lifescan, Milpitas, CA, U.S.A.). The measurements were repeated for each of three 5-min NMR acquisitions. The addition of CO2 was then discontinued and normocapnia reestablished.

In two rats, hypocapnia was induced after 2-DG-6-P microdialysis by increasing the ventilation rate from ~90 to 130 breaths/min and the ventilation volume from ~2.0 to 4.0 ml. Arterial blood was sampled until the Pco2 fell to <35 mm Hg. Three 5-min NMR acquisitions were collected along with measurements for arterial pH and blood gases. Normocapnia was then reestablished.

Following 2-DG-6-P microdialysis in five rats, 10 1-min control spectra were acquired. Ischemia was produced by rapidly injecting 2 ml of air into the femoral vein. Arterial blood pressure in the rat typically fell to near zero in <30 s, and 20 1-min ischemic spectra were collected.


In three rats, the extent of 2-DG-6-P diffusion away from a 2-mm microdialysis probe was assessed by 2-[14C(U)]DG-6-P autoradiography. 2-[14C(U)]DG-6-P was prepared by reacting 5 mCi of 2-[14C(U)]deoxyglucose (50 mCi/mM ; American Radiolabeled Chemicals, St. Louis, MO, U.S.A.) with a solution containing 50 mM Tris buffer (pH 8.1), 12 mM ATP, 12 mM MgCl2, and 36 U of hexokinase. Total volume was 270 ml. The reaction was allowed to proceed for at least 30 min. Then, 90 ml of the reaction mixture was added to 1.0 ml of 600 mM 2-DG-6-P and microdialyzed for 1 h under conditions identical to the NMR studies. The brain was then carefully removed, frozen, and stored at -70°C until further analysis. Frozen brains were cut into serial 20-μm-thick coronal sections, and 4 of every 10 sequential sections were exposed to Kodak Bio Max X-ray film for 6-8 days ; other selected sections were stained with thionin. The volume of tissue within the sampling zone of the NMR surface coil (2- to 3-cm radius) that was labeled by infusion of 2-[14C(U)]DG-6-P was determined by measuring the area of 14C-labeled tissue in each of the autoradiographic sections that corresponded to tissue under the coil, i.e., 4 mm in the rostral and caudal directions from the microdialysis probe ; the location of the probe was identified by tissue damage evident in the autoradiographs and histological sections. The area of labeled tissue and local 14C concentrations were determined with an MCID computer-assisted image analysis system (Imaging Research, St. Catharines, Ontario, Canada) calibrated with a distance standard (Imaging Research) and [14C]-microscales (Amersham, Arlington Heights, IL, U.S.A.). The mean area for each group of four serial sections was used to calculate the volume of 14C-labeled tissue for each group of 10 sections (i.e., 200-μm-thick tissue) ; these values were summed to obtain the volume of labeled tissue under the NMR surface coil. Labeled tissue beyond the radius of the NMR coil was excluded from analysis.


  1. Top of page
  2. Abstract
  4. NMR measurements
  7. Acknowledgements

Table 1 shows the physiological parameters monitored during hypercapnia. Addition of either 2.5, 5, or 10% CO2 to the ventilator gases caused the arterial Pco2 of the rat to increase to 51.3 ± 0.9, 62.1 ± 1.2, or 83.1 ± 1.2 mm Hg and arterial pH to decrease to 7.27 ± 0.01, 7.13 ± 0.01, or 7.08 ± 0.02, respectively. There was no significant change in arterial blood pressure or blood glucose during hypercapnia.

Table 1. Physiological parametersData are expressed as means ± SEM.
 pH Pco2 (mm Hg) Po2 (mm Hg)Blood pressure (mm Hg)Glucose (mg/%) Temp. (°C)
Control7.38 ± 0.0139.2 ± 1.2236 ± 40128 ± 4108 ± 436.9 ± 0.1
2.5% CO27.27 ± 0.0151.3 ± 0.9276 ± 44130 ± 4110 ± 636.8 ± 0.1
5% CO27.13 ± 0.0162.1 ± 1.2232 ± 27136 ± 4106 ± 737.6 ± 0.1
10% CO27.08 ± 0.0283.1 ± 1.2286 ± 45124 ± 11122 ± 1437.5 ± 0.1

In three rats, the cisternal bicarbonate values averaged 23.5 ± 0.1 mEq/L, whereas arterial pH averaged 7.40 ± 0.03 and arterial CO2 averaged 38.0 ± 2.1 mm Hg.

2-[14C(U)]DG-6-P microdialysis and autoradiography

In the autoradiograph shown in Fig. 1, the 60-min infusion of 2-[14C(U)]DG-6-P through a 2-mm microdialysis probe labeled 41 mm3 of tissue under the NMR surface coil, mainly cerebral cortex. In the coronal section shown in Fig. 1, the area of labeled tissue was 6% of the total tissue area, and the concentration of unlabeled 2-DG-6-P, estimated from the specific activity of the infused 2-[14C(U)]DG-6-P and measured tissue 14C level, was ~50 μmol/g of tissue. Although deeper placement with a 4-mm dialysis probe into brains of other animals labeled more tissue (111 ± 19 mm3, n = 3), there was a danger of placing the dialysis probe near the lateral ventricle, enabling the 2-DG-6-P to spread from the infusion site to other regions of brain via movement of the CSF. Labeling of periventricular tissue was evident in these autoradiographs (not shown).


Figure 1. Tissue labeling by microdialysis infusion of 2-[14C(U)]DG-6-P. 2-[14C(U)]DG-6-P was infused into the anterior right cerebral cortex for 60 min via a microdialysis probe (0.2-mm diameter). The location of the probe in the dorsomedial region of cerebral cortex is shown in the thionin-stained (thionin stains mainly nuclei) coronal section of brain (upper panel). Distortion of the microdialysis probe track is due to tissue damage and folding during removal of the probe from the brain and subsequent tissue processing (i.e., sectioning in the cryostat microtome and histological staining). The autoradiograph of an adjacent serial coronal section (lower panel) illustrates the area of cerebral cortex labeled by 2-[14C(U)]DG-6-P. Because the labeling was restricted to tissue surrounding the dialysis probe, the anatomical structures of the brain cannot be distinguished from background in the autoradiograph ; the dashed lines that outline the 14C-labeled tissue in the lower panel were therefore superimposed on the stained section in the upper panel to identify the approximate location of the labeled tissue that was assayed by NMR.

Download figure to PowerPoint

Microdialysis of 2-DG-6-P

Figure 2 shows a representative 5-min 31P spectrum of the rat brain after 45 min of 2-DG-6-P microdialysis. The 2-DG-6-P peak is clearly separated from the PME resonances. Figure 3 shows a representative 5-min collection of the PME region of the rat brain 31P spectra after 45 min of 2-DG-6-P microdialysis. It is compared with a control 5-min spectrum. It is generally agreed that PCr is undergoing isotropic motion in the cytosol. There is no reason to believe that the intracellular PCr and extracellular 2-DG-6-P resonance linewidths should be different. Figure 4 shows the results of subtracting the preinfusion PME spectrum from the PME spectrum after 45 min of 2-DG-6-P microdialysis. The resulting peak was fit with a single peak of the same linewidth as the PCr peak. This strongly suggests that 2-DG-6-P is present in a single pH compartment.


Figure 2. Representative 5-min 31P spectrum of the rat brain after 45 min of 2-DG-6-P microdialysis. NMR parameters are as described in Materials and Methods. 1, 2-DG-6-P ; 2, PMEs ; 3, Pi ; 4, PCr.

Download figure to PowerPoint


Figure 3. Representative 5-min 31P spectra of the rat brain. NMR parameters are as described in Materials and Methods. The spectra have been truncated to highlight the PME region either during control conditions (A) or after 45 min of 2-DG-6-P microdialysis (B).

Download figure to PowerPoint


Figure 4. A : The spectrum from Fig. 3B was subtracted from that in Fig. 3A and the resulting 2-DG-6-P peak was fit with a single peak (0.30 ppm wide). B : The PCr peak from the same spectrum with the scale for signal intensity was reduced four times to facilitate comparison and fit with a peak of the same linewidth. The solid line is the NMR spectrum ; the dashed line is the fitted peak.

Download figure to PowerPoint

The Pi peak was significantly wider than either PCr or 2-DG-6-P and probably represents several pHi compartments. No attempt was made to deconvolute the Pi peak. The average pHi was calculated by fitting the Pi region with a single broad peak. In seven rats, the average in vivo pHe was 7.24 ± 0.01 and the average pHi was 7.05 ± 0.01.

Arterial Pco2 vs. pHe

Figure 5 illustrates the effect of arterial the Pco2 on the pHe as calculated from the chemical shift of 2-DG-6-P. The points are fit to the Henderson-Hasselbalch equation, assuming a pKa for the bicarbonate buffer system of 6.1 and a solubility constant for CO2 of 0.031. The r2 for this fit is 0.91, which is consistent with the view that the bicarbonate buffer system and/or a buffer system with a similar pK dominates in the interstitial space.


Figure 5. Effect of varying the arterial Pco2 on the pHe was calculated from the chemical shift of 2-DG-6-P. In each of 10 rats, a control sample was followed by three consecutive samples during either hypercapnia (2.5, 5, or 10% CO2) or hypocapnia (CO2 < 30 mm Hg). Data were fit with the Henderson-Hasselbalch equation, assuming a pKa of 6.1, a solubility coefficient for CO2 of 0.031, and the bicarbonate ion as a variable.

Download figure to PowerPoint

Arterial Pco2 vs. pHi

Figure 6 shows the relationship between arterial Pco2 and pHi as calculated from the chemical shift of the largest peak in the inorganic phosphate region. These data cannot be properly described by the Henderson-Hasselbalch equation because arterial Pco2 is not directly related to pHi. Instead, the data are fit with a straight line to show that pHi decreases as Paco2 increases. This observation suggests that intracellular buffering during hypercapnia results from a complex system of physiochemical buffers and transport processes. Such processes might include the active extrusion of hydrogen ion from the cell via transmembrane Na+/H+ exchange.


Figure 6. Effect of varying the arterial Pco2 on the pHi was calculated from the chemical shift of Pi. In each of 10 rats, a control sample was followed by three consecutive samples during either hypercapnia (2.5, 5, or 10% CO2) or hypocapnia (CO2 <30 mm Hg). Data were fit with a linear equation (y = -0.0025× + 7.17 ; r2 = 0.60).

Download figure to PowerPoint


Figure 7 shows the change in pHe and pHi during 12 min of ischemia. Because each time point represents 1 min of summed data, the calculated pH values are assumed to approximate the pH value at the halfway point of collection. Following 30 s of ischemia, pHe has decreased by 0.28 pH unit, whereas pHi has decreased by only 0.06 pH unit. This supports the widely held view that cells buffer sudden changes in hydrogen ion concentrations at the expense of pHe. By 5.5 min of ischemia, the values for both pHe and pHi are not significantly different.


Figure 7. Effects of 12 min of total ischemia on brain pHe (•) and pHi (▪). pHe was calculated from the chemical shift of 2-DG-6-P, and pHi was calculated from the chemical shift of Pi in 1-min spectra collected from five rats. Data are expressed as means ± SEM.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  4. NMR measurements
  7. Acknowledgements


Normal interstitial fluid pH values determined by microelectrodes have been reported to range from 7.06 to 7.49 (Javaheri et al., 1983a ; Ohno et al., 1989 ; Katsura et al., 1991 ; Nedergaard et al., 1991). It is generally conceded that the normal interstitial fluid bicarbonate value is probably very similar to the bicarbonate value in CSF. The human CSF bicarbonate level is reported to be ~23 mEq/L (Altman and Dittmer, 1974 ; Fishman, 1980). Measurements in animals range from a low of 22 mEq/L for rabbits to a high of 25.8 mEq/L for dogs (Altman and Dittmer, 1974). Measurements on cisternal CSF samples in our rat model have yielded values of ~23.5 mEq/L (see Results). This would suggest that during this study, interstitial Pco2 ranged from 31 mm Hg for pH 7.49 to 84 mm Hg for pH 7.06. If we use our pHe value of 7.24 ± 0.01 in the Henderson-Hasselbalch equation, we calculate a Pco2 of 55 mm Hg for normal interstitial fluid.

If one ignores the Gibbs-Donnan membrane effect (this could cause pHe to decrease by 0.02 unit) and the H+ transported out of the cell (proton exchange is thought to be quiescent during normal metabolism), then the factor most responsible for defining interstitial fluid pH is Pco2. The Pco2 of the interstitial fluid should depend on the Pco2 of arterial plasma, the rate of CO2 production by the cell, and blood flow. From theoretical calculations based on the Krogh cylinder model, it has been concluded that interstitial Pco2 need not be much higher than the average of arterial and venous Pco2, a value substantiated by Pontén and Siesjö (1966) in a series of articles reporting studies in which a surface electrode was used to measure brain tissue Pco2. More recently, several authors have questioned the applicability of the Krogh model in brain tissue on anatomic (Davies et al., 1997) and theoretical (Van der Ploeg et al., 1994 ; Hellums et al., 1996) grounds. Others have reported CSF Pco2 values that were similar to jugular bulb blood Pco2 (Bradley and Semple, 1962 ; Pappenheimer et al., 1965) and were ~10 mm Hg higher than arterial blood Pco2. From this, one might conclude that interstitial fluid Pco2 was at least that of venous blood and possibly somewhat higher because of its close contact with the brain cells. Direct measurements of interstitial Pco2 by Hoffman et al. (1996), who placed a multiparameter sensor 4 cm into human brain, are in agreement with interstitial Pco2 being at least that of venous blood or somewhat higher (McKinley et al., 1996).

An important characteristic of NMR chemical shift pH measurements is that the uncertainty in pH calculation is minimal when pH = pKa but can increase at pH values near the titration endpoints. Ackerman et al. (1996) have performed extensive propagation-of-error analysis of a number of organic phosphate species used in pH calculations, including 2-DG-6-P. Their results indicate that to use 2-DG-6-P as a pH indicator in the pH range of interstitial space, an accurate determination of the chemical shift of the base form of 2-DG-6-P is critical. Our value for this parameter (7.56) is within the estimated precision in the carefully determined parameter value of Ackerman et al. (7.57 ± 0.01). Although we cannot eliminate the possibility of some error in parameter estimates because of differences between mock and in vivo interstitial fluids, we believe that the error is not large. Additionally, our NMR-based pH value agrees exactly with the pHe value (7.25) reported by Nedergaard et al. (1991). We believe the evidence suggests that our value for pH, and consequently the Pco2, comes closer to the actual values for interstitial fluid in vivo under control conditions.


Cerebral ischemia, produced by cardiac arrest in the rat, results in a decrease in pHe as measured by changes in the chemical shift of the 2-DG-6-P peak in the 31P spectrum (Fig. 7). The initial rapid fall in pHe reflects the fact that neither interstitial fluid bicarbonate nor the 2-DG-6-P has much buffering capacity at this pH because their pK values, 6.13 and 6.25, respectively, are barely within the range of optimal buffering (i.e., pK± 1.0). From ~ 0.5 to 5.5 min, the interstitial fluid buffers appear to resist the pH change caused by H+ efflux from the cells. From 5.5 to 12 min, interstitial [H+] is more or less in equilibrium with intracellular [H+]. Acidification of interstitial space during the early phase of ischemia probably includes a number of mechanisms including the rapid rise in interstitial Pco2 resulting from cessation of blood flow (Ljunggren et al., 1974) and the titration of available HCO3- and 2-DG-6-P by the extrusion of hydrogen ion from cells via the Na+/H+ antiporter (Katsura et al., 1994). Mutch and Hansen (1984), Kraig et al. (1983), and Katsura et al. (1992) have reported similar results using microelectrodes.

As ischemia continues, both pHe and pHi continue to decline until they stabilize at about a pH of 6.44. At lower pH values, determination of pHe via the 2-DG-6-P method is complicated by the overlapping of the PME peak and the 2-DG-6-P peak in the 31P spectrum. In this pH range, the slope of the 2-DG-6-P titration curve is considerably steeper than that for PME and the peaks become coincident at a pH of ~6.7 (Fig. 8). This limitation is minimized by microdialyzing enough 2-DG-6-P into the brain so that the resulting peak is significantly larger than the PME peak. That, together with the robust nature of the PeakFit software, allowed us to track the 2-DG-6-P peak throughout these studies.


Figure 8. Effect of pH on the chemical shift of either the 2-DG-6-P peak (•) or the PME peak (▪) relative to PCr. A titration curve for 2-DG-6-P was constructed by varying the pH of solutions containing 5mM 2-DG-6-P in artificial CSF at 37°C and the chemical shift of 2-DG-6-P relative to PCr measured as previously described (Fitzpatrick et al., 1996). The titration curve for PME is from data presented by Corbett et al. (1987). In this study, the pH of 5 mM phosphoethanolamine in artificial intracellular fluid at 38°C was varied and the chemical shift of PME relative to PCr measured. The calculated curve for 2-DG-6-P was pH = log[δs - 4.18)/(7.56 - δs)] + 6.25. The calculated curve for PME was pH = log[(δs - 3.19)/(6.95 - δs)] + 5.62.

Download figure to PowerPoint


We have shown that when 2-DG-6-P is microdialyzed into the interstitial space of rat brains, the resulting 31P-NMR peak can be successfully used to monitor in vivo pHe. In our model, we believe that the normal brain pHe is ~7.24. From this, we have also calculated a brain interstitial Pco2 of 55 mm Hg, a value that would seem to argue against the applicability of the Krogh cylinder model in brain tissue. The Krogh cylinder is too simple a model to describe the cerebral capillaries. It assumes that blood flows through the arterioles, capillaries, and then through the venules. The capillaries undergo many generations of branching (Davies et al., 1997), a factor not considered by the Krogh model. The result is an in vivo interstitial CO2 that is considerably higher than the simple Krogh model would predict (Lambertsen, 1960).

Measurement of pHe via the 2-DG-6-P method is limited by relatively poor temporal resolution as compared to microelectrode studies. Furthermore, there is the possibility of interference with the signals from the PME resonances, which must be carefully subtracted. These shortcomings are, however, offset by the ability to measure pHe in a volume of brain tissue at some distance from the site of probe entry, a technically simpler and less problematic methodology, and the ability to monitor pHe and pHi simultaneously. We believe that the 2-DG-6-P method for determination of pHe is a valuable tool and has the potential to greatly expand our understanding of cellular pH regulation in vivo, both in normal and in pathologically stressed brain tissue.


  1. Top of page
  2. Abstract
  4. NMR measurements
  7. Acknowledgements

The authors thank Nancy Cruz for her expert technical assistance and Donna Brackett for help in the preparation of the manuscript. This study was supported by the National Institute of Neurological Diseases and Stroke (grant NS-32059) and grants from the University of Wisconsin Departments of Anesthesiology and Neurological Surgery.

  • 1
    Ackerman J.J.H., Berkowitz B.A., Deuel R.K. (1984a) Phosphorus-31 NMR of rat brain in vivo with bloodless perfluorocarbon perfused rat.Biochem. Biophys. Res. Commun. 119,913919.
  • 2
    Ackerman J.J.H., Evelhoch J.L., Berkowitz B.A., Kichura G.M., Deuel R.K., Lown K.S. (1984b) Selective suppression of the cranial bone resonance from 31P NMR experiments with rat brain in vivo. J. Magn. Res. 56,318322.
  • 3
    Ackerman J.J.H., Soto G.E., Spees W.M., Zhu Z., Evelhoch J.L. (1996) The NMR chemical shift pH measurement revisited : analysis of error and modeling of a pH dependent reference.Magn. Res. Med. 36,674683.
  • 4
    Altman P.L. & Dittmer D.S. (1974) Biology Data Book, pp. 100101. Federation of American Societies for Experimental Biology, Bethesda.
  • 5
    Ammann D., Lanter F., Steiner R.A., Schulthess P., Shijo Y., Simon W. (1981) Neutral carrier based hydrogen ion selective microelectrode for extra- and intracellular studies.Anal. Chem. 53,22672269.
  • 6
    Benveniste H., Drejer J., Schousboe A., Diemer N.H. (1984) Elevation of the extracellular concentrations of glutamate and aspartate in rat hippocampus during transient cerebral ischemia monitored by intracerebral microdialysis.J. Neurochem. 43,13691374.
  • 7
    Bradley R.D. & Semple S.J. (1962) A comparison of certain acid-base characteristic of arterial blood, jugular venous blood and cerebrospinal fluid in man, and the effect on them of some acute and chronic acid-base disturbances.J. Physiol. (Lond.) 160,381391.
  • 8
    Bungay P.M., Morrison P.F., Dedrick R.L. (1990) Steady-state theory for quantitative microdialysis of solutes and water in vivo and in vitro.Life Sci. 46,105119.
  • 9
    Corbett R.J., Laptook A.R., Nunnally R.L. (1987) The use of the chemical shift of the phosphomonoester P-31 magnetic resonance peak for the determination of intracellular pH in the brains of neonates.Neurology 37,17711779.
  • 10
    Davies K.R., Finnegan W., Richardson G., Acuff V., Fenstermacher J.D. (1997) How many capillaries does it take to make a venule ?Soc. Neurosci. Abstr. 23,1574.
  • 11
    Duong T.Q., Ying H.S., Neil J.J. (1998) Evaluation of extra-and intracellular apparent diffusion in normal and globally ischemic rat brain via 19F NMR. Magn. Res. Med. 40,113.
  • 12
    Fishman R.A. (1980) Cerebrospinal Fluid in Diseases of the Nervous System, pp. 5051. Saunders, Philadelphia.
  • 13
    Fitzpatrick J.H.J, Kintner D., Anderson M., Westler W.M., Emoto S.E., Gilboe D.D. (1996) NMR studies of Pi-containing extracellular and cytoplasmic compartments in brain. J. Neurochem. 66,26122620.
  • 14
    Gilboe D.D., Kintner D., Anderson M.E., Fitzpatrick J.H., Emoto S.E., Markley J.L. (1993) Inorganic phosphate compartmentation in the normal isolated canine brain.J. Neurochem. 60,21922203.
  • 15
    Harris R.J. & Symon L. (1984) Extracellular pH, potassium, and calcium activities in progressive ischaemia of rat cortex.J. Cereb. Blood Flow Metab. 4,178186.
  • 16
    Hellums J.D., Nair P.K., Huang N.S., Ohshima N. (1996) Simulation of intraluminal gas transport processes in the microcirculation.Ann. Biomed. Eng. 24,124.
  • 17
    Hoffman W.E., Charbel F.T., Edelman G. (1996) Brain tissue oxygen, carbon dioxide, and pH in neurosurgical patients at risk for ischemia.Anesth. Analg. 82,582586.
  • 18
    Javaheri S., De Hemptinne A., Vanheel B., Leusen I. (1983a) Changes in brain ECF pH during metabolic acidosis and alkalosis : a microelectrode study.J. Appl. Physiol. 55,18491853.
  • 19
    Javaheri S., Weyne J., Demeester G. (1983b) Changes in the brain surface pH and cisternal cerebrospinal fluid acid-base variables in respiratory arrest.Respir. Physiol. 51,3143.
  • 20
    Katsura K., Ekholm A., Asplund B., Siesjö B.K. (1991) Extracellular pH in the brain during ischemia : relationship to the severity of lactic acidosis.J. Cereb. Blood Flow Metab. 11,597599.
  • 21
    Katsura K., Asplund B., Ekholm A., Siesjö B.K. (1992) Extra-and intracellular pH in the brain during ischaemia, related to tissue lactate content in normo- and hypercapnic rats.Eur. J. Neurosci. 4,166176.
  • 22
    Katsura K., Kristian T., Nair R., Siesjö B.K. (1994) Regulation of intra-and extracellular pH in the rat brain in acute hypercapnia : a re-appraisal.Brain Res. 651,4756.
  • 23
    Kauppinen R.A., Williams S.R., Busza A.L., Van Bruggen N. (1993) Applications of magnetic resonance spectroscopy and diffusion-weighted imaging to the study of brain biochemistry and pathology.Trends Neurosci. 16,8895.
  • 24
    Kraig R.P., Ferreira-Filho C.R., Nicholson C. (1983) Alkaline and acid transients in cerebellar microenvironment.J. Neurophysiol. 49,831850.
  • 25
    Lagerwall K., Daneryd P., Schersten T., Soussi B. (1995) In vivo 31P nuclear magnetic resonance evidence of the salvage effect of ascorbate on the postischemic reperfused rat skeletal muscle. Life Sci. 56,389397.
  • 26
    Lambertsen C.J. (1960) Carbon dioxide and respiration in acid-base homeostasis.Anesthesiology 21,642651.
  • 27
    Ljunggren B., Schutz H., Siesjö B.K. (1974) Changes in energy state and acid-base parameters of the rat brain during complete compression ischemia.Brain Res. 73,277289.
  • 28
    McKinley B.A., Morris W.P., Parmley C.L., Butler B.D. (1996) Brain parenchyma pO2, pCO2, and pH during and after hypoxic, ischemic brain insult in dogs.Crit. Care. Med. 24,18581868.
  • 29
    Moon R.B. & Richards J.H. (1973) Determination of intracellular pH by 31P magnetic resonance. J. Biol. Chem. 248,72767278.
  • 30
    Murphy-Boesch J. & Koretsky A.P. (1983) An in vivo NMR probe circuit for improved sensitivity.J. Magn. Res. 54,526532.
  • 31
    Mutch W.A. & Hansen A.J. (1984) Extracellular pH changes during spreading depression and cerebral ischemia : mechanisms of brain pH regulation.J. Cereb. Blood Flow Metab. 4,1727.
  • 32
    Nedergaard M., Kraig R.P., Tanabe J., Pulsinelli W.A. (1991) Dynamics of interstitial and intracellular pH in evolving brain infarct.Am. J. Physiol. 260,R581R588.
  • 33
    Ohno M., Obrenovitch T.P., Hartell N., Barratt S., Bachelard H.S., Symon L. (1989) Simultaneous recording of tissue pCO2, interstitial pH and potassium activity in the rat cerebral cortex during anoxia and the subsequent recovery period.Neurol. Res. 11,153159.
  • 34
    Pappenheimer J.R., Fencl V., Heisey S.R., Held D. (1965) Role of cerebral fluids in control of respiration as studied in unanesthetized goats.Am. J. Physiol. 208,436450.
  • 35
    Pontén U. & Siesjö B.K. (1966) Gradients of CO2 tension in the brain.Acta Physiol. Scand. 67,129140.
  • 36
    Portman M.A., Lassen N.A., Cooper T.G., Sills A.M., Potchen E.J. (1991) Intra-and extracellular pH of the brain in vivo studied by 31P-NMR during hyper- and hypocapnia. J. Appl. Physiol. 71,21682172.
  • 37
    Scheller D., Kolb J., Tegtmeier F., Lehmenkuhler A. (1992) Extracellular changes of inorganic phosphate are different during spreading depression and global cerebral ischemia of rats.Neurosci. Lett. 141,269272.
  • 38
    Siesjö B.K., Von Hanwehr R., Nergelius G., Nevander G., Ingvar M. (1985) Extra- and intracellular pH in the brain during seizures and in the recovery period following the arrest of seizure activity.J. Cereb. Blood Flow Metab. 5,4757.
  • 39
    Soto G.E., Zhu Z., Evelhoch J.L., Ackerman J.J.H. (1996) Tumor 31P NMR pH measurements in vivo : a comparison of inorganic phosphate and intracellular 2-deoxyglucose-6-phosphate as pH NMR indicators in murine radiation-induced fibrosarcoma-1. Magn. Res. Med. 36,698704.
  • 40
    Szerb J.C. & Redondo I.M. (1993) Astrocytes and the entry of circulating ammonia into the brain : effect of fluoroacetate.Metab. Brain Disord. 8,217234.
  • 41
    Taylor D.L., Obrenovitch T.P., Symon L. (1996) Changes in extracellular acid-base homeostasis in cerebral ischemia.Neurochem. Res. 21,10131021.
  • 42
    Van der Ploeg C.P., Dankelman J., Spaan J.A. (1994) Classical Krogh model does not apply well to coronary oxygen exchange.Adv. Exp. Med. Biol. 345,299304.
  • 43
    Waterfall A.H., Singh G., Fry J.R., Marsden C.A. (1995) Detection of the lipid peroxidation product malonaldehyde in rat brain in vivo.Neurosci. Lett. 200,6972.
  • 44
    Waterfall A.H., Singh G., Fry J.R., Marsden C.A. (1996) Acute acidosis elevates malonaldehyde in rat brain in vivo.Brain Res. 712,102106.