Biochem. J. (1978) 175,125-135Printed in Great Britain

Purification of the Hexokinases by Affinity Chromatography on

Sepharose-N-Aminoacylglucosamine Derivatives

DESIGN OF AFFINITY MATRICES FROM FREE SOLUTION KINETICS

By CHRISTINE L. WRIGHT, ARJUMAND S. WARSY, MICHAEL J. HOLROYDE*and IAN P. TRAYER

Department of Biochemistry, University of Birmingham, P.O. Box 363, Birmingham B15 2TT, U.K.

(Received 9 February 1978)

The purification is described of rat hepatic hexokinase type III and kidney hexokinasetype I on a large scale by using a combination of conventional and affinity techniquessimilar to those previously used for the purification of rat hepatic glucokinase [Holroyde,Allen, Storer, Warsy, Chesher, Trayer, Cornish-Bowden & Walker (1976) Biochem. J.153, 363-373] and muscle hexokinase type II [Holroyde & Trayer (1976) FEBS Lett. 62,215-219]. The key to each purification was the use of a Sepharose-N-aminoacylglucos-amine affinity matrix in which a high degree of specificity for a particular hexokinaseisoenzyme could be introduced by either varying the length of the aminoacyl spacer and/orvarying the ligand concentration coupled to the gel. This was predicted from a study ofthefree solution kinetic properties of the various N-aminoacylglucosamine derivatives used(N-aminopropionyl, N-aminobutyryl, N-aminohexanoyl and N-aminooctanoyl), synthe-sized as described by Holroyde, Chesher, Trayer & Walker [(1976) Biochem. J. 153,351-361]. All derivatives were competitive inhibitors, with respect to glucose, of thehexokinase reaction, and there was a direct correlation between the Ki for a particularderivative and its ability to act as an affinity matrix when immobilized to CNBr-activatedSepharose 4B. Muscle hexokinase type II could be chromatographed on the Sepharoseconjugates of all four N-aminoacylglucosamine derivatives, although the N-amino-hexanoylglucosamine derivative proved best. This same derivative was readily able tobind hepatic glucokinase and hexokinase type III, but Sepharose-N-amino-octanoyl-glucosamine was better for these enzymes and was the only derivative capable of bindingkidney hexokinase type I efficiently. Separate studies with yeast hexokinase showed thatagain only the Sepharose-N-amino-octanoylglucosamine was capable of acting as anefficient affinity matrix for this enzyme. Implications of these studies in our understandingof affinity-chromatography operation are discussed.

Lack of efficient purification procedures for themammalian hexokinases (ATP-D-hexose 6-phospho-transferases, EC 2.7.1.1) has hindered any in-depthinvestigations into their physiological control andmode of action. Until recently the only isoenzymeto be purified to homogeneity was type I from brain(Schwartz & Basford, 1967; Chou & Wilson, 1972)and pig heart (Easterby & O'Brien, 1973). Theselective extraction of the mitochondria to which thisisoenzyme is bound simplified the purificationprocess, but the other three isoenzymes could not beobtained so easily. However, Holroyde & Trayer(1974) and Holroyde et al. (1976a,b) succeeded inpurifying hepatic glucokinase (hexokinase type IV,

* Present address: Department of Physiology, Univer-sity of Cincinnati College of Medicine, Cincinnati,OH 45267, U.S.A.

Vol. 175

EC 2.7.1.2) to homogeneity by using, in conjunctionwith ion-exchange chromatography, a new affinitymatrix of Sepharose-N-(6-aminohexanoyl)-2-amino-2-deoxy-D-glucopyranose. Glucokinase differs fromthe other mammalian hexokinases in that it is onlyhalf their molecular weight (Holroyde et al., 1976b),has a slightly narrower hexose specificity and ahigher Km for glucose (Parry & Walker, 1966) anddisplays a sigmoidal dependence of enzyme rateagainst glucose concentration (Storer & Cornish-Bowden, 1976).

Initial experiments with this glucosamine affinitymatrix (Chesher et al., 1973) suggested that gluco-kinase bound selectively and reversibly, but thatlow-Km hexokinase activity in liver extracts (presumedto be types I, II and III) and yeast hexokinase wereunretarded. From the start this seemed puzzling,

125

C. L. WRIGHT, A. S. WARSY, M. J. HOLROYDE AND I. P. TRAYER

as it is well known that N-acetylglucosamine is a goodcompetitive inhibitor of all four isoenzymes (Walker,1966). It was eventually found, however, that, byadjusting the operating conditions slightly, hexo-kinase type II from rat muscle could be purified tohomogeneity (Holroyde & Trayer, 1976). Thisprompted a more thorough investigation of thereactions between the hexokinases and this affinitymatrix. Various glucosamine compounds weresynthesized, giving a series of derivatives in which thelength of the carbon chain attached to the aminogroup was varied. This had the effect of altering thelength of the spacer arm attaching the glucosamineto the Sepharose When the inhibition with respect toglucose of rat and yeast hexokinase isoenzymes bythese compounds was investigated, a definite cor-relation was found to exist between the K, of acompound in free solution and its effectiveness as anaffinity matrix when immobilized to Sepharose. Thus,by varying either the length of the aminoacyl spacermolecule and/or the concentration of ligand attachedto the gel, it has been possible to prepare a glucos-amine affinity matrix capable of purifying each of thehexokinase isoenzymes to homogeneity on a largescale. Some of this work has been reported in pre-liminary form (Wright et al., 1976).

Materials and Methods

ChemicalsCoenzymes, Dowex ion-exchange resins and

dithiothreitol were purchased from Sigma (London)Chemical Co., Kingston-upon-Thames, Surrey, U.K.Ethyl trifluoromethanecarbothiolate was obtainedfrom Pierce Chemical Co., Rockford, IL, U.S.A., andN- ethoxycarbonyl - 2 -ethoxy - 1, 2 - dihydroquinolinewas from Calbiochem, San Diego, CA, U.S.A.DEAE-cellulose (DE-52) was from Whatman Bio-chemicals, Maidstone, Kent, U.K., Sepharose 4B,DEAE-Sephadex A-50 and Sephadex G-25 were fromPharmacia (G.B.), London W.5, U.K., and CNBrwas from Aldrich Chemical Co., Gillingham, Dorset,U.K. Hydroxyapatite was obtained from BioRadLaboratories, Richmond, CA, U.S.A. N-Acetyl-glucosamine hydrochloride, glucosamine hydro-chloride, 3-aminopropionic acid, 4-aminobutyricacid, 6-aminohexanoic acid and 8-amino-octanoicacid were from BDH Chemicals, Poole, Dorset,U.K. All other chemicals were of analytical grade andused as supplied.

ProteinsYeast hexokinase [ATP-D-hexose 6-phospho-

transferase, EC 2.7.1.1; type C-302 (300 units/mg)]and glucose 6-phosphate dehydrogenase [D-glucose6-phosphate-NADP+ oxidoreductase, EC 1.1.1.49;from yeast, type VII (345 units/mg)] were purchased

from Sigma Chemical Co. Gifts of purified yeasthexokinase types A and B and a crude yeast extractafter acid precipitation (Ramel et al., 1971) weregenerously given by Professor E. A. Barnard,Department of Biochemistry, Imperial College ofScience and Technology, London S.W.7, U.K.

Hexokinase preparationsRat hepatic glucokinase was prepared as described

by Holroyde et al. (1976b) and rat skeletal-musclehexokinase type II as described by Holroyde &Trayer (1976). Hexokinase types I and III wereprepared from kidney and liver respectively. Theinitial stages of the preparation are described belowand are based on our method of preparation ofhepatic glucokinase and muscle hexokinase type II.Essentially the enzymes are extracted from thetissues, adsorbed batchwise on DEAE-cellulose andthen chromatographed on either DEAE-Sephadex(hexokinase type III) or hydroxyapatite (hexokinasetype I) before their final purification on the affinitycolumns described in the Results section.

Preparation of rat kidney hexokinase type IFrozen rat kidneys (400g) (purchased from Olac

(Northern), Bicester, Oxon., U.K.) were choppedfinely with scissors and homogenized in 2vol. of20mM-triethanolamine/HCl, pH7.0, containing0.2M-MgCl2, 10mM-KCl, 50mM-glucose, 1mM-EDTA, 0.5 mM-dithiothreitol and 5% (v/v) glycerol inan Ato-Mix blender for 2 min. The homogenate waskept for 45 min at 4°C before centrifuging at 23 5OOgfor 45min. The supernatant was adjusted to pH8 bythe slow addition of 2M-triethanolamine and thendialysed against several changes of the above buffer(at pH8 and with the MgCl2 omitted) until the con-ductivity of the non-diffusible material was approx.2mQ'-. The dialysed extract was recentrifugedas above and the supernatant adsorbed on DEAE-cellulose (approx. 35g of moist exchanger/100ml ofsupernatant) previously equilibrated with the pH8buffer. The DEAE-cellulose was collected on aBuchner funnel, washed with the pH8 buffer undersuction and packed into a column (35cm x 4.5cm).The column was washed further until the A280 of theemerging eluate was less than 0.1 and it was thendeveloped with a linear gradient to 0.4M-KCI(in pH 8 buffer) over 2 litres. The active fractions,eluted between 0.04M- and 0.O8M-KCI, were concen-trated by ultrafiltration to between 50 and 100ml anddialysed against three changes of 1.5 litres each of10mM-potassium phosphate buffer, pH7.0, con-taining 50mM-glucose, 1mM-EDTA, 0.5mM-dithio-threitol and 5 % (v/v) glycerol on a rocking dialyserfor about 3 h. The dialysis residue was then applied toa hydroxyapatite column (15 cm x 2.5cm) equili-

1978

126

PURIFICATION OF THE HEXOKINASES

brated in the above buffer. After a washing of thecolumn with this buffer, a linear gradient to 0.6M-potassium phosphate (over 2 litres) was applied. Theactive fractions, eluted between 0.2M- and 0.25M-potassium phosphate, were pooled, concentrated byultrafiltration and stored at 40C. This isoenzymeproved to be very stable, retaining 98% activity overa period of 6 months. The enzyme was obtained in75% yield with a specific activity of 11.6 units/mg(representing some 1400-fold purification at thisstage) and was free of other hexokinase activity asjudged by electrophoresis in 0.5 % agarose (see below).This material was purified to homogeneity bychromatography on Sepharose-N-(8-amino-octan-oyl)-glucosamine as described in the Results section(Fig. 5).Preparation of rat hepatic hexokinase type IIIThe 50 rats were starved for 24h and then fed with

carbon tetrachloride (20%, v/v) in liquid paraffin(0.5ml/lOOg body wt.) as described by Ueda et al.(1975). (This treatment destroys the hepatic gluco-kinase and enhances the low-Km hexokinase activity.)The rats were killed after an additional 48 h withoutfood, and their livers were removed and homogenizedin 2vol. of 10mM-potassium phosphate buffer,pH7.0, containing 10mM-glucose, 5mM-EDTA,0.5mM-dithiothreitol and 10% (v/v) glycerol for2min in an Ato-Mix blender. After centrifugation(23 5OOg for 1 h), the supernatant was adsorbed onDEAE-cellulose (about lOg of moist gel/lOOml ofsupernatant) equilibrated with the extraction buffer,and the DEAE-cellulose was then collected on aBuchner funnel. It was washed wvith the above buffercontaining 0.1 M-KCI (1 litre) under suction and thenpacked into a column (20cm x 4.5cm), and thewashing was continued until the A280 of the emergingfractions fell below 0.2. Hexokinase activity wasstep-eluted by raising the KCl concentration in thisbuffer to 0.4M. The active fractions were pooled anddialysed against two changes of 1.5 litres each of50mM-potassium phosphate buffer, pH7.0, con-taining 10mM-glucose, 1 mM-EDTA, 1 mM-MgCl2,0.5 mM-dithiothreitol and 10% glycerol (v/v) forabout 3 h until the conductivity of the non-diffusiblematerial was down to 5 mQ.-'. This material was thenapplied to a DEAE-Sephadex column (20cm x4.5 cm) equilibrated with the above buffer butcontaining 0.1M-potassium phosphate, pH7.0. Thecolumn was washed further with this buffer until theA280 fell to 0.2 and was then eluted with a lineargradient to 0.5 M-potassium phosphate, pH 7.0,containing the above ingredients, over 2 litres. Somehexokinase activity appeared in the washings beforeapplication of the gradient and was shown to behexokinase type II by electrophoresis (see below).This was occasionally saved and treated as describedin the Results section (Fig. 6) should this isoenzyme be

Vol. 175

required. The hexokinase activity eluted under thegradient contained both type II and type III iso-enzymes, which could be most easily resolved byaffinity chromatography (Fig. 6). The type IIIenzyme was not very stable at this stage of thepreparation and the affinity-chromatography stepwas performed immediately. At this stage, thehexokinase activity was about 40% of that in theoriginal liver extract. (Some of the 'loss' was hexo-kinase type I, which does not adsorb on the DEAE-cellulose at pH 7.0, and type II, some of which is lostat each ion-exchange step, although there is still acertain amount in the preparation at this stage.) Thespecific activity was 12.1 units/mg, which representsa 2500-fold purification.The kinetic experiments described in the early part

of the Results section were performed with homo-geneous samples of rat hepatic glucokinase, ratskeletal-muscle hexokinase type II and yeast hexo-kinase and with samples of hexokinase types I and IIIafter the two preliminary purification steps describedabove. Rat kidney hexokinase type I is pure withrespect to hexokinase type at this stage, but hepatichexokinase type III was obtained by pooling frac-tions from the higher-ionic-strength side of the peakeluted from the DEAE-Sephadex column andchecking for type-homogeneity by electrophoresis in0.5% agarose (see below).

Assay of hexokinase activityHexokinase activity was measured at 300C in a

total volume of 0.75ml by the coupled assay methodof Parry & Walker (1966) as described by Storer &Cornish-Bowden (1974). Glucokinase was assayedroutinely at a glucose concentration of 100mM, andthe low-Km hexokinases were assayed at either0.5mM-glucose or 0.1 mM-glucose (type III). For theinhibition studies glucose was removed from theenzyme solutions by passage down a Sephadex G-25column [usually a 1 ml sample applied to a 20cm x1 cm column operated in 20mM-triethanolamine/HCI,pH7.0, containing 0.2M-KCI, 0.5mM-dithiothreitol,lmM-EDTA, 1mM-MgCl2 and 5% (v/v) glycerol]and the enzyme used straight away.

Glucose was removed from the glucose 6-phos-phate dehydrogenase by overnight dialysis against20mM-Tris/HCl buffer, pH7.5. The competitiveinhibition constants (Ki values) were determined bythe method of Dixon (1953) and the uncompetitivecomponent (K') by the method of Cornish-Bowden(1974, 1976).One unit of hexokinase activity is defined as that

which catalyses the formation of 1,umol of glucose6-phosphate/min at 30°C.

Preparation ofglucosamine derivativesAn amide linkage was synthesized between the

amino group of glucosamine and the carboxy group

127

C. L. WRIGHT, A. S. WARSY, M. J. HOLROYDE AND I. P. TRAYER

of the following amino acids: 3-aminopropionicacid, 4-aminobutyric acid, 6-aminohexanoic acid and8-amino-octanoic acid (Holroyde et al., 1976a). Thefinal products were not crystallized, but gave singlespots on high-voltage paper electrophoresis at pH6.5(mobilities relative to glucosamine: glucosamine,1.0; products, 0.81-0.87) and on ascending paperchromatography in two solvent systems (Holroydeet al., 1976a). Positive reactions were obtained withboth ninhydrin and AgNO3 (Holroyde et al., 1976a).After acid hydrolysis (6M-HCl, 20h at 105°C)chromatographic analysis on an amino acid analysershowed only the appropriate amino acid and glucos-amine in a molar ratio of 1:1. The structures shownin Fig. 1 were therefore assigned to the products.

Preparation of Sepharose adsorbentsThe ligands were attached to CNBr-activated

Sepharose 4B under the conditions described byTrayer et al. (1974). The ligands were coupled at aconcentration of 10-12,umol/g of gel and character-ized as described by Holroyde et al. (1976a). The gelswere diluted to the required ligand concentration bydilution with unsubstituted Sepharose 4B (Holroydeet al., 1976a). The adsorbents are designated ase.g., Sepharose-C3-GlcN for glucosamine linked toSepharose by a C3 spacer.

Electrophoretic methodsEnzyme purity and estimations of molecular

weight were determined by polyacrylamide-gelelectrophoresis in the presence of sodium dodecylsulphate (Weber & Osborn, 1969). To identifyisoenzymes in the early part of this work, electro-phoresis was carried out in the absence of sodiumdodecyl sulphate in polyacrylamide gels and the gelswere stained for enzymic activity (Grossman &Potter,1974), but most frequently electrophoresis in 0.5%agarose was used (Allen & Walker, 1976). The latterprocedure gave greatly improved separation.

Analytical methodsConductivity measurements were made on a Radio-

meter type CDM conductivity meter (Radiometer,Copenhagen, Denmark). Protein determinationswere made by the turbidimetric tannin micro methodof Mejbaum-Katzenellenbogen & Dobryszycka(1959), with bovine serum albumin as standard.Amino acid analyses were performed with a Beckmanmodel 120B amino acid analyser, with correction forlosses as described by Holroyde et al. (1976a).

ResultsInhibitory properties of the N-aminoacylglucosaminederivativesThe general structure of the four glucosamine

derivatives synthesized is shown in Fig. 1. The

CH20H

t/OH \H,OHHO

I ~ ~~~+NH-C-[CH2].-NH3

Fig. 1. Structure of the N-aminoacylglucosamine deri-vatives

The synthesis of the various ligands [where n = 2,N-(3-aminopropionyl)-2-amino-2-deoxy-D-glucopy-ranose (C3-GlcN), n = 3, N-(4-aminobutyryl)-2-amino-2-deoxy-D-glucopyranose (C4-GIcN), n = 5,N- (6 -aminohexanoyl) -2 -amino -2-deoxy - D -glUCO -pyranose (C6-GlcN), and n = 7, N-(8-amino-octanoyl) - 2 -amino - 2 - deoxy - D -glucopyranose (C8 -GlcN)] is described in the Materials and Methodssection.

competitive inhibition constant (K1) of each iso-enzyme for each ligand was determined first withimpure enzyme fractions [but still homogeneous withrespect to hexokinase type as judged by electro-phoresis (see the Materials and Methods section)]and later, as the findings were extended, the experi-ments were repeated with pure isoenzymes. All theresults were in good agreement, and in separateassays for ADP production (Storer & Cornish-Bowden, 1977) the glucosamine derivatives could notbe shown to be substrates. Typical inhibition plotsare shown in Fig. 2, where the Ki in free solution ofhepatic hexokinase type III for the C3-GlcN deri-vative is shown to be 5.9mM. The inhibition is purelycompetitive, as indicated by Fig. 2(b), where theparallel lines in the plot of [S]/v against [I] show thatno uncompetitive inhibition component is detectable,i.e. Ki approaches infinity. However, the inhibition ofhexokinase type II by all four glucosamine derivatives,of hexokinase type III by the C6-GlcN and C8-GlcNderivatives, and of yeast hexokinase by the C8-GlcNderivative showed mixed inhibition (Cornish-Bowden,1974); all other combinations of enzymes andinhibitors showed purely competitive inhibition, asillustrated in Fig. 2. In the cases where mixed inhibi-tion occurred, this was demonstrated in the plot of[S]/v against [I], where the lines at different glucoseconcentrations intersected to the left of the ordinate.The results of the determinations of competitive

inhibition constant (KI) obtained from the plot of1/v against [I] for all the enzymes are shown inFig. 3 plotted against the number of carbon atoms inthe aminoacyl chain of the glucosamine derivative.Results for N-acetylglucosamine, which is generally abetter inhibitor of all the isoenzymes than the amino-

1978

128

PURIFICATION OF THE HEXOKINASES

-6 -3 0 3 6 9 1 2

[C3-GlcN] (mM)Fig. 2. Inhibition of rat liver hexokinase type III byN-(3-aminopropionyl) -2-amino-2-deoxy-D -glucopyranose

(C3-GlcN) at various glucose concentrationsInitial rates (v) were measured as described in theMaterials and Methods section at a constant concen-tration of MgATP2- (7.5mM). (a) The competitivecomponent of the inhibition was investigated byplotting 1/v against inhibitor concentration (Dixon,1953). (b) The uncompetitive component of inhibitionwas investigated by plotting [glucose]/v againstinhibitor concentration (Cornish-Bowden, 1974).The parallel lines indicate that this component is notdetectable, i.e. Ki approaches infinity. Concentrationsof glucose were 0.1 mm (A), 0.05mM (e) and 0.025mM(o).

acyl derivatives, are shown for comparison. Theintroduction of an additional aminomethyl moiety tothe N-acetylglucosamine side chain (i.e. the C3-GlcNderivative) considerably lessens its effectiveness as aninhibitor, but, as the number of carbon atoms in thechains was increased, so was the effectiveness of thevarious ligands as free solution inhibitors.

Small-scale affinity-chromatography experimentsA direct correlation was found to exist between the

K1 of a particular derivative and its ability to act as an

affinity matrix when immobilized on Sepharose. In a

series of small-scale experiments, some of which are

illustrated in Fig. 4, it became apparent that, provided

Vol. 175

12

8

4

° t 3 4 5 6 7 8GIcNAc No. of C atoms in aminoacyl chain

Fig. 3. Inhibition ofthe mammalian hexokinase isoenzymesby the various N-aminoacylglucosamine derivatives in free

solutionThe K1 values, estimated by plotting 1/v against in-hibitor concentration as described in Fig. 2, areplotted against the number of C atoms in the N-aminoacyl side chain of the glucosamine derivativesdepicted in Fig. 1. (Number of C atoms = n + 1 inFig. 1.) 0, Hexokinase type I (rat kidney); *, hexo-kinase type II (rat muscle); l, hexokinase type III(rat liver); U, glucokinase (hexokinase type IV, ratliver). GlcNAc, N-acetylglucosamine, shown forcomparison. The broken line at Ki = 2mM representsthe upper limit of Ki from which a successful affinitycolumn can be made.

that the K1 in free solution (in this context it can betaken as the dissociation constant prevailing in thesystem) was < 2mm then an efficient affinity-chrom-atography system could be designed when theglucosamine derivatives were attached to Sepharose.Thus the only really effective inhibitor of rat kidney

hexokinase type I was the C8-GlcN derivative(Ki 1.3 mM) and only the Sepharose conjugate of thisderivative proved a suitable affinity matrix (Fig. 5).If the C6-GIcN derivative was attached to Sepharoseat very high ligand concentrations (12,umol/g ofSepharose) the hexokinase type I (Ki 2.75mM) waspartially retained on the column, but not sufficientlywell to allow this to be used in a purification scheme.At lower ligand concentrations (Fig. 4b) the iso-enzyme passes through unretarded. Similarly musclehexokinase type II will bind efficiently to any of theimmobilized glucosamine derivatives, but liver

E

129

C. L. WRIGHT, A. S. WARSY, M. J. HOLROYDE AND I. P. TRAYER

U

-.~._

4-*:E

._

0c)

x

1.0

0.5 Ca0

0

0 4 8 12 16 20Fraction no.

Fig. 4. Chromatography of the mammalian hexokinase isoenzymes on various Sepharose-N-aminoacylglucosaminederivatives

All columns were the same dimensions (10cm x 0.8cm) and were operated in 20mM-triethanolamine/HCI, pH7.0,containing l0mM-KCI, 4mM-EDTA, 7.5mm-MgCl2, I mM-dithiothreitol and 5%/ (v/v) glycerol at 20ml/h, and 2mlfractions were collected. Hexokinase (activity 2 units in 2-3 ml) was applied to each column. At the arrow, 0.5 M-glucosewas included in the developing buffer. (a) Composite of four individual column profiles obtained by applying each ofthe isoenzymes in turn to a Sepharose-C4-GlcN matrix (ligand concentration: 8,mol/g wet wt. of packed gel). (b)Composite of four individual column profiles obtained by applying each of the isoenzymes in turn to a Sepharose-C6-GIcN matrix (ligand concentration: 4pmol/g wet wt. of packed gel). (c) Glucokinase and hexokinase type II wereapplied to a Sepharose-C6-GlcN matrix (6,umol/g) and eluted by a linear gradient formed from 20ml of the columnbuffer and 201n1 of 1 M-glucose dissolved in this buffer. o, Hexokinase type I (rat kidney); e, hexokinase type II (ratmuscle); El, hexokinase type III (rat liver); *, glucokinase (hexokinase type IV, rat liver); , glucose concentration.

hexokinase type III and glucokinase are best chrom-atographed on Sepharose conjugates of eitherC6-GlcN or C8-GlcN. Furthermore, in situationswhere the K1 values oftwo isoenzymes for a particularligand differed, these enzymes could be separatedchromatographically by either adjusting the ligandconcentration attached to the gel or using a glucose(or sometimes even a KCl) gradient. For example,

hexokinase type II (Ki 0.4mM) and glucokinase(Ki 0.75mM) could be separated on Sepharose-C6-GIcN in this way (Fig. 4c).

It should be emphasized that attachment of eitherglucosamine (Chesher et al., 1973) or any of thevarious amino acids (or their aminoalkane analogues)directly to CNBr-activated Sepharose did not providematrices that showed any significant interaction with

1978

130

PURIFICATION OF THE HEXOKINASES

the isoenzymes. Nor did the amino acids and theaminoalkanes inhibit the hexokinase activities in freesolution.The importance of choosing the most effective

ligand concentration attached to the gel (and how toadjust this) when designing a large-scale affinity-matrix operation has been reported earlier (Holroydeet al., 1976a,b; Trayer et al., 1978; Trayer, 1978;Harvey etal., 1974). Furthermore, better purificationsof these isoenzymes are obtained if a ligand concen-tration is chosen so that the enzyme is bound suffi-ciently tightly to allow the column to be thoroughlywashed with as high a KCI concentration as possible,which, in addition to removing further unwantedprotein, 'loosens' the enzyme-immobilized ligandinteraction. Elution of the required enzyme can thenbe effected specifically in a small volume by includingglucose in the developing buffer.

Large-scale preparation of hexokinase types I and III

The large-scale purifications of hexokinase type Ifrom rat kidney (Fig. 5) and hepatic hexokinase typeIII (Fig. 6) were both carried out on a Sepharose-C8-

GlcN affinity matrix; neither enzyme had beenpreviously purified to homogeneity. After priorchromatography of the kidney extracts on DEAE-cellulose and hydroxyapatite (see the Materials andMethods section), the fractions containing hexokinasetype I were applied to a Sepharose-C8-GIcN column;which was operated as described in Fig. 5. Washing ofthe column with 70mM-KCl removed -unwantedprotein, and the enzyme was specifically eluted byinclusion of 200mM-glucose in the developing buffer.At this stage the hexokinase fraction gave one majorand one minor band on polyacrylamide-gel electro-phoresis in sodium dodecyl sulphate. A furtherpassage down the affinity column removed the minorcontaminant, leaving a band migrating in a positionequivalent to-a mol.wt. of 96000. Identical molecularweights have been recorded for hexokinase type Ifrom heart (Easterby & O'Brien, 1973; Easterby,1975) and brain (Chou & Wilson, 1972; Chakrabarti& Kenkare, 1974). The firnal product had a specificactivity of about 180 units/mg, which represented a21 500-fold purification. The two affinity-chromato-graphy steps were performed in greater than 85%yield.

0.6

0.4

0.2

_

0.

C.)

E

06-c)

0

x

0 50 100 150

Fraction no.Fig. 5. Purification of rat kidney hexokinase type I on Sepharose-C8-GlcN

The extracts of 400g of rat kidneys were partially purified as described in the Materials and Methods section. Thesample was dialysed against 20mM-triethanolamine/HCl, pH 7.0, containing lOmM-KCI, 4mM-EDTA, 0.5 mM-dithio-threitol and 5% (v/v) glycerol and applied directly to the affinity column (25cm x 3 cm, adjusted to 5,umol of ligandper g wet wt. of packed gel as described under 'Preparation of Sepharose adsorbents') equilibrated in the same buffer,The column w4s operated at 40ml/h, and 6ml fractions were collected. At arrow A the KC1 concentration in the bufferwas increased to 70mM, and at arrow B 200mM-glucose was also included. e, A280; o, hexokinase activity.

'Vol. 175

131

0

C. L. WRIGHT, A. S. WARSY, M. J. HOLROYDE AND I. P. TRAYER

1-1

E 1.28.

0.

I._

0.

v0

<,- 0.4_4

20 40

Fraction no.60

0.2

0.1 X

0

Fig. 6. Separation and purification of hexokinase types II and IIIfrom rat liver on Sepharose-C8-GIcNThe extracts of the livers of 50 carbon tetrachloride-treated rats were partially purified by ion-exchange chromato-graphy as described in the Materials and Methods section. The sample, containing only hexokinase types II and IIIactivities, was dialysed against the buffer described in Fig. 6 and applied directly to the affinity column (20cm x 3 cm,adjusted to 3.85,pmol of ligand/g of gel as described under 'Preparation of Sepharose adsorbents') equilibrated tothis buffer. The column was washed with this buffer until the A280 of the emerging solution was * 0.1 (not shown).At arrow A the KCl concentration of the buffer was increased to 140mM, and at arrow B 50mM-glucose was added tothis. The column was operated at 40ml/h, and 4ml fractions were collected. e, A280; o, hexokinase activity. Electro-phoretic analysis (Allen & Walker, 1976) indicated that the protein eluted by the 140mM-KCI alone contained onlythe hexokinase type II isoenzyme together with other protein material, whereas the KCI/glucose eluate containedhomogeneous hexokinase type III.

Hepatic hexokinase type III could have beenpurified to homogeneity on either the C6-GlcN or theC8-GIcN derivatives attached to Sepharose; theC8-GlcN derivative was chosen since hexokinase typeII is a major contaminant after the two ion-exchangesteps (see the Materials and Methods section) andthis glucosamine derivative best distinguishes be-tween them (see Fig. 3). Fig. 6 shows the results ofapplying the partially purified hexokinase type IIIpreparation to Sepharose-C8-GlcN (3.85pmol ofligand/g of gel). After washing of the column withthe application buffer (containing lOmM-KCI) (notshown), the column was developed with a buffercontaining 140mM-KCl. This eluted hexokinase typeII quantitatively together with unwanted protein. Thetype III enzyme was then eluted specifically with50mM-glucose. If required, the hexokinase type IIeluted with the 140mM-salt could be purified by

passage directly down a Sepharose-C6-GlcN deri-vative (Holroyde & Trayer, 1976). The enzyme wouldbind in this buffer and could be subsequently eluted byglucose.

Hexokinase type III was found to be very unstablein our hands even on the addition to the buffers ofglucose, dithiothreitol and glycerol, ingredients thathelp stabilize glucokinase (Holroyde et al., 1976b).The freshly eluted enzyme had a specific activity ofabout 200 units/mg and gave a single migratingspecies on polyacrylamide gels in the presence ofsodium dodecyl sulphate (mol.wt. 98000). Thepreparation appeared to contain trace amounts ofproteinases, since the band migrating at a positionequivalent to a mol.wt. of 98000 gradually dis-appeared on standing over a 2-3-week period to bereplaced by faster-migrating bands with a concomi-tant loss of enzymic activity. We have not observed

1978

132

PURIFICATION OF THE HEXOKINASES

24

16

i

No. of C atoms in aminoacyl chain

1.0T0

In

-

O. 5 *-

Cd(U

°xe

O0 3 6 9 12 15

Fraction no.

Fig. 7. Affinity-chromatography studies on yeast hexo-kinase

(a) Inhibition of yeast hexokinase by the variousN-aminoacylglucosamines in free solution. Theexperiments were performed as described in Figs.2 and 3 with a pure sample of yeast hexokinase type B(see the Materials and Methods section). Virtuallyidentical results were obtained with the type-A iso-enzyme. (b) Crude (see the Materials and Methodssection) yeast hexokinase (0.1 ml; AA340/min = 7.0)was applied to a column of Sepharose-C8-GlcN(5cm x 0.9cm, containing 5pmol of glucosamineligand/g of gel) operated in the buffer described inFig. 6. At arrow A the KCI concentration wasincreased to 20mM, and at arrow B 200mM-glucosewas added to this. The column was operated at20ml/h, and 2ml fractions were collected. *, A280;o, hexokinase activity.

this problem with any of our other hexokinasepreparations.

Studies on yeast hexokinaseThe usefulness of the kinetic data in helping to

predict affinity-chromatography behaviour was veryobvious in our studies on yeast hexokinase. Our early

Vol. 175

attempts to chromatograph this enzyme on Sepha-rose-C6-GlcN (Chesher et al., 1973) were unsuc-cessful, an observation confirmed by others (E. A.Barnard, personal communication). The K1 values forthe C3-GlcN, C4-GlcN and C6-GlcN derivativesincreased rapidly as the length of the spacer moleculewas increased (Fig. 7a). However, the addition oftwomethylene groups to the spacer molecule had adramatic effect on the inhibitory properties of thisglucosamine derivative (cf. Ki 24.2mM for C6-GlcNand Ki 0.9mM for C8-GlcN). It was therefore possibleto chromatograph successfully a crude yeast extracton Sepharose-C8-GlcN (Fig. 7b) and obtain a20-25-fold purification. Although this is only shownon a small scale here, our experience with the mam-malian hexokinases suggests that it is also possible toscale up this method.

Discussion

For the first time it has been shown possible topurify to homogeneity all four mammalian hexo-kinases by a combination of conventional andaffinity techniques. The key to each purification wasthe use of a glucosamine affinity matrix in which ahigh degree of specificity for a particular hexokinaseisoenzyme could be introduced by either varying thelength of the spacer molecule and/or varying theligand concentration coupled to the gel. In principlethe method should be applicable to the purificationof any hexokinase inhibited by N-acetylglucosamine.It was particularly noteworthy that the choices ofmatrix, ligand concentration and operating con-ditions were largely predicted from free solutionkinetic data. These results explain our earlier puzzlingobservations that, when liver extracts were applied toSepharose-C6-GlcN, only glucokinase was boundand low-Km hexokinase activity was unretarded(Chesher et al., 1973). Under the conditions of theseexperiments, hexokinase type I, the major low-Kmhexokinase in liver, would not be expected to bind tothis affinity matrix. In this respect the report byRijksen & Staal (1976) on the purification of humanerythrocyte hexokinase type I is particularly note-worthy. These authors used an immobilized glucos-amine derivative (Sepharose-C6-GlcN in our nomen-clature) as the final step of their purification scheme.They showed that, although the enzyme did not bindvery efficiently, it was sufficiently retarded to achievethe required purification. Although they did notreport a ligand concentration attached to the gel,their mode of preparation suggests that it was atleast 10,umol/g of gel, conditions where we observeda retardation (but not binding) of the rat kidneyhexokinase type I. Our experience suggests that animmobilized C8-GlcN conjugate would provide amore efficient affinity matrix. There are considerable

133

0

134 C. L. WRIGHT, A. S. WARSY, M. J. HOLROYDE AND I. P. TRAYER

differences, however, between the two type I enzymes;the rat hexokinase is more active (specific activity of180 units/mg at 30°C compared with 25-50 units/mgat 37°C for the human enzyme) and smaller (mol.wt.96000 as opposed to 132000).The exact nature of the interaction of enzyme with

immobilized ligand is difficult to assess. Increasingthe length of the spacer molecule, in general terms,increases the strength of binding of the enzymes tothe matrices, but it would be naive to suggest anincreased steric availability of the glucosamine asbeing responsible for this. Indeed, the fact thathexokinase type II will bind to the immobilizedconjugates of all the N-aminoacylglucosamine de-rivatives tried and, in more general terms, the strongcorrelation between the kinetic and chromatographicdata argue for an altogethermore specific explanation.Nevertheless it was somewhat unexpected that theK1 values that such a closely related family ofenzymesexhibit towards the N-aminoacylglucosamine deri-vatives should differ so markedly as the carbon chainlength is increased. This must reflect differences intheir active-site conformations, and it is probable thatthe binding of the ligands to the enzymes involves acombination of steric, hydrophobic and electrostaticinteractions. In this respect it should be emphasizedthat the positively charged amino function of theN-aminoacylglucosamine derivatives under the con-ditions of the kinetic studies would be retained in theisQurea generated when these derivatives werecoupled to CNBr-activated agarose (Wilchek,1974).

It did not escape our notice that in most caseswhere the Ki value was less than 2mM these valueswere obtained from mixed-inhibition systems (asopposed to purely competitive), although only thecompetitive component was used in the constructionof Fig. 3. Since this increased inhibition generallyoccurs with the longer polymethylene chains, it couldbe interpreted as an increased hydrophobic com-ponent in the enzyme-ligand interaction that mayhelp to reinforce the affinity effects (Barry & O'Carra,1973). These affinity columns were always operatedwith a very high yield ofhexokinase, usually90-100 %.If the total yield of applied protein was calculated,however, it was always much less than this and variedwith the purity of the applied sample. These effectshave been noticed in our previous studies (Holroydeet al., 1976a,b). In addition, the 'capacity' of theaffinity columns for these enzymes also depended onthe purity ofthe applied sample: the purer the sample,the higher the 'capacity'. This phenomenon can beexpected in low-affinity systems where non-specificadsorption of protein can readily interfere withspecific interactions. Thus the more impure thefraction to be chromatographed the larger thecolumn required (see also Paulson et al., 1977;Schwyzer & Hill, 1977). The low-affinity systems

described here do have the one advantage that adjust-ment of ligand concentration can dramatically affectthe subsequent chromatography. In very-high-affinity systems such an adjustment of ligand concen-tration is likely to run into steric/accessibilityproblems.

It has been suggested that the important quantita-tive differences in the behaviour of the hexokinaseisoenzymes towards these various affinity matricesreflect (slight) differences in their active-site confor-mations. By this criterion, the active site of the yeastenzyme appears to differ most from the others, aprediction borne out by affinity-labelling experimentswith N-bromoacetylglucosamine (cf. Otieno et al.,1975, 1977; Connolly & Trayer, 1978).

The authors are grateful to Professor D. G. Walker andDr. A. Cornish-Bowden for helpful discussions and to theMedical Research Council for financial support. M. J. H.was the recipient of a Science Research Council TrainingAward.

References

Allen, M. B. & Walker, D. G. (1976) Biochem. Soc. Trans.4,1057

Barry, S. & O'Carra, P. (1973) Biochem. J. 135, 595-607Chakrabarti, U. & Kenkare, U. W. (1974) J. BioL. Chem.

249, 5984-5988Chesher, J. M. E., Trayer, I. P. & Walker, D. G. (1973)

Biochem. Soc. Trans. 1, 876Chou, A. C. & Wilson, J. E. (1972) Arch. Biochem. Biophys.

151, 48-55Connolly, B. A. & Trayer, I. P. (1978) Biochem. Soc. Trans.

6, 586-587Cornish-Bowden, A. (1974) Biochem. J. 137, 143-144Cornish-Bowden, A. (1976) Principles ofEnzyme Kinetics,

pp. 92-94, Butterworths, LondonDixon, M. (1953) Biochem. J. 55, 170-171Easterby, J. S. (1975) Eur. J. Biochem. 58, 231-235Easterby, J. S. & O'Brien, M. J. (1973) Eur. J. Biochem.38,201-211

Grossman, S. H. & Potter, V. R. (1974) Anal. Biochem. 59,54-62

Harvey, M. J., Lowe, C. R., Craven, D. B. & Dean,P. D. G. (1974) Eur. J. Biochem. 41, 335-340

Holroyde, M. J. & Trayer, I. P. (1974) Biochem. Soc.Trans. 2, 1310-1311

Holroyde, M. J. & Trayer, I. P. (1976) FEBS Lett. 62,215-219

Holroyde, M. J., Chesher, J. M. E., Trayer, I. P. & Walker,D. G. (1976a) Biochem. J. 153, 351-361

Holroyde, M. J., Allen, M. B., Storer, A. C., Warsy, A. S.,Chesher, J. M. E., Trayer, I. P., Cornish-Bowden, A. &Walker, D. G. (1976b) Biochem. J. 153, 363-373

Mejbaum-Katzenellenbogen, W. & Dobryszycka, W. M.(1959) Clin. Chim. Acta 4, 515-522

Otieno, S., Bhargava, A. K., Barnard, E. A. & Ramel,A. H. (1975) Biochemistry 14,2403-2410

Otieno, S., Bhargava, A. K., Serelis, D. & Barnard, E. A.(1977) Biochemistry 16, 4249-4255

1978

PURIFICATION OF THE HEXOKINASES 135

Parry, M. J. & Walker, D. G. (1966) Biochem. J. 99,266-274

Paulson, J. C., Beranek, W. E. & Hill, R. L. (1977) J. Biol.Chem. 252,2356-2362

Ramel, A. H., Rustum, Y. M., Jones, J. G. & BarnardE. A. (1971) Biochemistry 10, 3499-3508

Rijksen, G. & Staal, G. E. J. (1976) Biochim. Biophys. Acta445,330-341

Schwartz, G. P. & Basford, R. E. (1967) Biochemistry 6,1070-1079

Schwyzer, M. & Hill, R. L. (1977) J. Biol. Chem. 252,2338-2345

Storer, A. C. & Cornish-Bowden, A. (1974) Biochem. J.141, 205-209

Storer, A. C. & Cornish-Bowden, A. (1976) Biochem. J.159,7-14

Storer, A. C. & Cornish-Bowden, A. (1977) Biochem. J.165,61-69

Trayer, I. P. (1978) in Techniques in the Life Sciences(Tipton, K., ed.), vol. 1, North-Holland, Amsterdam,in the press

Trayer, I. P., Trayer, H. R., Small, D. A. P. & Bottomley,R. C. (1974) Biochem. J. 139, 609-623

Trayer, I. P., Holroyde, M. J., Small, D. A. P., Trayer,H. R. & Wright, C. L. (1978) in Chromatography ofSynthetic and Bio-polymers (Epton, R., ed.), vol. 2,pp. 159-178, Ellis Horwood, Chichester

Ueda, M., Taketa, K. & Kosaka, K. (1975) Clin. Chim.Acta 60, 77-84

Walker, D. G. (1966) Essays Biochem. 2, 33-67Weber, K. & Osborn, M. (1969) J. Biol. Chem. 244,4406-4412

Wilchek, M. (1974) in Immobilized Biochemicals andAffinity Chromatography (Dunlap, R. B., ed.), pp. 15-31,Plenum Press, New York

Wright, C. L., Holroyde, M. J., Warsy, A. S. & Trayer,I. P. (1976) Biochem. Soc. Trans. 4, 1052-1053

Vol. 175