Fructose metabolism in the adult mouse opticnerve, a central white matter tract

Paul J Meakin1, Maxine J Fowler1, Alex J Rathbone1, Lynne M Allen1, Bruce R Ransom2,David E Ray1 and Angus M Brown1,2

1MRC Applied Neuroscience Group, School of Biomedical Sciences, Queens Medical Centre, University ofNottingham, Nottingham, UK; 2Department of Neurology, University of Washington School of Medicine,Seattle, Washington, USA

Our recent report that fructose supported the metabolism of some, but not all axons, in the adultmouse optic nerve prompted us to investigate in detail fructose metabolism in this tissue, a typicalcentral white matter tract, as these data imply efficient fructose metabolism in the central nervoussystem (CNS). In artificial cerebrospinal fluid containing 10 mmol/L glucose or 20 mmol/L fructose,the stimulus-evoked compound action potential (CAP) recorded from the optic nerve consistedof three stable peaks. Replacing 10 mmol/L glucose with 10 mmol/L fructose, however, causeddelayed loss of the 1st CAP peak (the 2nd and 3rd CAP peaks were unaffected). Glycogen-derivedmetabolic substrate(s) temporarily sustained the 1st CAP peak in 10 mmol/L fructose, as depletionof tissue glycogen by a prior period of aglycaemia or high-frequency CAP discharge renderedfructose incapable of supporting the 1st CAP peak. Enzyme assays showed the presence of bothhexokinase and fructokinase (both of which can phosphorylate fructose) in the optic nerve. Incontrast, only hexokinase was expressed in cerebral cortex. Hexokinase in optic nerve had lowaffinity and low capacity with fructose as substrate, whereas fructokinase displayed high affinity andhigh capacity for fructose. These findings suggest an explanation for the curious fact that the fastconducting axons comprising the 1st peak of the CAP are not supported in 10 mmol/L fructosemedium; these axons probably do not express fructokinase, a requirement for efficient fructosemetabolism.Journal of Cerebral Blood Flow & Metabolism (2007) 27, 86–99. doi:10.1038/sj.jcbfm.9600322; published online 3 May 2006

Keywords: compound action potential; fructokinase; fructose; glucose; hexokinase; CNS

Introduction

Although it is widely agreed that glucose is the mainenergy substrate of the adult mammalian brain underresting conditions (McIlwain and Bachelard, 1985), ithas been known for over half a century, however, thatin vitro brain preparations can survive on nonglucosesubstrates in the total absence of exogenously appliedglucose (McIlwain, 1953). This implies that it is thelimiting permeability of the blood–brain barrier,rather than the brain’s inability to metabolise sub-strates, that restricts the metabolic substrates the brainuses. It is clear, therefore, that glucose-derived energysubstrates generated within the brain parenchyma, orblood–brain barrier-permeable substrates present in

the systemic circulation in sufficient quantities, couldcontribute to fuel brain function.

One of the first nonglucose substrates shown tosupport the in vitro brain slice was the monosac-charide fructose (Bernheim and Bernheim, 1941;Klein, 1944; McIlwain, 1953). Fructose is absorbedfrom the small intestine (Blakemore et al, 1995) viathe specialised glucose transporter, GLUT5 (Rand etal, 1993), and the majority cleared by the liver viafructokinase (Van den Berghe, 1994), although thekidney also plays a significant role in fructosemetabolism (Mayes, 1993), hence circulating fruc-tose levels are low, 50 to 100 mmol/L, although theycan reach up to 0.5 mmol/L (Darakhshan et al, 1998).Fructokinase phosphorylates fructose to fructose-1-phosphate, which is converted by aldolase B todihydroxyacetone and glyceraldehyde. The glycer-aldehyde is metabolised by triose isomerase toglyceraldehyde-3-phosphate, which then enters theglycolytic pathway (Michal, 1999).

In order for metabolism of residual fructose inbrain to occur, fructose must cross the blood brain–barrier and be taken up into brain cells, which in

Received 1 December 2005; revised and accepted 27 March 2006;published online 3 May 2006

Correspondence: Dr AM Brown, MRC Applied NeuroscienceGroup, School of Biomedical Sciences, Queens Medical Centre,University of Nottingham, Nottingham NG7 2UH, UK.E-mail: ambrown@nottingham.ac.uk

This work was supported by the MRC.

Journal of Cerebral Blood Flow & Metabolism (2007) 27, 86–99& 2007 ISCBFM All rights reserved 0271-678X/07 $30.00

www.jcbfm.com

turn must efficiently metabolise fructose. Systemi-cally administered fructose did not restore corticalactivity in rabbits made hypoglycaemic by hepatec-tomy (Maddock et al, 1939), data supported byfindings that there was minimal detection of radio-labelled fructose uptake into the brain (Oldendorf,1971). This view of negligible fructose uptake intothe brain is being reassessed. Because the GLUT5transporter is found in tissues that metabolisefructose (Burant et al, 1992), its presence in theblood–brain barrier (Mantych et al, 1993) suggestsfructose entry into the brain. For cells to metabolisefructose, they must be capable of taking up fructose,that is, the cells must possess the GLUT5 transporter.Until recently, GLUT5 was detected only in micro-glia in the central nervous system (CNS), its functionunknown (Payne et al, 1997; Sasaki et al, 2004), butstrengthening the case for the presence of fructosein the brain. Recently, the GLUT5 transporter hasbeen shown in cerebellar Purkinje and granule cellsin human fetus (Nualart et al, 1999) and Purkinjecells in mouse cerebellum (Funari et al, 2005). Thefunction is not yet known but their presence doesimply active fructose metabolism in this tissue.

In vitro cortical and hippocampal brain slicescan metabolise fructose (Bernheim and Bernheim,1941; Klein, 1944; McIlwain, 1953), although notas efficiently as glucose. In the cerebral cortex,fructose shows an equivalent oxidation to 58% ofglucose (Weber et al, 2001), and in hippocampalslice 10 mmol/L fructose can support the energylevels (ATP and phosphocreatine), but results in adecrease in the population spike (Bachelard et al,1984; Saitoh et al, 1994; Wada et al, 1998; Yamane etal, 2000). Increasing the fructose concentration to20 mmol/L, both fully supports the energy levelsand the population spike (Bachelard et al, 1984).

In this present paper, we have investigated theability of fructose to support the stimulus evoked CAPin the adult mouse optic nerve (MON). The differentialresponse of the peaks contributing to the CAP impliesa significant difference in the metabolic properties ofneighbouring axons. The ability of the 1st CAP peak tosurvive in 20 mmol/L, but not 10 mmol/L fructose,implies that the ability of fructose to be efficientlymetabolised is not because of deficiencies in fructosetransport, but rather because of selective expression offructokinase in axons, a particularly appealing propo-sition as fructokinase activity has not previously beenshown in central tissue.

Materials and methods

All procedures were performed in accordance with theAnimals (Scientific Procedures) Act, 1986, under appro-priate authority of project and personal licences.

Electrophysiology

Adult male CD-1 mice (30 to 35 g) were obtained fromCharles Rivers, UK. The mice were killed by cervical

dislocation and then decapitated. Optic nerves weredissected free and cut at the optic chiasm and behindthe orbit. The optic nerves were gently freed from theirdural sheaths and placed in an interface perfusionchamber (Medical Systems Corp., Greenvale, NY, USA)and maintained at 371C and superfused with artificialcerebrospinal fluid (aCSF) containing (in mmol/L): NaCl126, KCl 3.0, CaCl2 2.0, MgCl2 2.0, NaH2PO4 1.2, NaHCO3

26, and glucose 10. In some experiments, fructose wassubstituted for glucose, and appropriate osmotic compen-sations made when required. The chamber was continu-ously aerated by a humidified gas mixture of 95% O2/5%CO2. After dissection, optic nerves were allowed toequilibrate in standard aCSF for about 30 mins beforebeginning an experiment. Suction electrodes back-filledwith the appropriate aCSF were used for stimulation andrecording. One electrode was attached to the rostral end ofthe nerve for stimulation and the second suction electrodewas attached to the proximal end of the nerve to record thecompound action potential (CAP), thus all recordingswere orthodromic. Stimulus pulse (30ms duration)strength (Grass S88 dual output square pulse stimulatorin combination with an SIU5 RF stimulus isolation unit;Grass, Astro-Med Inc., Slough, UK) was adjusted to evokethe maximum CAP possible and then increased another25% (i.e., supramaximal stimulation). During an experi-ment, the supramaximal CAP was elicited every 10 secs.The recording electrode was connected to an Axoprobe 1Aamplifier, whose conditioned output (� 10) was amplified100� (Tektronix D13 Dual Beam Conditioning Oscillo-scope with 5A18N Dual Trace Amplifier), filtered at30 kHz and acquired at 20 kHz.

Data Analysis and Curve Fitting

Optic nerve axon function was monitored quantitatively asthe area under the CAP, which represents the best measureof the number of active axons as currents generated byindividual axons within a fibre tract are considered to sum(Cummins et al, 1979; Stys et al, 1991). Data were acquiredonline (Digidata 1200A; Axon Instruments) using proprie-tary software (Axotape; Axon Instruments, MolecularDevices, Union City, CA, USA). The three individualpeaks of the CAP were described by individual Gaussianfits of the form (Clampfit 8.0; Axon Instruments):

y ¼ A

wffiffiffiffiffiffiffiffip=2

p e�2ðtime�cÞ2

w2 ð1Þ

where A is the peak area, w is the width of peak at half-maximum amplitude, and c the latency to maximumamplitude of the peak as described previously (Allen et al,2006). Data are presented as means and standard deviation.Significance was determined using Student’s t-test, whereP < 0.05 was taken to indicate statistical significance.

Enzyme Assays

Tissue for enzyme assay was removed and immediatelysnap frozen in liquid nitrogen. Tissue was weighed anddiluted to a concentration of 1:100 (brain) or 1:10 (MON)in the appropriate buffer.

Fructose metabolism in MONPJ Meakin et al

87

Journal of Cerebral Blood Flow & Metabolism (2007) 27, 86–99

Hexokinase assay: The hexokinase assay has previouslybeen described in detail (Bachelard, 1989). Briefly, thetypical assay contained 39 mmol/L triethanolamine buffer(pH 7.6), 7.8 mmol/L MgCl2, 1.1 mmol/L bNADP (nicoti-namide adenine dinucleotide phosphate), 0.74 mmol/LATP, 2.5 U of glucose-6-phosphate dehydrogenase plussample, and varying concentrations of glucose. Absor-bance was measured at 340 nm at 301C.

Fructokinase assay: Tissue was removed from theanimal and snap frozen in liquid nitrogen and stored at�801C. Mouse optic nerves were diluted 1:10 in 50 mmol/L triethanolamine buffer (pH 7.6), whereas brain and liversamples were diluted 1:100. Sample homogenate (5mL) (intriplicate) was added to a 96-well plate (total volume260ml) containing 100mL of 250 mmol/L 4-(2-hydro-xyethyl)-1-piperazineethanesulphonic acid (HEPES, pH7.1), 10 mL of 250 mmol/L KCl, 10 mL of 10 mmol/L Nafluoride (to inhibit phosphoenolpyruvate conversion to 2-phosphoglycerate), 10 mL of mmol/L K2PO4, 10 mL of0.25 mmol/L phosphoenolpyruvate, 20 mL of 0.15 mmol/Lnicotinamide adenine dinucleotide (reduced form)NADH, 75 mL fructose (0.25 to 100 mmol/L), 5 mL lactatedehydrogenase (B5 U, EC 1.1.1.27), and 5 mL pyruvatekinase (B1 U, EC 2.7.1.40). The plate was preincubated at371C for 30 mins, and then the reaction was initiated bythe addition of 10mL of ice-cold 2.5 mmol/L MgATP. Theassay was recorded over 30 mins at 371C and the reactionsequence is displayed below. Enzyme activities wereexpressed as nmol/min per mg protein using an extinctioncoefficient (NADH) of 6.22 mmol/L per cm (Hagopian et al,2005). Disappearance of absorption was measured at340 nm at 301C. The protein content was estimated usingthe Bio-Rad DC assay (based on methods of Lowry).Control experiments were performed in which substrateswere omitted. In these cases, there was no measuredactivity. In the case where the enzymes were omitted, thebackground reading (less than 5% of the test reading) wassubtracted from test experiments to control for ATPases.

Fructose

ATP Fructokinase (in sample)

Fructose-1-phosphate

ADP

Phosphoenolpyruvate

Pyruvate kinase

Pyruvate

NADH Lactate dehydrogenase

ATP

NAD Lactate

Estimates of Km and Vmax were made using best-fittingregression of Michaelis–Menton kinetics in the form:

V0 ¼ Vmax�½S�Km þ ½S� ð2Þ

where V0 is enzyme activity, Km is the Michaelis constantdefined as the substrate concentration at which enzymevelocity is half-maximal, Vmax is the maximal reaction rate,and [S] is the substrate concentration.

Western Blotting

Cytosolic fractions were prepared: 10% (w/v) whole-cell(WC) homogenate was centrifuged at 200,000g for 2 hours at41C, the supernatant was taken as the cytosolic fraction,and the samples were spun through a 10 kDa filter(Millipore, Hatters Lane, Watford, UK). To concentrate theproteins in the fraction, proteins were precipitated usingice-cold acetone:ether (2:1) at 5� sample volume. Theywere then spun at 5,000 r.p.m. on a bench top centrifuge(Eppendorf 5415c) for 5 mins at room temperature. Thesupernatant was discarded and the pellet washed 3� usingice-cold ether: Industrial methylated spirits (IMS):H2O(10:7:1) at 5� original sample volume. This was achievedby resuspending the pellet in the wash buffer, spinning at5,000 r.p.m. for 5 mins at room temperature and discardingthe supernatant. After the final wash, the remaining pelletwas air dried for > 15 mins in a fumehood to let anyremaining wash buffer evaporate. Samples prepared byprotein precipitation were resuspended in 10 mmol/L Tris-HCl (pH 8.0), and the protein content estimated using theBio-Rad DC assay (based on methods of Lowry). Sampleswere diluted with sodium dodecyl sulphate (SDS) samplebuffer (25 mmol/L Tris, 25 mmol/L dithiothretiol (DTT), 2%(w/v) SDS, 10% (v/v) glycerol, and 0.01% (w/v) bromo-phenol blue) to give protein content of 40mg/lane in a 20mLsample for liver, kidney, and brain, but 150mg/lane for opticnerve. The samples were boiled for 5 mins. Samples werethen run on a 5% to 20% SDS polyacrylamide gel for2 hours at 40 mA. Proteins were then electrotransferredonto a nitrocellulose membrane (Amersham, AmershamPlace, Buckinghamshire, UK), using standard blottingbuffer (25 mmol/L Tris, 200 mmol/L glycine in 20% (v/v)methanol), run at 40 mA overnight. The blot was blockedusing 4% Marvel in 1� TBS (Tris-buffered saline-10 mmol/L Tris and 150 mmol/L NaCl, pH 7.5) for 1 hourat room temperature, and then incubated with the primaryantibody (anti-fructokinase immunoglobulin (Ig g, 1:5,500dilution; Genway Biotech Inc., San Diego, CA, USA) for2 hours at room temperature in 4% Marvel in 1�TBS. Itwas then washed with TBS containing 0.005% Tween-20,then incubated with the secondary antibody (rabbit poly-clonal to chicken, 1:1,000 dilution; Abcam, Milton Road,Cambridge, UK) for 1.5 hours at room temperature in 4%Marvel in 1�TBS. The membrane was developed using anECL detection kit (Amersham).

Immunohistochemistry

Sections (10mm) were cut on a cryostat, thaw-mountedonto superfrost slides (Fischer Scientific, Bishop Meadow

Fructose metabolism in MONPJ Meakin et al

88

Journal of Cerebral Blood Flow & Metabolism (2007) 27, 86–99

Road, Loughborough, UK), and stored at �801C. Sectionswere then warmed at room temperature for 15 to 30 minsbefore being processed for immunohistochemistry. Tissuesections were fixed in 100% ethanol for 15 mins, floodedwith phosphate-buffered saline (PBS; phosphate-bufferedsaline tablets BR00146; Oxoid Ltd, Basingstoke, Hamp-shire, UK), and then rinsed in PBS twice, for 5 mins eachtime, then rinsed three times for 3 mins in Buffer 1 (PBS +1% bovine serum albumin + 0.002% Tween-20). Sectionswere then blocked in Buffer 1 plus normal goat serum(NGS) and normal rabbit serum (NRS) at room temperaturefor 40 mins. The tissue sections were then incubated inappropriate antibody diluted in Buffer 1 at room tempera-ture for 1.5 hours. Primary antibodies: anti-ketohexokinase(fructokinase) 1:200 dilution chicken IgY (Genway BiotechInc., San Diego, CA, USA); anti-GFAP (glial fibrillaryacidic protein; SIGMA G3893) 1:1000 mouse ascites(Sigma-Aldrich Company Ltd, Poole, Dorset, UK); anti-neurofilament (SIGMA N0142) 1:400 mouse monoclonalClone No. N52; and anti-tubulin (SIGMA T5168) 1:2,000mouse ascites Clone No. 8-5-1-2. The tissue sections wererinsed for three times 3 mins in Buffer 1. Secondaryantibodies: Rabbit polyclonal to chicken IgY heavy andlight (fluorescein isothiocyanate-labelled 488 nm, Abcamab6749) diluted 1:500. Goat anti-mouse IgG (tetramethylrhodamine isothiocyanate labelled 568 nm, MolecularProbes A-11031) diluted 1:500 incubated for 1 hour atroom temperature. The tissue sections were then rinsedthree times with Buffer 1 and two times in PBS followedby a 15 mins rinse in PBS, coverslipped with CitifluorAF1, sealed with nail polish, and viewed with a Leica TCSSP2SE confocal microscope running Leica confocal soft-ware. Control sections were incubated in NRS and NGS, orthe primary and secondary antibody was omitted, and nospecific staining was observed.

Results

Effect of Fructose on the Compound Action Potential

Supramaximal stimulus of the MON evoked a triplepeaked CAP typical of the rodent preparation(Figure 1B), which could be accurately fit by thesum of three Gaussian peaks (Allen et al, 2006), andwas stable of over several hours when perfused withaCSF containing 10 mmol/L glucose, as describedpreviously (Figure 1A; Brown et al, 2003). Theindividual peaks of the CAP recorded at 60 and180 mins in the presence of 10 mmol/L glucose aCSFshowed no significant differences (1st CAP peak0.9470.18 versus 1.0470.25 mV ms, 2nd CAP peak2.0870.57 versus 2.3270.53 mV ms, and 3rd CAPpeak 1.9870.77 versus 1.9470.65 mV ms, n = 6;Figure 2A). Replacing 10 mmol/L glucose with20 mmol/L fructose had no significant effect on theCAP, and a stable CAP was maintained for 2 hours inthe presence of 20 mmol/L fructose (Figures 1C and1D). The individual peak areas of the CAP recordedat 60 and 180 mins (after 2 hours exposure to20 mmol/L fructose) were: 1st CAP peak 0.7170.10

versus 0.7870.21 mV ms, 2nd CAP peak 1.7370.31versus 1.8270.47 mV ms, and 3rd CAP peak 1.0670.36 versus 1.0970.37 mV ms, n = 4 (Figure 2B).Substitution of 10 mmol/L glucose with 10 mmol/Lfructose, however, resulted in a delayed decrease inthe CAP area (Allen et al, 2006), although even after2 hours in the presence of 10 mmol/L fructose thetotal CAP area was 84.8%728.4% of the baselinevalue (Figures 1E and 1F). Closer investigationrevealed that the 1st CAP peak fell after about30 mins and continued to decrease from that pointuntil it was completely lost, whereas the 2nd and3rd CAP peaks were not significantly affected by theaddition of 10 mmol/L fructose (Figure 1F) asreported previously (Allen et al, 2006). The indivi-dual peaks of the CAP recorded at 60 and 180 mins(after 2 hours exposure to 10 mmol/L fructose), were:1st CAP peak 0.6270.19 versus 0 mV ms, 2nd CAPpeak 1.6270.60 versus 1.6170.49 mV ms, and 3rdCAP peak 1.0470.38 versus 0.9670.46 mV ms, n = 7(Figure 2C). Replacing 10 mmol/L glucose with5 mmol/L fructose resulted in a more rapid loss ofthe 1st CAP peak, combined with a small butinsignificant decrease in the 2nd and 3rd CAP peaks(Figures 1G and 1H): 1st CAP peak 0.7570.14 versus0 mV ms, 2nd CAP peak 1.9070.34 versus 1.5570.40 mV ms (P = 0.317, NS), and 3rd CAP peak1.6170.54 versus 1.6770.43 mV ms, n = 4 (Figure2D). Thus, in the presence of aCSF containing 5 to10 mmol/L fructose, the 2nd and 3rd CAP peaks canbe supported for extended periods of time, but the1st CAP peak is rapidly lost.

Fructose Cannot Support the 1st Compound ActionPotential Peak in the Absence of Glycogen

These data suggest that an endogenous compoundwithin the MON is sustaining the 1st CAP peakduring the initial 30 mins in 10 mmol/L fructose. Aprime candidate for this compound is glycogen,which is located exclusively in astrocytes (Cataldoand Broadwell, 1986), as we have previously shownthat glycogen can support MON function duringboth hypo- and aglycaemia (Brown et al, 2003). Theglycogen is metabolised during aglycaemia to L-lactate, which is subsequently shuttled out of theastrocyte and taken up by the axon, where it ismetabolised (Baltan Tekkok et al, 2005; Brown et al,2003). However, once the glycogen levels aredepleted during aglycaemia, the CAP fails, thus anindication of exhaustion of glycogen content is theonset of CAP failure (Brown et al, 2003). We exposedthe MONs to glucose-free aCSF for 25 mins andduring this time the CAP was maintained for 20.673.8 mins (n = 5), but it then fell from this point tozero (Figure 3A). Subsequent addition of 10 mmol/Lglucose aCSF resulted in almost complete recoveryof CAP function to 95.9%74.4% of baseline levels(Figure 3A). The 1st CAP peak recovered to 1.8570.42 versus 2.1070.40 mV ms (NS), the 2nd CAP

Fructose metabolism in MONPJ Meakin et al

89

Journal of Cerebral Blood Flow & Metabolism (2007) 27, 86–99

peak recovered to 4.6871.22 versus 5.1871.01 mV ms (NS), and the 3rd CAP peak recoveredto 1.5471.13 versus 2.3070.75 mV ms (NS, n = 5;Figure 3B). Recovery in 20 mmol/L fructose after a25-min period of aglycaemia produced results thatwere qualitatively identical to those in 10 mmol/Lglucose (Figure 3C). The 1st CAP peak recovered to0.8570.24 versus 0.7470.22 mV ms (NS), the 2ndCAP peak recovered to 2.4970.85 versus 2.087

0.81 mV ms (NS), and the 3rd CAP peak recovered to1.3070.60 versus 1.1970.66 mV ms (NS, n = 5;Figure 3D). However, in 10 mmol/L fructose com-pared with 10 mmol/L glucose or 20 mmol/L fruc-tose, the CAP recovery was slower and lesscomplete, 78.3%78.9% of control (n = 5, Figure3E), significantly less than the recovery in10 mmol/L glucose (P < 0.005). This was because ofthe total loss of the 1st CAP peak (0.7470.13 versus

0

0.5

1

1.510 mM fructose10 mM gluc

0

0.5

1

1.510 mM glucose

A

C

B

E

0

0.5

1

1.510 mM gluc 20 mM fructose

0

0.5

1

1.5

0 30 60 90 120 150 180Time (mins)

0 30 60 90 120 150 180Time (mins)

0 30 60 90 120 150 180Time (mins)

0 30 60 90 120 150 180Time (mins)

Nor

mal

ized

CA

P a

rea

Nor

mal

ized

CA

P a

rea

Nor

mal

ized

CA

P a

rea

Nor

mal

ized

CA

P a

rea

10 mM gluc 5 mM fructose

F

G

D

H

Figure 1 Effect of various concentrations of fructose on the compound action potential (CAP). (A) In the presence of 10 mmol/Lglucose, the CAP was stable for 3 hours. Compound action potentials sampled every 5 mins. (B) Representative CAPs from the timepoints indicated by the arrows in (A), plotted in sequential order with the CAPs recorded earliest towards the rear of the stack. Scalebars are 1 ms (applies to D, F, and H) and 2 mV. (C) In 20 mmol/L fructose, the CAP is stable and maintained for 2 hours.(D) Representative CAPs as described in (B). Scale bar 2 mV. (E) In the presence of 10 mmol/L fructose, the total CAP area fell afterabout 30 mins. (F) Representative CAPs illustrating the progressive decrease and ultimate loss of the 1st CAP peak. Scale bar 2 mV.(G) In 5 mmol/L fructose, the total CAP area fell more rapidly than in 10 mmol/L fructose. (H) Representative CAPs illustrating therapid loss of the 1st CAP peak, but maintenance of the 2nd and third CAP peaks. Scale bar 2 mV.

Fructose metabolism in MONPJ Meakin et al

90

Journal of Cerebral Blood Flow & Metabolism (2007) 27, 86–99

0 mV ms), but the 2nd (2.1470.29 versus 1.8170.40 mV ms, NS) and 3rd CAP peaks (1.0770.27versus 0.6870.33 mV ms, NS) were not significantlydifferent from the baseline levels (n = 5; Figure 3F).

These data indicate that once glycogen levels aredepleted, 10 mmol/L fructose cannot support func-tion in the axons contributing to the 1st CAP peak.

Increased Energy Demand Reduces the Ability ofFructose to Support the Compound Action Potential

To further test the role of glycogen in supporting the1st CAP peak in the presence of 10 mmol/L fructose,we increased the metabolic demand on the tissue,which we have previously shown to significantlydeplete glycogen content (Brown et al, 2003). A 15-min period of 100 Hz stimulation was imposed onthe nerve followed by a recovery period in either10 mmol/L glucose or 10 mmol/L fructose. In MONsbathed in 10 mmol/L glucose 100 Hz stimuli did nothave a significant effect on the total CAP area, whichwas maintained at baseline levels for the duration ofthe stimulus. During the recovery period in10 mmol/L glucose, the total CAP area was main-tained at baseline levels (Figures 4A and 4B),indicating that no irreversible injury had occurred.This was borne out when comparisons were made ofthe CAP peaks before and after the stimulus; therewas no significant effect on the peak areas (1st CAPpeak 0.9570.19 versus 0.7970.24 mV ms, NS; 2ndCAP peak 2.1870.57 versus 2.5370.63 mV ms, NS;3rd CAP peak 1.7770.40 versus 1.5770.39 mV ms,NS, n = 5; Figure 4C). However, when 10 mmol/Lglucose was replaced with 10 mmol/L fructose20 mins before the stimulus, the total CAP fellsignificantly once the stimulus was imposed (Figure4D) and recovery was only partial as the 1st CAPpeak area was significantly reduced from 0.9370.33to 0.1170.21 mV ms (P < 0.005). The 2nd (2.5270.88versus 2.3570.49 mV ms) and 3rd CAP peaks (1.5070.52 versus 1.3570.57 mV ms) were not signifi-cantly affected (n = 5; Figures 4E and 4F).

Hexokinase and Fructokinase Activity

To determine the pathway of fructose metabolism inthe MON, we performed enzyme assays to identifythe activity of specific enzymes. Hexokinase assaysof brain (occipital cortex) were performed for thesake of comparison, as the kinetic parameters ofbrain hexokinase are well established (McIlwain andBachelard, 1985). The concentration dependence ofbrain hexokinase with glucose as substrate yieldeda Km of 31.571.5mmol/L and a Vmax of 344.7717.5 nmol/min per mg protein (n = 3; Figure 5A),which is consistent with previously publishedresults (McIlwain and Bachelard, 1985; Newsholmeand Leech, 1983). As hexokinase can phosphorylatefructose to fructose-6-phosphate, as well as glucose toglucose-6-phosphate, we determined the relationshipbetween fructose concentrations and hexokinaseactivity and found the Km increased to 1.9070.45 mmol/L with a Vmax of 392.6713.6 nmol/minper mg protein (n = 3; Figure 5B). This increase in

10 mM glucose

0

1

2

3

1 3CAP peak

1 3CAP peak

CA

P a

rea

(mV

ms)

20 mM fructose

0

0.5

1

1.5

2

2.5

1CAP peak

CA

P a

rea

(mV

ms)

10 mM fructose

0

0.5

1

1.5

2

2.5

CA

P a

rea

(mV

ms)

5 mM fructose

0

1

2

3

CA

P a

rea

(mV

ms)

A

B

C

D

2

2

1 3CAP peak

2

32

Figure 2 Effect of various fructose concentrations the individualcompound action potential (CAP) peak areas. (A) There was nosignificant effect on the CAP peaks areas after 60 or 180 minsincubation in 10 mmol/L glucose. Open columns represent10 mmol/L glucose after 60 mins; grey columns represent10 mmol/L glucose after 180 mins. This scheme also appliesto the data illustrated in (B–D). (B) Similarly, there was no effecton the CAP peak areas after 60 mins in 10 mmol/L glucose, or120 mins in 20 mmol/L fructose. (C) Incubation in 10 mmol/Lfructose for 2 hours resulted in a loss of the 1st CAP peak,compared with mouse optic nerves bathed in 10 mmol/Lglucose for 60 mins. (D) Similarly in MONs bathed in 5 mmol/L fructose for 2 hours, the 1st CAP peak was lost compared withcontrol MONs.

Fructose metabolism in MONPJ Meakin et al

91

Journal of Cerebral Blood Flow & Metabolism (2007) 27, 86–99

the Km of hexokinase with fructose as substrate hasbeen reported previously (McIlwain and Bachelard,1985; Newsholme and Leech, 1983; Sols and Crane,1954). We next sought to determine if there was anyfructokinase activity in the brain. Fructokinaseassays in cortical brain tissue with fructose assubstrate yielded a similar kinetic profile (Km = 1.7370.99 mmol/L; Vmax 339.3744.1 nmol/min per mgprotein, n = 3; Figure 5C) to the hexokinase assaywith fructose as substrate. As the principal behindthe fructokinase assay is dephosphorylation of ATPto ADP, which can occur with either hexokinase(fructose-fructose-6-phosphate) or fructokinase(fructose-fructose-1-phosphate), we concludedthat there was little to no fructokinase activity incortical tissue.

We performed similar experiments in the MON.With glucose as a substrate the Km and Vmax forhexokinase were 104.0734.0 mmol/L and 18.672.3 nmol/min per mg protein, respectively (n = 3;

Figure 5D), whereas with fructose as a substrate, theKm and Vmax were 2.8270.08 mmol/L and 17.372.42 nmol/min per mg protein, respectively (n = 3;Figure 5E). However fructokinase assays revealed asignificant difference in the ability of the respectiveenzymes to phosphorylate fructose. The Km was211.0711.3 mmol/L and the Vmax 144.0710.4 nmol/min per mg protein (n = 3; Figure 5F), that is,fructose was ‘efficiently’ metabolised in MON.

To further investigate the capacity of fructokinaseand hexokinase to phosphorylate fructose in thebrain or MON, we performed assays using asaturating concentration of the substrate (20 mmol/L fructose). In the cortical brain tissue, there was nosignificant difference in the reaction rate for hex-okinase or fructokinase, respectively (302.57116.7versus 307.97153.3 nmol/min per mg, n = 9; Figure6). However in MON, the Vmax was far greater forfructokinase than for hexokinase (111.9745.6 ver-sus 9.977.4 nmol/min per mg, n = 9), again reinfor-

20 40 60 80 100 120

0 mM glucose

10 mM fructose10 mM glucose

01234567

1 3CAP peak

CA

P a

rea

(mV

ms)

CA

P a

rea

(mV

ms)

CA

P a

rea

(mV

ms)

A

C

0

0.5

1

1.5

20 40 60 80 100 120Time (mins)

Time (mins)

Time (mins)

Nor

mal

ized

CA

P A

rea

0

0.5

1

1.5

Nor

mal

ized

CA

P A

rea

0

0.5

1

1.5

Nor

mal

ized

CA

P A

rea

0 mM glucose

10 mM glucose 10 mM glucose

0

1

2

3

E

0 20 40 60 80 100 120 140

0 mM glucose20 mM fructose10 mM glucose

0

1

2

3

4

B

D

F

2

1 3CAP peak

2

1 3CAP peak

2

Figure 3 Effect of depleting glycogen content on the ability of fructose to support the compound action potential (CAP). (A) A 25-minperiod of exposure to 0 mmol/L glucose resulted in delayed loss of the CAP. However, the CAP fully recovered on subsequentreintroduction of 10 mmol/L glucose artificial cerebrospinal fluid (aCSF). Insets are representative CAPs recorded at the time pointsindicated by the arrows. A similar scenario applies to (C and E). (B) Histogram illustrating the CAP peak areas at 60 and 120 minsafter recovery in 10 mmol/L glucose. There was no significant difference in the peak areas. (C) Recovery in 20 mmol/L fructose after a25-min period of aglycaemia fully restored the CAP peak areas to baseline levels. (D) Recovery in 20 mmol/L fructose fully restoredthe CAP peaks, but recovery in 10 mmol/L fructose (F) illustrated the loss of the 1st CAP peak. (E) Recovery in 10 mmol/L fructoseresulted in incomplete restoration of the total CAP area because of the loss of the 1st CAP peak. The smaller (B) trace indicates theglucose data from (C) for comparative purposes.

Fructose metabolism in MONPJ Meakin et al

92

Journal of Cerebral Blood Flow & Metabolism (2007) 27, 86–99

cing efficient fructose metabolism in the MONcompared with the cortex.

Fructokinase Expression in Optic Nerve

Western blotting revealed that fructokinase wasdetected in the liver and kidney as expected (Mayes,1993), but was absent from the brain (Figure 7).However, fructokinase was detected in the opticnerve, although at a lower density than in liver orkidney.

Immunohistochemical labellings were performedto confirm the presence of fructokinase in the MONaxons. As a preliminary step sections of liver weredoubly labelled with anti-tubulin and anti-fructoki-nase antibodies to highlight the pattern of hepato-cyte fructokinase expression. The outline of thehepatocytes could clearly be seen and the fructoki-nase was expressed within the cell membrane and

throughout the cytoplasm, but as expected, wasabsent from the nucleus of the cell (Figure 8A).Double labelling with anti-GFAP and anti-fructoki-nase antibodies showed an astrocytic presence offructokinase (Figure 8B). As with the hepatocytes,the fructokinase was localised at the cell membraneand throughout the cytoplasm, but was absent fromthe nucleus. In MON sections co-labelled with anti-neurofilament and anti-fructokinase, there was adegree of co-localisation as indicated by the regionsof orange (Figure 8C). The red staining in the middlepanel indicates axons cut in cross-section. Whilethere is considerable co-localisation indicatingthe presence of fructokinase on axons, fructokinasestaining was also present in a cell in the left centreof the left panel, most likely an astrocyte. Higherpower images reveal that there were large axonspresent (diameter B1 mm) that did not label withfructokinase (Figure 8D, right panel), which corre-sponds well with our functional and assay data,

A

B

C

0

0.5

1

1.5

2

0 20 40 60 80 100Time (mins)

0 20 40 60 80 100Time (mins)

Nor

mal

ized

CA

P a

rea

0

0.5

1

1.5

2

Nor

mal

ized

CA

P a

rea

100 Hz10 mM glucose

0

1

2

3

1CAP peak

CA

P A

rea

(mV

ms)

100 Hz

10 mM fructose10 mM gluc

0

1

2

3

4C

AP

Are

a (m

V m

s)

*

32 1CAP peak

32

D

E

F

Figure 4 Effect of increased tissue energy demand on the ability of fructose to support the compound action potential (CAP).(A) Imposing a 15-min period of 100 Hz stimulus on the mouse optic nerve (MON) did not significantly affect the CAP. (B) CAPs frombefore (black) and after (grey) the period of 100 Hz stimulus show the full recovery of the CAP in the presence of glucose. Scale barsare 1 ms and 2 mV. (C) Plotting the individual peak areas shows the full recovery of each of the three CAP peaks in 10 mmol/Lglucose. (D) The 15-min period of 100 Hz stimulus in the presence of 10 mmol/L fructose resulted in an almost total loss of the CAPfollowed by incomplete recovery of the total CAP area. The smaller (}) trace indicates the glucose data from Figure 3C forcomparative purposes. (E) CAPs from before (black) and after (grey) the period of 100 Hz stimulus show the loss of the 1st CAP peakin the presence of fructose. Scale bars are 1 ms and 2 mV. (F) Plotting the individual peak areas illustrates the loss of the 1st CAPpeak but the full recovery of the second and third peaks in 10 mmol/L fructose. *P < 0.005.

Fructose metabolism in MONPJ Meakin et al

93

Journal of Cerebral Blood Flow & Metabolism (2007) 27, 86–99

which led us to hypothesise that fructokinase waspresent in a select population of axons, those ofsmaller diameter contributing to the 2nd and 3rdCAP peaks.

Discussion

The results presented in this paper illustrate thatMON axons can survive with fructose as sole energy

substrate, if present at a sufficiently high concentra-tion. However, if the bathing concentration offructose is lowered, a subset of large axons( > 0.75 mm diameter) contributing to the 1st CAPpeak are lost after a 30 mins delay, indicating thatthese axons cannot efficiently metabolise fructose(Allen et al, 2006). This temporary maintenance ofthe 1st CAP peak in the presence of 10 mmol/Lfructose is because of the metabolism of glycogen,which provides supplementary energy substrate,probably in the form of L-lactate, to support axonfunction. However, once the glycogen is depletedthe 1st CAP peak is lost. We have shown a keydifference between cortical brain tissue and theMON, a central white matter tract with regard to

Fructokinase

-1000

100200300400500

-20 0 20 40 60 80 100[Fructose] (mM)

V (

nmol

/min

/mg)

Km = 1.73 mMVmax = 339.3 nmol/min/mg

Hexokinase

-1000

100200300400500

-20 0 20 40 60 80 100[Fructose] (mM)

V (

nmol

/min

/mg)

Km = 1.90 mMVmax = 392.6 nmol/min/mg

Hexokinase

-100

0100200300400

-1 0 2[Glucose] (mM)

V (

nmol

/min

/mg)

Km = 31.5 µMVmax = 344.7 nmol/min/mg

A

B

C

Hexokinase

-505

101520

[Glucose] (mM)

Km = 104.0 µMVmax = 18.6 nmol/min/mg

Fructokinase

-500

50100150200

-20 0 20 40 60 80 100[Fructose] (mM)

V (

nmol

/min

/mg)

V (

nmol

/min

/mg)

V (

nmol

/min

/mg)

Km = 211.0 µMVmax = 144.0 nmol/min/mg

Hexokinase

-505

10152025

-20 0 20 40 60 80 100[Fructose] (mM)

Km = 2.82 mMVmax = 17.3 nmol/min/mg

D

E

F

Brain MON

1 543 -1 0 21 543

Figure 5 Kinetic analysis of hexokinase and fructokinase in cortical brain tissue (A–C) and mouse optic nerve (MON) (D–F). Incortical brain tissue hexokinase activity versus substrate concentrations with either glucose (A) or fructose (B) as the substrate, andfructokinase activity with fructose as the substrate (C). In MON hexokinase activity versus substrate concentrations with eitherglucose (D) or fructose (E) as the substrate, and fructokinase activity with fructose as the substrate (F).

0

100

200

300

400

500

600

HK FK HK FK

V (

nmol

/min

/mg)

Brain

MON

Figure 6 Enzyme activities of fructose metabolism in the brainand mouse optic nerve (MON). Enzyme activities weremeasured from brain and MON tissue, with 20 mmol/L fructoseas substrate. HK = hexokinase; FK = fructokinase.

1 2kDa

38

543

Figure 7 Representative Western blot of fructokinase content inmouse liver, kidney, brain, and optic nerve. All lanes are fromthe same gel and exposed to the film for the same length oftime. Redundant lanes have been cropped and omitted. Lane1—optic nerve; lane 2—blank; lane 3—brain; lane 4—liver;and lane 5—kidney.

Fructose metabolism in MONPJ Meakin et al

94

Journal of Cerebral Blood Flow & Metabolism (2007) 27, 86–99

Figure 8 Immunohistochemical illustration of fructokinase expression in liver and mouse optic nerve (MON). (A) Liver sections co-labelledwith antitubulin antibodies (red) and the fructokinase antibody (green) show the widespread expression of fructokinase in hepatocytes. Scalebar 5mm. (B) Coexpression of fructokinase (green) and anti-GFAP antibodies (red) reveals astrocytic expression of fructokinase in transversesections of MON. The arrow illustrates cytoplasmic fructokinase expression in an astrocyte. The dotted circle encloses an area illustratingdiscreet pinpoint fructokinase expression less than 1mm in diameter, which is not colocalised with anti-GFAP antibodies, and presumablyreflects axonal fructokinase expression (see below). Scale bar 5mm. (C) Coexpression of fructokinase (green) and antineurofilament antibodies(red) in MON shows axonal expression of fructokinase. Middle panel illustrates axons cut in cross-sections with axon diameters of >1mm.Left panel illustrates fructokinase expression in astrocytes (arrow) as well as axons. Right panel is a merged image with orange dots illustratingcoexpression of antineurofilament and fructokinase antibodies. Scale bar 25mm. (D) Higher power image illustrating that certainneurofilament expressing axons do not express fructokinase (arrows). Scale bar 5mm.

Fructose metabolism in MONPJ Meakin et al

95

Journal of Cerebral Blood Flow & Metabolism (2007) 27, 86–99

fructose metabolism. The cortex lacks the enzymefructokinase, thus all phosphorylation of fructoseoccurs via hexokinase, but in the MON bothhexokinase and fructokinase are present, althoughthe differences in their kinetic profile suggests thatfructose is preferentially metabolised by fructoki-nase. Our results suggest that fructokinase isexpressed only in the axons that contribute to the2nd and 3rd CAP peaks, and that the remainingMON axons do not express fructokinase, hence theirinability to thrive in modest concentrations offructose.

The Role of Glycogen in Supporting the 1st CompoundAction Potential Peak in the Presence of Fructose

As aglycaemia-induced injury in MON is due totoxic influx of Ca2 + , the limited duration ofaglycaemia imposed in these experiments was notof sufficient duration to permit significant Ca2 +

influx (Brown et al, 2001a). This is a key point as25 mins of aglycaemia, while depleting tissue ofglycogen and resulting in CAP failure, does not leadto significant axon death, that is, the effects ofaglycaemia on CAP area are fully reversible ifglucose is reintroduced. Similarly, if 20 mmol/Lfructose is introduced after the period of aglycaemia,the CAP fully recovers. However, when 10 mmol/Lfructose was introduced after the period of aglycae-mia, both the rate and degree of CAP recovery wereadversely affected resulting in complete loss of the1st CAP peak. This loss was not because ofirreversible injury of these axons for the reasonsmentioned above. Increasing tissue energy demandby imposing a 15-min period of 100 Hz stimulus onthe MON significantly decreases glycogen levels(Brown et al, 2003). In MONs bathed in 10 mmol/Lglucose, there was no significant decrease in thetotal CAP area below baseline during the stimulus,which we have shown is because of the ambientglucose being supplemented by glycogen-derivedL-lactate, which is transported to the axons (Brownet al, 2004, 2003, 2005). On cessation of the stimulusthe baseline was unchanged. However, in thepresence of 10 mmol/L fructose, which was added20 mins before the stimulus to ensure both dissipa-tion of glucose and equilibration of fructosethroughout the nerve, a different scenario unfolded.At the onset of stimulus, the CAP area fell, and bythe end of the stimulus had fallen to zero, but onlythe 2nd and 3rd CAP peaks fully recovered afterstimulus: the 1st CAP peak was lost. Both sets ofexperiments show that in the presence of 10 mmol/Lfructose, but in the absence of glycogen, the 2nd and3rd CAP peaks can be fully supported but the 1stCAP peak cannot. Two clear points emerge fromthese experiments. In the presence of 10 mmol/Lfructose: (1) the axons contributing to the 2nd and3rd CAP peaks can both take up and efficientlymetabolise sufficient fructose to sustain baseline

axon conduction for extended periods of time, and(2) glycogen-derived metabolites are supplied toaxons contributing to the 1st CAP peak, in a mannersimilar to that which occurs during a/hypoglycae-mia where glycogen supports axon function whereinsufficient glucose is available to support the CAP(Brown et al, 2003).

Effectiveness of Fructose as a Metabolic Substrate inthe Mouse Optic Nerve

With the exception of one paper (Brown et al,2001b), all previous publications on the ability offructose to support CNS function in in vitropreparations have concentrated on grey matter, andin particular on the hippocampal slice (Bachelard etal, 1984; Wada et al, 1998; Yamane et al, 2000). Thefirst experiments to investigate the ability of fructoseto support CNS function were performed over 50years ago and showed that at a sufficiently highconcentration fructose could maintain normal ratesof oxygen uptake (McIlwain, 1953). Experimentswhere comparison was made of fructose’ ability tomaintain biochemical status and synaptic activitywere initially puzzling. Fructose (10 mmol/L) couldnot support the population spike and also led to asmall decrease in the excitatory postsynaptic poten-tial (EPSP). In addition, 10 mmol/L fructose couldnot support baseline levels of phosphocreatine orATP (Bachelard et al, 1984). However, increasing theconcentration of fructose to 20 mmol/L could fullysupport both phosphocreatine and ATP levels, andsupport the EPSP, but still resulted in a decrease inthe population spike (Bachelard et al, 1984). Similarresults have been found in other studies with thekey factor being the ability of fructose to sustainenergy levels in hippocampal slice while failing tofully support synaptic function, with the populationspike more severely affected than the EPSP (Wadaet al, 1998; Yamane et al, 2000). This implies anenergy demand that cannot be met by fructose,despite its ability to maintain phosphocreatine andATP at baseline levels. Accumulating evidencepoints to the importance of the key glycolyticenzyme phosphofructokinase in supporting func-tion, as hippocampal areas with the least phospho-fructokinase activity were most susceptible to loss ofsynaptic activity (Li et al, 2000). Thus, there may bea requirement for glycolytically produced ATP tosupport tissue energy requirements in a manner thatoxidative metabolism cannot match, perhaps be-cause of the proximity of glycolytic enzymes to themembrane bound Na +-K+-ATPase (Yamane et al,2000). This is supported by the inability of oxidativesubstrates such as lactate and pyruvate to fullysupport synaptic activity in the absence of glucose(Chih et al, 2001; Kanatani et al, 1995; Wada et al,1998; Yamane et al, 2000). However, it appearsunlikely that this occurs in MON as the MON isdevoid of glutamatergic synapses. That 10 mmol/L

Fructose metabolism in MONPJ Meakin et al

96

Journal of Cerebral Blood Flow & Metabolism (2007) 27, 86–99

fructose can fully restore the 2nd and 3rd CAP peaksfrom zero in the total absence of a glycogen (orglucose) indicates that fructose is a suitable meta-bolic substrate for at least some MON axons underbaseline conditions (although during periods ofhigh-intensity stimulus fructose is not as effectiveas glucose at maintaining function).

Fructose Metabolism

In isolated CNS tissue such as cerebellar cortex andhippocampus, fructose is metabolised by hexoki-nase (Bachelard et al, 1971; McIlwain and Bache-lard, 1985). The Km for glucose is reported as 50 to100 mmol/L (Bachelard et al, 1971; DiPietro andWeinhouse, 1959; McIlwain and Bachelard, 1985;Sols and Crane, 1954), whereas the Km with fructoseas substrate is several orders of magnitude greaterthan this (Sols and Crane, 1954). However, there isno significant difference in the Vmax with glucose orfructose as substrate, and it has even been reportedthat the Vmax for fructose is greater (Newsholme andLeech, 1983; Sols and Crane, 1954). As the circulat-ing level of glucose in the brain is in the lowmillimolar range, and the presumed fructose levelsare in the micromolar range, under physiologicconditions glucose will be about 50% phosphory-lated, whereas fructose will be less than 1%phosphorylated (Newsholme and Leech, 1983).Thus, there is minimal if any fructose phosphory-lated by hexokinase in the presence of physiologicconcentrations of glucose (DiPietro and Weinhouse,1959). Under experimental in vitro conditions wherethe brain tissue is placed in a superfusion chamberand the blood–brain barrier is circumvented, wewould expect the hexokinase activity with either10 mmol/L fructose or 10 mmol/L glucose to besimilar, given that each substrate will have freeaccess to the brain tissue, and will be present inconcentrations well in excess of the Km. In the MONthe situation is different. The Vmax for hexokinasewith glucose as substrate is much lower than inbrain, possibly reflecting the decreased glucoseutilisation in white matter compared with greymatter (Clarke and Sokoloff, 1999). It has beenreported in rabbit optic nerve that the Km forphosphorylation of glucose by hexokinase is in thelow millimolar range implying that glucokinase ispresent (Iannello et al, 1994, 1996), but we could notreplicate these results. The Vmax of hexokinase inMON remains low with fructose as substrate but theKm increases reflecting decreased affinity. Thus,hexokinase in both the brain and MON displays alow Km with glucose as substrate, but orders ofmagnitude higher for fructose. That fructokinase,which does not phosphorylate glucose (Mayes,1993), is present in the MON is a surprise, especiallygiven its absence in the cortex, and its absence fromcultured astrocytes, although these were of corticalorigin (Bergbauer et al, 1996). Its low Km and high

Vmax, similar to the values in liver (Hagopian et al,2005), suggest that if any fructose is metabolised inthe MON, it will be via fructokinase and nothexokinase. It is interesting to note that the levelsof fructose in the brain (10 to 50 mmol/L) are justbelow the Km for fructokinase, suggesting thatfructose could act as an alternate fuel when glucoseis low. Thus, a possible explanation for the ability of10 mmol/L fructose to support activity in the 2ndand 3rd CAP peaks, but not the 1st CAP peak couldbe because of the presence or absence of fructoki-nase in axons. Our immunohistochemistry datasupport this theory. We were able to show thatfructokinase is indeed expressed in axons, but alsoshowed the absence of fructokinase staining in somelarge axons. However, our immunohistochemicalresults cannot be used as evidence of a cleardichotomy between larger axons and intermediate/smaller axons, and it is also possible there is acontinuous distribution in the ability of axons toutilise fructose, diminishing as axons get larger, andincreasing as they get smaller. Of interest, but notdirectly relevant to the issues advanced in thiscurrent paper is the presence of fructokinase inGFAP-expressing astrocytes, whose function iscurrently unknown. If we hypothesise that fructoki-nase is selectively expressed in MON axons, then itis easy to imagine fructokinase efficiently metabo-lising sufficient fructose in the 2nd and 3rd CAPpeak axons given its Km of 211 mmol/L and thepresumed millimolar concentration of fructose inthe extracellular space. It should be noted thatbecause of diffusion, 10 mmol/L fructose in theaCSF would probably result in only low millimolarconcentration in the extracellular space (BaltanTekkok et al, 2003). However, the significantlyhigher Km of hexokinase in MON with fructose assubstrate in the 1st CAP peak axons will notphosphorylate sufficient fructose to support func-tion in MONs bathed in 10 mmol/L fructose.Hexokinase is obviously expressed in all MONaxons contributing to the CAP, as shown by theability of glucose to fully rescue the CAP afteraglycaemia. The fact that hexokinase in these axonscannot efficiently use fructose as an energy substratesuggests the following scenarios: (1) fructose cannotefficiently enter the axons. This is unlikely to be thecase as increasing fructose to 20 mmol/L fullysupported the CAP implying that fructose can crossthe axonal membranes when present in sufficientquantities, and even at lower concentrations(5 mmol/L) can still support the 2nd and 3rd CAPpeaks. It is unlikely there are differences in theability of axons to take up fructose. The GLUT5transporter is present in CNS tissue, and our resultssuggest that the GLUT5 or equivalent fructosetransporter is present in MON axons. (2) Fructoki-nase is expressed in 2nd and 3rd CAP peak axonsbut not in the 1st CAP peak axons, and althoughhexokinase is present in these axons, it cannotefficiently metabolise fructose.

Fructose metabolism in MONPJ Meakin et al

97

Journal of Cerebral Blood Flow & Metabolism (2007) 27, 86–99

Is Fructose Metabolised in the Optic Nerve Under InVivo Conditions?

Fructose is not generated within the body and anyfructose present in the system must occur fromingestion from one three sources. Naturally occur-ring in fruit, from the disaccharide sucrose, or viathe actions of sorbitol dehydrogenase on sorbitol,which may in turn be obtained naturally in food, orvia the action of aldose reductase on glucose(Salway, 1999). Thus, fructose is unlike glucoseand will have a circulating level dependent onlevels of ingested fructose. In the eye this is not thecase, however, as there is evidence of the productionof sorbitol and fructose via the polyol pathway(Salway, 1999). This pathway is particularly im-portant in uncontrolled type II diabetes where bothsorbitol and fructose levels are increased in the eye.Increasing evidence points to the accumulation ofsorbitol and its subsequent conversion to fructose askey initiators of diabetic retinopathy (Asnaghi et al,2003; Chung et al, 2003; Dagher et al, 2004; Lee andChung, 1999). The mechanism of damage is hy-pothesised as alterations of nicotinamide adeninedinucleotide/NADH ratios resulting in the produc-tion of reactive oxygen species (Mitka, 2005). Therole of fructose in diabetic retinopathy is unknown,but the fructose generated in the eye could be usedas an energy substrate to fuel metabolism, but thisawaits further study.

Acknowledgements

We thank Jon Banks and Alistair Mathie for thegenerous donation of equipment. We thank HermanBachelard for stimulating discussions on fructosemetabolism, Rob Layfield and Catherine Hughes fortechnical help with the Western blots, SusanAnderson for help with confocal imaging, and LizWhelband for help with image analysis.

References

Allen LM, Anderson S, Wender R, Meakin PJ, Ransom BR,Ray DE, Brown AM (2006) Fructose supports energymetabolism of some, but not all, axons in adult mouseoptic nerve. J Neurophysiol 95:1917–25

Asnaghi V, Gerhardinger C, Hoehn T, Adeboje A, LorenziM (2003) A role for the polyol pathway in the earlyneuroretinal apoptosis and glial changes induced bydiabetes in the rat. Diabetes 52:506–11

Bachelard HS (1989) Measurement of carbohydrates andtheir derivatives in neural tissues. In: Carbohydratesand energy metabolism (Boulton A, Baker G, Butter-worth R, eds), Clifton, N J: Humana Press, 133–54

Bachelard HS, Clark AG, Thompson MF (1971) Cerebral-cortex hexokinase. Elucidation of reaction mechanismsby substrate and dead-end inhibitor kinetic analysis.Biochem J 123:707–15

Bachelard HS, Cox DW, Drower J (1984) Sensitivity ofguinea-pig hippocampal granule cell field potentials to

hexoses in vitro—an effect on cell excitability? J Physiol352:91–102

Baltan Tekkok S, Brown AM, Ransom BR (2003) Axonfunction persists during anoxia in mammalian whitematter. J Cerebl Blood Flow Metab 23:1340–8

Baltan Tekkok S, Brown AM, Westenbroek RE, Pellerin L,Ransom BR (2005) Transfer of glycogen-derived lactatefrom astrocytes to axons via specific monocarboxylatetransporters supports mouse optic nerve activity.J Neurosci Res 81:644–52

Bergbauer K, Dringen R, Verleysdonk S, Gebhardt R,Hamprecht B, Wiesinger H (1996) Studies on fructosemetabolism in cultured astroglial cells and controlhepatocytes—lack of fructokinase activity and immu-noreactivity in astrocytes. Dev Neurosci 18:371–9

Bernheim F, Bernheim MLC (1941) Note on the oxidationof various sugars by brain tissue. J Biol Chem 140:441–4

Blakemore SJ, Aledo JC, James J, Campbell FC, Lucocq JM,Hundal HS (1995) The GLUT5 hexose transporter isalso localized to the basolateral membrane of thehuman jejunum. Biochem J 309:7–12

Brown AM, Baltan Tekkok S, Ransom B (2004) Energytransfer from astrocytes to axons—the role of CNSglycogen. Neurochem Int 45:529–36

Brown AM, Baltan Tekkok S, Ransom BR (2003) Glycogenregulation and functional role in mouse white matter.J Physiol 549.2:501–12

Brown AM, Sickmann HM, Fosgerau K, Lund TM,Schousboe A, Waagepetersen HS, Ransom BR (2005)Astrocyte glycogen metabolism is required for neuralactivity during aglycaemia or intense stimulation inmouse white matter. J Neurosci Res 79:74–80

Brown AM, Wender R, Ransom BR (2001a) Ionic mechan-isms of aglycemic axon injury in mammalian centralwhite matter. J Cereb Blood Flow Met 21:385–95

Brown AM, Wender R, Ransom BR (2001b) Metabolicsubstrates other than glucose support axon function incentral white matter. J Neurosci Res 66:839–43

Burant CF, Takeda J, Brot-Laroche E, Bell GI, Davidson NO(1992) Fructose transporter in human spermatozoa andsmall intestine is GLUT5. J Biol Chem 267:14523–6

Cataldo AM, Broadwell RD (1986) Cytochemical identifi-cation of cerebral glycogen and glucose-6-phosphataseactivity under normal and experimental conditions. I.Neurons and glia. J Electro Micro Technol 3:413–37

Chih CP, He J, Sly TS, Roberts EL, Jr (2001) Comparison ofglucose and lactate as substrates during NMDA-in-duced activation of hippocampal slices. Brain Res893:143–54

Chung SS, Ho EC, Lam KS, Chung SK (2003) Contributionof polyol pathway to diabetes-induced oxidative stress.J Am Soc Nephrol 14:S233–6

Clarke DD, Sokoloff L (1999) Circulation and energymetabolism of the brain. In: Basic neurochemistry—molecular, cellular, and medical aspects (Siegel GJ,Agranoff BW, Albers RW et al, eds), 6th edn. Philadel-phia: Lippincott-Raven Publishers, 637–69

Cummins KL, Perkel DH, Dorfman LJ (1979) Nerve fiberconduction-velocity distributions. I. Estimation basedon the single-fiber and compound action potentials.Electroencephalogr Clin Neurophysiol 46:634–46

Dagher Z, Park YS, Asnaghi V, Hoehn T, Gerhardinger C,Lorenzi M (2004) Studies of rat and human retinaspredict a role for the polyol pathway in human diabeticretinopathy. Diabetes 53:2404–11

Darakhshan F, Hajduch E, Kristiansen S, Richter EA,Hundal HS (1998) Biochemical and functional char-

Fructose metabolism in MONPJ Meakin et al

98

Journal of Cerebral Blood Flow & Metabolism (2007) 27, 86–99

acterization of the GLUT5 fructose transporter in ratskeletal muscle. Biochem J 336:361–6

DiPietro D, Weinhouse S (1959) Glucose oxidation in ratbrain slices and homogenates. Arch Biochem Biophys80:228–82

Funari VA, Herrera VL, Freeman D, Tolan DR (2005) Genesrequired for fructose metabolism are expressed inPurkinje cells in the cerebellum. Brain Res Mol BrainRes 142:115–22

Hagopian K, Ramsey JJ, Weindruch R (2005) Fructosemetabolizing enzymes from mouse liver—influence ofage and caloric restriction. Biochim Biophys Acta1721:37–43

Iannello S, Campione R, Volpicelli G, Belfiore F (1994)Rabbit optic nerve phosphorylates glucose through aglucokinase-like enzyme—studies in normal and spon-taneously hyperglycemic animals. Biochem Med MetabBiol 53:122–9

Iannello S, Campione R, Volpicelli G, Belfiore F (1996)Rabbit lens and retina phosphorylate glucose through aglucokinase-like enzyme—study in normal and spon-taneously hyperglycemic animals. J Diab Compl 10:68–77

Kanatani T, Mizuno K, Okada Y (1995) Effects of glycolyticmetabolites on preservation of high energy phosphatelevel and synaptic transmission in the granule cells ofguinea pig hippocampal slices. Experientia 51:213–6

Klein R (1944) Oxidation of fructose by brain in vitro.J Biol Chem 153:295–300

Lee AY, Chung SS (1999) Contributions of polyol pathwayto oxidative stress in diabetic cataract. FASEB J 13:23–30

Li X, Yokono K, Okada Y (2000) Phosphofructokinase, aglycolytic regulatory enzyme has a crucial role formaintenance of synaptic activity in guinea pig hippo-campal slices. Neurosci Lett 294:81–4

Maddock S, Hawkins JE, Holmes E (1939) The inadequacyof substances of the ‘glucose cycle’ for maintenance ofnormal cortical potentials during hypoglycemia pro-duced by hepatecomy with abdominal evisceration.Am J Physiol 125:551–65

Mantych GJ, James DE, Devaskar SU (1993) Jejunal/kidneyglucose transporter isoform (Glut-5) is expressed in thehuman blood–brain barrier. Endocrinology 132:35–40

Mayes PA (1993) Intermediary metabolism of fructose. AmJ Clin Nutr 58:754S–65S

McIlwain H (1953) Substances which support respirationand metabolic response to electrical impulses inhuman cerebral tissues. J Neurol Neurosurg Psychiat16:257–66

McIlwain H, Bachelard HS (1985) Biochemistry and thecentral nervous system. London: Churchill Livingstone,660pp

Michal G (1999) Biochemical pathways—an atlas ofbiochemistry and molecular biology. New York: JohnWiley and Sons Inc., 277pp

Mitka M (2005) Diabetic retinopathy mechanism probed.JAMA 293:148–9

Newsholme EA, Leech AR (1983) Biochemistry for themedical sciences. New York: John Wiley and Sons Inc.,952pp

Nualart F, Godoy A, Reinicke K (1999) Expression of thehexose transporters GLUT1 and GLUT2 during theearly development of the human brain. Brain Res824:97–104

Oldendorf WH (1971) Brain uptake of radiolabeled aminoacids, amines, and hexoses after arterial injection.Am J Physiol 221:1629–39

Payne J, Maher F, Simpson I, Mattice L, Davies P (1997)Glucose transporter GLUT5 expression in microglialcells. Glia 21:327–31

Rand EB, Depaoli AM, Davidson NO, Bell GI, Burant CF(1993) Sequence, tissue distribution, and functionalcharacterization of the rat fructose transporter GLUT5.Am J Physiol 264:G1169–76

Saitoh M, Okada Y, Nabetani M (1994) Effect of mannose,fructose and lactate on the preservation of synapticpotentials in hippocampal slices. Neurosci Lett171:125–8

Salway JG (1999) Metablism at a glance. Oxford: BlackwellScience, 111pp

Sasaki A, Yamaguchi H, Horikoshi Y, Tanaka G, NakazatoY (2004) Expression of glucose transporter 5 bymicroglia in human gliomas. Neuropathol Appl Neu-robiol 30:447–55

Sols A, Crane RK (1954) Substrate specificity of brainhexokinase. J Biol Chem 210:581–95

Stys PK, Ransom BR, Waxman SG (1991) Compoundaction potential of nerve recorded by suction elec-trode—a theoretical and experimental analysis. BrainRes 546:18–32

Van den Berghe G (1994) Inborn errors of fructosemetabolism. Annu Rev Nutr 14:41–58

Wada H, Okada Y, Uzuo T, Nakamura H (1998) The effectsof glucose, mannose, fructose and lactate on thepreservation of neural activity in the hippocampalslices from the guinea pig. Brain Res 788:144–50

Weber MH, de Oliveira KR, Valle SC, Schweigert ID, RottaLN, Fagundes I, Kruger AH, Souza K, Souza DO, PerryML (2001) Study of developmental changes on hexosesmetabolism in rat cerebral cortex. Neurochem Res26:161–6

Yamane K, Yokono K, Okada Y (2000) Anaerobic glyco-lysis is crucial for the maintenance of neural activity inguinea pig hippocampal slices. J Neurosci Methods103:163–71

Fructose metabolism in MONPJ Meakin et al

99

Journal of Cerebral Blood Flow & Metabolism (2007) 27, 86–99