Introduction

There is increasing evidence that astrocytes and neurons constitute a tightly coupled metabolic unit. Already early studies of brain metabolism using ¹⁴C-labeled precursors such as glucose or acetate led to the concept of two distinct tricarboxylic acid (TCA) cycle compartments, that is, two separate pools for glutamate and glutamine (Berl and Frigyesi, 1969; Cremer, 1970; Minchin and Beart, 1975; Van den Berg et al, 1969). Later histochemical work on the restricted localization of glutamine synthetase to glial cells (Martinez-Hernandez et al, 1977; Norenberg and Martinez-Hernandez, 1979) further supported the idea on the glial and neuronal site of these distinct pools. Simplification of these data has sometimes led to the wrong assumption that the oxidative activity of astrocytes is negligible. However, studies using mainly nuclear magnetic resonance spectroscopy have suggested that astrocytic energy metabolism indeed contains a substantial oxidative component which may account for up to 30% of total oxidative metabolism in brain under resting conditions (for review see Hertz and Kala, 2007; Hertz et al, 2007; Hyder et al, 2006). Recent transcriptome analysis has revealed a pattern of gene expression in astrocytes consistent with a significant oxidative activity (Lovatt et al, 2007).

Another question is how astrocytic metabolism behaves during activation. A recent study using ex vivo autoradiography with ¹⁴C-labeled acetate at the carbon 2 position suggested that astrocytic oxidative metabolism increases with brain activity (Cruz et al, 2005). Labeled acetate is a promising candidate to investigate astrocytic oxidative metabolism in vivo because it is selectively taken up and metabolized by astrocytes (Hassel et al, 1992; Muir et al, 1986; Waniewski and Martin, 1998). Earlier studies in the heart muscle have further revealed that quantitative measures for oxygen consumption can be derived from the washout rate of label after injection of radiolabeled acetate (Buck et al, 1991). Acetate can be labeled in the 1- or 2-carbon position. Whole-body studies using ¹³C-labeled acetate demonstrated that labeled CO₂ is produced during the second or third turn in the TCA cycle, depending on whether 1-¹³C-acetate or 2-¹³C-acetate was used (Wolfe and Jahoor, 1990). However, these results reflect whole-body metabolism, brain-specific reactions cannot be directly derived from them. In brain, with 1- and 2-labeled acetate there was substantial labeling of amino acids, such as glutamate, glutamine, and aspartate and further TCA cycle intermediates as shown in older and more recent literature (Badar-Goffer et al, 1990; Berl and Frigyesi, 1969; Lebon et al, 2002; Tyce et al, 1981; Tyson et al, 2003; Van den Berg et al, 1969; Van den Berg and Ronda, 1976). However, the static measurement of metabolites at single time points yields only limited information on astrocytic oxidative metabolism.

In this study we applied 1-¹¹C-acetate and dynamic measurements to investigate astrocytic oxidative metabolism. Specific questions were (1) what is the extraction fraction (EF) of the tracer, (2) what is the kinetic behavior at baseline and during stimulation, (3) how does dichloroacetate, an agent that increases the production of acetyl-coenzyme A (CoA) from pyruvate in the cell, change the kinetics of 1-¹¹C-acetate, and (4) is the loss of radiolabel dependent on blood flow. Studies were performed in rats and humans. In rats data were acquired using a -scintillator, in humans we applied positron emission tomography (PET).

Materials and methods

Automated Radiosynthesis of Sodium 1-¹¹C-Acetate

¹¹C-CO₂ was produced by the ¹⁴N(p,)¹¹C nuclear reaction using a nitrogen gas (N₂/0.5% O₂) target. ¹¹C-CO₂ was bubbled at room temperature in a mixture of methylmagnesium bromide (3 mol/L solution in diethyl ether; Fluka, 70 L, 0.21 mmol) and dry diethyl ether (3 mL) placed in the reactor, where the chemical trapping of the ¹¹C-CO₂ by C-carboxylation of the Grignard reagent takes place instantly. After delivery (10 mins) the reaction was quenched by automated addition of nearly equimolar amount of acid (1 mL 0.2 mol/L aqueous HCl, 0.2 mmol HCl) and the diethyl ether from the bilayer mixture was removed by evaporation in a stream of nitrogen (100 mL/min) for approximately 10 mins. The residual water solution of 1-¹¹C-acetic acid was automatically drawn into the loop of the preparative chromatograph (Merck-Hitachi chromatograph equipped with L-4000A UV detector at 254 nm, isocratic pump L-6000A, 5 mL loop, and radiation detector Radiation Monitor; Eberline Instrument Corporation, Santa Fe, NM, USA), injected into a semipreparative reversed phase column (Polymerx, 10 m, 250 10 mm, Phenomenex) and eluted with a mixture of sterile salt physiologic solution as the mobile phase. The fraction corresponding to the 1-¹¹C-acetic acid (t_R=7 mins) was collected to give up to 10 GBq activity of no-carrier-added sodium 1-¹¹C-acetate with over 99% radiochemical purity and specific activity of approximately 400 GBq/mol at the end of the synthesis in 10 mL sterile isotonic, buffered at physiological pH solution. The radiolabeling was performed in a lead shielded cell using a computer-assisted homemade automatic apparatus controlled by LabVIEW software (National Instruments) via Modulink and RS232 interfaces.

Radiotracer Experiments

Experiments were performed in animals using the -scintillator and in humans using PET. All animal experiments were approved by the local veterinary authorities and were performed in accordance with their guidelines. The experiments were performed by licensed investigators. The animals were kept in cages in a ventilated cabinet with standardized conditions of light (night/day cycle 12/12 h) and temperature and free access to food and water. Before the experiment the animals were fasted overnight. A total of 29 Sprague–Dawley rats (5 for the EF determination, 24 for the different experimental paradigms (for details see below)) weighing 31041 g (means.d.) were included in the study.

The human study population was composed of 6 healthy male volunteers aged between 24 and 36 years without any known neurological impairments. All subjects were studied at the University Hospital Zurich. Informed written consent was obtained from all subjects before the beginning of the experiment. The protocol was approved by the Local Ethics Committee.

Magnetic resonance scanning was performed on each subject (Twin Speed; GE Medical Systems, Milwaukee, WI, USA) to rule out any cerebral pathologies and for anatomical coregistration of the PET scan.

Animal Preparation

Surgery was performed under isoflurane anesthesia and involved the placement of an arteriovenous shunt from the right femoral artery to the right femoral vein, tracheotomy for mechanical ventilation, and exposition of the skull with subsequent thinning of the bone above the primary somatosensory cortex using a dental drill. The actual experiment was then performed under urethane anesthesia (1400 mg/kg, i.p.). The arteriovenous shunt was used for the injection of the tracer, collection of blood samples for metabolite analysis, the monitoring of arterial blood pressure, and the continuous measurement of total arterial ¹¹C activity. For the latter purpose, the shunt was run through a coincidence counter (GE Medical Systems) that stored the whole blood radioactivity concentration at 1-sec intervals. The online arterial sampling procedure is described in detail elsewhere (Weber et al, 2002).

-Scintillator

The -scintillator (Swisstrace, Zurich, Switzerland) has been described in detail earlier (Weber et al, 2003; Wyss et al, 2007). In short, it consists of a scintillation tip (Bicron, BF12, Newbury, OH, USA) with a length of 1 mm and a diameter of 0.25 mm attached to a high numerical aperture glass fiber. The scintillator was made light tight by applying a uniform coating of silver particles. The scintillations were measured using a photomultiplier tube and counting electronics (Perkin Elmer, MA, USA). The limited range of -particles within biological tissues leads to a limited detection volume centered around the scintillating tip of the probe. Monte Carlo simulations demonstrated that for C-11 the distance required to detect 90% of the -particles around the probe is 2.0 mm (Pain et al, 2002). The sensitivity of the used -scintillator was in the range of 0.036 to 0.047 cps/kBq/cc.

The scintillator was inserted into the cortex using a stereotactic frame (David Kopf Instruments, Tujunga, CA, USA). At the insertion point, which was approximately (according to Bregma) 1 mm posterior and 5 mm lateral, the thinned bone was removed and the dura was carefully incised. The scintillator tip was decreased 1.4 mm below the dura. In the stimulation experiments, the correct position was verified by Laser Doppler flowmetry (Periflux System 5000; Perimed AB, Järfälla, Sweden). This position was adjusted to avoid large superficial blood vessels. Only one scintillator penetration was performed per animal. The count rate was stored on a personal computer using a bin width of 1 sec yielding tissue time–activity curves.

Positron Emission Tomography

All subjects were scanned on a whole-body PET/computer tomography-scanner (Discovery ST-RX; GE Medical Systems, Waukesha, WI, USA). This is a scanner with an axial field of view of 14.6 cm and a reconstructed in-plane resolution of approximately 7 mm. Before the positioning of the subjects in the scanner, a radial artery was cannulated for timed arterial blood sampling. An additional catheter for administration of the radiotracers was placed in the contralateral antecubital vein. At the beginning of each study, a low-dose computer tomogram for attenuation correction was performed.

Experimental Protocols

Cerebral blood flow measurements using H₂¹⁵O: First Pass Extraction Fraction

The first pass EF of 1-¹¹C-acetate at baseline was determined according to the relationship EF=K₁/CBF, where K₁ is the transport parameter of the one-tissue compartment model and CBF is cerebral blood flow. Cerebral blood flow was determined in five animals using ¹⁵O-labeled water injected before measurement of 1-¹¹C-acetate kinetics. In the human study CBF was measured in all subjects 10 mins before the 1-¹¹C-acetate injections. The visual stimulation was interrupted between CBF and acetate measurements. Approximately 300 MBq H₂¹⁵O was injected intravenously using an automatic injection device. After the arrival of the bolus in the brain, the acquisition of a series of 18 frames (10 secs each frame) was started.

Measurements of 1-¹¹C-Acetate Kinetics During Different Conditions

In the animal experiments, two injections spaced 30 mins apart were administered. Data were acquired for 20 mins after the injection (slow injection over a period of 30 secs) of 230 to 260 MBq (diluted in 0.5 mL) of radiotracer. This part of the study included 24 animals, which were distributed among the following experimental subgroups: (1) in this group the baseline was followed by electrostimulation (bas–stim); (2) the order was reversed (stim–bas). Groups (1) and (2) were examined with respect to a potential order effect. (3) To evaluate the reproducibility two consecutive baseline measurements were performed (bas–bas). (4) In four animals the effect of acetazolamide (Diamox, 66 mg/kg), a carbonic anhydrase inhibitor increasing CBF but not oxygen consumption (CMRO₂; Okazawa et al, 2001), on the kinetics of 1-¹¹C-acetate was examined. After 20 mins of baseline acquisition, acetazolamide was intravenously injected over 1 min. Ten minutes later the same amount of 1-¹¹C-acetate was injected for the second time. In a last group (5) the effect of dichloroacetate was investigated. Four hours after the injection of dichloroacetate (50 mg/kg, Sigma-Aldrich Product No. 347795, Buchs, Switzerland) a baseline study was followed by electrostimulation (dichloroacetate challenge). Dichloroacetate was dissolved in physiological saline and was adjusted to pH 7.4.

To activate the whisker-to-barrel pathway the infraorbital branch of the trigeminal nerve was electrically stimulated with two stainless steel electrodes. The cathode was inserted through the infraorbital hiatus and the anode was positioned in the masticatory muscle on the ipsilateral side. The current was adjusted to 2 mA, and 2 Hz stimulation frequency was chosen because previous measurements using ¹⁸F-fluorodeoxyglucose-autoradiographies demonstrated this frequency to be optimal eliciting a maximal increase in glucose consumption of 70% to 80% (unpublished data). The stimulation started 30 secs before the tracer injections. To check the effectivity of the applied stimulation and the effect of acetazolamide in group (4) animals, a Laser Doppler flowmetry probe was fixed just above the primary somatosensory cortex to follow regional CBF changes.

In humans, PET was performed under baseline conditions with eyes closed and during visual stimulation using different video clips presented on a screen (Charlie's Angels and Rolling Stones). In three subjects the baseline examinations preceded the stimulation studies and in the other three subjects the order was reversed to rule out any order effect. The visual stimulation was started 20 secs before scan start and continued during complete PET acquisition. After the intravenous injection of 250 MBq 1-¹¹C-acetate, administered as a slow bolus over approximately 60 s, data were recorded for 20 mins (25 frames: 6 20, 3 40, and 16 60 secs).

Time–activity curves were derived from volumes of interest defined on three consecutive slices on the anatomical magnetic resonance scans and subsequently transferred to the PET data. Volumes of interest were drawn in the visual cortex as the target region and in the frontal cortex and cerebellum as references.

Statistics

The Wilcoxon signed rank test for dependent samples was used to evaluate differences between within-subject conditions and the Mann–Whitney U-test was used for group differences. The criterion for significance was set at P<0.05. Values are expressed as means.d.

Data Acquisition and Analysis

The details are described in the appendix.

Results

Animal Studies

Metabolites of 1-¹¹C-acetate in the blood

High-pressure liquid chromatography identified ¹¹C-CO₂ as the only metabolite in blood. The time course of the fraction of authentic ¹¹C-acetate in arterial plasma is demonstrated in Figure 1. The pH and arterial blood gases were all in the physiological range (pH=7.35 to 7.45; pCO₂=35 to 45 mm Hg; pO₂=70 to 100 mm Hg). Plasma lactate levels were also determined as lactate competes with acetate for the uptake into brain. They ranged between 1.4 and 2.1 mmol/L. However, no relationship between plasma lactate levels and changes in the kinetics of 1-¹¹C-acetate could be found. Only in group 5 (dichloroacetate challenge) there was a statistically significant difference in plasma levels between the two different modalities (see below).

Figure 1.

In the upper panel the time course of ¹¹C radioactivity in whole blood (), plasma (), and brain tissue () acquired over 20 mins is shown. In addition the model fit is demonstrated (straight line, K₁=0.085 mL/min/mL tissue, k₂=0.015 per minute). The goodness-of-fit is demonstrated by the random distribution of the residuals, demonstrated in the insert. In the lower panel the fraction f of true 1-¹¹C-acetate in plasma in animal and human studies is plotted. The filled circles represent data points from animal studies (anesthetized rats) and the solid line is the corresponding fit of the function f=100-(1-exp(-t)). Least squares fitting yielded =93% and =0.11 per minute. In the human study (open circles; conscious humans), and were taken from previous experiments which had yielded very similar values (=91%, =0.13 per minute) (Buck et al, 1991).

Full figure and legend (99K)

First pass extraction fraction

CBF was measured in 5 baseline experiments and was 5616 mL/min/100 g. The first pass EF of 1-¹¹C-acetate calculated in these experiments was 18%5%.

Reproducibility of the 1-¹¹C-acetate measurements

The reproducibility of the time–activity curves and of the parameters K₁ and k₂ is demonstrated in the top graphs of Figures 2 and 3. In contrast to trigeminal nerve stimulation, the tissue time–activity curves in the two sequential baseline experiments look identical. The difference of the mean K₁ (K_1experiment2-K_1experiment1) was -0.001, and for k₂ the difference was 0.002 per minute.

Figure 2.

The top and middle graph demonstrate time–activity curves of ¹¹C activity in rat brain tissue normalized to injected activity per g bodyweight yielding standardized uptake values (SUV). The curves represent mean data from five experiments. In the baseline–baseline experiments (top panel) the curves appear identical. In contrast, the washout of ¹¹C activity is remarkably larger during stimulation (middle panel). The bottom panel illustrates the residuals between fit and data if only the first 5 mins of data are fitted. Such a short fit yields substantially lower initial washout rates than a 20 mins fit, demonstrating that the onset of the washout of radiolabel from tissue is delayed.

Full figure and legend (33K)

Figure 3.

Summary of animal data. Panels (A and B) demonstrate the reproducibility of parameters K₁ and k₂. There was no significant change of K₁ and k₂ between sequential baseline experiments. Panels (C and D) show the increase of K₁ and k₂ during infraorbital nerve stimulation. The values of k₂ after the administration of dichloroacetate (DAC, 50 mg/kg) at baseline and during infraorbital nerve stimulation are demonstrated in panel (E). The P-values were calculated using the Wilcoxon signed rank test. Values are given as means.d.

Full figure and legend (45K)

1-¹¹C-Acetate kinetics during infraorbital nerve stimulation

The electrical infraorbital nerve stimulation increased the washout of 1-¹¹C-acetate as shown in the middle graph of Figure 2. The order of the experiments (baseline–stimulation or stimulation–baseline) had no significant effect on the changes of the kinetic parameters (Figures 3C and 3D). Therefore the two groups (1) and (2) (bas–stim and stim–bas) were pooled. During stimulation, k₂ increased by 93% from 0.0140.007 to 0.0270.006 per minute (P<0.01) in the primary somatosensory cortex. In parallel, K₁ increased from 0.0800.015 at baseline to 0.0950.017 (P<0.01). The efficacy of the stimulation was confirmed by the increase of the laser Doppler flowmetry signal by 20 to 25% (data not shown).

To check for a delayed onset of the radioactivity washout, the first 5 mins of data were refitted with only k₂ as a free parameter. In this dataset, k₂ decreased from 0.0140.007 to 0.0010.002 in the baseline experiments and from 0.0270.006 to 0.0030.004 in the stimulation experiments. An example of a 5 mins fit is illustrated at the bottom of Figure 2. The low k₂ of 0.001 fits the initial data well, then the washout is clearly increased.

Effect of acetazolamide on k₂

In the 4 experiments k₂ was 0.0160.002 at baseline and 0.0160.002 per minute 10 mins after acetazolamide injection. Laser Doppler flowmetry showed a signal increase of 61.1%8.1%.

Effect of dichloroacetate on k₂ during baseline and stimulation

Dichloroacetate pretreatment decreased arterial plasma lactate from 1.50.1 to 1.250.2 mmol/L (means.d.; P<0.05) but exerted no change of arterial blood gases which stayed in the physiological range. Four hours after intravenous dichloroacetate administration, k₂ was less at baseline and during stimulation, compared with the experiments without dichloroacetate. At baseline the values were on average 36% less (0.0090.002 compared with 0.0140.007), however, this difference failed to reach significance (n=5; P=0.15). During stimulation k₂ was 26% less (n=5; P<0.05) than that in the experiments without administration of dichloroacetate (0.0200.004 compared with 0.0270.006). There was still a significant 122% increase of k₂ from baseline to electrostimulation (from 0.0090.002 to 0.0200.004; n=5; P<0.05; Figure 3E).

Relationship of the tissue activity and changes of the washout rate k₂

This relationship is a measure of the parameter identifiability. A larger change of the tissue activity with a unit change of k₂ is associated with a better parameter identifiability. The sensitivity analysis demonstrated that at 20 mins, a 10% change in k₂ led to a 2.1% change in tissue activity at baseline (k₂=0.014 per minute) and to a 3.8% change during stimulation (k₂=0.027 per minute). Furthermore, the activity change in tissue was linear in the investigated range 0 to 30% k₂ change.

Human Studies

1-¹¹C-Acetate kinetics present the same characteristics as in animal studies

In all subjects the blood gas parameters were always in the physiological range (pH=7.35 to 7.45; pCO₂=35 to 45 mm Hg; pO₂=70 to 100 mm Hg). An illustration of the effect of the visual activation on the kinetic of 1-¹¹C-acetate is demonstrated in Figure 4 and the quantitative values are shown in Figure 5. The calculated EF of 1-¹¹C-acetate in visual cortex was 194% at baseline and 133% during stimulation. The visual stimulation increased mean CBF in visual cortex by 64% (P<0.05). In the control regions, 'cerebellum' and 'frontal cortex,' no significant change of the CBF was noticed. In visual cortex mean K₁ and k₂ values of 1-¹¹C-acetate were 11% and 62% (P<0.05 for both) higher during stimulation as shown in Figures 5F and 5I. No significant changes of K₁ or k₂ were noted in frontal cortex and cerebellum. However, in frontal cortex K₁ and k₂ data are somewhat distorted by a single subject, which showed a decrease of the values in this region in response to visual stimulation.

Figure 4.

Time–activity curves of 1-¹¹C-acetate uptake in the human visual cortex normalized to injected activity per kg bodyweight yielding standardized uptake values (SUV). The curves represent mean data from all six human experiments. During visual stimulation the clearance of ¹¹C activity is larger compared with baseline. Note also the modestly faster wash-in of radioactivity during stimulation.

Full figure and legend (10K)

Figure 5.

Summary of human data. The change of CBF from baseline to visual stimulation in the three examined brain areas is demonstrated in the top panels (A–C), the one of K₁ and k₂ of 1-¹¹C-acetate in the middle (D–F) and bottom panels (G–I), respectively. The P-values were calculated using Wilcoxon signed rank test. Values are given as means.d.

Full figure and legend (69K)

Discussion

It was previously suggested that the astrocyte-specific substrate acetate allows to measure metabolism in human brains in vivo (Hosoi et al, 2004; Muir et al, 1986). However, to our knowledge this is the first study using label clearance from brain after 1-¹¹C-acetate administration for the investigation of astrocytic oxidative metabolism in vivo. Labeled acetate was already used in various studies to illuminate certain aspects of astrocytic metabolism. Some groups used autoradiographical determination of ¹⁴C-acetate uptake (Cruz et al, 2005; Muir et al, 1986), other groups measured the metabolic fate of acetate by measuring the appearance of labeled compounds as the label traveled through the TCA cycle (e.g., Badar-Goffer et al, 1990; Berl and Frigyesi, 1969; Lebon et al, 2002; Van den Berg and Ronda, 1976).

Our study demonstrates for the first time the changes of 1-¹¹C-acetate kinetics during stimulation in vivo, which are thought to be correlated with the changes of astrocytic oxidative metabolism. Although total cerebral CMRO₂ can be successfully measured in vivo using PET and inhaled ¹⁵O₂ (Mintun et al, 1984; Yee et al, 2006), a direct in vivo measurement of CMRO₂ in astrocytes alone is not available.

K₁ and Extraction Fraction

Compared with the heart where 1-¹¹C-acetate is almost 100% extracted, the cerebral EF is approximately 5-fold less. Nevertheless, brain uptake is large enough to allow quantification. One consequence of the lower EF is the blunted increase of K₁ compared with CBF during stimulation. In the human study K₁ increased 11% compared with a 64% increase in CBF (visual cortex).

Quantitative Relationship Between k₂ and Astrocytic CMRO₂

In the following discussion about the meaning of the rate constant k₂ we provide arguments that k₂ may be strongly related to CMRO₂. However, it is clear that a direct proof is not given and would be difficult to establish, as there is actually no data available to prove this conclusion directly. One major reason is that a direct measurement of astrocytic CMRO₂ is yet impossible in vivo.

We hypothesize that k₂ is directly related to CMRO₂ in the astrocytes as has been demonstrated in the heart for the myocytes (Buck et al, 1991). Prerequisites for this hypothesis are (1) the washout of radioactivity from the volume of interest is indeed because of the loss of ¹¹C-CO₂ and (2) the rate-limiting step determining k₂ is the production of ¹¹C-CO₂ and not its removal.

Prerequisite (1)

There seems to be no doubt that at least part of the radioactivity loss is because of ¹¹C-CO₂, because the production of labeled CO₂ from exogenous 1-¹⁴C-acetate has been proven in astrocyte cultures (Waniewski and Martin, 1998). However, a quantitative proof of prerequisite (1) would be difficult. One theoretical possibility would be the measurement of the arterio-venous difference of ¹¹C-CO₂. However, a first order estimation of this difference demonstrates that it would be in the order of 4% of the total blood counts. The conclusive demonstration of such a small difference would require a prohibitively large amount of animals.

Prerequisite (2)

It is well known that CO₂ diffuses practically freely across the blood–brain barrier (Paulson, 2002). It could therefore theoretically be possible that the removal of CO₂ by the blood is a limiting factor for the washout of CO₂. However, the lack of a change of k₂ after the acetazolamide-induced blood flow increases demonstrates that the removal of the radioactivity from the volume of interest is not a limiting factor.

Other arguments which point at a strong relationship between k₂ and astrocytic CMRO₂ are (1) the increase of k₂ during stimulation, (2) the delay of the onset of the label washout and the decrease of the K₁/k₂ ratio during stimulation, (3) the behavior of k₂ under dichloroacetate, and (4) reasonable values of astrocytic CMRO₂ can be obtained by applying the calibration established in the dog heart.

Changes of k₂ During Increased Brain Activity

Parameter k₂ significantly increased during stimulation. In the animal study, k₂ almost doubled from baseline to infraorbital nerve stimulation, indicating a substantial increase in oxygen consumption. In the human study, the increase in the visual cortex was 63%. Although not significant, there also seemed to be a trend toward higher values in cerebellum and frontal cortex. In fact, omitting the subject with a decreased value during stimulation, the increase became significant, not surprising considering the connectivity in the brain.

Delay of the Washout Onset and the Decrease of the K₁/k₂ Ratio During Stimulation

Another result pointing at ¹¹C-CO₂ as the washout agent is the delay of the washout onset (bottom graph of Figure 2). Such a delay for the appearance of labeled CO₂ was demonstrated by Waniewski and Martin (1998) in cultured astrocytes and is explained by the fact that labeled CO₂ is only produced in the second turn of the label in the TCA cycle. If the loss of label was substantially because backdiffusion of labeled acetate, one would not expect a delay. Furthermore, if backdiffusion of labeled acetate was the main factor determining k₂, one would expect the same effect of stimulation on K₁ and k₂, that is, the ratio K₁/k₂ would remain constant. However, the data demonstrate a markedly lower ratio during stimulation. In the anesthetized rat studies, the mean ratio dropped 44% (P=0.08), in the studies with dichloroacetate the decrease was 53% (P=0.001) and in the human visual cortex the ratio dropped 44% (P=0.048). This behavior suggests that the labeled compounds entering and leaving the brain tissue are different. The mentioned points are all consistent with the hypothesis that the washed out compound is indeed ¹¹C-CO₂. However, we cannot completely exclude that some label loss is because of release of other labeled substrates, particularly because of efflux of glutamine, which is quickly labeled after injection of radiolabeled acetate (Cremer, 1970). Despite the fact that the blood–brain barrier is closed for the passage of glutamate and glutamine from blood to brain, the reverse direction seems possible. Glutamine is taken up by endothelial cells and intracellularly converted by the enzyme glutaminase to glutamate. When the glutamate concentration within endothelial exceeds the one in plasma net transport of glutamate takes place through facilitative transporters (Hawkins et al, 2006). The high number of steps involved in this process and the fact that normal plasma concentration of glutamate is approximately 10 times higher than glutamate concentration in the extracellular fluid of the brain (Hawkins et al, 2006) indicate that this process is slow and probably not of relevance during our acquisition period of 20 mins.

Pharmacological Interventions Using Dichloroacetate and Acetazolamide

The dichloroacetate experiments are also consistent with the hypothesis that k₂ is related to CMRO₂. This compound activates pyruvate dehydrogenase (Abemayor et al, 1984) by inhibiting its phosphorylation by the PDH kinase (Whitehouse et al, 1974). As a consequence more pyruvate is converted to acetyl-CoA and less to lactate and the concentration of acetyl-CoA is expected to increase. If the rate constants in the following chain of acetyl-CoA oxidation, which are expected to be related to k₂, remained the same, the rate of the acetyl-CoA metabolism would increase and with it most likely CMRO₂. However, as the energetic demand is unchanged there is no reason why CMRO₂ should increase. Consequently one would expect a decrease in k₂ and this is exactly what was observed, at baseline as well as during stimulation. A good candidate for regulation at this stage is citrate synthase, which condenses acetyl-CoA and oxaloacetate into citrate. It has been documented that this enzyme is inhibited by increased concentrations of acetyl-CoA (Bayer et al, 1981). This inhibition could be a major reason for the observed decrease of k₂ after dichloroacetate. Furthermore, the suggested increase of the cold acetyl-CoA concentration under dichloroacetate is itself indirect evidence for the ability of astrocytes to completely oxidize glucose, as is the work by Lovatt et al (2007), which confirmed significantly higher relative expression of almost all TCA-cycle enzymes in freshly isolated astrocytes than neurons and showed that the oxidative pathways were active in astrocytes.

Analogy with Previous Heart Studies

The idea to use the washout rate k₂ of 1-¹¹C-acetate as a marker of CMRO₂ was inspired by earlier studies performed in the heart (Armbrecht et al, 1990; Buck et al, 1991; Lindner et al, 2006). In those studies it was clearly established that k₂ reflects myocardial oxygen consumption in a linear fashion. One rough indication that this approach may also be valid in the brain is derived from the following estimation. In the dog heart, the relationship between CMRO₂ and k₂ was CMRO₂ (mL/min/100 g)=48.8 mL/100 g k₂, (per minute). Application of this relationship to this study yields an astrocytic CMRO₂ of 0.7 mL/min per 100 g in rats and humans at baseline. This amounts to approximately 20% of whole brain oxygen consumption (3.40.28 mL/min per 100 g in humans (Kety and Schmidt, 1948)). This value compares favorably to the published values for the fraction of astrocytic ATP generation by oxidative processes (Hertz and Kala, 2007; Hertz et al, 2007; Oz et al, 2004) which are in the range of 14% to 30%.

Is k₂ Specific for Astrocytic Metabolism?

One important observation in this regard is that labeled acetate is taken up only by astrocytes and not neurons (Waniewski and Martin, 1998). However, a fraction of the label will eventually end up in neurons as labeled glutamine/glutamate (via the glutamate/glutamine cycle), which was produced via the -ketoglutarate-glutamate step in the astrocytes. Nevertheless, the transfer of labeled glutamine to neurons is slow relative to the 20 mins acquisition time of this study. Lebon et al (2002) investigated astrocytic energy metabolism using nuclear magnetic resonance spectroscopy and 2-¹³C-acetate in humans. They concluded that the labeled glutamine/glutamate observed during the first 20 mins after the start of the 2-¹³C-acetate infusion reflected astrocytic glutamate only. It therefore seems reasonable to assume that most of the ¹¹C activity measured in our study was originating from astrocytes. Even if a small fraction of labeled glutamine/glutamate had reached the neurons, the contribution of neuronal ¹¹C-CO₂ production could be expected to have been small. Hertz et al (1988) demonstrated in cell cultures, that glutamate oxidation of neurons was at most 25% of the one in astrocytes. In addition, a recent work found a 4.3-fold higher expression of the enzyme glutamate dehydrogenase Glud1 in astrocytes, also suggesting a substantially higher capacity for astrocytes to metabolize glutamate (Lovatt et al, 2007).

Comparison of the k₂ Method with Autoradiography

Cruz et al (2005) investigated the accumulation of 2-¹⁴C-acetate during acoustic stimulation in rats using autoradiography performed 5 mins after tracer injection. They reported a 15 to 18% increase of 2-¹⁴C-acetate uptake in the stimulated compared with the contralateral structures. These results are in line with our experiments. At 5 mins, the uptake of 1-¹¹C-acetate was 14% higher in the stimulation compared with the baseline experiments (Supplementary Figure S1). This value was calculated by simulating tissue time–activity curves with a mean input curve and the mean K₁ and k₂ values of the stimulation and baseline experiments. Here, it is crucial to point to an important difference in the methodology. By evaluating just one time point as in autoradiographic studies, it is not easy to determine to what degree the differential uptake is influenced by increased blood flow, although this influence seems to be low for acetate (Dienel et al, 2007), but probably not negligible as our demonstration of a small K₁ increase during stimulation suggests. To separate delivery and washout, a kinetic approach as chosen in this study seems advantageous. However, autoradiographic studies deliver images and, when using the long-lived ¹⁴C, allow an easier determination of metabolites.

Using the short-lived isotope ¹¹C we simply concentrated on two aspects of the label kinetics, delivery, and washout. We cannot deliver any data concerning the fate of the label between these steps, which may be very complex. An indication of this complexity is demonstrated in the model of Figure 6. Previous work by Berl and Frigyesi (1969) demonstrated (1) that 5 mins after injection of 1-¹⁴C-acetate most label is found in amino acids and (2) that there is a decline of their concentration afterwards. This may indicate that most acetate oxidation happens indirect via amino acids. More extended information on the fate of the label are found elsewhere (Dienel et al, 2007; Berl and Frigyesi, 1969). Nevertheless, still more studies are needed to identify the contribution of acetate and TCA cycle metabolites as direct substrates for oxidative metabolism.

Figure 6.

One-tissue compartment models used in animal and human studies to analyze CBF using H₂¹⁵O (A) and the kinetics of 1-¹¹C-acetate respectively (B). K₁ describes the transfer of acetate from blood to tissue whereas k₂ corresponds to the flux of radiolabel from tissue to blood. Labeled acetate in tissue (C_ac-tiss) is converted to acetyl-CoA, catalyzed by acetyl-CoA synthetase. Acetyl-CoA then enters the TCA cycle, where the label can either be transferred to CO₂ or metabolites (C_m-tiss) such as glutamine and glutamate which were produced via -keto-glutarate. Some labeled metabolites will reenter the TCA cycle at the -keto-glutarate level and in this way yield more labeled CO₂. In theory, k₂ denotes the backdiffusion of all labeled species from tissue. However, as discussed in the main text, we argue that the loss of label is mainly because of backdiffusion of ¹¹C-CO₂, and that the rate of this label loss, described by k₂, is reflecting the production of CO₂ and is therefore closely related to oxidative metabolism.

Full figure and legend (53K)

General Remarks

Because of the delay of the onset of the label washout, k₂ is smaller during the first few minutes after injection. This effect was not taken into account in the tracer kinetic modeling, for example, k₂ was assumed constant. To estimate the resulting error we calculated simulated tissue time–activity curves with k₂=0 (during the first 5 mins) and 0.027 per minute thereafter. The refit of these curves using a constant k₂ demonstrated only a 1.5% change of k₂.

Another point concerns parameter identifiability. The sensitivity analysis demonstrated that the dependence of tissue activity on k₂ is more pronounced at the higher values, implying that the k₂ values during baseline are less reliable than during stimulation. This aspect and the fact that k₂ at baseline is relatively small may lead to a relatively large error margin in the ratio k_2simulation/k_2baseline. We also did not consider the effect of blood-borne ¹¹C-CO₂ on K₁ and k₂. Most of the CO₂ in blood is converted to HCO₃^-, which does not cross the blood–brain barrier. Approximately 5% of total HCO₃^-/CO₂ is present as dissolved ¹¹C-CO₂ (Brooks et al, 1984) which freely penetrates the brain. We constructed simulated tissue time–activity curves using the one-tissue compartment model and the blood-borne ¹¹C-CO₂ as a second input curve. K₁ and k₂ values for this fraction of CO₂ were taken from Brooks et al (1984). These tissue time–activity curves were then refitted using the standard one-tissue compartment model not considering blood borne ¹¹C-CO₂. The deviation of the K₁ and k₂ values was less than 1.5%, demonstrating that the effect of blood-borne ¹¹C-CO₂ on K₁ and k₂ is negligible.

Summary

1-¹¹C-acetate seems a promising tracer to investigate astrocytic oxidative metabolism in animals and humans. If the washout rate indeed represents the production of ¹¹C-CO₂, then its increase during stimulation would point to a substantially higher astrocytic oxidative metabolism during brain activation. However, the quantitative relationship between k₂ and CMRO₂ needs to be determined in future experiments.

References

1. Abemayor E, Kovachich GB, Haugaard N (1984) Effects of dichloroacetate on brain pyruvate dehydrogenase. J Neurochem 42:38–42 | Article | PubMed | ChemPort |
2. Alpert NM, Eriksson L, Chang JY, Bergstrom M, Litton JE, Correia JA, Bohm C, Ackerman RH, Taveras JM (1984) Strategy for the measurement of regional cerebral blood flow using short-lived tracers and emission tomography. J Cereb Blood Flow Metab 4:28–34 | PubMed | ChemPort |
3. Armbrecht JJ, Buxton DB, Schelbert HR (1990) Validation of [1-¹¹C]acetate as a tracer for noninvasive assessment of oxidative metabolism with positron emission tomography in normal, ischemic, postischemic, and hyperemic canine myocardium. Circulation 81:1594–1605 | PubMed | ChemPort |
4. Badar-Goffer RS, Bachelard HS, Morris PG (1990) Cerebral metabolism of acetate and glucose studied by ¹³C-n.m.r. spectroscopy. A technique for investigating metabolic compartmentation in the brain. Biochem J 266:133–139 | PubMed | ChemPort |
5. Bayer E, Bauer B, Eggerer H (1981) Evidence from inhibitor studies for conformational changes of citrate synthase. Eur J Biochem 120:155–160 | Article | PubMed | ChemPort |
6. Berl S, Frigyesi TL (1969) The turnover of glutamate, glutamine, aspartate and GABA labeled with [1-¹⁴C]acetate in caudate nucleus, thalamus and motor cortex (cat). Brain Res 12:444–455 | Article | PubMed | ChemPort |
7. Brooks DJ, Lammertsma AA, Beaney RP, Leenders KL, Buckingham PD, Marshall J, Jones T (1984) Measurement of regional cerebral Ph in human-subjects using continuous inhalation of (Co2)-C-11 and positron emission tomography. J Cereb Blood Flow Metab 4:458–465 | PubMed | ChemPort |
8. Buck A, Wolpers HG, Hutchins GD, Savas V, Mangner TJ, Nguyen N, Schwaiger M (1991) Effect of carbon-11-acetate recirculation on estimates of myocardial oxygen consumption by PET. J Nucl Med 32:1950–1957 | PubMed | ChemPort |
9. Cremer JE (1970) Selective inhibition of glucose oxidation by triethyltin in rat brain in vivo. Biochem J 119:95–102 | PubMed | ChemPort |
10. Cruz NF, Lasater A, Zielke HR, Dienel GA (2005) Activation of astrocytes in brain of conscious rats during acoustic stimulation: acetate utilization in working brain. J Neurochem 92:934–947 | Article | PubMed | ISI | ChemPort |
11. Dienel GA, Schmidt KC, Cruz NF (2007) Astrocyte activation in vivo during graded photic stimulation. J Neurochem 103:1506–1522 | Article | PubMed | ChemPort |
12. Hassel B, Paulsen RE, Johnsen A, Fonnum F (1992) Selective inhibition of glial cell metabolism in vivo by fluorocitrate. Brain Res 576:120–124 | Article | PubMed | ChemPort |
13. Hawkins RA, O'Kane RL, Simpson IA, Vina JR (2006) Structure of the blood-brain barrier and its role in the transport of amino acids. J Nutr 136:218S–226S | PubMed | ISI | ChemPort |
14. Hertz L, Drejer J, Schousboe A (1988) Energy metabolism in glutamatergic neurons, GABAergic neurons and astrocytes in primary cultures. Neurochem Res 13:605–610 | Article | PubMed | ISI | ChemPort |
15. Hertz L, Kala G (2007) Energy metabolism in brain cells: effects of elevated ammonia concentrations. Metab Brain Dis 22:199–218 | Article | PubMed | ChemPort |
16. Hertz L, Peng L, Dienel GA (2007) Energy metabolism in astrocytes: high rate of oxidative metabolism and spatiotemporal dependence on glycolysis/glycogenolysis. J Cereb Blood Flow Metab 27:219–249 | Article | PubMed | ISI | ChemPort |
17. Hosoi R, Okada M, Hatazawa J, Gee A, Inoue O (2004) Effect of astrocytic energy metabolism depressant on ¹⁴C-acetate uptake in intact rat brain. J Cereb Blood Flow Metab 24:188–190 | Article | PubMed | ChemPort |
18. Hyder F, Patel AB, Gjedde A, Rothman DL, Behar KL, Shulman RG (2006) Neuronal-glial glucose oxidation and glutamatergic-GABAergic function. J Cereb Blood Flow Metab 26:865–877 | Article | PubMed | ChemPort |
19. Kety SS, Schmidt CF (1948) The nitrous oxide method for the quantitative determination of cerebral blood flow in man—theory, procedure and normal values. J Clin Invest 27:476–483 | Article | PubMed | ISI | ChemPort |
20. Lebon V, Petersen KF, Cline GW, Shen J, Mason GF, Dufour S, Behar KL, Shulman GI, Rothman DL (2002) Astroglial contribution to brain energy metabolism in humans revealed by ¹³C nuclear magnetic resonance spectroscopy: elucidation of the dominant pathway for neurotransmitter glutamate repletion and measurement of astrocytic oxidative metabolism. J Neurosci 22:1523–1531 | PubMed | ISI | ChemPort |
21. Lindner O, Sorensen J, Vogt J, Fricke E, Baller D, Horstkotte D, Burchert W (2006) Cardiac efficiency and oxygen consumption measured with ¹¹C-acetate PET after long-term cardiac resynchronization therapy. J Nucl Med 47:378–383 | PubMed |
22. Lovatt D, Sonnewald U, Waagepetersen HS, Schousboe A, He W, Lin JHC, Han X, Takano T, Wang S, Sim FJ, Goldman SA, Nedergaard M (2007) The transcriptome and metabolic gene signature of protoplasmic astrocytes in the adult murine cortex. J Neurosci 27:12255–12266 | Article | PubMed | ChemPort |
23. Martinez-Hernandez A, Bell KP, Norenberg MD (1977) Glutamine synthetase: glial localization in brain. Science 195:1356–1358 | Article | PubMed | ChemPort |
24. Mikolajczyk K, Szabatin M, Rudnicki P, Grodzki M, Burger C (1998) A JAVA environment for medical image data analysis: initial application for brain PET quantitation. Med Inform (Lond) 23:207–214 | Article | PubMed | ChemPort |
25. Minchin MC, Beart PM (1975) Compartmentation of amino acid metabolism in the rat dorsal root ganglion; a metabolic and autoradiographic study. Brain Res 83:437–449 | Article | PubMed | ChemPort |
26. Mintun MA, Raichle ME, Martin WR, Herscovitch P (1984) Brain oxygen utilization measured with O-15 radiotracers and positron emission tomography. J Nucl Med 25:177–187 | PubMed | ISI | ChemPort |
27. Muir D, Berl S, Clarke DD (1986) Acetate and fluoroacetate as possible markers for glial metabolism in vivo. Brain Res 380:336–340 | Article | PubMed | ChemPort |
28. Norenberg MD, Martinez-Hernandez A (1979) Fine structural localization of glutamine synthetase in astrocytes of rat brain. Brain Res 161:303–310 | Article | PubMed | ISI | ChemPort |
29. Okazawa H, Yamauchi H, Sugimoto K, Toyoda H, Kishibe Y, Takahashi M (2001) Effects of acetazolamide on cerebral blood flow, blood volume, and oxygen metabolism: a positron emission tomography study with healthy volunteers. J Cereb Blood Flow Metab 21:1472–1479 | Article | PubMed | ChemPort |
30. Oz G, Berkich DA, Henry PG, Xu Y, LaNoue K, Hutson SM, Gruetter R (2004) Neuroglial metabolism in the awake rat brain: CO2 fixation increases with brain activity. J Neurosci 24:11273–11279 | Article | PubMed | ISI | ChemPort |
31. Pain F, Laniece P, Mastrippolito R, Pinot L, Charon Y, Glatigny A, Guillemin MT, Hantraye P, Leviel V, Menard L, Valentin L (2002) SIC, an intracerebral radiosensitive probe for in vivo neuropharmacology investigations in small laboratory animals: prototype design, characterization, and in vivo evaluation. IEEE Trans Nucl Sci 49:822–826 | Article | ChemPort |
32. Paulson OB (2002) Blood-brain barrier, brain metabolism and cerebral blood flow. Eur Neuropsychopharmacol 12:495–501 | Article | PubMed | ISI | ChemPort |
33. Tyce GM, Ogg J, Owen Jr CA (1981) Metabolism of acetate to amino acids in brains of rats after complete hepatectomy. J Neurochem 36:640–650 | Article | PubMed | ChemPort |
34. Tyson RL, Gallagher C, Sutherland GR (2003) C-Labeled substrates and the cerebral metabolic compartmentalization of acetate and lactate. Brain Res 992:43–52 | Article | PubMed | ChemPort |
35. Van den Berg CJ, Krzalic L, Mela P, Waelsch H (1969) Compartmentation of glutamate metabolism in brain. Evidence for the existence of two different tricarboxylic acid cycles in brain. Biochem J 113:281–290 | PubMed | ChemPort |
36. Van den Berg CJ, Ronda G (1976) The incorporation of double-labelled acetate into glutamate ad related amino acids from adult mouse brain: compartmentation of amino acid metabolism in brain. J Neurochem 27:1443–1448 | Article | PubMed | ChemPort |
37. Waniewski RA, Martin DL (1998) Preferential utilization of acetate by astrocytes is attributable to transport. J Neurosci 18:5225–5233 | PubMed | ISI | ChemPort |
38. Weber B, Burger C, Biro P, Buck A (2002) A femoral arteriovenous shunt facilitates arterial whole blood sampling in animals. Eur J Nucl Med Mol Imaging 29:319–323 | Article | PubMed | ChemPort |
39. Weber B, Spath N, Wyss M, Wild D, Burger C, Stanley R, Buck A (2003) Quantitative cerebral blood flow measurements in the rat using a beta-probe and H₂¹⁵O. J Cereb Blood Flow Metab 23:1455–1460 | Article | PubMed |
40. Whitehouse S, Cooper RH, Randle PJ (1974) Mechanism of activation of pyruvate dehydrogenase by dichloroacetate and other halogenated carboxylic acids. Biochem J 141:761–774 | PubMed | ISI | ChemPort |
41. Wolfe RR, Jahoor F (1990) Recovery of labeled CO₂ during the infusion of C-1- versus C-2-labeled acetate: implications for tracer studies of substrate oxidation. Am J Clin Nutr 51:248–252 | PubMed | ChemPort |
42. Wyss MT, Ametamey SM, Treyer V, Bettio A, Blagoev M, Kessler LJ, Burger C, Weber B, Schmidt M, Gasparini F, Buck A (2007) Quantitative evaluation of ¹¹C-ABP688 as PET ligand for the measurement of the metabotropic glutamate receptor subtype 5 using autoradiographic studies and a beta-scintillator. Neuroimage 35:1086–1092 | Article | PubMed |
43. Yee SH, Lee K, Jerabek PA, Fox PT (2006) Quantitative measurement of oxygen metabolic rate in the rat brain using microPET imaging of briefly inhaled ¹⁵O-labelled oxygen gas. Nucl Med Commun 27:573–581 | Article | PubMed | ChemPort |

Appendices

Appendix

Details of -Probe Experiments and Human PET Measurements

Blood and Tissue Processing

(1) Physiological blood parameters: Several blood gas variables (pH, pCO₂, pO₂) relevant to cerebral energy metabolism were measured with a pH/blood gas analyzer (AVL, Compact 3, Roche Diagnostics, Switzerland) before the start of tracer kinetic measurements. Plasma lactate levels were determined before the first tracer injection in 24 animals (Ektachem DT, Kodak, USA). In animals of group 5 (dichloroacetate challenge) lactate levels were determined twice, before and 4 h after injection of dichloroacetate respectively (Ektachem DT, Kodak, USA). In five animals there were technical problems with the analyzer rendering a measurement impossible.

(2) Determination of the arterial plasma input curve: For the CBF measurements arterial blood was continuously drawn during scanning and the activity in the catheter was measured in a coincidence counter (GE Medical Systems), which was cross calibrated with the -scintillator and the PET scanner respectively. During the 1-¹¹C-acetate scan arterial radioactivity was also measured online for the total scan time in the animal experiments (for an example of acquired time–activity curves see Supplementary Figure S2). In the human study, it was acquired online only for the first 5 mins so as not to miss the peak of the arterial input curve, thereafter for the remaining 15 mins 4 single samples were taken manually at different time points (around 3, 6, 15 and 25 mins) and processed as described below. Whole-blood activity was then corrected for (1) a different tracer concentration in whole blood and plasma and (2) the build up of labeled metabolites. The ratio '¹¹C_plasma/¹¹C_{whole blood}' was determined in all animals at different time points (three blood samples per injection). The data of all animals were then pooled and the time course of the ratio was approximated by fitting a quadratic polynomial to the data. This function was then used to convert counts in whole blood to counts in plasma. In humans, the ¹¹C concentration in whole blood (C_B) was first converted to ¹¹C concentration in plasma (C_PL) using the same conversion function as determined in rats.

Secondly, plasma activity was corrected for the build-up of labeled metabolites using data from previous 1-¹¹C-acetate studies in the human heart (Buck et al, 1991).

(3) Metabolite analysis in blood: Three blood samples (approximately 400 L each) were collected at different time points during each 20-min acquisition for analysis of authentic tracer and metabolites. These samples were first centrifuged for 3 mins at 2000g. Proteins were then precipitated with 75 L acetonitrile in 50 L plasma. After a second centrifugation for 3 mins at 2000g, the composition of the ¹¹C-derived radioactivity in the supernatant (80 L) was analyzed by high-pressure liquid chromatography on a polymeric column (PRP-1, 5-m, 250 4.1 mm i.d.; Hamilton, USA) with 3 mmol/L phosphoric acid in water (pH 2.67) as the mobile phase (1 mL/min). The amount of authentic tracer was expressed as a fraction of total plasma ¹¹C counts (f).

The fraction data of all animals was then pooled and the time-course of the fraction was approximated by a decaying exponential function of the form function f=1-(1-exp(-t)). This function was subsequently used to convert the total plasma activity to the time course of authentic 1-¹¹C-acetate (=input curve).

Data Analysis

In the animal and human studies the analytical and statistical methods were similar as described below. All calculations were performed using the software package PMOD (Mikolajczyk et al, 1998; PMOD Technologies, Adliswil, Switzerland).

(1) Cerebral blood flow measurements: The basis of the calculation for CBF measurements was the one-tissue compartment model including a partition coefficient for H₂¹⁵O (see Figure 6A). The change of the radioactivity concentration in tissue C_T is then defined by the following differential equation

where C_A is the arterial tracer concentration and p is the tissue partition coefficient, for example, that fraction of tissue that is permeable for H₂¹⁵O. In this configuration C_TISS is the concentration of H₂¹⁵O in 1 mL of brain and it is assumed that H₂¹⁵O immediately reaches a homogeneous concentration in permeable space and no division into a vascular and a tissue compartment is necessary. The analytical solution of Equation 1 is

where means mathematical convolution.

Equation 2 was fitted to the data using least squares fitting (Marquardt algorithm) implemented in the software PMOD (Mikolajczyk et al, 1998). Before that data analysis tissue TACs and arterial input curves were corrected for physical decay of ¹⁵O.

With the human H₂¹⁵O PET data we calculated quantitative parametric maps representing regional CBF using the integration method described by Alpert et al (1984). This method yielded maps of K₁ and k₂ which represent regional CBF and regional CBF/p respectively (p=partition coefficient). Before calculation of the parametric maps, the input curve was corrected for delay and dispersion.

(2) 1-¹¹C-acetate experiments: The investigated method consisted of standard compartmental modeling using an arterial input function. Tracer kinetic modeling was performed using the one-tissue compartment model depicted in Figure 6B.

K₁ describes uptake of tracer across the blood–brain barrier. K₁ is related to the first pass EF and CBF through the following equation: K₁=EF CBF. This equation was used to determine EF in the animal and human experiments.

k₂ represents back diffusion of radioactivity from tissue to vascular space.

Tracer exchange between the compartments is described by the following differential equation:

As the total activity measured in a region is composed of counts from tissue and blood, all models contained a parameter () correcting for blood activity.

where C_VOI, in the human PET study is the ¹¹C activity concentration in the defined volume of interest, in the animal study it is the ¹¹C activity measured by the tip of the -scintillator; , percentage of intravascular space in tissue; C_T, activity in the extra vascular compartment; C_B, total blood activity.

C_T was calculated by numerical integration of the differential equations. The correction for the contribution of blood in the brain () was fixed at 0.05. This value considers that signals originating from large superficial pial vessels impact the -scintillator measurements.

Delay of Onset of Radiolabel Washout

To check whether the start of the radioactivity washout from tissue was delayed, the first 5 mins of data after tracer injection were refitted with only k₂ as a free parameter and K₁ fixed to the full length fit.

Acknowledgements

The authors acknowledge the important laboratory and editorial help by Prateep Beed and John Mc Clacken, respectively, and thank Tibor Cservenyàk and Konstantin Drandarov for radiosynthesis of the tracer. The authors also thank Rolf Grütter for valuable discussions.

Supplementary Information accompanies the paper on the Journal of Cerebral Blood Flow & Metabolism website (http://www.nature.com/jcbfm)