Glucose transport rates are estimated noninvasively in physiological and pathological states by kinetic imaging using PET. The glucose analog most often used is 18F-labeled 2FDG. Compared with glucose, 2FDG is poorly transported by intestine and kidney. We examined the possible use of 6FDG as a tracer of glucose transport. Lacking a hydroxyl at its 6th position, 6FDG cannot be phosphorylated as 2FDG is. Prior studies have shown that 6FDG competes with glucose for transport in yeast and is actively transported by intestine. Its uptake by muscle has been reported to be unresponsive to insulin, but that study is suspect. We found that insulin stimulated 6FDG uptake 1.6-fold in 3T3-L1 adipocytes and azide stimulated the uptake 3.7-fold in Clone 9 cells. Stimulations of the uptake of 3OMG, commonly used in transport assays, were similar, and the uptakes were inhibited by cyclochalasin B. Glucose transport is by GLUT1 and GLUT4 transporters in 3T3-L1 adipocyte and by the GLUT1 transporter in Clone 9 cells. Cytochalasin B inhibits those transporters. Rats were also imaged in vivo by PET using 618FDG. There was no excretion of 18F into the urinary bladder unless phlorizin, an inhibitor of active renal transport, was also injected. 18F activity in brain, liver, and heart over the time of scanning reached a constant level, in keeping with the 6FDG being distributed in body water. In contrast, 18F from 218FDG was excreted in relatively large amounts into the bladder, and 18F activity rose with time in heart and brain in accord with accumulation of 218FDG-6-P in those organs. We conclude that 6FDG is actively transported by kidney as well as intestine and is insulin responsive. In trace quantity, it appears to be distributed in body water unchanged. These results provide support for its use as a valid tracer of glucose transport.
- positron emission tomography
- renal transport
structural requirements for the active transport of glucose by intestine and for insulin stimulation of glucose transport into muscle were defined many years ago (15, 16, 32, 35). 2-Deoxy-d-glucose (2DG) is responsive to insulin but not transported by intestine. 3-O-methyl-d-glucose (3OMG) and galactose are insulin responsive and actively transported by intestine. Whereas α-methyl-d-glucoside is actively transported, 6-O-methyl-d-glucose and 6-iodo-deoxy-d-galactose are not. Since the sodium/glucose cotransporters are similar in intestinal brush border and kidney proximal tubules (10, 31), the structural requirements for the transport of glucose by intestine and kidney are presumed similar.
Increased attention has recently been directed to quantifying rates of glucose transport into tissues and organs, particularly by positron emission tomography (PET), using glucose analogs as tracers. For example, [18F]2-deoxy-2-fluoro-d-glucose (218FDG) was used in a PET dynamic study to estimate the effect of insulin on the transport of glucose across the blood-brain barrier (12) and on glucose transport into muscle in obesity and diabetes (34). That 2-fluoro-2-deoxy-d-glucose (2FDG) is phosphorylated as well as transported must be taken into account in estimating the transport parameters.
Scanning with 218FDG has also become the preferred diagnostic imaging modality for a variety of human cancers. However, 218FDG is not handled by cells precisely as glucose (10). Thus, accumulation of 18F activity in urine, because of 2FDG's limited transport by renal tubules, can interfere with visualization of pelvic and sometimes abdominal abnormalities (20, 30). For better visualization, PET centers sometimes dilute or remove by catheter bladder contents (21).
Despite the very short half-life of 11C, [11C]3OMG has been used to measure changes in brain glucose metabolism during hypoglycemia (2), [11C]3OMG and [14C]3OMG to examine rates of glucose transport in muscle (6, 24), and [11C]methyl-d-glucoside to trace glucose transport in kidney (7). [123I]6-deoxy-6-iodo-d-glucose (6123IDG) was also recently proposed as a possible tracer (13). It was reported to be responsive to insulin in in vitro preparations, although not in isolated rat heart, and able to detect a defect in glucose transport in diabetic mice (25), although not in diabetic rats (17). Additionally, 6123IDG was excreted in large amounts into the bladder within minutes after injection (17). Since intestinal and renal transport likely have similar structural specificity, and 6-iodo-6-deoxygalactose is not transported by intestine (10, 35), 6123IDG would not be expected to be transported by the proximal tubules.
Studies suggest that 6-fluoro-6-deoxy-d-glucose (6FDG), which has no hydroxyl at its carbon 6, and therefore cannot be phosphorylated, may be transported as is glucose (3–5). Thus, 6FDG is actively transported by intestinal sacs of hamsters (35). It competitively inhibits fermentation of glucose by intact yeast, but not yeast extracts (4). Kinetic studies support 6FDG and 2DG competing with glucose at the same transport site and 6FDG not affecting any metabolic pathway within the yeast cell (5).
We now report studies intended to help assess 6FDG's potential use as a tracer of glucose transport. To determine whether 6FDG is transported via facilitated diffusion pathways, the uptake of 6FDG was compared with that of 3OMG in Clone 9 cells that express glucose transporter 1 (GLUT1) and in 3T3-L1 adipocytes which express GLUT1 and GLUT4 (29). 3OMG, commonly used in glucose transport assays, is transported but not phosphorylated. As a further test of whether 6FDG uptake is mediated by the GLUTs, its uptake with and without cytochalasin B (CB), a specific inhibitor of the GLUTs, was measured in both cell types under basal and stimulated conditions. Furthermore, 618FDG uptake, in the absence and presence of phlorizin, an inhibitor of sodium-dependent glucose transport, and compared with 218FDG uptake, was imaged by PET in rats.
MATERIALS AND METHODS
Radioactive glucose analogs.
[1-3H]3OMG, specific activity 6.9 Ci/mmol, was purchased from NEN Life Science Products (Boston, MA). [1-3H]Glucose, specific activity 20 Ci/mmol, purchased from Moravek Biochem (Brea, CA), was converted on a microscale to [1-3H]6FDG (3). The tosyl group in the intermediate was replaced by fluoride ion using tetrabutyl ammonium fluoride, and then the isopropylidene blocking group was removed by acid hydrolysis. The [1-3H]6FDG was purified by descending paper chromatography using a butanol-acetic acid water system (23). Glucose was well separated from the 6FDG, the mobility of which was established using guide spots of 6FDG purchased from Sigma Chemical (St. Louis, MO). The 6FDG eluted from the paper was further purified by HPLC using a Bio-Rad HPX-87P column (Bio-Rad Laboratories, Hercules, CA) with water at 80°C as solvent, giving a single radioactive peak with a specific activity of 0.5 μCi/μmol. Glucose oxidase oxidized 6FDG at ∼6% the rate of glucose assayed using a YSI 2300 glucose analyzer (Yellow Springs, OH). 618FDG was prepared as we described elsewhere (22).
Clone 9 cell culture.
Clone 9 cells, subcultured in 60-mm culture dishes at a dilution of 1:6 in Dulbeco's Modified Eagle Medium (DMEM) containing 10% calf serum as previously described (18, 29), were utilized between passages 27 and 55. Upon confluence and prior to study, the medium was replaced with medium devoid of serum for 24 h. Sodium azide (5 mM) or diluent were added to the cells 90 min prior to measurement of glucose transport.
3T3-L1 fibroblasts culture and differentiation.
3T3-L1 fibroblasts were also cultured as previously described (26). Briefly, 3T3-L1 fibroblasts were cultured in DMEM (25 mM glucose) with 10% fetal bovine serum (FBS) for 4 days and then cultured in DMEM-FBS containing 1.7 μM insulin, 0.25 μM dexamethasone, and 0.5 mM 3-isobutyl-1-methylxanthine for 4 days to initiate differentiation. The medium was then changed to DMEM-FBS containing insulin alone, cultured for 3 more days and then changed to DMEM-FBS alone for 2–3 days. Under these conditions, >95% of the cell population exhibited the morphological characteristics of adipocytes. For uptake experiments, 3T3-L1 adipocytes were differentiated in 24-well plates, serum starved for 4 h, washed twice, and then incubated for 30 min in glucose-free DMEM. The 3T3-L1 adipocytes were then incubated with or without 10 nM insulin for 20 min.
Measurement of [3H]3OMG and [3H]6FDG uptake.
To measure glucose analog uptake, cells were incubated, as previously described (11), for 90 s in an uptake medium consisting of 0.5–1 ml of glucose-free DMEM containing 5 mM 3OMG, with 1 μCi of [3H]3OMG or 6FDG with 1 μCi [3H]6FDG, in the presence or absence of 50 μM CB. Uptake was terminated by removal of the medium followed by five rapid washes with ice-cold phosphate-buffered saline containing 0.1 mM phloretin. Cells were harvested in 0.5 to 1 ml of water, and their protein content was determined by a modification of the method of Lowry using a test kit manufactured by Bio-Rad Laboratories. Their radioactivity content was determined by scintillation spectrometry. CB-inhibitable 3OMG and 6FDG uptake, assayed in parallel, were calculated as the difference between the mean uptake in each of replicate plates or wells in the absence (−) of CB and presence (+) of CB (11). Uptakes during the 90-s incubations are expressed in counts per minute per milligram of cell protein. The fold stimulation of uptake of 3OMG and 6FDG by insulin and azide was calculated by dividing the uptakes in the presence of insulin (I) and azide (A) by the uptakes in the basal state (B), i.e., [(−CB) − (+CB)]I or A/[(−CB) − (+CB)]B.
Male rats of the Sprague-Dawley strain, weighing 150–300 g, were fasted overnight. A 30-gauge catheter was inserted into a tail vein of the rat. It was anesthetized with 2% isoflurane, taped to a plastic slab, and transported to a small animal PET scanner (Micro-PET R4 Siemens, formerly Concorde; Micro Systems, Knoxville, TN). Anesthesia was maintained at an isoflurane concentration of 1.5%. A three-part attenuation scan was done, with each part covering an 8-cm field of view and 7 min in length, to traverse the entire rat. The three parts were combined to create a whole body attenuation scan. The emission scan protocol that was used was for continuous bed motion at a speed of 16 s/pass; the scan was 60 min in duration. Synchronous with the emission scan (time 0), 618FDG or 218FDG, 0.5–1.0 mCi depending upon rat weight, was injected via catheter. The rat after the scan was returned to its cage to recover from the anesthesia.
Eight rats were scanned with 618FDG, six of which were scanned with, as well as without, phlorizin administration. Four rats were scanned with 218FDG, all with and without phlorizin. Rats were randomly selected to receive phlorizin in their first or second scan. The two scans were performed ≥36 h apart. Phlorizin was injected intraperitoneally in 0.44 ml of saline at a concentration of 0.8 mg/kg 1 h before the transmission scan. During the emission scan the rats were infused intravenously with phlorizin at that same concentration at a rate of 1.5 ml/h. All images were reconstructed by iterative ordered subsets expectation maximization methods (14). Volumes of interest were drawn on the bladder, brain, heart, left and right kidney, and liver for each rat. ASIpro VM microPET Data Analysis software (Concorde Microsystems) was used to quantitatively reconstruct and analyze the images. The tissue time activity curves (TTACs) were decay corrected, normalized by weight and dose, and expressed as 18F activity concentrations in μCi/cc per μCi/g.
Uptakes in the cell incubation are presented as means ± SE. A paired t-test was used to compare differences in uptakes. P values were obtained from the corresponding t values. The TTACs obtained on PET scanning were compared statistically in two ways. The activity concentrations from the last two frames in each scan, the last 10 min of scanning, were averaged, and those averages, taken as a measure of the end activity concentrations, were analyzed for significant differences. The area under the curve (AUC) for each TTAC, over the 60 min of scanning, except for the first 5 min, taken as the measure of the total activity concentrations, was also analyzed for significant differences. The AUC was estimated using the trapezoidal method in Matlab software (Mathworks, Natick, MA). Paired t-tests were used to compare differences between 618FDG and 618FDG with phlorizin and between 218FDG and 218FDG with phlorizin. Between 618FDG and 218FDG and again between 618FDG with phlorizin and 218FDG with phlorizin, an unpaired t-test was used. P values were again obtained from the corresponding t values.
Figure 1 depicts the uptake of 3OMG and 6FDG in 3T3-L1 adipocytes under basal conditions and following 20 min of incubation in the presence of 10 nM insulin in both the absence and presence of CB. Under basal conditions, CB inhibited glucose transported, measured using 3OMG by 75% and using 6FDG by 76%. In the presence of insulin, CB inhibited uptake by 74% using 3OMG and by 78% using 6FDG. The stimulation by insulin of GLUT-mediated glucose transport was measured as 1.5-fold using 3OMG (basal vs. insulin, P = 0.05), not different from the 1.6-fold stimulation measured using 6FDG (basal vs. insulin, P < 0.01).
Figure 2 summarizes the results for the experiments performed with Clone 9 cells under basal conditions and following 90 min of exposure to 5 mM sodium azide. The uptake of 3OMG was inhibited by 58% and 6FDG by 81%, under basal conditions, and 3OMG by 84% and 6FDG by 93% in the presence of azide. The 4.3-fold stimulation of CB-inhibitable glucose transport measured using 3OMG was not different from the 3.7-fold stimulation measured using 6FDG (basal vs. azide, P < 0.001).
Figure 3 presents images from two coronal slices and one sagittal slice, representative of those seen in the scannings. When 218FDG was injected (Fig. 3A) without and (Fig. 3B) with phlorizin, 18F was concentrated in heart, kidney, and bladder, with lesser amounts in brain and liver. Activity was prominent in the heart wall but not heart cavity. When 618FDG was injected (Fig. 3C) without phlorizin, 18F was distributed throughout those organs to a similar extent, except for its absence in bladder. On phlorizin injection, activity was in the bladder while appearing less prominent in brain and liver and more prominent in kidney.
The TTACs on injections of 218FDG and 618FDG without and with phlorizin are shown (Figs. 4 –7) for bladder, brain, heart, kidney (the average of the right and left kidney), and liver. The activity concentrations from which the curves were constructed are the means of the activity concentrations measured in the scans of the four rats given 218FDG without and with phlorizin and the eight rats given 618FDG, six without and with phlorizin. Within 5 min after 218FDG was injected (Figs. 4 and 5), 18F activity spiked in kidney, i.e., there was an increase and then a decline, rapid and then gradual. Within a few minutes, activity appeared in the bladder, dramatically increasing over the remainder of the hour. There was also an early spike of activity in heart and liver. During the rest of the hour, activity increased in brain and heart but not liver. Among those three organs, activity concentration was highest in heart and least in liver.
When 618FDG was injected (Figs. 6 and 7), early spikes in activity occurred in liver, heart, and kidney, as for 218FDG, but they were not followed by increases in activity in any of the organs during the remainder of the hour, and the activity concentrations in them were not different. No 18F was detected in the bladder on injecting 618FDG in the absence of phlorizin, but on phlorizin injection activity rapidly increased in the bladder, as was the case for 218FDG both with and without phlorizin. The end concentration activity of ∼3 μCi/cc per uCi/g in all organs (Fig. 6) is in accord with 618FDG being well distributed in body water by the end of the scan. The mean dose of 618FDG injected was 530 μCi, and the mean weight of the rats was 226 g. Assuming the organs were 70% water, with a body water then of 0.7 × 226 = 158 cc, the activity expected would be 530/158 = 3.2 μCi/cc per μCi/g. In accord with the increase in 18F activity excretion on phlorizin administration, end activity concentrations in brain, heart, and liver were not different, but less (Fig. 7), ∼2 μCi/cc per μCi/g, in agreement with the 18F again being distributable in body water.
In comparing differences between Figs. 4 and 5 with those between Figs. 6 and 7, the effect of phlorizin on 6FDG is larger than that on 2FDG. However, both 218FDG and 618FDG with phlorizin have a greater excretion of the tracer, causing an overall drop of activity in tissues with phlorizin compared to without.
Figs. 8 –11 provide a comparison of the TTACs for brain, heart, liver, and kidney. Where the P values for the difference between two curves are bracketed, the first P value is the comparison between the end concentration activities and the second between the AUCs. In brain (Fig. 8), the end concentration and AUC were significantly greater with 218FDG than with 618FDG (P = 0.04, 0.01). When phlorizin was given, 218FDG activity was greater than 618FDG activity (P = 0.04, 0.01). The activity concentrations injecting 618FDG were less with than without phlorizin (P = 0.03, 0.01). In heart (Fig. 9), concentrations were more when 218FDG than 618FDG was injected both without (P = 0.001, 0.001) and with phlorizin (P = 0.005, 0.02). Again, when 618FDG was injected, concentrations declined when phlorizin was also given (P = 0.003, 0.0004). For liver (Fig. 10), end concentrations were less given 218FDG than 618FDG (P = 0.009). End concentrations giving 618FDG were again greater without than with phlorizin (P = 0.001, 0.002), and that was also the case for 218FDG (P = 0. 03). In kidney (Fig. 11), the AUC was significantly more for 218FDG than 618FDG without (P = 0.02) but not with phlorizin. That is in accord with renal transport of 618FDG being blocked by phlorizin and 218FDG being poorly transported with and without phlorizin. Renal transport of 218FDG appears lessened on phlorizin addition (P = 0.04).
In the absence of azide, the inhibition by CB of 75% of 6FDG uptake in Clone 9 cells is good evidence for that uptake being mediated by GLUT1. Exposure to azide resulting in a severalfold stimulation of CB-inhibitable 6FDG uptake, similar to the result observed using 3OMG as a control, is evidence for 6FDG transport in the cells being via a facilitative GLUT1 transporter. Inhibition by CB of 75% of the basal transport of 6FDG in 3T3-L1 adipocytes, and stimulation of that uptake by insulin, again a result paralleling that observed with 3OMG, is evidence that 6FDG is insulin responsive and provides further support for 6FDG being a valid tracer of glucose transport.
Wick et al. (33) concluded 6FDG was not responsive to insulin, but they concluded this using a fed rabbit preparation in which an insulin effect may have been masked. Thus, they noted the rapid uptake of 6FDG in the absence of added insulin. Drury and Wick (8) reported that 3OMG was also nonresponsive using that preparation, but we found that it was responsive (16), and many investigators have also since found that to be so. The negligible renal excretion of 6FDG, except on administering phlorizin, an inhibitor of renal glucose transport, provides good evidence for the active transport of 6FDG by kidney tubules. That finding is in accord with 6FDG's active transport by intestine and contrasts with findings for 2DG and 2FDG.
The 18F peaks in heart, liver, and kidney, shortly after the administrations of 618FDG and 218FDG, presumably reflect the period when 618FDG and 218FDG blood concentrations were highest in those organs, with the declines occurring as the sugars were distributed in body water. The spikes are most marked in heart, presumably because of the blood volume it contains, and most prolonged in the kidney because of urine formation and excretion. The areas under those peaks during the first 5 min of scanning, then differing among the organs, were excluded from the calculation of the AUC (see Statistical analysis) to obtain a better measure of the difference among the organs in the distribution of the 18F after its entrance into the organ volume. The gradual increase in activity in brain and heart over the remaining hour on 218FDG administration is in accord with a gradual accumulation of 218FDG 6-phosphate in those organs. That is also in accord with 18F concentrating in heart wall compared with cavity when 218FDG, but not 618FDG, was injected. Glucose-6-phosphatase, catalyzing the hydrolysis of the 6-phosphate, presumably prevented accumulation in liver.
The failure of 18F from 618FDG to accumulate during the period following the peaks is in accord with 6FDG not being converted to any significant extent to metabolite(s) retained in those organs but rather being distributed unchanged in body water. In accord with that is the apparent same constant activity in all of the organs (Figs. 6 and 7). The lower activity in the heart, brain, and liver with time, when phlorizin was given along with 618FDG, then reflects the excretion of the 6FDG into the bladder. The lowest retention of 18F in the kidney occurring when 6FDG was given in the absence of phlorizin can be explained by 618FDG then being reabsorbed by the proximal tubules and reentering the systemic circulation. The blocking of that reabsorption then explains the greater concentration of 18F in kidney volume when phlorizin was given. The overall effect of phlorizin on 618FDG, due to being well transported by kidney, is greater than that on 218FDG. Moran et al. (21) provided evidence that, whereas 2DG is not actively transported by kidney, 2FDG is to some extent. That is in accord with the end concentration in kidney suggesting less transport when 218FDG was injected with than without phlorizin (P = 0.04).
Although 6FDG cannot be phosphorylated at its 6th position, there are reports suggesting that 6FDG can be metabolized in some manner, at least when given in substrate quantity. Serif and Wick (28) reported that inhibition of glucose utilization by kidney slices approached 100% at high concentrations of 2DG but only 40% at high concentrations of 6FDG. They suggested that 6FDG might inhibit a specific pathway of glucose oxidation or transport. Bessell et al. (1) reported that 6FDG inhibited the growth of a transplanted lymphoma in mice. 6-Deoxy-, 6-chloro-, and 6-bromodeoxyglucose and the corresponding galactose derivatives were without effect. The LD50 of 6FDG was ∼200 mg/kg and of 2FDG ∼600 mg/kg. A single dose of 6FDG of 500 mg/kg ip in rats produced degeneration and necrosis of centrilobular hepatocytes in 24–48 h with restitution to normal structure in 6 days. After injecting 1 g/kg, 6FDG's concentration in blood at 30 min was three times that of glucose.
Competitive inhibition of glucose transport was postulated to explain the cytotoxic effect of 6FDG, but a large dose of glucose before giving 6FDG did not alter the toxicity (1). A toxic metabolite of the 6FDG was considered a possible explanation, but 6FDG was then known only to be a substrate for glucose dehydrogenase and at a very high Km (19). A highly toxic compound, such as fluoroacetate, formed by some unknown metabolic pathway, was considered unlikely. A single dose of 6FDG (120 mg/kg) inhibited spermatogenesis in rats (9). The rats became infertile, but most recovered. The testes hypotrophied, and tubules were damaged. Doses to 960 mg/kg did not affect fertility in female mice. Glycolysis in rat seminiferous tubules was inhibited by incubation with 10 mM 6FDG. 6FDG is reported to be a substrate for aldose reductase (27) as well as glucose dehydrogenase (19). The polyol formed on reduction is a substrate for sorbitol dehydrogrenase (27). However, the Kms are much higher for the reductase, as well as the dehydrogenase, than the concentrations of 6FDG experienced in our administrations.
Developing a tracer that solely follows glucose transport could create a significant opportunity in radiopharmaceutical research (10). The dose 618FDG needed in adult humans for PET may be expected to be in a range used for 218FDG, i.e., between 5 and 10 mCi. With a specific activity of 200–2,000 mCi/μmol in its preparation (22), the amount of 6FDG injected, including carrier, would be at most only about 10 μg, also many magnitudes less than what LD50 reports in mice of 200 mg/kg. A sugar whose kinetics trace solely glucose transport could aid in the understanding of the separate contributions of glucose transport, glucose phosphorylation, and the downstream pathways of phosphorylated glucose utilization in physiological and pathological states. 6FDG may fulfill that need, reflecting glucose transport not only in insulin-responsive tissues but also in nonresponsive tissues.
This study was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants DK-14507 to B. R. Landau and DK-61994 to F. Ismail-Beigi.
The guidance of Dr. Paul K. Jones, Case Department of Epidemiology and Biostatistics, in the statistical analysis is much appreciated. We also thank Pravesh Asthana, Yu-Hua Fang, and Cristian Salinas, Case Department of Biomedical Engineering, for their help during the PET experiments.
↵† Deceased 24 March 2007.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 by American Physiological Society