Functional expression of SGLTs in rat brain

Abstract

This work provides evidence of previously unrecognized uptake of glucose via sodium-coupled glucose transporters (SGLTs) in specific regions of the brain. The current understanding of functional glucose utilization in brain is largely based on studies using positron emission tomography (PET) with the glucose tracer 2-deoxy-2-[F-18]fluoro-d-glucose (2-FDG). However, 2-FDG is only a good substrate for facilitated-glucose transporters (GLUTs), not for SGLTs. Thus, glucose accumulation measured by 2-FDG omits the role of SGLTs. We designed and synthesized two high-affinity tracers: one, α-methyl-4-[F-18]fluoro-4-deoxy-d-glucopyranoside (Me-4FDG), is a highly specific SGLT substrate and not transported by GLUTs; the other one, 4-[F-18]fluoro-4-deoxy-d-glucose (4-FDG), is transported by both SGLTs and GLUTs and will pass through the blood brain barrier (BBB). In vitro Me-4FDG autoradiography was used to map the distribution of uptake by functional SGLTs in brain slices with a comparable result from in vitro 4-FDG autoradiography. Immunohistochemical assays showed that uptake was consistent with the distribution of SGLT protein. Ex vivo 4-FDG autoradiography showed that SGLTs in these areas are functionally active in the normal in vivo brain. The results establish that SGLTs are a normal part of the physiology of specific areas of the brain, including hippocampus, amygdala, hypothalamus, and cerebral cortices. 4-FDG PET imaging also established that this BBB-permeable SGLT tracer now offers a functional imaging approach in humans to assess regulation of SGLT activity in health and disease.

glucose is the primary metabolic substrate for the brain, and a continuous supply of glucose is required for normal neuronal function (28). Glucose is transported into brain across the blood-brain barrier (BBB) and made available to neurons and glia cells via specific transporters. Glucose transporters are divided in two major groups: the energy-independent facilitated transporters, GLUTs (22), whose isoform GLUT1 is expressed in the BBB and glial cells and GLUT3 in neurons (12); and sodium-coupled glucose transporters, SGLTs, which were first identified in intestine and kidney (37, 38). The SGLTs are known to avidly accumulate not-metabolized glucose analogs in cells due to the presence of inward sodium electrochemical potential gradients across the cell membrane (11). The physiological importance of SGLTs in brain is unknown, and our goals here were to determine whether SGLTs are normally expressed in rat brain and to develop tracers to image SGLT activity in vivo using positron emission tomography (PET).

Various radiolabeled sugar analogs can be envisioned as potential substrates for in vivo assessment of brain SGLT activity using PET imaging. However, a successful SGLT-mediated cerebral molecular imaging probe must be a good substrate not only for neuronal SGLTs but also for GLUT1, the major transporter responsible for d-glucose entry through the BBB. On the basis of earlier work from our laboratories we reasoned that specific SGLT radiolabeled imaging probes, such as α-methyl-4-[F-18]-fluoro-4-deoxy-d-glucopyranoside (Me-4FDG) (11), are not expected to penetrate the BBB efficiently and therefore cannot be used for in vivo assessment of brain SGLT activity, but a second tracer, 4-[F-18]fluoro-4-deoxy-d-glucose (4-FDG, Fig. 1A), is a good candidate. 4-FDG a priori meets a key qualification in that it is an excellent substrate for both SGLT and GLUT transporters (Fig. 1B) (11, 14). Thus 4-FDG will cross the BBB and be presented to neurons expressing SGLTs and accumulate as a result of active transport against the glucose gradient (see Fig. 9). However, since 4-FDG is not metabolized via hexokinase (5) (or other enzymes) there will be no metabolic sink to accumulate this tracer in cells expressing GLUTs.

Fig. 1.

Glucose transporter imaging probes and their transport cascades into cells. A: structure of glucose analog probes. B: graphical interpretation of glucose utilization via glucose transporters (GLUT) and sodium-coupled glucose transporters (SGLT). Glucose (Glc) is transported into cells by both GLUT and SGLT. To discriminate between the various glucose transport pathways, probes were designed to be substrates for either GLUT [2-deoxy-2-[F-18]fluoro-d-glucose (2-FDG)] or SGLT [α-methyl-4-[F-18]fluoro-4-deoxy-d-glucopyranoside (Me-4FDG)], or both [4-[F-18]fluoro-4-deoxy-d-glucose (4-FDG)]. 2-FDG can be phosphorylated by hexokinase (HK) but cannot be further metabolized, resulting in metabolic trapping in cells. Neither 4-FDG nor Me-4FDG is substrate for HK, so their accumulation in cells is driven by the sodium/glucose cotransport (SGLT) mechanism. 6P, 6-phosphate.

Fig. 9.

Graphical interpretation of glucose utilization via SGLT probed by 4-FDG in intact brain in vivo. 4-FDG is reversibly transported from blood into the brain via GLUT1 at the BBB and accumulated in those neurons expressing SGLTs by secondary active transport.

With these two tracers we investigated the functional expression of SGLTs in rat brain using in vitro biochemical uptake assays on brain slices, and their regional functional distribution was demonstrated by two independent imaging approaches—in vitro autoradiography and ex vivo autoradiography. Finally, microPET imaging was used to test the feasibility of using 4-FDG PET to study in human subjects.

MATERIALS AND METHODS

Radiochemical Synthesis

4-FDG and Me-4FDG were synthesized from the corresponding acetylated galactose triflate by radiofluorination with cyclotron-produced [F-18]fluoride, followed by deprotection under acidic and basic conditions, respectively. The probes were synthesized with high radiochemical yields (50–86%) with chemical and radiochemical purities of >97% and a specific radioactivity >2,000 (Ci/mmol) at end of synthesis.

In Vitro Substrate Affinity for Hexokinase

The reaction was initiated by adding 0.1 units of hexokinase (catalog no. 376811, Calbiochem, Gibbstown, NJ) into a tube containing ∼5 μCi of the probe (2-FDG, 4-FDG, or Me-4FDG) with 10 mM ATP, 10 mM MgCl₂, and 160 mM Tris·HCl buffer (pH 7.4) in a total volume of 5 ml. One-milliliter aliquots were taken out after a 60-min incubation. The reaction was stopped by adding 0.5 ml of MeOH, and the solution was kept on ice for HPLC analysis (polar bonded-phase Zorbax NH₂ column, Agilent Technologies, Santa Clara, CA; mobile phase: 55% 100 mM potassium phosphate, pH 7, and 45% MeOH ; flow rate: 1 ml/min). Fractions were collected every 18 s, and total collection time was 9 min. The activity in each fraction was analyzed in a gamma well counter.

Animal Preparation for Molecular Imaging

All animal experiments in this study were performed with authorization from the Chancellor's Animal Research Committee of University of California at Los Angeles (UCLA). Sprague-Dawley female rats (250 to 370 g, 3 to 5 mo old) were purchased from Charles River Laboratories and maintained in a climate-controlled room on a 12-h light-dark cycle with food and water available ad libitum.

In Vivo Metabolism of 2-FDG, 4-FDG, and Me-4FDG

Rats were euthanized by intravenous injection of pentobarbital sodium 1 h after administration of the probe intravenous injection. A blood sample (∼2 ml) was collected and cleared by centrifugation. Supernatant plasma (0.5 ml) was mixed with the same volume of HPLC buffer (55% 100 mM potassium phosphate, pH 7, and 45% methanol) and filtered through a 0.2-μm filter (Millipore). The brain was removed and homogenized in HPLC buffer (∼3 ml), cleared by centrifugation and filtration. The filtered supernatants of plasma and brain were analyzed by HPLC analysis using the conditions established for in vitro metabolite determinations.

Immunohistochemistry

The SGLT1-specific antibody (69F) was produced in rabbit against a synthetic peptide (Peptidomics Core, UCLA/CURE Digestive Diseases Research Center) corresponding to residues 563–575 of hSGLT1 (LRNSKEERIDLDA) and affinity purified using the immunizing peptide coupled to CNBr-Sepharose 4B (Sigma). Antibody specifically bound to the peptide was eluted in NaCl-acetic acid, pH 2.5, and immediately neutralized with Tris base. The ability of the antibody to recognize specifically the SGLT1 protein was verified on exogenously expressed hSGLT1 protein in Xenopus laevis oocytes histological sections using the antigenic peptide as a control.

Cryoprotected (25% sucrose infused) paraformaldehyde perfusion-fixed rat brains were cryosectioned for immunoprobe as free floating slices (∼25 μm). The brain was perfused by 3.5% phosphate-buffered saline (PBS) solution of formaldehyde with a blunt 23-gauge needle placed in the left ventricle. Endogenous peroxidases were inactivated with 0.3% H₂O₂ (30 min), washed in PBS (pH 7.4) and blocked with 5% normal donkey serum in PBS. Slices were exposed to the affinity-purified SGLT1-specific antibody 69F (1:100) in PBS and normal donkey serum overnight. After being washed in PBS, the bound antibody was detected with donkey anti-rabbit secondary antibody using the avidin-biotin complex method (Vector Labs) and visualized with diaminobenzidine. Specificity of the antibody was verified by incubating adjacent slices in antibody preabsorbed with 0.5 (μg/ml) of the immunizing peptide.

Specimens were processed at the Microscopic Techniques Laboratory, Brain Research Institute, UCLA. Images were acquired using the Aperio ScanScope at the Translational Pathology Core Laboratory, Department Pathology and Laboratory Medicine, UCLA.

In Vitro Transport Assays and Autoradiography

Coronal sections of the rat brain (300 μm) were cut with the aid of a Vibratome (series 1000, Ted Pella) (36). Tissue slices were stored in ice-cold oxygenated HEPES buffer A [134 mM NaCl, 2.5 mM KCl, 3 mM CaCl₂, 1 mM MgCl₂, 0.34 mM Na₂HPO₄, 10 mM HEPES, and 10 mM glucose, pH 7.3 (4)] before use. For uptake and autoradiographic experiments, slices were washed free of glucose (replaced with 10 mM mannitol) and incubated with each radioactive probe (10 μCi/ml in 100 μM sugar), in the absence or presence of inhibitors, in oxygenated HEPES buffer B (HEPES buffer A with substitution of mannitol for glucose) at 22°C for 20 min. In all cases the radiolabeled probes (e.g., 100 μM nonradioactive 2-FDG, 4-FDG, or Me-4FDG) were incubated 1) with no inhibitor as control; 2) with 10 μM of phlorizin (Sigma), an inhibitor of SGLTs (25); 3) with 5 μM of cytochalasin B (Sigma), an inhibitor of GLUTs (34); and 4) with both inhibitors, as a representation of nonspecific tissue retention. After incubation, tissue slices were washed by ice-cold HEPES buffer B with cytochalasin B and phlorizin (Sigma), classic inhibitors of GLUTs and SGLTs, respectively, to minimize any further probe transport through GLUTs or SGLTs during tissue processing. Radioactivity in brain slices was counted by a gamma counter or exposed on an image plate at −80°C (Fujifilm BAS-2025). The image plate was read by an image reader at room temperature (Fujifilm BAS-5000) and displayed by Multi Gauge V3.0 software (Fujifilm Medical Systems). To distinguish the uptake from SGLT1 and SGLT2, Me-4FDG uptake was measured as above but in the presence of 25 mM galactose or 25 mM glucose. SGLT1 has the same K_m and maximum rate of transport for glucose and galactose (11), but galactose is not transported by SGLT2 (16).

Ex Vivo Autoradiography

Ex vivo is defined as the in vivo tracer uptake analyzed by in vitro autoradiography. The molecular imaging probe (2-FDG, 4-FDG, or Me-4FDG; 5–8 mCi) was administered to the rat via tail veil injection. After 1 h, animals were euthanized with pentobarbital sodium through intravenous injection. The brain was removed immediately and frozen in 2-methylbutane at −40°C (31). Coronal sections (20 μm thick of the midbrain) were cut on a cryostat (Leica CM3050) at −22°C and exposed on the image plate (Fujifilm BAS-2025) for 90 min at room temperature.

In Vivo MicroPET Scans

A 10-min static scan in a small animal PET scanner (microPET Focus 220; CTI Concorde Microsystems) was initiated at 60 min post intravenous administration of 4-FDG (∼3 mCi). After microPET scan, a 10-min computerized tomography (CT) scan was acquired with a MicroCAT II tomograph (ImTek) for attenuation correction of the microPET images (8). Images were reconstructed using the maximum a posteriori algorithm (7). Images were subsequently displayed using AMIDE (20). Three-dimensional cryosection images were coregistered with the microPET and microCT images to provide anatomical information of the brain substructures (33).

RESULTS

Functional Activity of SGLTs in Brain

SGLT brain functional activity was first tested by in vitro transport assays on 300 μm thick mid-brain slices (36) with a specific SGLT transport tracer, Me-4FDG (Fig. 1A). Me-4FDG was selected on the basis of the SGLT specificity demonstrated by α-methyl-d-glucopyranosides (Table 1). The uptake of 100 μM Me-4FDG in brain slices was abolished by a specific SGLT inhibitor (phlorizin; 10 μM). In contrast, the specific GLUT inhibitor, cytochalasin B (5 μM), did not inhibit Me-4FDG uptakes (Fig. 2A). When 100 μM 4-FDG was used, the uptake was partially inhibited by both phlorizin and cytochalasin B (Fig. 2B), and the effect of each inhibitor was additive in that 4-FDG uptake in the presence of both inhibitors was inhibited ∼100%. The phlorizin-sensitive 4-FDG uptake was approximately equal to the total Me-4FDG uptake (Fig. 2, A and B). Moreover, with uptakes of 100 μM 4-FDG, the phlorizin component and Na-dependent component are comparable, 10 ± 2 nmol·g⁻¹·20 min⁻¹ and 17 ± 2 nmol·g⁻¹·20 min⁻¹ (means ± SE, n = 8), respectively. These results provide the evidence for functional SGLT expression in the brain. As expected, the uptake of 100 μM 2-FDG was not affected by the SGLT inhibitor, phlorizin, but was blocked by the GLUT inhibitor, cytochalasin B (Fig. 2C), confirming that 2-FDG is solely transported by GLUTs.

Table 1.

Kinetics of hexose membrane transport and phosphorylation

	SGLT1			GLUT1			Hexokinase
Hexose	Km, mM	Turnover, s−1	Reference	Km, mM	Turnover, s−1	Reference	Km, mM	Vmax	Reference
2-FDG	>100	UD	11	3.2	1,000–13,000*	3	0.19	0.5	5
4-FDG	0.07	160†	11	NA	1,000–13,000*		84	0.1	5
Me-4FDG	0.1‡	160†		UD§	∼0		UD§	UD§
Me-glc	0.7	160	11, 21, 26	UD	∼0	3	UD	UD	32
d-Glucose	0.5	160	11, 21, 26	6.3	1,000–13,000	3, 6	0.17	1	5

SGLT1, sodium-coupled glucose transporter 1; GLUT1, glucose transporter 1; 2-FDG, 2-deoxy-2-[F-18]fluoro-d-glucose; 4-FDG, 4-[F-18]fluoro-4-deoxy-d-glucose; Me-4FDG, methyl-4-FDG; UD, undetectable; NA, not available.

^*Estimated to be identical to d-glucose at 38°C.

^†Estimated to be identical to α-methyl-d-glucopyranoside (Me-glc) at 38°C.

^‡Mean of two experiments as in Ref. 11.

^§This work.

Fig. 2.

Functional activity of SGLT transporters in brain slices. The transport of the tracer (100 μM Me-4FDG, 4FDG, and 2-FDG) was measured for 20 min at 22°C with or without inhibitor, phlorizin (Pz, an inhibitor of SGLT) and cytochalasin B (Cyto B, an inhibitor of GLUT). A: accumulation of Me-4FDG in brain slices is repressed by phlorizin but not cytochalasin B. B: accumulation of 4-FDG was reduced by both phlorizin and cytochalasin B. C: accumulation of 2-FDG is blocked by cytochalasin B but not phlorizin, showing that 2-FDG is only transported by GLUT and not able to probe glucose utilization via SGLT. Each error bar is ± 1 SE (n = 7). The variance was corrected by error propagation.

Regional Expression of SGLT1 Transporter in Brain

Immunohistochemical staining demonstrated the expression of SGLT1 with the highest levels in the cerebral cortices, hippocampus, amygdala, and hypothalamus (Fig. 3a, left; also see Ref. 2). Regional differences in SGLT1 expression in neurons was observed by comparing cresyl violet staining of neurons (Fig. 3c) and immunostaining of the SGLT1 protein (Fig. 3d). SGLT1 immunoreactivity in the CA1 region of hippocampus was primarily observed on glutamatergic pyramidal neurons located in the stratum pyramidale but not in the stratum radiatum. For these CA1 neurons in hippocampus, SGLT1 immunoreactivity was observed in neuron cell bodies, axons, and dendrites (Fig. 3e). No obvious SGLT1 staining was observed in capillaries forming the BBB identified by SGLT1 immunoreactivity (2) and hematoxylin and eosin staining (A. S. Yu and B. A. Hirayama, unpublished observations). Antibodies for other SGLTs are not generally available, but SGLT2 protein has been observed in regions of the rat brain, including the cerebellum and hippocampus, but not the BBB (H. Koepsell and I. Sabolic, personal communication). The functional expression of SGLT2 was examined in rat brain slices using functional assays. Galactose, a poor SGLT2 substrate (16), does not completely inhibit Me-4FDG uptake into rat brain slices, suggesting that SGLT2 contributes to Me-4FDG uptake (Fig. 4).

Fig. 3.

Distribution of functional SGLT transporters in midbrain. a, left: the location of SGLT1 protein was probed by immunohistochemistry with the use of an SGLT1-specific antibody. Functional SGLT transporters were probed by in vitro autoradiography with Me-4FDG (a, right) and 4-FDG (b). Brain slices were exposed at −80°C. The highlighted regions are cortex (CX), hippocampus (HP), hypothalamus (HTH), thalamus (TH), and amygdala (AM). Comparison of cresyl violet staining of neurons (c) and immunostaining of SGLT1 protein (d) shows SGLT1 expression in pyramidal neurons located in the stratum pyramidale of the CA1 region of hippocampus but not the stratum radiatum (SR). e: SGLT1 protein was detected throughout the cell body and terminals by immunostaining (scale bars are 5 mm for a and b and 100 μm for c–e).

Fig. 4.

Detection of SGLT2 activity in midbrain. Me-4FDG (100 μM) uptake was reduced 95% by 25 mM glucose but only 75% by 25 mM galactose. This suggests that SGLT2 accounts for 20% of the total Me-4FDG uptake activity. Each error bar is ± 1 SE (n = 6).

Regional Brain Distribution of Functional SGLTs

The distribution of functional SGLT isoforms was first evaluated using in vitro autoradiography by incubating fresh 300-μm brain slices with Me-4FDG, the specific probe for SGLT transporters. Me-4FDG accumulation was observed in cerebral cortices, hippocampus, amygdala, and hypothalamus consistent with immunohistochemistry results (Fig. 3a). Similar pattern of accumulation was observed with 4-FDG (Fig. 3b). The accumulation of both tracers in these regions of the brain was sensitive to phlorizin (A. S. Yu and B. A. Hirayama, unpublished observations).

Subsequently, in vivo regional brain glucose uptake via SGLTs and GLUTs was examined using ex vivo autoradiography with 4-FDG and 2-FDG, respectively, in the intact rat brain (Fig. 5). Ex vivo autoradiography allows measurement of in vivo uptake of radioprobes under normal physiological conditions while delivering high image resolution in comparison with microPET imaging. Figure 5 shows that 4-FDG is accumulated in the hippocampus, especially in CA1 region, whereas 2-FDG is not. MicroPET experiments were also conducted to test the feasibility of imaging SGLT activity in vivo (Fig. 6). This demonstrated that 4-FDG, a substrate for GLUT1, crosses the BBB efficiently and accumulates in the hippocampus where SGLT highly expresses. The PET images of the small rat brain are of lower resolution (1.75 mm at the center of the field of view) than the ex vivo autoradiography. As expected from its low affinity for GLUT1 (Table 1), ex vivo autoradiography and microPET studies showed that Me-4FDG did not cross the BBB (Fig. 7).

Fig. 5.

Regional brain distribution of 4-FDG and 2-FDG in normal physiological conditions by ex vivo autoradiography. The glucose pathway via SGLTs and via GLUTs are demonstrated by 4-FDG (A) and by 2-FDG (B), respectively. Atlas: structures of the midbrain, cortex, hippocampus, hypothalamus, thalamus, and amygdala.

Fig. 6.

4-FDG microPET image of SGLT activity in midbrain of rat. 4-FDG was injected intravenously into the rat, and the microPET image was taken 1 h after injection. A and B: MicroPET image (A) is a 0.2-mm slice of rat midbrain, and the corresponding cryosection (B) is provided to define the brain substructure. Note that the estimated resolution of the microPET image at the center of the field is 1.75 mm.

Fig. 7.

Me-4FDG is not able to cross the blood-brain barrier (BBB) and be transported by SGLTs. A and B: ex vivo autoradiographic (A) and microPET (B) images of Me-4FDG. No uptake of Me-4FDG was observed in most regions of the brain, except hypothalamus, where the BBB is known to be more permeable (40).

Probe Metabolism

A critical factor in the quantitative interpretation of radiolabeled substrate uptake in tissues is the metabolic fate of the tracer. This was investigated by measuring 1) in vitro enzymatic phosphorylation by hexokinase; 2) in vivo metabolic fate in rat brain tissue extracts; and 3) in vivo metabolic fate in rat plasma.

Enzymatic phosphorylation.

As expected, hexokinase produced one major hexokinase-mediated product with 2-FDG, 2-FDG-6-phosphate (Fig. 8). In contrast, both 4-FDG and Me-4FDG were not phosphorylated by this enzyme (Fig. 8), which is consistent with the reported poor specificity of hexokinase for 4-fluorosugars (Table 1) (5).

Fig. 8.

Tracer metabolite analysis. Top: in vitro metabolic transformation of 2-FDG, 4-FDG, and Me-4FDG with hexokinase. Hexokinase phosphorylates 2-FDG, but not 4-FDG or Me-4FDG. Middle: for all three probes, the parent molecular imaging probe was recovered intact from plasma with no other metabolites found. Bottom: in vivo metabolism of the probes in rat brain tissue. With 2-FDG, the major metabolite found in brain was 2-FDG-6-phosphate (fractions 24 and 25), whereas for 4-FDG only the parent probe was detected in brain. No uptake of Me-4FDG in brain tissue was observed. The peak at fraction 10 is the parent probe (2-FDG, 4-FDG, or Me-4FDG) as determined using standard samples.

Metabolic fate in rat brain tissue extracts.

2-FDG-6-phosphate was the major product observed in rat brain tissue extracts 1 h following intravenous administration of 2-FDG (Fig. 8) (13). In parallel experiments, 4-FDG was recovered intact from rat brain tissue extracts, with no detectable transformation to its 6-phosphate derivative or any other product. The lack of metabolic conversion of 4-FDG in brain extracts provides additional support for SGLT-mediated accumulation of this probe in neurons. Moreover, no Me-4FDG was detected in brain 60 min after intravenous injection in rats (Fig. 8).

Peripheral metabolism.

For all probes, only the parent molecule (>97.5%) was detected in plasma up to 1 h after intravenous probe administration (Fig. 8).

These experiments show that unlike 2-FDG, neither 4-FDG nor Me-4FDG is metabolized to any detectable extent by hexokinase or by rat tissues.

DISCUSSION

This work provides the evidence that glucose uptake into a number of specific regions in the normal brain occurs via the SGLT pathway in vivo. Our use of 4-FDG as an in vivo SGLT PET imaging probe was validated by transport assays, three different molecular imaging approaches, and immunostaining.

The presence of the specific facilitated transporter (GLUT1) as the sole glucose carrier through the BBB presents a notable challenge for in vivo measurement of brain glucose utilization via SGLT transporters present in neuron membranes (Table 1). For probe design, consideration was given to the fact that equatorial −OH groups of d-glucose at C-2 and C-3 are required for transport via SGLT1, but the presence or absence of −OH groups at C-4 and C-6 are relatively unimportant (11, 15). Molecular requirements for GLUT1 transport are different in that the −OH groups at C-1, C-3, and C-6 are essential for recognition, whereas the −OH group at C-4 is of minor importance (9). Therefore, position C-4 of d-glucose was, a priori, an optimal choice for fluorine-18 labeling for preservation of substrate properties for both GLUT1 at the brain capillaries and neuronal SGLT transporters. Carbon-1 methyl substitution of d-glucose, α-methyl-d-glucopyranoside, results in a poor substrate for GLUT1 (11, 39), with the resulting inability to cross the BBB. Parenthetically, the lack of in vivo Me-4FDG uptake into the brain from blood (Fig. 7), along with the absence of SGLT antibody staining in brain capillaries, establishes the absence of functional SGLT pathway in the BBB under normal physiological conditions. SGLT activity has been detected previously in cell cultures of brain endothelial cells, but it is known that gene expression in isolated cultured cells (23, 35) may be different from that in native cells. Furthermore, in the recent report with an “improved antibody,” SGLT1 was no longer detected in brain capillaries (2). Moreover, since Me-4FDG is the substrate for SGLT1 and SGLT2 and does not penetrate the BBB in vivo in our experiment, it is also possible to discard the presence of SGLT2 in brain capillaries.

Although Me-4FDG is not suitable as a brain SGLT probe for microPET imaging or ex vivo determinations (Fig. 7), its high specificity to SGLTs makes it an excellent probe for measuring the activity and regional distribution of SGLTs using biochemical assays (Fig. 2) and in vitro autoradiography (Fig. 3a, right) and a control for 4-FDG. The phlorizin-sensitive uptake of Me-4FDG into cortices, hippocampus, amygdala, and hypothalamus in brain slices (Fig. 3a) demonstrates that SGLT transporters are not only abundant in these regions (Fig. 3), but are functional. Parallel in vitro autoradiography results obtained with 4-FDG, and confirmed with ex vivo autoradiography, indicate that 4-FDG can also target functional SGLT activity in vitro and in vivo.

There is significant uptake both Me-4FDG and 4-FDG in thalamus as judged by in vitro autoradiography, and this seems higher than expected from the SGLT1 immunohistochemistry (Fig. 3, a and b). This may suggest the presence of other SGLT isoforms, e.g., SGLT2 and SGLT3, whose mRNAs have been earlier demonstrated in thalamus (19). According to the in vitro transport assay on brain slices (Fig. 4), there is SGLT2 activity. The precise distribution of SGLT2 and other SGLTs in rat brain is yet unknown and this needs to be further investigated as antibodies become available. However, the regional distribution of SGLT mRNAs has been reported by the Allen Institute for Brain Sciences (Seattle, WA) (19).

At least five lines of evidence support 4-FDG accumulation via SGLTs in rat brain: 1) the knowledge that SGLT transporters are able to accumulate intracellular glucose to orders of magnitude higher than the extracellular concentration (18); 2) inhibition of a fraction of 4-FDG transport by SGLT-specific phlorizin (Fig. 2B); 3) a fraction of 4-FDG uptake is Na⁺ dependent; 4) demonstration that no enzymatic transformation of 4-FDG occurs indicates that no metabolic sink is responsible for 4-FDG brain accumulation (Fig. 8); and 5) transport of nonmetabolized 4-FDG via GLUTs would not produce probe accumulation above its extracellular level. Therefore, we propose a SGLT-dominant biochemical pathway of 4-FDG accumulation in specific regions of the brain (Fig. 9), where 4-FDG distributes across three spaces: plasma, extracellular space, and neuronal intracellular space. The lack of Me-4FDG permeability in the intact brain indicates that GLUT1 (but not SGLT) in the BBB mediates 4-FDG transport between plasma and extracellular space. 4-FDG accumulation indicates neuronal SGLT mediated transport against a concentration gradient from extracellular to the intracellular space in neurons.

A critical question to be asked is: If the brain has abundant access to circulating glucose via facilitated transporters (e.g., GLUTs), why would it need energetically expensive glucose via SGLT transporters? The role of SGLTs in intestine and kidney is focused on the recovery of glucose for metabolic utilization in the whole body, but in the brain this role may differ. As an example, the exquisite localization of SGLT1 in pyramidal neuronal cell bodies in hippocampus suggests a potential protective role in anoxia and hypoglycemia. SGLT1 pumps glucose into cells, so under conditions of ischemia (hypoglycemia and anoxia), neurons could maintain their membrane potential by scavenging glucose from the extracellular fluid, thus enabling ATP synthesis through glycolysis (1). Another viable interpretation is that SGLT1 in the hippocampus is acting as a glucose sensor similar to that demonstrated for SGLT3 (10). In this case, changes in extracellular glucose concentrations cause changes in the membrane potential, intracellular calcium concentrations (24), and a physiological response.

The hippocampus has particular relevance for being the brain substructure with earliest involvement in Alzheimer's disease and other forms of dementia (27, 29). The early vulnerability of hippocampus neurons in Alzheimer's disease (17), coupled with the high SGLT1 densities in these sensitive neuronal bodies and terminals (Fig. 3e), is indicative of the critical role of SGLTs in maintaining neuronal health. Since 4-FDG probes SGLT expression, its value in detecting neuronal functional changes in hippocampus is predicted, possibly providing very early detection in asymptomatic stages of neuronal damage. It is also expected that 4-FDG would be a powerful probe for investigating SGLT functional expression in other neuropathological conditions such as epilepsy, stroke, brain trauma, and other dementias in human (30). The 4-FDG microPET image of the rat brain (Fig. 6) demonstrates the feasibility of 4-FDG PET to study the human brain in health and disease.

In conclusion, this work advances the critical observation that SGLT transporters are expressed in a functional form in specific regions of the rat brain under normal physiological conditions. Mapping of this distribution and alterations in SGLT expression and function are possible in the intact brain using 4-FDG, a molecular imaging probe with many desirable characteristics for quantitative assessment of brain SGLT activity (e.g., high affinity, selective specificity, and no metabolism). Utilizing both 2-FDG and 4-FDG (or Me-4FDG for peripheral organs) in humans, as an extension of this investigation, would help provide a comprehensive insight into the very basic biochemical mechanism of brain glucose utilization via GLUT and SGLT transporters in health and disease states.

GRANTS

Financial support from the National Institutes of Health (Grants RO1-DK077133 and P01-AG-025831) is acknowledged. J. R. Barrio also gratefully acknowledges the support of the Elizabeth and Thomas Plott Chair Endowment in Gerontology.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.