Address correspondence and reprint requests to Gerald A. Dienel, PhD Department of Neurology, Slot 830, University of Arkansas for Medical Sciences, 4301 W. Markham St., Shorey Bldg., Room 715, Little Rock, AR 72205, USA. E-mail: firstname.lastname@example.org
α-Syntrophin is a component of the dystrophin scaffold-protein complex that serves as an adaptor for recruitment of key proteins to the cytoplasmic side of plasma membranes. α-Syntrophin knockout (KO) causes loss of the polarized localization of aquaporin4 (AQP4) at astrocytic endfeet and interferes with water and K+ homeostasis. During brain activation, release of ions and metabolites from endfeet is anticipated to increase perivascular fluid osmolarity, AQP4-mediated osmotic water flow from endfeet, and metabolite washout from brain. This study tests the hypothesis that reduced levels of endfoot AQP4 increase retention of [14C]metabolites during sensory stimulation. Conscious KO and wild-type mice were pulse-labeled with [6-14C] glucose during unilateral acoustic stimulation or bilateral acoustic plus whisker stimulation, and label retention was assayed by computer-assisted brain imaging or analysis of [14C]metabolites in extracts, respectively. High-resolution autoradiographic assays detected a 17% side-to-side difference (p < 0.05) in inferior colliculus of KO mice, not wild-type mice. However, there were no labeling differences between KO and wild-type mice for five major HPLC fractions from four dissected regions, presumably because of insufficient anatomical resolution. The results suggest a role for AQP4-mediated water flow in support of washout of metabolites, and underscore the need for greater understanding of astrocytic water and metabolite fluxes.
Fluxes of glucose through the energy-producing and biosynthetic pathways in neurons and astrocytes are closely linked to rates of neurotransmission. However, the biochemical and physiological processes that underlie increases in glucose utilization during functional activation are not established in sufficient detail to adequately understand the cellular basis of metabolic activity and roles of metabolite trafficking. Brain activation in normal conscious subjects frequently up-regulates glycolysis compared with oxidative metabolism, thereby increasing lactate production by unidentified cells and lactate release from the stimulated structures (Dienel 2012a, b). Rapid release of substantial quantities of lactate underlies the discordant images of brain activation obtained with 14C-labeled deoxyglucose (DG) and glucose (Dienel 2012b), and is consistent with a small magnitude of cell-to-cell lactate shuttling coupled to its oxidation by nearby activated cells, as predicted by metabolic modeling (Mangia et al. 2011).
The [14C]DG method assays the hexokinase step and measures total glucose utilization (Sokoloff et al. 1977), whereas assays with [6-14C]glucose rely on label trapping in the TCA cycle-derived amino acid pools and reflect mainly glucose oxidation. Many laboratories have shown that stimulus-induced increases in total glucose utilization assayed in normal conscious rats with [14C]DG are much higher to those registered with [1- or 6-14C]glucose (Dienel 2012b). Our previous studies have demonstrated that lactate is quickly labeled by [14C]glucose, rapidly dispersed into the astrocytic syncytium via gap junction-mediated metabolite trafficking, and released to blood within minutes of pulse labeling; inhibition of lactate transporters and blockade of gap junctions increase label retention in the activated tissue (Adachi et al. 1995; Cruz et al. 1999, 2007). In adult rat brain slices, astrocytes have a much higher initial rate of and capacity for lactate uptake from extracellular fluid compared with neurons, and lactate diffusion through gap junctions to other astrocytes greatly exceeds lactate shuttling to nearby neurons (Gandhi et al. 2009). Astrocytes are extensively coupled by gap junctions, and dye transfer from a single impaled astrocyte in the inferior colliculus labels as many as 10 000 cells within 5 min and causes extensive dye labeling of gap junction-coupled perivascular endfeet (Ball et al. 2007). Astrocytic endfeet face the vasculature and pia, and contain monocarboxylic acid (Rafiki et al. 2003; Bergersen 2007) and glucose transporters (Simpson et al. 2007), and channels for water (aquaporin4, AQP4) and K+ (Kir4.1 and BK [large-conductance Ca2+-sensitive channels]) (Price et al. 2002; Amiry-Moghaddam et al. 2004a; Nagelhus et al. 2004) (Fig. 1). Release of 14C-lactate and other metabolites to perivascular space facilitates their washout by discharge to blood (Cruz et al. 1999) and the lymphatic drainage system, along with small molecules and proteins (e.g., albumin and amyloid-β) in interstitial fluid (Bradbury and Cserr 1985; Ball et al. 2010).
Perivascular fluid flow is driven by aortic pulsations and it moves along the vasculature to spinal lymph nodes and through the cribriform plate of the nose to cervical lymph nodes (Bradbury and Cserr 1985; Rennels et al. 1985; Johnston et al. 2005; Carare et al. 2008; Ball et al. 2010). The peripheral lymphatic drainage system accounts for a large fraction of cerebrospinal fluid (CSF) absorption (Koh et al. 2005; Nagra et al. 2006), and sealing the cribriform plate reduces the passage of [125 I]albumin injected into the lateral ventricle to cervical lymph by 87% (Bradbury and Westrop 1983) and doubles resting intracranial pressure (Mollanji et al. 2002). The lymphatic drainage system is functionally important because ligation or extirpation of cervical lymph nodes causes edema in the brain and eye, increased intracranial pressure, behavioral abnormalities, severe swelling of astrocytic endfeet, mitochondrial swelling, filling of the space between the basement membranes with fluid, appearance of vacuoles in endothelial cells of brain capillaries, myelin damage, increased protein concentration, and elevated macrophage number (Csanda et al. 1963; Csillik and Foldi 1967; Foldi et al. 1967, 1968a,b; Casley-Smith et al. 1976). Together, the above findings indicate that other fluid efflux pathways (e.g., arachnoid villi) do not adequately compensate for lymphatic blockage. The recent, important finding that clearance of mannitol, dextran, and amyloid-β from brain is markedly reduced in AQP4 knockout (KO) mice suggests that AQP4-mediated water flow supports lymphatic drainage and solute removal (Iliff et al. 2012). Differentiated astrocytes cultured from newborn AQP4-null mice have about a two-fold reduction in water permeability when assayed at 37°C (Solenov et al. 2004), raising the possibility that water flux changes of this magnitude may be sufficient to influence tracer clearance in vivo, assuming that astrocytic plasma membrane water permeability is similar in cultured cells and adult brain.
AQP4 is the major brain water channel that is localized in astrocytes, not endothelium, and its polarized distribution at endfeet requires the presence of α-syntrophin, a component of the dystrophin scaffold-protein complex that serves as an adaptor for recruitment of membrane channels, receptors, kinases, and other proteins to the cytoplasmic side of plasma membranes (Albrecht and Froehner 2002; Amiry-Moghaddam et al. 2004a; Bragg et al. 2006; Haj-Yasein et al. 2011). Deletion of the α-syntrophin gene disrupts targeting of AQP4 to endfeet (Yokota et al. 2000; Neely et al. 2001), and AQP4 redistributes to other regions of the astrocytic plasma membrane. AQP4 mis-localization is associated with endfoot swelling during basal conditions, slower clearance of extracellular K+ after electrical stimulation, and attenuated brain edema after transient ischemia or hypo-osmotic challenge (Amiry-Moghaddam et al. 2003a,b, 2004b). Endfoot swelling suggests that tortuosity of extracellular space limits contributions of compensatory water efflux pathways in α-syntrophin KO mice. Endfoot swelling was ascribed to reduced efflux of water generated by glucose oxidation and ATP synthesis, with 6CO2 and 6H2O formed from each glucose and one H2O per ATP formed from ADP + Pi (i.e., 32–36 ATP/glucose) (Amiry-Moghaddam et al. 2003a). However, this magnitude of metabolic water efflux appears to be overestimated because ATP levels are constant under steady state conditions, and an equivalent quantity of water would be consumed by concomitant ATP hydrolysis. Thus, for a resting brain glucose oxidation rate of 1 μmol/g/min, the calculated oxidative water formation rate is 0.11 μL/g/min [i.e., 6 μmol/g/min × 18 μg/μmol × 1 nL/1 μg (density of water) = 108 nL/min/g], which is 30–50% of the rate of interstitial fluid drainage from rat brain [~0.18–0.3 μL/min/g, (Bradbury and Cserr 1985)]. CSF drains from brain via the arachnoid and lymphatic systems (Koh et al. 2005; Nagra et al. 2006), and CSF formation rate in pentobarbital-anesthetized rats is 3.4 μL/min (Chodobski et al. 1995). If astrocytic endfeet are involved in oxidative water and CSF clearance, perivascular fluid-lymphatic water flow may be reduced by AQP4 mis-localization or elimination.
Release of osmotically active compounds (e.g., K+, lactate, and other substances) from astrocytic endfeet into perivascular space during brain activation is posited to stimulate AQP4-mediated osmotic water flow and contribute to lactate washout, and this water-metabolite flux is hypothesized to be reduced in α-syntrophin KO mice (Fig. 1). As a first step toward analysis of factors that contribute to lactate washout from activated brain in vivo, the objective of this study was to determine whether greater quantities of 14C-metabolites of [6-14C]glucose are retained in brain of α-syntrophin KO mice during sensory stimulation. Our previous studies in conscious rats demonstrated that a unilateral broadband acoustic stimulus increased glucose utilization in the activated inferior colliculus, an auditory processing structure, by about 70% when assayed with the routine [14C]DG method (Cruz et al. 2005). A single tone stimulus activates fewer cells and produces a smaller side-to-side difference, about 47% with [14C]DG that contrasts the 18% difference when assayed with [1- or 6-14C]glucose (Cruz et al. 2007). Thus, if AQP4 mis-localization increases retention of labeled metabolites of [14C]glucose, the side-to-side difference should rise from an anticipated 18% in wild-type mice toward 70% in KO mice.
Materials and methods
Male α-syntrophin KO on a C57bl6J background (10 generations) (Adams et al. 2000) and male wild-type C57bl6J littermate mice (3–4 months old) were housed under a standard 12 h light-dark cycle and given free access to food and water. On the day of the experiment, the non-fasted mice were anesthetized with isofluorane (1–2%, to effect), and catheters were inserted into a femoral artery and vein of all animals. Metabolic labeling assays using [6-14C]glucose were carried out in conscious mice using either of two brain activation protocols, unilateral acoustic stimulation or bilateral acoustic plus whisker stimulation. When assayed with [14C]DG, glucose utilization increases in the inferior colliculus by 45–70% during acoustic stimulation of conscious rats (Cruz et al. 2005, 2007) and by 50–60% in barrel cortex of conscious mice given whisker stimulation (Melzer et al. 1985; Melzer and Smith 1998; Esaki et al. 2005).
In the first paradigm study, mice were given unilateral acoustic stimulation, and computer-assisted autoradiographic image analysis was used to evaluate side-to-side differences in total 14C concentrations in regions of interest. To facilitate monaural activation, one tympanic membrane of the anesthetized mice was punctured and the external auditory meatus was plugged with bone wax. This sensory stimulation paradigm was carried out in a sound-insulated box so the animals were minimally exposed to other external stimuli or stresses.
The second protocol was used for determination of levels of unlabeled and 14C-labeled metabolites in extracts of dissected tissue samples from funnel-frozen brain. The small size of mouse brain structures required pooling of tissue from both hemispheres, and the mice were, therefore, given bilateral broadband acoustic stimulation and along with bilateral whisker stimulation to activate the whisker barrel somatosensory cortex. This paradigm involved alerting, visual, and motor responses of the mice to a novel situation involving brushing of the face, whiskers, and paws with soft paint brushes, in addition to the acoustic stimulus. This procedure enabled analysis of four brain regions in each animal. To preserve labile intermediary metabolites, the brains were funnel-frozen in situ at the end of the experimental period, and these mice were prepared as previously described (Cruz and Dienel 2002; Dienel et al. 2002; Cruz et al. 2005). After preparative surgery, all mice were restrained via hind limb plaster casts and allowed to recover for at least 3 h in a shelter box. Special care was taken to minimize noise and other stimuli throughout the preparative and experimental procedures. During the recovery and experimental periods rectal temperature was monitored and maintained at 37°C. Immediately prior to the experiment, mean arterial blood pressure was measured with a calibrated Micro-Med Analyzer (Louisville, KY, USA), and arterial blood was drawn for assay of PCO2, PO2, and pH (CIBA-Corning Model 248 pH/Blood Gas Analyzer, CIBA-Corning, Medfield, MA, USA), hematocrit, and plasma glucose and lactate levels (Yellow Springs Instruments, Model 2700 Dual Channel Analyzer, Yellow Springs, OH, USA). All animal use procedures were in strict accordance with the NIH Guide for Care and Use of Laboratory Animals, and were reviewed and approved by the local animal care and use committee.
Sensory stimulation and metabolic labeling
For the autoradiographic imaging study, a unilateral acoustic stimulus was presented to conscious wild-type and α-syntrophin knockout mice in a sound-insulated box using a Grass Instruments (Astro-Med, West Warwick, RI, USA) S10CTCM Click-Tone Module and two Grass Model 10H2S Audiometric Headphones [broadband click, 40 Hz-8 kHz tone with an intensity setting of 103 dB to obtain an intensity of about 88 dB at the location of the animal (Cruz et al. 2005)]. The acoustic stimulus was initiated 10 min before the intravenous pulse of d-[6-14C]glucose (100 μCi/kg; 56 mCi/mmol; Amersham Biosciences, Piscataway, NJ, USA) and maintained throughout the labeling period. The catheters were accessible from outside the shelter box, and 11 samples of arterial blood were drawn at timed intervals for assay of plasma levels of glucose, lactate, and 14C (Packard Model 2550 liquid scintillation counter, Packard Instruments Company, Meridian, CT, USA). These data were used to calculate the specific activity of glucose at each time point and the integrated specific activity of the precursor pool of [14C]glucose in arterial plasma (i.e., the area under the time-specific activity curve). At the end of the experimental interval (~5 min), the mice were killed with pentobarbital (200 mg/kg, iv), their brains were quickly removed, frozen in isopentane at ~−40°C, and stored at −80°C. Serial coronal sections (20 μm-thick) were cut at about −20°C, dried onto glass coverslips at 55°C, and exposed to x-ray film along with calibrated 14C methacrylate microscale standards (Amersham). Total 14C levels in regions of interest were determined by computer-assisted densitometry (InterFocus Imaging Ltd., Cambridge, England), with a spatial resolution of ~100–200 μm (Sokoloff et al. 1989). Comparison of side-to-side differences in each animal minimized any effects of animal-to-animal variations on the calculated results and increased sensitivity to detect side-to-side differences between groups.
Acoustic and whisker stimulation
For combined acoustic-whisker stimulation of conscious wild-type and α-syntrophin knockout mice, the shelter box was removed immediately prior to stimulation onset. The acoustic stimulus described above plus bilateral brushing of the whiskers with soft paintbrushes were initiated 1 min prior to the intravenous [6-14C]glucose injection, after which 11 timed arterial blood samples were drawn and analyzed as described above. This stimulus paradigm presented auditory, whisker, and visual stimuli to the mice, and they moved the upper body and forelimbs in response to the novel, bilateral whisker brushing. At the end of the experimental period, mice were anesthetized with an intravenous bolus of the fast-acting anesthetic thiopental (25 mg/kg, i.v.), and their brains were frozen in situ by applying liquid nitrogen to the intact skull (within about 30 s. after the thiopental bolus) via a funnel (Ponten et al. 1973). The heads were stored at −80°C until dissection of whisker barrel field cortex, cerebral cortex located medial to the barrel cortex (including portions of association cortex, fore- and hind-limb sensory cortex, and motor cortex), visual cortex, and inferior colliculus. Dissection and weighing of samples in a cryo-glovebox at about −20°C prevents postmortem ischemic glycolytic metabolism that would result in consumption of glucose, glycogen, and glycolytic intermediates and elevation of lactate level (Lowry et al. 1964). Dissections were based on a mouse atlas, and whisker barrel cortex was determined by means of measured distances from landmarks in the skull (lambda and bregma) and from the midline after the skull was removed. Unfortunately, the anatomical accuracy of the grossly dissected tissue samples does not have the same precision as computer-assisted image analysis of autoradiographs. For example, it was difficult to remove adjacent tissue from regions of interest (e.g., subcortical white matter) because of the small size of the tissue samples. Technical constraints also prevented use of thicker cryo-sections to obtain tissue punches of regions of interest. Ethanol extracts of each frozen tissue sample were prepared (Cruz and Dienel 2002; Dienel et al. 2002), and unlabeled glucose and lactate levels were assayed with a Yellow Springs Instruments 2700 Select biochemistry Analyzer. Labeled compounds in the extracts were separated by high pressure liquid chromatography (HPLC) using a Dionex (Sunnyvale, CA, USA) anion exchange chromatography system and Dionex IonPac AS11-HC analytical column. Elution times of selected unlabeled standards were determined by use of a suppressor (to lower background by acid-base neutralization of the eluant) and conductivity detection (Wang and Dienel 1994; Cruz et al. 2005). For analysis of labeled metabolites in the brain tissue extracts, the suppressor was removed from the system to prevent loss of labeled compounds that cross the suppressor membrane and are discharged to the waste line (Wang and Dienel 1994). Timed fractions of the eluate were collected and their 14C contents determined by liquid scintillation counting. 14C Recoveries in the column eluates for all samples averaged 97.1 ± 9.7% (mean ± SD, n = 52), and the 14C contained in the six major HPLC fractions (see Table 3) accounted for 97.6 ± 2.1% of the total recovered 14C.
Net incorporation of [6-14C]glucose into labeled compounds retained in brain at the end of the 5 min labeling interval were calculated in two ways. For the mice given unilateral acoustic stimulation, total 14C concentrations in regions of interest in each hemisphere were determined by computer-assisted quantitative autoradiography, and the 14C level in the right (activated) hemisphere was normalized to that in the left (non-activated) hemisphere of the same animal; each hemisphere was exposed to the same integrated specific activity in the arterial plasma precursor pool. For the animals given combined acoustic plus whisker stimulation, the total 14C concentration (nCi/g) in each HPLC fraction was divided by the integrated specific activity of [14C]glucose in arterial plasma ([nCi/μmol glucose]min) for that animal. This accounts for animal-to-animal differences in the precursor pool and yields a minimal calculated rate of glucose utilization with units of μmol plasma glucose/[g min]. These rate estimates are minimal values for two reasons, (i) the integrated specific activity of the [6-14C]glucose precursor pool arterial plasma exceeds that in brain because of restricted passage of [14C]glucose across the blood-brain barrier, and (ii) rapid release of labeled metabolites of glucose, especially [14C]lactate, from activated tissue to blood, other brain regions, and lymphatic drainage, causes incomplete recovery of 14C-metabolites and underestimation of the total metabolite pool size (Adachi et al. 1995; Cruz et al. 1999, 2007).
Comparisons of ratio data were made with the one-sample t-test to determine if side-to-side differences were statistically significantly different from 1.0 (i.e., no difference because of the stimulation paradigm). Comparisons between group means were made with the unpaired t-test. Multiple comparisons among groups were made with analysis of variance (anova) followed by the post-hoc Bonferroni multiple comparison test.
Body weights and temperature, blood pressure, hematocrit, blood gases and pH, and arterial plasma glucose and lactate levels in all experimental animals were within the normal range and were similar in the four experimental groups (Table 1). However, in the acoustic plus whisker stimulation group, the baseline glucose and lactate levels in the knockout mice were somewhat lower than the wild-type mice, whereas during stimulation they were not statistically significantly different from wild-type. The mice were not fasted overnight prior to the experiment, and their startle responses to the abrupt onset of novel stimuli and their physical movement of the forelimbs, head, and torso caused arterial plasma glucose and lactate levels to rise during the 5 min labeling period, presumably because of liver glycogen-dependent release of glucose to blood and activity-dependent release of lactate from muscle. Lactate levels rose at onset of the stimulus, then tended to decrease with time during the assay period, whereas glucose levels tended to rise slightly with time. The percent increases in arterial plasma lactate levels during the stimulation interval ranged from 50 to 130% in the different experimental groups, whereas the percent rise in plasma glucose level was much smaller, ranging from 5 to 25%. Indices of stress responses were not assayed, but liver glycogenolysis is enhanced by stress, and this would cause plasma glucose levels to rise; the small increases in glucose level are consistent with relatively low stress.
Table 1. Physiological variables in conscious α-syntrophin knockout (KO) and wild-type mice
Body weight (g)
MABP (mm Hg)
Rectal Temp (oC)
PO2 (mm Hg)
PCO2 (mm Hg)
Values (means ± SD for the number of mice indicated in parentheses) were determined immediately before the metabolic labeling experiment. MABP and Hct. denote mean arterial blood pressure and hematocrit, respectively. pH, PO2, and PCO2 were determined in arterial blood. Glucose and lactate levels were assayed in arterial plasma prior to (baseline) and at timed intervals during the 5 min labeling interval after the intravenous bolus injection of [6-14C]glucose. The glucose and lactate levels during stimulation are the group means of the mean values of the samples drawn during metabolic labeling for each animal.
p < 0.05 compared with wild-type animals in same experimental group (t-test).
Selective increase in labeling of inferior colliculus in KO mice during acoustic stimulation
Metabolic activation of the auditory pathway in conscious wild-type and α-syntrophin KO mice was first assayed by quantitative autoradiography after unilateral acoustic stimulation. Autoradiography registers total 14C and does not distinguish between unmetabolized [14C]glucose and labeled metabolites. However, our previous studies in conscious rats showed that ~80–90% of the label in brain at 5–7 min after intravenous pulse labeling with [6-14C]glucose is recovered in metabolites, and images reflect mainly metabolism (Adachi et al. 1995; Dienel et al. 2002). The wild-type mice did not exhibit left-right differences in total 14C concentration in the auditory or non-auditory structures (Fig. 2, Table 2). Failure of [6-14C]glucose labeling to register effects of unilateral auditory activation was unexpected because increases were detectable in conscious rats (Cruz et al. 2005, 2007). In contrast, the α-syntrophin knockout mice did have significantly higher levels of 14C in the activated inferior colliculus compared with its contralateral structure (Fig. 2, Table 2). The integrated specific activity of [6-14C]glucose in arterial plasma was similar in the wild-type and knockout mice, indicating that exposure of brain to precursor [14C]glucose was equivalent in the two groups (legend, Table 2).
Table 2. Selective increase in retention of 14C-labeled compounds derived from [6-14C]glucose in the inferior colliculus of α-syntrophin knockout (KO) mice during unilateral acoustic stimulation
Right/left ratio for total 14C
Wild-type (n = 6)
α-Syntrophin KO (n = 5)
Conscious mice were given unilateral broadband acoustic stimulation to activate auditory pathways in the right hemisphere, pulse-labeled with an intravenous injection of tracer amounts of [6-14C]glucose, and total 14C concentrations were determined in the activated and contralateral regions of interest in each animal by quantitative autoradiography. Values (mean ± SD for the indicated number of animals) are ratios of total 14C concentration in the activated compared with contralateral region of interest. The integrated specific activities of [14C]glucose in arterial plasma ([nCi/μmol glucose] min) were 357 ± 194 and 270 ± 108 for the wild-type and KO mice, respectively.
p < 0.05 (one-sample t-test to determine if the right/left ratio is different from 1.0).
Labeling of metabolite pools in brain during bilateral auditory and whisker stimulation
Because the increase in retained label in knockout mice (Table 2) could reflect, in part, higher levels of unmetabolized [14C]glucose, [14C]metabolites were analyzed in four dissected brain structures of mice given bilateral whisker brushing plus acoustic stimulation. Unlabeled glucose and lactate levels in different brain regions of wild-type and knockout mice given the combined stimulation paradigm ranged from about 1–2 μmol/g, with the lowest levels of about 1 μmol/g in the medial cortex (Table 3). There were no statistically significant differences in glucose or lactate levels between the wild-type and knockout mice within each region or among regions.
Table 3. Regional metabolism of [6-14C]glucose during bilateral stimulation of whisker-to-barrel and acoustic pathways in conscious wild-type and α-syntrophin knockout (KO) mice
Brain glucose (μmol/g)
Brain lactate (μmol/g)
[6-14C]glucose incorporation into brain metabolites (μmol/g/min)
Mice were pulse-labeled with [6-14C]glucose during whisker and acoustic stimulation, and levels of unlabeled and labeled compounds were assayed in extracts of grossly dissected brain regions of funnel-frozen brain (See Methods). The 14C recovered in each fraction (nCi/g wet wt.) from each animal was normalized by dividing by the integrated specific activity (ISA; [nCi/μmol glucose] min) of arterial plasma glucose for that animal (see Methods); mean ISA values were 302 ± 86 and 313 ± 37 for the wild-type and KO mice, respectively. Values are means ± SD (n = 7/group except the visual cortex of the KO group, n = 4).
The total column corresponds to overall minimal glucose utilization rates in each structure and does not include the 14C in the 3 min fraction that contains the unmetabolized precursor pool of [14C]glucose (and perhaps 14C-labeled neutral amino acids).
p < 0.05 versus wild-type in the same brain structure (t-test).
p < 0.05 versus whisker barrel cortex and visual cortex in KO mice (anova and Bonferroni's multiple comparison test).
Brain glucose concentration represents the balance between glucose supply and demand, and it varies with plasma glucose concentration and metabolic rate. We previously showed that the steady state brain:plasma glucose distribution ratio in conscious, resting rats was relatively stable and equal to about 0.20 when the glucose concentration exceeded 9 mmol/L in arterial plasma or 1.5 μmol/g in brain (Dienel et al. 1991; Holden et al. 1991). The brain : plasma distribution ratio for glucose at the end of the 5 min stimulation interval averaged 0.13 for the wild-type and knockout mice, and it ranged from about 0.05 to 0.23 in various brain regions, with the lowest values in medial cortex. These ratios are unlikely to represent a new steady state during stimulation because of the relatively short experimental period and a 1.2–1.6 min half-life for brain glucose in resting rat brain (Savaki et al. 1980). When plotted against plasma glucose concentration, the glucose distribution ratios for the stimulated mice were not clustered within a relatively narrow range when plasma glucose was 10–15 mmol/L (Fig. 3a), as was observed at steady-state in resting rats; instead there was a rather large range of values, extending from about 0.05 to 0.22 (Fig. 3a). In contrast, the glucose distribution ratios during brain activation were linearly correlated with brain glucose content (Fig. 3b) and were much lower than the corresponding steady-state values in resting rats. Thus, during brief activation, plasma glucose levels need not be predictive of those in brain. The rate of glucose utilization, particularly the medial cortex, probably exceeded its delivery, causing the brain:plasma distribution ratio to fall.
Brain lactate concentration is the net result of the fluxes into (uptake from plasma and generation within brain from glucose or glycogen) and from (efflux to blood and metabolism in brain) the lactate pool. Regional brain:plasma distribution ratios for lactate were similar in both groups of stimulated mice and ranged from about 0.3 to 1.3, with one high outlier. The lactate distribution ratio was lowest in the tissue samples with the highest plasma lactate concentration (Fig. 3c, Tables 1 and 3), suggesting some uptake of lactate from blood to brain in these mice.
Labeled compounds in four brain regions of mice given acoustic plus whisker stimulation were separated into six major fractions by anion exchange HPLC (Table 3). Glucose was recovered in the 3 min fraction that may also contain other labeled compounds, e.g., neutral amino acids (Wang and Dienel 1994). Lactate eluted at 11.5–12.5 min, and initial experiments showed that most of the label was removed from this fraction by pre-treatment of the extract with lactate oxidase. The 22–23 min fraction had the highest 14C content and probably contains the glutamate, aspartate, and other acidic compounds; glutamate is the most highly labeled metabolite in rat brain at 5 min after pulse labeling with [6-14C]glucose (Dienel et al. 2002). Later-eluting fractions would contain phosphorylated and other anionic metabolites (Cruz et al. 2005). The 3 min fraction contained about 20% of the total 14C in the extracts, and the other fractions contained the following percentages of the sum of the 14C in all fractions excluding the 3 min glucose fraction: 5–7 min, 19%; 11.5–12.5 min, 8%; 22–23 min, 50%; 24–26 min, 6%; and 28–42 min, 16%. Labeling of the five major metabolite fractions and the total calculated rate of glucose utilization (i.e., sum of all fractions except the 3 min) were similar in the wild-type and knockout mice in each brain region (Table 3). The lactate fraction contained a small proportion of the 14C (8%) and did not differ in the wild-type and knockout mice. Total calculated glucose utilization did not differ among the four dissected brain regions in the wild-type mice, whereas medial cortex was significantly lower than that of barrel cortex and visual cortex in the knockout mice (p < 0.05, anova, Bonferroni test).
The biochemical and physiological changes underlying energetics of the neurovascular unit (comprised of astrocytes, neurons, and endothelial cells) during brain activation are complex and inadequately understood. Functional imaging and metabolic studies using [14C]DG, [14C]glucose, [14C]acetate, and other tracers have been useful tools to identify and quantify pathways activated during sensory stimulation of normal conscious rats, and such studies have shown that astrocytes are not solely glycolytic during sensory stimulation; they also increase the fluxes through the glycogenolytic and oxidative pathways (Dienel et al. 2001, 2002, 2007a, b; Cruz and Dienel 2002; Cruz et al. 2005; Dienel 2012b). Our studies with spreading cortical depression demonstrated up-regulation of glycolysis with lactate efflux to blood, and [14C]lactate accounted for almost all of the label released from brain to blood (Adachi et al. 1995; Cruz et al. 1999). However, lactate release to blood explained only half of the magnitude of the 50% underestimate of metabolic rate when assayed with [6-14C]glucose compared with [14C]DG, suggesting the presence of another major route for metabolite washout from brain. Subsequently, we identified lactate transporters and metabolite diffusion through astrocytic gap junctions as key elements in the rapid dispersal of lactate and labeled metabolites from activated regions via the astrocytic syncytium and perivascular endfeet (Ball et al. 2007; Cruz et al. 2007; Gandhi et al. 2009). Recovery of [14C]lactate and other labeled compounds in the meninges after microinfusion of 14C-labeled glucose or lactate into the inferior colliculus, with 34% of the glucose-derived label recovered in the meninges compared with 60% in the infused inferior colliculus, brought our attention to lactate discharge via perivascular-lymphatic drainage system (Ball et al. 2010). Because [6-14C]glucose registers small increases in oxidative metabolism compared with total glucose utilization during activation and lactate does not appreciably accumulate in activated brain, more lactate must be generated and most of the lactate produced must be released (Dienel 2012a, b). Glucose utilization in the inferior colliculus of the rat rose from 0.7 to 1.2 μmol/g/min during broadband acoustic stimulation (Cruz et al. 2005). About 30% of this increase is estimated to be oxidized (0.5 × 0.3 = 0.15) based on trapping of products of [14C]glucose [see Fig. 4d in (Dienel 2012b)], so glycolysis rises by 0.35 μmol glucose/g/min, equivalent to 0.7 μmol lactate/g/min. During 5 min stimulation, 3.5 μmol lactate/g is formed, and brain lactate level rises by ~1 μmol/g, so the quantity of lactate washed out is 2.5 μmol/g, or 0.5 μmol/g/min over the 5 min assay interval. Continuous lactate release from endfeet and lactate transport along the vasculature by pulsatile flow may serve an important long-range signaling function to increase blood flow and glucose delivery to activated tissue because lactate (Hein et al. 2006; Yamanishi et al. 2006; Gordon et al. 2008), K+, and other substances released from endfeet are vasodilators (Attwell et al. 2010; Dunn and Nelson 2010). Release of K+ from BK/Kir4.1 channels to perivascular space is vasodilatory when [K+]e rises from ~3 to < 20 mM (Filosa et al. 2006; Girouard et al. 2010). If K+ release increases perivascular fluid osmolality, water fluxes would rise and may involve endfoot AQP4 channels. MCT-mediated co-transport of water with lactate/H+ released from endfeet and glucose influx from blood would also modulate an endfoot/perivascular space osmotic gradient (MacAulay and Zeuthen 2010).
In this study, we sought to extend our knowledge of brain activation by asking whether disruption of water flow in astrocytic endfeet because of mis-localization of AQP4 in α-syntrophin knockout mice would impair metabolite washout during sensory stimulation (Fig. 1). When total 14C levels in regions of interest in mice given unilateral acoustic stimulation were assayed by brain imaging that has high anatomical resolution, the 14C level was 17% higher in the activated compared with the contralateral inferior colliculus in knockout, but not wild-type, mice (Table 2). Surprisingly, left-right differences were not detected by image analysis in the auditory structures in the wild-type mice (Fig. 2, Table 2), whereas they were readily detected in rats given a similar stimulation protocol (Cruz et al. 2007), raising the possibility that release of labeled metabolites is much faster in mice than rats. Metabolite analysis of extracts of grossly dissected tissue showed that unlabeled glucose levels and the 14C recovered in the glucose fraction in each brain region were similar in the knockout and wild-type mice, suggesting that [14C]glucose does not explain the higher levels of total 14C in autoradiographs from knockout mice (Fig. 2). The biochemical analyses did not, however, reveal lactate accumulation or higher levels of label in downstream pools, such as glutamate, in any of the four structures in the knockout compared with wild-type mice (Table 3). The reasons for discordant results in the imaging and biochemical analysis remain to be established, but inclusion of adjacent tissue with the dissected regions of interest is likely. Accurate dissections of small brain structures in a cryo-glovebox are technically difficult, and the presence of non-activated tissue would blunt and obscure the relatively small differences identified by autoradiographic image analysis.
The similarity of labeling of the five major metabolite fractions and the overall calculated rate of glucose utilization in four brain regions indicates that there are no large differences in metabolism of glucose and metabolite retention in the knockout and wild-type mice (Table 3). However, in both groups, the calculated rates using [6-14C]glucose are much lower than values anticipated from [14C]DG studies, and are similar to values in resting animals. Unilateral sensory stimulation of conscious mice and rats produces side-to-side differences in the range of 50–70% corresponding to rates of up to 1.2 μmol/g/min (Ginsberg et al. 1987; Melzer and Smith 1998; Cruz et al. 2005, 2007; Esaki et al. 2005; Dienel et al. 2007a). Lower calculated rates obtained with [6-14C]glucose (Table 3) compared with [14C]DG and failure to detect side-to-side differences in wild-type mice by brain imaging are consistent with underestimates because of rapid [14C]metabolite efflux in both groups of mice and to use of plasma integrated specific activity.
The brain:plasma distribution ratios for glucose (Fig. 3) are unlikely to have reached a new steady state during the brief activation, but many values, particularly in medial cortex of both groups and visual cortex of wild-type mice were lower than anticipated, based from our previous studies in resting rat brain at steady state (Dienel et al. 1991, 1997; Holden et al. 1991). Mice may have a lower resting steady-state glucose distribution ratio compared with rats, but the lowest ratios probably arose from a glucose supply-demand mismatch during brain activation. Postmortem metabolism of brain glucose is considered an unlikely cause for the low distribution ratios because conversion of glucose to lactate would have increased both the level of unlabeled lactate and the label recovered in the lactate fractions (11.5–12.5 min) above those in the corresponding glucose fractions (3 min), and this was not the case (Table 3). In fact, the lactate fractions contained only about 20–30% of the amount of 14C in the glucose-containing fractions (Table 3). In spite of the lower brain glucose levels in medial cortex compared with the other three regions, the mean tissue glucose concentration was 1 μmol/g (Table 3), which is 20 times the Km of hexokinase for glucose [0.05 mmol/L, (Wilson 2003)]. Thus, even if metabolic demand exceeds supply, glucose is continuously delivered from arterial plasma, which has an adequate concentration of glucose (Table 1) to sustain increased brain metabolism. In rats, glucose delivery exceeds demand by 60% over a wide range of metabolic rates and plasma and brain glucose levels [see Fig. 3e in (Dienel 2012b)]. Based on Michaelis-Menten kinetics, the rate of glucose phosphorylation by hexokinase will not fall until brain glucose level is below 0.5 μmol/g, so the medial cortex still has a two-fold margin of safety.
In conclusion, the higher side-to-side differences detected by autoradiographic brain imaging (Fig. 2, Table 2) raise the possibility that elimination of AQP4-mediated water flow at the astrocytic endfeet is associated with reduced metabolite washout from activated brain. More work is, however, required to establish a role for endfoot water fluxes in perivascular fluid flow and the lymphatic drainage system. Brain water homeostasis is quite complex and poorly understood, and water transport among different compartments is linked to local osmotic changes and water co-transport with lactate, glucose, glutamate, and electrolytes (MacAulay and Zeuthen 2010, 2012). Previous studies with α-syntrophin knockout mice revealed that loss of the polarized localization of AQP4 at endfeet is associated with endfoot swelling and reduced brain edema after transient ischemia or hypo-osmotic challenge, whereas localization of other proteins was not grossly disrupted, e.g., endfoot potassium channel Kir4.1 was stated to be only modestly decreased, and the levels of monocarboxylic acid transporter 1 (MCT1), glucose transporter 1 (GLUT1), and excitatory amino acid transporter 2 (EAAT2) were unaffected (Neely et al. 2001; Amiry-Moghaddam et al. 2003a, b, 2004a). Although knockout of α-syntrophin slowed K+ clearance from extracellular fluid and the mice were more sensitive to hyperthermia-induced seizures, the gene deletion did not alter the amplitude of stimulus-induced population spikes (Amiry-Moghaddam et al. 2003b) or grossly affect obvious behavior or glucose metabolism (Table 3). Deletion of AQP4 influences stimulus-induced extracellular volume dynamics without altering major aspects of synaptic function and plasticity, although burst stimulus-induced long-term potentiation is altered (Skucas et al. 2011; Haj-Yasein et al. 2012).
Four observations, (i) control of water flux at the blood-brain interface by astrocytic endfeet in glial-targeted AQP4-null mice (Haj-Yasein et al. 2011), (ii) swelling of endfeet in the α-syntrophin knockout mice (Amiry-Moghaddam et al. 2003a), (iii) reduced perivascular tracer clearance in AQP4-null mice (Iliff et al. 2012), and (iv) reduced solute clearance, severe swelling of endfeet, and increased intracranial pressure when cervical lymphatic drainage is blocked (Csanda et al. 1963; Foldi et al. 1967, 1968a, b; Casley-Smith et al. 1976; Bradbury and Westrop 1983; Mollanji et al. 2001) suggest that alternative water flux pathways involving arachnoid villi (Nagra et al. 2006) or interstitial fluid flow through clefts between astrocytic endfeet (Mathiisen et al. 2010) and across the vasculature do not adequately compensate for loss of water channels at endfeet and obstruction of lymphatic drainage. Nevertheless, interpretation of the consequences of altered physiology of the α-syntrophin and AQP4-null mice must be conservative and take into account the possibility of contributions from unidentified adaptive changes. For example, deletion of AQP4 reduced the perivascular localization of α-syntrophin and dystrophin (Eilert-Olsen et al. 2012), and knockdown of AQP4 with small interfering RNA reduced expression of connexin43 (Cx43) protein and gap junctional coupling in cultured mouse (but not rat or human) astrocytes, whereas cytoskeletal and morphological changes occurred in rat astrocytes subjected to AQP4 knockdown (Nicchia et al. 2005). Cx43 is a major gap junction protein in astrocytes, and altered Cx43 protein levels and reduced metabolite trafficking throughout the astrocytic syncytium would also reduce rates of metabolite dispersal and discharge during brain activation. In addition, gene knockout or knockdown can cause widespread changes in the brain transcriptome, as shown for Cx43 knockdown or knockout astrocytes. Cx43 down-regulation alters the expression of more than 250 gene products, disrupting intercellular communication and altering the expression (up or down) of genes in many broad categories of cellular function, including transcription, energy and metabolism, cell junctions, adhesion, and extracellular matrix, cell signaling, transport processes, and cell cycle (Iacobas et al. 2005, 2008). Thus, adaptive changes to α-syntrophin or AQP4 gene deletion may contribute to maintaining washout of metabolites from activated brain tissue. Because, the effect of α-syntrophin knockout on retention of labeled metabolites of [6-14C]glucose is measurable but small, new in vivo experimental approaches with high temporal-spatial resolution are needed to assay water and metabolite movement through astrocytes, perivascular fluid, and lymphatic drainage. These methods would also be important to evaluate pathophysiological states, such as diabetes and cardiovascular disease, where structural changes involving the neurovascular unit and thickening of the basement membrane may affect fluid flow and metabolite trafficking in the perivascular-lymphatic system.
This study was supported by National Institutes of Health grants NS038230 and DK081936 to GD and NS33145 to SCF. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Neurological Diseases, Stroke or National Institute of Diabetes and Digestive and Kidney Diseases, or the National Institutes of Health. The funding sources had no role in study design; collection, analysis, and interpretation of data; writing of the report; and the decision to submit the article for publication. The authors declare no conflict of interest.