Glucose supply to the brain in health and disease
It has been well-established that the brain has a high energy demand for the maintenance of normal physiological function; this is primarily supplied by glucose (1–3). Glucose however does not cross membranes easily, particularly in the case of the blood brain barrier (BBB), which consists of endothelial cells joined by tight junctions (4). Since glucose transport has been linked with conditions including Alzheimer’s disease and diabetes, many studies have investigated the nature of glucose delivery to the brain.
As indicated by Fick’s law, for passive diffusion of to be the mechanism of glucose transport blood glucose and cerebral glucose concentrations must be directly proportional. Crone et al. found that in canines, brain glucose uptake was dependent on plasma glucose concentration (5). This indicated that brain glucose supply occurs through a transporter mechanism and not via passive diffusion.
Figure 1: Graph showing 14C glucose uptake at various plasma glucose concentrations in canine brains. Dashed line: Expected result if mediated via passive diffusion; solid line: obtained results. (Adapted from Crone et al. 1965.)
Figure 1 not only demonstrates that brain glucose uptake cannot occur via diffusion but also suggests that individuals with hypoglycaemia, a physiological state commonly associated with diabetes, would experience higher rates of brain glucose uptake than healthy individuals. The graph shows that glucose uptake is highest at the lowest blood plasma glucose concentrations, perhaps to maintain high rates of glucose utilization, even during periods of starvation. This is supported by Robinson and Rapoport, who concluded that at half the normal plasma glucose levels in rats, brain hexokinase, an enzyme responsible for the phosphorylation of glucose within the brain, is still over 95% saturated (6). This indicates that cerebral glucose transport does not limit cerebral glucose metabolism in health and implies the existence of a mechanism to adjust cerebral glucose levels with varied blood plasma glucose concentrations. Despite the significance of cerebral transport regulation, especially considering the high cerebral metabolic demand, its mechanism remains unclear.
In addition to facilitated transport, the study previously discussed by Crone et al. suggested that glucose transport was not dependent on the presence of insulin (5). The pancreas from the studied canines was removed and the canines’ blood glucose-fructose ratios measured. They were later given an injection of insulin and the brain glucose-fructose ratios were measured again. In both cases, brain glucose-fructose ratio remained the same, indicating that brain glucose supply is insulin independent. Despite this conclusion, the study did not record the results for this investigation as cerebral blood glucose concentrations independent of fructose. If in the presence of insulin, both glucose and fructose concentrations had changed significantly yet proportionally in these canines, the cerebral glucose-fructose ratio would remain the same. Therefore, the reliability of this conclusion is questionable.
A more recent study investigated the effect of insulin on BBB glucose transport in humans (7). Rather than eliminating insulin and then replacing it as done by Crone et al., Hasselbalch et al. studied their subjects in conditions of hypoinsulinemia and hyperinsulinemia. Using an intravenous double-indicator technique, the study found no significant difference in the D-glucose blood to brain unidirectional clearance among the two groups (ml/g/min: 0.070±0.018 vs. 0.072±0.025). They also measured unidirectional clearance of fluorodeoxyglucose (FDG) from the blood to the brain using proton electron transfer (PET) and found no significant difference between hypoinsulinemic and hyperinsulinemic subjects (ml/g/min: 0.101±0.023 vs. 0.099±0.037). This data supported Crone et al. and demonstrated that brain glucose transport is insulin independent.
Facilitated glucose transporter proteins type 1 and 3 (GLUT1 and GLUT3) have been recognised as the main transporters responsible for brain glucose supply (3,8,9). As summarised in figure 2, these transporters have similar structures, with 12 transmembrane domains and terminal amide and carboxylic acid groups (10,11).
Figure 2: A simplified schematic showing GLUT1 and GLUT3 structures. (Adapted from Mueckler et al. 1985)
GLUT1 is expressed at the BBB and in brain endothelial and epithelial-like barriers while GLUT3 is expressed in the neurons (3). A study of embryonic and adult rats used various techniques to investigate GLUT1 distribution during development and found through electron microscopy that upregulation appeared to be strongly associated with the BBB (12). Investigation by light microscopic immunohistochemistry found that at all stages of development, brain blood vessels were GLUT1 immunoreactive, with restriction of the marker to capillaries within the brain parenchyma and vessels surrounding neurones. Having indicated the presence of GLUT1 in the BBB, the study also showed through electron microscopy that development of brain capillary networks corresponded with changes in GLUT1 concentration. A clear increase in luminal and abluminal immunogold particle concentration was observed with development, as shown in figures 3A and 3B.
Figure 3: Box plots showing the distribution of immunoreactive GLUT1 in A: the luminal membrane and B: the abluminal membrane of BBB capillary networks. (Taken from Bolz et al. 1996)
The immunogold particle acted as a secondary antibody used to detect the primary anti-GLUT1 antibody, therefore increases in immunogold detection corresponded to increases in immunoreactive GLUT1. This study not only confirmed that GLUT1 is distributed in the BBB but suggested upregulated GLUT1 expression during development.
Nagamatsu et al. studied GLUT3 in developing rats via Northern blot analysis and found that GLUT3 levels were initially low before significantly increasing to high levels in adults (13). This is supported by a more recent study using by McCall et al. who found GLUT3 in neuronal cell lines using Western blot analysis and used light microscopy to show the appearance of GLUT3 localisation to ‘synaptically rich’ regions of the brain, such as the cortex. This suggests that GLUT3 supplies glucose to these neuropil for the purpose of fuelling synaptic transmission (14).
Other transporters associated with brain glucose transport include GLUT2 in the astrocytes, GLUT 4 in the hippocampus and cerebellum, as well as GLUT6 in the neurons and GLUT8 in blastocytes. In addition to the facilitative transporters, there are sodium dependent glucose transporters; SGLT1, SGLT2, SGLT3, SGLT4 and SGLT6 also contribute to delivery of glucose to the brain, as discussed in a review (3).
Transporter deficiency has been shown to result in impaired brain glucose delivery. Though symptoms have been described as variable, patients with GLUT1 deficiency syndrome (GLUT1 DS) may present with seizures, developmental delay and motor problems including dystonia and ataxia (15). Compared to individuals without GLUT1 DS, patients with GLUT1 DS have been found to have significantly lower glucose brain uptake. De Vivo et al. compared the case reports of two children with defective glucose transport to 39 healthy subjects and found that the children with known defective transport had significantly lower CSF glucose levels than normal values (16). At 6 months old, the patients had CSF/blood glucose ratios of 0.22 and 0.33, both values being lower that the normal value of 0.65. In a study of mice, Tang et al. modelled GLUT1 deficiency by inactivating one copy of the Slc2a gene, which codes for the transporter protein. They found that the capillary network of 8 week and 20-week old mice with the deficiency was smaller than age-matched controls by approximately 30% and 40% respectively (17).
Corresponding with the concentration-dependent manner of cerebral glucose transport, kinetics are often described using the Michaelis-Menten (MM) model (18,19).
Figure 4: MM model; a graph depicting the shape of the graph plotted from the MM equation. (Equations from Feinendegen et al. 1986; adapted from Robinson et al. 1986)
Km represents the MM constant- a description of substrate concentration at half of the maximal reaction velocity. Vmax refers to the maximum velocity of any enzymatic reaction. With respect to human cerebral glucose transport, Km has been found to be 4-5mM, and Vmax has been found to range from 0.4- 2.0µmol/g/min, as discussed in many PET and NMR investigations of glucose transport kinetics (8,18–20).
Variations of this model have been used to explain glucose transport across the BBB, including the symmetric MM model included in papers by Feinendegen et al. and Gruetter et al. (8,18). These models include time-activity curves for both plasma glucose and cerebral glucose concentrations, which appear to better characterise unidirectional BBB glucose transport. A later study by Gruetter et al. proposed a reversible MM model to account for glucose being delivered into the BBB before entering the brain (21). This kinetics model identified the BBB as another potential compartment in the multi-step transport process of brain glucose delivery and proposed that glucose movements are not as straightforward as passing through the BBB. The main difference between the reversible and standard MM models are that reversible MM has a lower Km, making glucose uptake closer to Vmax.
Gruetter et al. found that at 5% cerebral blood volume, Km (measured as Kt) was 0.06mM in the reversible MM model and -0.09mM in the standard MM model.
A review by Lund-Andersen et al. summarised that the similar glucose concentrations in the cerebral intracellular and extracellular spaces allow them to be described as a single functional compartment with a ‘large dispersion volume’. This dispersion volume may justify the more recent finding by Gruetter et al. that glucose transport within the brain itself is more rapid than across the BBB; the rate limiting step in brain glucose transport was concluded to be the BBB (8).
Many studies have investigated disease effects on the rate of glucose transport. A marked decrease in the rate of glucose transport has also been observed in patients with Alzheimer’s disease (AD). A PET investigation of glucose transport in AD patients found that FDG transport was significantly lower in the temporal cortex compared to the healthy age-matched controls (0.126 vs. 0.161, p<0.05)(22). This may be justified by another study by Kalaria and Harik which used the inhibitory ability of cytochalasin B (CB) to localise hexose transporters in different parts of the brain (23). Compared with controls, AD patients showed significantly lower concentrations of CB binding (pmol/mg protein) in the cerebral microvessels (20.2±3.9 vs. 42.0±3.0, p<0.001), frontal cortex (3.81±0.6 vs. 8.79±1.2, p<0.001), temporal cortex (3.69±0.7 vs. 7.10±0.8, p<0.01) and hippocampus (3.73±0.8 vs. 7.03±0.7, p<0.01). It is important to note that the investigation had the limitation of not being able to differentiate hexose transporters between those that transport glucose and other hexose sugar transporters. Despite this, these results show a significant decrease in the presence of hexose transporter in the BBB, which would explain the lower rate of glucose transport previously found via PET investigation.
Another disease associated with impaired cerebral glucose transport is Huntington’s disease (24). Grade 3 patients showed significantly decreased GLUT 1 (10.2±0.8 vs. 30.1 ±6.4, p=0.03) and GLUT 3 (18±4.8 vs. 70.2±8.5, p=0.001) immunoreactivity on Western blot analysis.
In a study of rats, Pardridge et al. found that diabetic rats had a lower rate of glucose clearance than controls, (175±18 vs. 294±38 p<0.01) (25). Choi et al. studied BBB glucose transporter mRNA and found expression in diabetic rats appeared higher than controls in Northern blot analysis (26). One would expect that the decreased rate of glucose transport would correspond with decreased expression of glucose transporter; in the same study however, increased expression of transporter was also not associated with increased transporter activity. Vannucci et al. also found via in situ hybridisation that GLUT1 mRNA levels in the BBB were higher in the thalamus and hypothalamus for diabetic rats than controls at both 5 and 10 weeks of age, p<0.05 (27). Vannucci et al. also found that the diabetic rats had higher brain/blood plasma glucose ratios at both 5 (0.135±0.003 vs. 0.099±0.004, p<0.05) and 10 (0.125±0.003 vs. 0.099±0.004, p<0.05) weeks compared to controls, however had significantly lower cerebral glucose utilization rates at 10 weeks (µmol/g/min: 58.77±2.55 vs. 77.53±1.53). This data suggests that while there may be greater cerebral glucose uptake in diabetic individuals, it may not be as efficiently metabolised as in healthy individuals.
In conclusion, glucose is transported into the brain through facilitated diffusion which is mainly mediated by GLUT1 and GLUT3 (3,8,9). Though most reports have described cerebral glucose transport using the MM model, the mechanism for ensuring sufficient rate of glucose transport to meet high cerebral metabolic demands remains unclear. Transporter deficiency results in variable symptoms associated with cerebral dysfunction as well as low cerebral glucose concentration (15). Diabetes, Alzheimer’s and Huntington’s disease have also been significantly associated with lower rates of cerebral glucose transport, however studies have not found a correlation between transport rate and transporter expression in diabetes. (3,22,23,25).
Word Count: 1952