Biochem. J. (1985) 230, 53-60 53Printed in Great Britain

The purification and properties of human liver ketohexokinaseA role for ketohexokinase and fructose-bisphosphate aldolase in the metabolic production of oxalate

from xylitol

Renze BAIS, Heather M. JAMES, Allan M. ROFE and Robert A. J. CONYERS*Metabolic Research Group, Division of Clinical Chemistry, Institute of Medical and Veterinary Science,

Adelaide, S.A. 5000, Australia

(Received 27 December 1984/11 April 1985; accepted 24 April 1985)

Ketohexokinase (EC 2.7.1.3) was purified to homogeneity from human liver, andfructose-bisphosphate aldolase (EC 4.1.2.13) was partially purified from the samesource. Ketohexokinase was shown, by column chromatography and polyacrylamide-gel electrophoresis, to be a dimer of Mr 75 000. Inhibition studies with p-chloromercuribenzoate and N-ethylmaleimide indicate that ketohexokinase containsthiol groups, which are required for full activity. With D-xylulose as substrate,ketohexokinase and aldolase can catalyse a reaction sequence which formsglycolaldehyde, a known precursor of oxalate. The distribution of both enzymes inhuman tissues indicates that this reaction sequence occurs mainly in the liver, to alesser extent in the kidney, and very little in heart, brain and muscle. The kineticproperties of ketohexokinase show that this enzyme can phosphorylate D-xylulose asreadily as D-fructose, except that higher concentrations of D-xylulose are required.The kinetic properties of aldolase show that the enzyme has a higher affinity for D-xylulose 1-phosphate than for D-fructose 1-phosphate. These findings support a rolefor ketohexokinase and aldolase in the formation of glycolaldehyde. The effect ofvarious metabolites on the activity of the two enzymes was tested to determine theconditions that favour the formation of glycolaldehyde from xylitol. The resultsindicate that few of these metabolites affect the activity of ketohexokinase, but thataldolase can be inhibited by several phosphorylated compounds. This work suggeststhat, although the formation of oxalate from xylitol is normally a minor pathway,under certain conditions of increased xylitol metabolism oxalate production canbecome significant and may result in oxalosis.

There has been renewed interest in the pathwaysfor the metabolism ofsugars and sugar alcohols. Forexample, Williams and co-workers (Williams,1980; Williams et al., 1984; Bleakley et al., 1984)have stimulated intense debate (Rognstad et al.,1982; Landau & Wood, 1983) with their formula-tion of a second pentose phosphate pathway. Theenzymes for the metabolism of sugar alcohols inhuman brain have been described (O'Brien &Schofield, 1980; O'Brien et al., 1983). Xylitol isgenerally considered to be metabolized via theglucuronate-xylulose and pentose phosphate path-

Abbreviation used: SDS, sodium dodecyl sulphate.* To whom correspondence and reprint requests

should be addressed.

ways, but its metabolism, both in vivo and in vitro,has been associated with the production of oxalateand known oxalate precursors (Hannett et al.,1977; Rofe et al., 1977, 1979, 1980; Hauschildt &Brand, 1979).Only recently has a satisfactory metabolic

pathway been proposed to explain oxalate produc-tion from xylitol. D-Xylulose, which can be formedfrom xylitol by the enzyme polyol dehydrogenase(Hollman & Touster, 1957; Arsenis & Touster,1969; O'Brien et al., 1983), can be phosphorylatedto xylulose 1-phosphate by ketohexokinase (ATP:D-fructose 1-phosphotransferase, EC 2.7.1.3) (com-monly called 'fructokinase' in the earlier litera-ture), and this phosphate ester can then be cleavedby fructose-bisphosphate aldolase (D-fructose-1,6-

Vol. 230

R. Bais, H. M. James, A. M. Rofe and R. A. J. Conyers

bisphosphate D-glyceraldehyde-3-phosphate-lyase,EC 4.1.2.13) to dihydroxyacetone phosphate andglycolaldehyde. Glycolaldehyde is a known oxalateprecursor, and we have shown that ketohexokinaseand aldolase, partially purified from human liver,can produce glycolaldehyde from xylulose (Jameset al., 1982). Independently, it has also been shownthat the same reactions can be catalysed by bovineliver ketohexokinase and rat liver aldolase and thatxylulose 1-phosphate is produced in isolatedhepatocytes that are metabolizing xylulose (Barn-grover et al., 1981; Barngrover & Dills, 1983).The present paper describes the purification of

ketohexokinase and aldolase from human liver andexamines their properties. In addition, the activityof these enzymes has been studied to determine theconditions under which the metabolism of xylitolwill favour the production of the oxalate precursor,glycolaldehyde.

Materials

Human tissues were obtained from the Divisionof Tissue Pathology in our Institute. The tissueswere macroscopically free of disease and taken,within 24h of death, from cadavers which hadbeen stored at 4°C. The tissues were used immedi-ately on receipt.ATP, ADP, phosphoenolpyruvate, fructose 1-

phosphate, fructose 1,6-bisphosphate and suspen-sions of pyruvate kinase, lactate dehydrogenase,triosephosphate isomerase, glycerol-3-phosphatedehydrogenase, glycerokinase, sorbitol dehydro-genase, aldehyde dehydrogenase and aldolase werepurchased from Boehringer Mannheim, Sydney,N.S.W., Australia. Xylulose, glycolaldehyde, SDS,phenylmethanesulphonyl fluoride, dithiothreitoland glycollate oxidase from spinach were fromSigma Chemical Co., St. Louis, MO, U.S.A. N-Ethylmaleimide and p-chloromercuribenzoatewere from Calbiochem-Behring, Sydney, N.S.W.,Australia. Sephacryl S-200, DEAE-Sepharose,CM-Sephadex, Chromatofocusing resin, BlueSepharose and Polybuffer were obtained fromPharmacia (South Seas) Pty. Ltd., Sydney,N.S.W., Australia. Acrylamide, bisacrylamide andNNN'N'-tetramethylethylenediamine were fromEastman Kodak Co., Rochester, N.Y., U.S.A. Allradiochemicals were obtained from Amersham(Australia) Pty. Ltd., Sydney, N.S.W., Australia.Other chemicals were of the highest purity avail-able. All sugars and sugar alcohols were of theD-form.

Methods

Preparation of D-xylulose 1-phosphateXylulose 1-phosphate was prepared by a modifi-

cation of the method of Byrne & Lardy (1954). The

eluate from the Amberlite CG-400 (200 mesh)column was assayed for total and inorganicphosphate (Ames, 1966), xylulose 1-phosphate(Barngrover et al., 1981), fructose 1 ,6-bisphosphate(Michal & Beutler, 1974) and glycolaldehyde(Datta & Racker, 1961). The fractions containingxylulose 1-phosphate were pooled and dried at45°C under vacuum on a rotary evaporator. Thedry residue was washed with water in the rotaryevaporator and made up to 4 ml with water,neutralized with KOH and stored at - 200C. Theyield of xylulose 1-phosphate was 80% fromfructose 1,6-bisphosphate, and the final productcontained less than 2% Pi and no detectable triosephosphate.

Enzyme assaysKetohexokinase activity was measured either

radiochemically or spectrophotometrically as de-scribed by Adelman et al. (1967). Control activitieswith the spectrophotometric assay were measuredby omitting ATP (for polyol dehydrogenase activ-ity), omitting fructose (for ATPase activity) andomitting fructose and ATP (for NADH dehydro-genase activity) as described previously (James etal., 1982). The activity was also measured with1 mM-fructose or 1 mM-xylulose as the substrate.

Aldolase activity was measured, with fructose1,6-bisphosphate as the substrate, by using theCalbiochem-Behring (Australia) Aldolase Stat-Pak, which is based on the method of Pinto et al.(1969). The activity of aldolase with fructose 1-phosphate or xylulose 1-phosphate was determinedby a similar method, except that the triosephos-phate isomerase was omitted. Control assays didnot contain substrate.The activity of the coupled reaction sequence

involving ketohexokinase and aldolase was exam-ined by measuring the formation of dihydroxyace-tone phosphate with glycerol-3-phosphate dehy-drogenase and NADH or by measuring theformation of glycolaldehyde with alcohol dehydro-genase and NADH (James et al., 1982).

All enzyme assays were carried out at 37°C.

Homogenates for enzyme assaysTissues were rinsed with cold 0.9% (w/v) NaCl

to remove blood, cut into thin slices and placed in50mM-triethanolamine/HCI buffer, pH7.5, con-taining 250mM-sucrose, 5mM-EDTA, 10mM-MgCl2 and 25mM-2-mercaptoethanol. Liver, kid-ney and brain extracts were prepared with threepasses of a loosely fitting Teflon pestle in a glasshomogenizer, and heart and skeletal muscle werehomogenized for 2 x 15 s in a Sorvall Omni-Mixer.Assays were performed on the supernatant frac-tions which were obtained by immediate centrifu-gation of the homogenate at 15000g for 20min at

1985

54

Ketohexokinase and xylitol metabolism

4°C. For the measurement of ketohexokinaseactivity in these crude supernatant fractions, theassay contained, in 20mM-triethanolamine/HClbuffer, pH7.5: lOOmM-KCl, 20mM-MgCl2, 15mM-ATP, 0.6mM-NADH, 16mM-NaF (to inhibitATPase activity), 1Ounits of aldolase/ml, 8 units ofglycerol-3-phosphate dehydrogenase/ml, 2 units ofglycerokinase/ml and 60 units of alcohol dehydro-genase/ml. Depending on which substrate is used,the assay measures the glyceraldehyde or glycol-aldehyde formed with alcohol dehydrogenase andNADH. However, because dihydroxyacetonephosphate inhibits aldolase, this is removed byusing glycerol-3-phosphate dehydrogenase andglycerokinase. This modified assay system wasnecessary because, if the ATP depletion wasmonitored, there are several reactions which wouldcontribute to the apparent ketohexokinase activ-ity, including hexokinase and phosphofructokin-ase if fructose was the substrate and xylulokinase ifxylulose was the substrate.

Protein assaysProtein concentration was determined by the

procedure of Lowry et al. (1951), except that theprotein was first precipitated with 5vol. of 10%(w/v) trichloroacetic acid to prevent interferenceby sucrose, EDTA or 2-mercaptoethanol. Therelative amounts of protein in the column effluentswere estimated from the A280.

Purification of human liver ketohexokinaseHuman liver ketohexokinase was prepared by a

modification of the methods used to prepare the rat(Sanchez et al., 197 la) and ox (Raushel & Cleland,1977) liver enzymes (see Table 1). All purificationsteps were carried out at 4°C. The Chromatofocus-ing purification step was performed as describedby Sluyterman & Elgersma (1978). The enzymewas then concentrated with an Amicon concentra-tor and dialysed against l0mM-Tris/HCl buffer,pH6.0. The dialysis residue was loaded on a BlueSepharose column (21 cm x 1.3cm) and the keto-

hexokinase activity was eluted with the samebuffer. The fractions containing ketohexokinasewere pooled and the pH was adjusted to 7.2. Theywere then concentrated with an Amicon concen-trator and stored at - 20°C.

Purification of human liver aldolaseHuman liver aldolase was prepared by the

procedure described by Eagles & Iqbal (1973), withthe following modification. After passage throughthe CM-cellulose 11 column, the protein wasprecipitated with 60%-satn. (NH4)2SO4 and theprecipitate chromatographed on a Sephacryl S-200column (96cm x 3.4cm), which was equilibratedwith l0mM-triethanolamine/HCl buffer, pH7.3.The fractions containing aldolase activity werepooled, concentrated with an Amicon concen-trator and stored frozen at - 20°C.

Results

Purification of ketohexokinase and aldolaseThe results for the preparation of ketohexokin-

ase from human liver are shown in Table 1. Thefinal specific activity of the ketohexokinase issimilar to that reported for both the ox (Raushel &Cleland, 1977) and rat (Sanchez et al., 1971a) liverenzymes. The enzyme appears to be homogeneouswhen examined by polyacrylamide-gel electro-phoresis under both denaturing and non-denatur-ing conditions (Fig. 1). It is necessary to performmore column-chromatography steps in this pre-paration than were described for the ox liverenzyme. This is because a greater amount ofprotein is obtained in the initial step of the humanliver preparation than in the ox liver extractionwithout there being any increase in the totalketohexokinase activity.The aldolase preparation was examined by

polyacrylamide-gel electrophoresis, and is nothomogeneous (results not shown), but, usingSephacryl S-200 column chromatography, we havefound that the enzyme has a Mr of 150000, which is

Table 1. Purification of ketohexokinase from human liver

Purification step

Crude extractpH step33%-satn.-(NH4)2SO4 fraction55%-satn.-(NH4)2SO4 fractionSephacryl S-200 column eluateDEAE-Sepharose column eluateCM-Sephadex column eluateChromatofocusing column eluateBlue-Sepharose column eluate

TotalVolume activity(ml) (units)

1450135815114392108884413

15711272954715611488456307141

Totalprotein(mg)

241692123688307741007022733972327915

Specificactivity

(units/mg)

0.00650.010.0310.0710.271.21.973.99.4

Yield Purification(%) (fold)

100816146393129209

25

11421853026001446

Vol. 230

55

R. Bais, H. M. James, A. M. Rofe and R. A. J. Conyers .

similar to that reported for aldolase from othersources (Peanasky & Lardy, 1958; Gracy et al.,1969). Other physical properties of the enzyme arethe same as have been described previously (Eagles& Iqbal, 1973).These two enzyme preparations were assayed for

the following enzymic activities, which couldpossibly interfere with the system that we aremeasuring: glycerol-3-phosphate dehydrogenase,polyol dehydrogenase, triosephosphate isomerase,ATPase, glucokinase, phosphofructokinase, alco-hol dehydrogenase and lactate dehydrogenase.There was no detectable activity of any of theseenzymes in either enzyme preparation.

Physical properties of ketohexokinaseThe M, of the enzyme as determined by column

chromatography on Sephadex G-100 and Sepha-cryl S-200 is 75000. Polyacrylamide-gel electro-phoresis in non-denaturing conditions (Davis,1964) and electrophoresis in the presence of SDS(Weber & Osborn, 1969) both indicate that theenzyme runs as a single protein band (Fig. 1). Theprotein migrates between malate dehydrogenase(37000) and ovalbumin (43000) on SDS gels andcorresponds to a Mr of 39000. Thus the resultsindicate that the human liver enzyme is a dimerwith subunits of M, 39000. The pI obtained on theChromatofocusing column for the enzyme is 5.5(Fig. 2) and is similar to the value of 5.7 obtainedfor the ox liver enzyme by isoelectric focusing inpolyacrylamide gels (Raushel & Cleland, 1977). Nometal ions and, in particular, no zinc, magnesium,copper or calcium could be detected in theketohexokinase preparation by either EnergyDispersion Analysis of X-rays (Woldseth, 1973) oratomic-adsorption spectrophotometry.The effect of thiol reagents (dithiothreitol, N-

acetylcysteine and 2-mercaptoethanol) and EDTAon the stability of ketohexokinase was studied bydiluting the partially purified enzyme in 20mM-triethanolamine/HCI buffer, pH 7.2, containing10mM effector. We found that the enzyme wasstable for at least 48 h, after which time the activitydecreased in the presence of N-acetylcysteine and2-mercaptoethanol, but to a lesser extent than thatwhich occurred with buffer alone. Ketohexokinasecan be inhibited by p-chloromercuribenzoate(Adelman et al., 1967; Sanchez et al., 1971a), andthis indicates that the enzyme has an absoluterequirement for a thiol group. However, the resultsobtained by Adelman et al. (1967), using iodoace-tate and iodoacetamide, are contradictory and donot support this conclusion. To study whetherthere is an essential thiol group present in theenzyme, we used not onlyp-chloromercuribenzoatebut also N-ethylmaleimide, which is considered tobe a more specific inactivator of thiol groups

1 2 3 4Fig. 1. Polyacrylamide-gel electrophoresis ofketohexokin-

ase purified from human liverThe polyacrylamide gels are (1) the pH 8.6 systemdescribed by Davis (1964) and (2-4) the SDS systemdescribed by Weber & Osborn (1969). The samplesare (1) and (2) ketohexokinase, (3) ovalbuminmarker (Mr 43000) and (4) malate dehydrogenasesubunit marker (M, 37000). For the non-denaturinggel electrophoresis, 60ug of protein was diluted withgel buffer containing 20% sucrose and then loadedon the gel. For the SDS/polyacrylamide-gel electro-phoresis, 60pg of protein was incubated overnightwith an equal volume of electrophoresis buffercontaining I mM-2-mercaptoethanol. Electrophore-sis conditions were as described in the text.

(Morell et al., 1964). We found that both thesecompounds inhibited ketohexokinase, but that p-chloromercuribenzoate was more effective at thesame concentration. Partial protection against thisinhibition was afforded by the presence of 10mM-fructose, but the other substrate for ketohexokin-ase, MgATP, and the activator, K+, did not protectagainst the inhibition by either p-chloromercuri-benzoate or N-ethylmaleimide.

Kinetic properties of ketohexokinase and aldolaseThe Km and relative Vmax. values for human liver

ketohexokinase and aldolase with various sub-

1985

56

.k"

Ketohexokinase and xylitol metabolism

7

6

5

2.5 3

1-1

;::r.

0 20 40 60 80 100 120Tube number

Fig. 2. Purification of human liver ketohexokinase on a Chromatofocusing columnThe partially purified protein was loaded on the Chromatofocusing column, which had been equilibrated with25mM-imadazole/HCl buffer, pH7.0. The activity was eluted with Polybuffer, pH4.0. The flow rate was 40ml/h,and 4ml fractions were collected. Protein was monitored by measuring A280 (@). Ketohexokinase activity (A) wasmeasured by the spectrophotometric assay as described in the text. *, pH.

Table 2. Apparent kinetic constants Jbr various substrates of ketohexokinase and aldolase from human liverInitial rates were calculated from progress curves by using the assay mixtures as described in the text. All assays wereperformed at 37°C.

Enzyme Substrate

Relative specificactivity

Km(app.) [ Vmax.(app.)](mM) (%)

Ketohexokinase

Aldolase

D-FructoseD-XyluloseD-Fructose 1,6-bisphosphateD-Fructose 1-phosphateD-Xylulose I-phosphate

strates were determined (Table 2). The Km valuefor ketohexokinase with xylulose as the substrate ismuch greater than that with fructose, but the Vmax.values are the same. The similar Vmax. values forthe human liver enzyme are in contrast with thosefound for the enzyme from other species, in whichthe activity with xylulose was lower (Adelman etal., 1967; Raushel & Cleland, 1977).

For aldolase, the Km values for fructose 1-phosphate, xylulose 1-phosphate and fructose 1,6-bisphosphate differ considerably, but the Vmax.values are similar. The Km values for both fructose1-phosphate and fructose 1,6-bisphosphate aresimilar to those reported previously for the humanliver enzyme (Gurtler et al., 1971; Eagles & Iqbal,1973), but the ratio of the Vmax values for these twosubstrates is slightly higher.

Tissue distribution of ketohexokinase and aldolaseThe distribution of ketohexokinase and aldolase

in various human tissues is shown in Table 3.Ketohexokinase activity is highest in the liver andkidney, with very little activity in the other tissuestested. Aldolase, however, is more widely distrib-uted, with considerable activity in muscle. Theactivity of ketohexokinase and aldolase in the liveris similar to that previously reported (Heinz et al.,1968).

Coupled reaction sequence of ketohexokinase andaldolase

It has previously been shown that xylitol can bemetabolized via a series of reactions in whichketohexokinase and aldolase form glycolaldehyde,

Vol. 230

0.866.00.00372.00.016

1009910099110

57

00

xll

R. Bais, H. M. James, A. M. Rofe and R. A. J. Conyers

Table 3. Distribution of ketohexokinase and aldolase in human tissuesActivities are expressed as means+ S.D. (n = 5). The assay mixtures are described in detail in the text.

Activity (umol/min per g of tissue)

Enzyme Substrate Liver Kidney Brain Heart Muscle

Ketohexokinase D-Fructose 2.8 +1 .1 2.3 +1.2 0.22 + 0.12 0.14 + 0.08 0.17 + 0.08D-Xylulose 2.5+1.7 2.1 +1.9 0.13+0.18 0.16+0.07 0.12+0.09

Aldolase D-Fructose 1,6-bisphosphate 11.7 + 5.4 4.7+1.4 4.6+0.7 4.9+1.8 58 + 15D-Fructose 1-phosphate 4.8+4.4 1.7+0.5 0.22+0.10 0.21 +0.03 3.3+2.7D-Xylulose 1-phosphate 6.8 +6.1 2.2+0.8 0.5+0.17 0.53+0.30 6.1+5.5

which is a direct precursor of oxalate (James et al.,1982; Barngrover et al., 1981). The validity of thispathway has been shown by measuring theformation of the products from xylulose andfructose (James et al., 1982). There are severalenzymes with the potential to interfere with thisseries of reactions (see above), but we could notdetect them in our enzyme preparations. The timecourse for the coupled reaction of ketohexokinaseand aldolase with either fructose or xylulose as theinitial substrate is shown in Fig. 3. The results arenon-linear, which is to be expected for a coupledenzyme system (McClure, 1969; Garcia-Carmonaet al., 1981), and the size of the lag can be varied bychanging the aldolase concentration.

In addition, we have shown, in vitro, theformation of oxalate from xylitol. This sequence ofreactions was carried out with enzymes purifiedfrom several different sources, namely sheep liversorbitol dehydrogenase, human liver ketohexokin-ase and aldolase, yeast aldehyde dehydrogenase,spinach glycollate oxidase and rabbit musclelactate dehydrogenase. As cofactors, 10mM-ATP,5mM-NAD+ and 5mM-FMN were added and,although the incorporation of radioactivity intooxalate from labelled xylitol was low, it wascomparable with the amount of oxalate producedfrom xylitol by isolated hepatocytes (Rofe et al.,1977).We have measured the effect of a range of

metabolites (10mM) on the activity of bothketohexokinase and aldolase. Initial investigationswith ketohexokinase were carried out by thespectrophotometric assay, with fructose (25 and1 mM) and xylulose (1 mM) as substrates. Whenthere was a change in the activity, the radioactiveassay, with fructose as the substrate, was used, asthis eliminates any possible effects of the metabo-lites on the auxiliary enzymes. For aldolase withfructose 1,6-bisphosphate, fructose 1-phosphateand xylulose 1-phosphate (all at 20 x Km) assubstrates, the activity can be measured with eitheralcohol dehydrogenase or glycerol-3-phosphatedehydrogenase as the linking enzyme. If there wasany change in the activity of the aldolase assay, the

0.5

0 5 10 15Time (min)

Fig. 3. Progress curves br reactions oJfructose or xylulosewith ketohexokinase and aldolase purifiedfrom human liverThe reaction mixtures were essentially as describedin the text with the initial substrate concentrationbeing 10mM-fructose (-) or -xylulose (M). Reac-tions were initiated by the addition of substrate andmonitored at 340nm.

effect was tested by both of these auxiliary enzymesystems.These studies showed that, of the intermediates

involved in the sequence of reactions which formoxalate from xylulose (James et al., 1982), onlyglyoxylate had any effect on either ketohexokinaseor aldolase. Amino acids and C5 and C6 aldoses,ketoses and polyols had no effect on the activity ofeither enzyme. Of the other compounds tested,Ca2+, ADP and citrate had a significant effect onthe ketohexokinase activity. The Ca2+ probablycompetes with Mg2+ for the ATP, citrate complex-es essential metal ions (e.g. K+) (Sanchez et al.,1971b), and ADP inhibition has been reportedpreviously (Raushel & Cleland, 1977). We alsoconfirmed the last authors' observation of inhibi-tion by fructose 1-phosphate, but the inhibitionconstant is high and probably has no physiologicalsignificance.Our inhibitor studies with human liver aldolase

gave similar results to those described by Woods etal. (1970). They used purified rat liver aldolase and

1985

58

Ketohexokinase and xylitol metabolism

found that several phosphorylated metabolitesinhibit the enzymic activity and that, whenfructose 1-phosphate (or, as in our studies, xylulose1-phosphate as well) is a substrate, aldolase is moresensitive to inhibition than when fructose 1,6-bisphosphate is a substrate. The phosphorylatedintermediates that caused marked inhibition in-cluded glucose 1-phosphate, glucose 6-phosphate,fructose 6-phosphate, ribose 5-phosphate, eryth-rose 4-phosphate, glycerol 2,3-bisphosphate, ATP,ADP and IMP. Aldolase was also inhibited byascorbic acid, with a concentration of 1 mMcausing 50% inhibition. Consideration of thestructure of ascorbic acid and the substrates ofaldolase suggest that this effect is due to a sterichindrance by ascorbic acid. It is unlikely, however,that ascorbic acid concentration in vivo would bethis high, even in association with mega-doseintakes. The other inhibitor of the activity isglyoxylate, but its mechanism of action is unclear.

Discussion

The procedure reported here for the purificationof ketohexokinase from human liver results in theisolation of a homogeneous enzyme preparation.Column chromatography and polyacrylamide-gelelectrophoresis indicate that the enzyme is a dimerof Mr 75000. Other workers have shown rat liverketohexokinase to be a monomer of M, 28000(Sanchez et al., 1971a) and the ox liver enzyme tobe a dimer of M, 56000 (Raushel & Cleland, 1977).However, in their studies with the rat liver enzyme,Sanchez et al. (1971a) detected at least two proteinbands by polyacrylamide-gel electrophoresiswhich corresponded to the enzymic activity. Thusit may be that the rat liver enzyme can exist in anumber of active forms. When we chromato-graphed rat liver ketohexokinase on Sephacryl S200 under the same conditions as described for thehuman liver enzyme, the activity corresponded toan M, of 75000. One explanation for the discre-pancy in the results is that various buffers used inthe three studies result in different active forms ofthe enzyme. We have tried to locate differentactive forms of the human liver enzyme by usingh.p.l.c. with a variety of buffers, but have beenunsuccessful. One problem that we found with thiswork was that, if the preparation was nothomogeneous, there were several apparent activitypeaks. In our experience, however, ketohexokin-ase could only be shown to be present in the peaksif the spectrophotometric assay included all thecontrol assays described in the Methods section, orthe radiochemical assay was used.

Stability studies on crude extracts of ketohexo-kinase and the partially purified enzyme givedifferent results and showed that only the purified

enzyme could be stabilized by N-acetylcysteine or2-mercaptoethanol. One explanation for thesecontradictory results is that in the crude extracts,proteinases, which are known to be present in theliver at high concentrations, are released duringthe extraction procedure and are activated by thethiol reagents or the EDTA. To minimize thisproblem, phenylmethanesulphonyl fluoride, a pro-teinase inhibitor, was added to the extractionbuffer. With the purified enzyme the presence ofthiol groups is required for full catalytic activity,and these groups are most likely to be at the activesite of the enzyme (or immediately adjacent to it),because the inactivation could be partially pre-vented by the substrate, fructose.

Ketohexokinase is most abundant in the liver.The kidney has about half the activity present inthe liver, and there are very low activities presentin the brain, heart and muscle. These findingsagree with the results obtained for the distributionof the enzyme in the rat (Adelman et al., 1967), andour results for the activity in human liver aresimilar to that found by Heinz et al. (1968). In allthe human organs tested, the activities with bothfructose and xylulose as substrates are very similar,and the liver must be the major organ for xylitolmetabolism. It has been claimed by several authorsthat xylitol can be metabolized by tissues otherthan liver and kidney, but it is very difficult toascertain from results of experiments in vivowhether one is monitoring the metabolism ofxylitol itself or the metabolism of glucose after theliver has converted xylitol into glucose.

Xylulose is the oxidation product of the hepaticNAD-dependent xylitol dehydrogenase, and undernormal circumstances this compound can bephosphorylated by xylulokinase to xylulose 5-phosphate (Arsenis & Touster, 1969), which entersthe pentose phosphate pathway, and then theglycolytic-gluconeogenic pathway to form glu-cose, pyruvate and lactate.The results described above indicate that xylitol

can, in addition, be metabolized through xylulose1-phosphate to glycolaldehyde, a known oxalateprecursor. Barngrover & Dills (1983) concludedthat the production of xylulose from xylitol is soslow that xylulokinase never becomes saturatedand that the xylulose concentration would notreach values sufficient for extensive phosphoryla-tion by ketohexokinase. The relatively high Km ofliver ketohexokinase for xylulose would alsosupport this conclusion. However, xylulose 1-phosphate and glycolaldehyde are produced inisolated hepatocytes treated with xylulose, indicat-ing that this is a minor pathway of xylitolmetabolism. The proposed pathway can explainthe production of oxalate from xylitol observed inisolated hepatocytes, especially since the amount

Vol. 230

59

60 R. Bais, H. M. James, A. M. Rofe and R. A. J. Conyers

of oxalate produced is less than 0.2% of the xylitolutilized (Rofe et al., 1980).These studies are important in that they provide

a biochemical explanation for the clinical observa-tion that a number of patients receiving xylitol, asan intravenous nutrient, developed adverse reac-tions, including the deposition of calcium oxalatein the kidney, brain, lung and vascular tissues, andthis resulted in the death of some of these patients(Thomas et al., 1972; Schr6der, 1980). However,the liver is the major site of oxalate production(Liao & Richardson, 1972; Richardson, 1973) and,in xylitol toxicity, it would appear that the oxalateis produced in the liver, and transported to the non-hepatic tissues in which the deposition occurs. It isclear that the production of oxalate from xylitol is arelatively minor pathway, but in situations ofexcessive xylitol intake, such as parenteral infu-sion, it does become important.

We gratefully acknowledge the technical assistance ofJudith Nairn, the assistance of Dawn Burbidge in thepreparation of the manuscript, and the support of RocheProducts (Australia).

ReferencesAdelman, R. C., Ballard, F. J. & Weinhouse, S. (1967) J.

Biol. Chem. 242, 3360-3365Ames, B. N. (1966) Methods Enzymol. 7, 115-118Arsenis, C. & Touster, 0. (1969) J. Biol. Chem. 244,

3895-3899Barngrover, D. A. & Dills, W. L., Jr. (1983) J. Nutr. 113,

522-530Barngrover, D. A., Stevens, H. C. & Dills, W. L., Jr.

(1981) Biochem. Biophys. Res. Commun. 102, 75-80Bleakley, P. A., Arora, K. K. & Williams, J. F. (1984)

Biochem. Int. 8, 491-500Byrne, W. L. & Lardy, H. A. (1954) Biochim. Biophys.

Acta 14, 495-501Datta, A. G. & Racker, E. (196 1)J. Biol. Chem. 236, 617-

623Davis, B. J. (1964) Ann. N.Y. Acad. Sci. 121, 404-427Eagles, P. A. M. & Iqbal, M. (1973) Biochem. J. 183, 429-439

Garcia-Carmona, F., Garcia-Canovas, F. & Lozano,J. A. (1981) Anal. Biochem. 113, 286-291

Gracy, R. W., Lacko, A. G. & Horecker, B. L. (1969) J.Biol. Chem. 244, 3913-3919

Gurtler, B., Bally, C. & Leuthardt, F. (1971) Hoppe-Seyler's Z. Physiol. Chem. 352, 1455-1462

Hannett, B. Thomas, D. W., Chalmers, A. H., Rofe,A. M., Edwards, J. B. & Edwards, R. G. (1977) J.Nutr. 107, 458-465

Hauschildt, S. & Brand, K. (1979) Biochem. Med. 21, 55-61

Heinz, F., Lamprecht, W. & Kirsch, J. (1968) J. Clin.Invest. 47, 1826-1832

Hollman, S. & Touster, 0. (1957) J. Biol. Chem. 225, 87-102

Jamnes, H. M., Bais, R., Edwards, J. B., Rofe, A. M. &Conyers, R. A. J. (1982) Aust. J. Exp. Biol. Med. Sci.60, 117-122

Landau, B. R. & Wood, H. G. (1983) Trends Biochem.Sci. 8, 292-296

Liao, L. L. & Richardson, K. E. (1972) Arch. Biochem.Biophys. 153, 438-448

Lowry, 0. H., Rosebrough, N. J., Farr, A. L. & Randall,R. J. (1951) J. Biol. Chem. 193, 265-275

McClure, W. R. (1969) Biochemistry 8, 2782-2786Michal, G. & Beutler, H. 0. (1974) in Methods of

Enzymatic Analysis (Bergmeyer, H. U., ed), pp. 1314-1319, Academic Press, New York and London

Morell, S. A., Ayers, V. E., Greenwalt, T. J. & Hoffman,P. (1964) J. Biol. Chem. 239, 2696-2705

O'Brien, M. M. & Schofield, P. J. (1980) Biochem. J. 187,21-30

O'Brien, M. M., Schofield, P. J. & Edwards, M. R.(1983) Biochem. J. 211, 81-90

Peanasky, R. J. & Lardy, H. A. (1958) J. Biol. Chem. 233,371-373

Pinto, P. V. C., Kaplan, A. & Van Dreal, P. A. (1969)Clin. Chem. 15, 349-360

Raushel, F. M. & Cleland, W. W. (1977) Biochemistry 16,2169-2175

Richardson, K. E. (1973) Toxicol. Appl. Pharmacol. 24,530-538

Rofe, A. M., Thomas, D. W., Edwards, R. G. &Edwards, J. B. (1977) Biochem. Med. 18, 440-451

Rofe, A. M., Conyers, R. A. J., Bais, R. & Edwards, J. B.(1979) Aust. J. Exp. Biol. Med. Sci. 57, 171-176

Rofe, A. M., James, H. M., Bais, R., Edwards, J. B. &Conyers, R. A. J. (1980) Aust. J. Exp. Biod. Med. Sci.58, 103-116

Rognstad, R., Wals, P. & Katz, J. (1982) Biochem. J. 208,851-855

Sanchez, J. J., Gonzales, N. S. & Pontis, H. G. (1971a)Biochim. Biophys. Acta 227, 67-78

Sanchez, J. J., Gonzales, N. S. & Pontis, H. G. (1971b)Biochim. Biophys. Acta 227, 79-85

Schroder, R. (1980) Dtsch. Med. Wochenschr. 105, 997-1001

Sluyterman, L. A. fE. & Elgersma, 0. (1978) J.Chromatogr. 150, 17-30

Thomas, D. W., Edwards, J. B., Lawrence, J. R. &Edwards, R. G. (1972) Med. J. Aust. 1, 1238-1246

Weber, K. & Osborn, M. (1969) J. Biol. Chem. 244,4406-4412

Williams, J. F. (1980) Trends Biochem. Sci. 5, 315-318Williams, J. F., Arora, K. K. & Longenecker, J. P. (1984)

Trends Biochem. Sci. 9, 275-277Woldseth, R. (1973) X-ray Energy Spectrometry, Kevex

Corp., Burlingame, CAWoods, H. F., Eggleston, L. V. & Krebs, H. A. (1970)

Biochem. J. 119, 501-510

1985