Transgalactosylation by thermostable glycosidases … and 2-propanol, respectively. Therefore, the...

Eur. J. Biochem. 267, 5055±5066 (2000) q FEBS 2000

Transgalactosylation by thermostable b-glycosidases fromPyrococcus furiosus and Sulfolobus solfataricusBinding interactions of nucleophiles with the galactosylated enzyme intermediate make

major contributions to the formation of new b-glycosides during lactose conversion

Inge Petzelbauer1, Andreas Reiter2, Barbara Splechtna1, Paul Kosma2 and Bernd Nidetzky1

1Division of Biochemical Engineering, Institute of Food Technology and 2Institute of Chemistry, UniversitaÈt fuÈr Bodenkultur Wien, Vienna, Austria

The hyperthermostable b-glycosidases from the Archaea Sulfolobus solfataricus (SsbGly) and Pyrococcus

furiosus (CelB) hydrolyse b-glycosides of d-glucose or d-galactose with relaxed specificities pertaining to the

nature of the leaving group and the glycosidic linkage. To determine how specificity is manifested under

conditions of kinetically controlled transgalactosylation, the major transfer products formed during the hydrolysis

of lactose by these enzymes have been identified, and their appearance and degradation have been determined in

dependence of the degree of substrate conversion. CelB and SsbGly show a marked preference for making new

b(1!3) and b(1!6) glycosidic bonds by intermolecular as well as intramolecular transfer reactions. The

intramolecular galactosyl transfer of CelB, relative to glycosidic-bond cleavage and release of glucose, is about

2.2 times that of SsbGly and yields b-d-Galp-(1!6)-d-Glc and b-d-Galp-(1!3)-d-Glc in a molar ratio of

< 1 : 2. The partitioning of galactosylated SsbGly between reaction with sugars [kNu (m21´s21)] and reaction

with water [kwater (s21)] is about twice that of CelB. It gives a mixture of linear b-d-glycosides, chiefly

trisaccharides at early reaction times, in which the prevailing new glycosidic bonds are b(1!6) and b(1!3) for

the reactions catalysed by SsbGly and CelB, respectively. The accumulation of b-d-Galp-(1!6)-d-Glc at the end

of lactose hydrolysis reflects a 3±10-fold specificity of both enzymes for the hydrolysis of b(1!3) over b(1!6)

linked glucosides. Galactosyl transfer from SsbGly or CelB to d-glucose occurs with partitioning ratios,

kNu /kwater, which are seven and . 170 times those for the reactions of the galactosylated enzymes with

1-propanol and 2-propanol, respectively. Therefore, the binding interactions with nucleophiles contribute chiefly

to formation of new b-glycosides during lactose conversion. Likewise, noncovalent interactions with the glucose

leaving group govern the catalytic efficiencies for the hydrolysis of lactose by both enzymes. They are almost

fully expressed in the rate-limiting first-order rate constant for the galactosyl transfer from the substrate to the

enzyme and lead to a positive deviation by < 2.5 log10 units from structure±reactivity correlations based on the

pKa of the leaving group.

Keywords: b-glycosidase; oligosaccharides; Pyrococcus; specificity; Sulfolobus.

Enzymes that catalyse the hydrolysis of glycosides withretaining stereochemistry at the anomeric carbon are thoughtto operate by a two-step reaction mechanism. The first irre-versible step of the mechanism involves the formation of aconfigurationally inverted, covalent glycosyl±enzyme inter-mediate followed by the departure of the leaving group. Inthe second step, this intermediate is hydrolysed, again withinversion, thus completing the reaction with retention ofconfiguration (reviewed in [1±3]). In a well known variant

of the second step of the glycosylase mechanism, theglycosylated enzyme may be intercepted by nucleophilesother than water and form so-called transglycosylation products.These products are typical kinetic intermediates which will behydrolysed by the enzyme if allowed to react for sufficientlylong incubation times [1,4].

A classical example of the occurrence of transglycosylationis the enzymatic hydrolysis of lactose in which the transfer of agalactose residue to the substrate or to products is commonlyobserved [4,5]. Transgalactosylation is described to involveboth intermolecular and intramolecular reactions. Intramolecularor direct galactosyl transfer to d-Glc yields regio-isomers oflactose and is thought to occur after glycosidic bond cleavage inlactose has taken place, but before the intercepting d-Glc hasdiffused away into solution (Scheme 1). That is the mechanismby which allolactose is formed by b-galactosidases even in theabsence of significant concentrations of free d-Glc [4±6].Intermolecular or indirect transgalactosylation presents theroute by which disaccharides, trisaccharides and tetrasaccharides,and eventually longer oligosaccharides are produced from lactose[4±6].

Correspondence to B. Nidetzky, Institut fuÈr Lebensmitteltechnologie,

UniversitaÈt fuÈr Bodenkultur, Muthgasse 18, A-1190 Vienna, Austria.

Fax: 143 1 36006 6251, Tel.: 143 1 36006 6274,

E-mail: [email protected]

Abbreviations: SsbGly, b-glycosidase from Sulfolobus solfataricus;

CelB, b-glycosidase from Pyrococcus furiosus; oNPGlc,

4-nitro-phenyl-b-d-glucoside; oNPGal, 4-nitro-phenyl-b-d-galactoside;

CE, capillary electrophoresis; HPAEC, high-performance anion-exchange

chromatography; DSS, 3-(trimethylsilyl)-1-propanesulfonic acid; TMS,

tetramethylsilane.

(Received 8 May 2000, accepted 12 June 2000)

The b-glycosidases from the Archaea Sulfolobus solfataricus(SsbGly) [7,8] and Pyrococcus furiosus (CelB) [9,10] arehyperthermostable enzymes that hydrolyse b-glycosides com-posed of two or more glycosyl units, with very broadspecificities regarding the sugar bound to subsite 1 1 (fornomenclature, see [11]), the cleaved glycosidic linkage and thenature of the leaving group [10,12]. Lactose, b-d-Galp-(1!4)-d-Glc, is a reasonably good substrate of SsbGly and CelB [13].Its complete hydrolysis into d-Gal and d-Glc and its conversioninto galactose-rich oligosaccharides (GalOS) at high tempera-tures present possible technological applications of the two`thermozymes' [13±15].

The aim of this work was to understand better how thecatalytic efficiencies of SsbGly and CelB for the hydrolysis ofdifferent b-glycosides are manifested during the process oflactose conversion, i.e. under conditions of kinetically con-trolled transgalactosylation. It is known that the distribution ofglycosidic linkages in GalOS produced from lactose by usingdifferent b-glycosidases is not uniform, and the spectrum ofproducts obtained is critically dependent on the source of theenzyme and changes with increasing extent of substratedepletion (reviewed in [4]). However, a causal relationshipbetween the specificity for the hydrolysis of b-glycosides, thestructures of the transgalactosylation products, and the forma-tion and degradation of GalOS during lactose conversion hasrarely been studied in detail [4±6]. Generally, the formation oftransglycosylation products by b-glycosidases is thought toreflect the interplay of binding interactions between thenucleophile and the active site of the enzyme, and the intrinsicchemical reactivity of the intercepting nucleophile [1,2]. We areinterested to distinguish between these two possibilities for thetransfer reactions catalysed by SsbGly and CelB.

We report here the identification of the major GalOSproduced by SsbGly and CelB during the conversion oflactose, and show the characteristic profiles of concentration vs.reaction time for each main by-product. A number of nucleo-philic acceptors differing in noncovalent-bonding capabilitywith the enzyme have been used to determine how galacto-sylated (or glucosylated) CelB and SsbGly partition betweenthe reaction with one of these acceptors, and with water. Theresults suggest a major role of nucleophile binding during theintermolecular glycosyl transfer catalysed by CelB and SsbGly.Likewise, binding interactions with the leaving group governthe catalytic efficiencies for the hydrolysis of b-galactosidescatalysed by SsbGly and CelB. Upon transgalactosylation, newglycosidic linkages are formed with a preference that clearlyreflects the specificity constants for the hydrolysis of thecorresponding glycosidic bonds. The intramolecular trans-galactosidase reaction with lactose is shown to contribute tothe GalOS produced by both enzymes, however, especiallyCelB. The results provide the basis for understanding thedifferences in the product spectrum obtained during lactoseconversion by the thermostable b-glycosidases.

M A T E R I A L S A N D M E T H O D S

Materials

o-nitrophenyl-b-d-glucopyranoside (oNPGlc), o-nitrophenyl-b-d-galactopyranoside (oNPGal), 6-O-b-d-galactopyranosyl-d-galactose, and methyl-b-d-galactoside were from Sigma. The4-O-b-d-galactopyranosyl-d-galactose and 3-O-b-d-galacto-pyranosyl-d-galactose were obtained as a mixture fromMegazyme (Sydney, Australia) and used after further purifica-tion. Allolactose (6-O-b-d-galactopyranosyl-d-glucose) was akind gift of S. Riva (CNR, Milan, Italy). The 3-O-b-d-galactopyranosyl-d-glucose was synthesized (A. Reiter &P. Kosma, unpublished data) by Ag-triflate promoted couplingof 2,3,4,6-tetra-O-acetyl-d-galactopyranosyl bromide with1,2; 5,6-di-isopropylidene-a-d-glucofuranose followed bystandard deprotection. Laminaribiose (3-O-b-d-glucopyranosyl-d-glucose), gentiobiose (6-O-b-d-glucopyranosyl-d-glucose),cellobiose (4-O-b-d-glucopyranosyl-d-glucose) and sophorose(2-O-b-d-glucopyranosyl-d-glucose) were from Sigma.

Enzymes

The recombinant b-glycosidases from S. solfataricus and P.furiosus were overexpressed in Escherichia coli [16,17] andpurified by: (a) the thermoprecipitation of mesophilic bacterialprotein at 80 8C for 30 min; and (b) anion exchange chromato-graphy on a MonoQ column (Amersham-Pharmacia). Alcoholoxidase (from Pichia pastoris) and galactose dehydrogenasewere from Sigma and used without further purification.

Assays

The measurement of b-galactosidase activity using lactoseor oNPGal as the substrates was carried out as describedpreviously [13]. d-Glc was routinely determined by anenzymatic assay based on using glucose oxidase and peroxidase[18]. d-Gal can interfere with this assay because of the minutebut significant activity of glucose oxidase for oxidizing d-Gal.Therefore, when the concentration of d-Gal exceeded by< 500-fold the expected concentration of d-Glc in the sample,an alternative enzymatic assay was used; it is based on the useof hexokinase and glucose 6-phosphate dehydrogenase, andthe NADH produced on oxidation of glucose 6-phosphate(Glc 6-P) is usually measured [18]. However, the absorbance at340 nm of the commercial oNPGlc hampers the spectro-photometrical measurement of the produced NADH at thiswavelength. Therefore, by using a second reaction in whichNADH reduces 3-(4,5-dimethylthiazolyl-2)-2,5-diphenyltetra-zolium bromide when the electron carrier 5-methylphenaziniummethyl sulphate is present, the released NADH could bedetermined colorimetrically at 550 nm. d-Gal was measuredby using an enzymatic assay with NAD-dependent galactosedehydrogenase.

Reaction with methyl-b-D-galactoside

The reaction of the b-glycosidases with methyl-b-d-galactosidewas studied discontinously at 80 8C using screw-capped tubesto prevent evaporation. Samples were taken after a 10-minincubation, and the methanol released was quantified byenzymatic oxidation catalysed by alcohol oxidase. It wasensured that the depletion of substrate was , 8% in all assaysand, therefore, the formation of methanol with time was linear.

Scheme 1. The mechanism of lactose hydrolysis by b-galactosidases

and b-glycosidases, according to [4]. The formation of transgalactosyl-

ation products by intramolecular and intermolecular reactions is shown.

E, Enzyme; Lac, lactose; Nu, nucleophile.

5056 I. Petzelbauer et al. (Eur. J. Biochem. 267) q FEBS 2000

To a reaction mix (1 mL), containing 2 mm 2,2 0-azino-bis(3-ethylbenzthiazoline-6-sulfonic acid) in 100 mm potassium phos-phate buffer pH 7.5, were added peroxidase (75 U mL21; 10 mLadded) (Sigma) and alcohol oxidase (100 U mL21; 10 mLadded). After preincubation of the enzymes for 2±3 min at25 8C, the reaction was started by adding 100 mL of thesample. The absorbance at 405 nm was recorded continuouslyup to < 25 min, and calibration was with methanol standards.As the commercial methyl-b-d-galactoside contains contami-nating methanol, it was necessary to record a blank for eachsubstrate concentration and values reported are corrected forthese blank readings. A control reaction lacking methanol wasalso recorded, and corrections were made accordingly.

Transferase assay

Initial velocities were determined in 50 mm sodium phosphatebuffer pH 6.5, at 80 8C using oNPGal or oNPGlc as the sub-strate. The rates of formation of oNP (voNP) and galactose (vGal)or glucose (vGlc) were measured in the absence and presence ofdifferent acceptors [13]. The concentration of the substratevaried between 3 and 7.5 mm. The acceptor concentrationswere as follows: 1-propanol (0.33±1.67 m), 2-propanol(0.33±3.3 m), d-Gal (0.05±0.4 m) and d-fucose (0.05±0.7 m).d-Glc and d-Gal were determined as described in Assays.Transfer constants were determined from a plot of velocityratio, voNP /vGal or voNP /vGlc vs. acceptor concentration. Calcu-lation of transfer constants requires that there be a lineardependence of the velocity ratio on the acceptor concentrationand that the maximum concentration of the acceptor inhibit therelease of oNP by , 20% compared with the control lacking theacceptor.

Lactose hydrolysis and production of the GalOS mixture

Hydrolysis of lactose was carried out at 70 8C in batch-wisereaction, using a lactose concentration of 0.8 m dissolved in20 mm sodium citrate buffer pH 5.5. Screw-capped bottles witha working volume of 50 mL were used, and agitation was at400 r.p.m. in a rotary shaker (Infors, Bottmingen, Switzerland).The enzyme concentration was 20 U´mL21, based on the assaywith lactose as the substrate [13]. Samples (0.5 mL) were takenevery hour during the reaction, ultrafiltered to remove theenzyme and used for further analyses. When the depletion oflactose was < 80%, the remaining reaction volume waslyophilized.

Partial purification of GalOS produced by SsbGly

The lyophilisate (200 mg sugar) was dissolved in < 1 mL of a1 : 1 mixture of water and EtOAc/EtOH/H2O (7 : 2 : 1).Partial fractionation of sugar components was achieved bycolumn chromatography (10 � 1.5 cm) using silica gel 60(0.040±0.063 mm; Baker, the Netherlands) and elution with400 mL EtOAc/EtOH/H2O (7 : 2 : 1), 180 mL EtOAc/EtOH/H2O (5 : 3 : 1), and 120 mL EtOAc/EtOH/H2O (3 : 3 : 1).Monosaccharides were eliminated by this procedure. Pooledfractions containing the main by-products were evaporated todryness, redissolved in water and purified further by preparativeTLC (silica gel 60 F254, Merck, Darmstadt, Germany) using2-methyl-1-propanol/pyridine/water (6 : 4 : 3) as the eluent.Single bands were cut out from the unstained plates andextracted with 10 mL water. These samples were centrifugedfor 1 h at 4 8C to remove remaining silica gel, lyophilized again

and redissolved in water, and used for analysis by capillaryelectrophoresis.

Isolation of GalOS produced by CelB

The major GalOS were purified in a first step by columnchromatography on silica gel (46 � 4 cm) using methanol/chloroform/H2O (10 : 10 : 3) as the eluent. The main fractionswere O-acetylated by using a standard protocol (pyridine,Ac2O) and separated in a second step by column chromato-graphy (30 � 1 cm) using toluene/EtOAc/EtOH (10 : 10 : 1)as the eluent. Fractions of 1 mL were collected and analysed byTLC. The fractions which appeared to be pure were charac-terized by NMR. The pure compounds were de-O-acetylatedwith 0.1 m methanolic NaOMe, characterized by NMR andused as authentic standards for further analyses.

Sugar analyses

Capillary electrophoresis (CE). This was carried out on aHP-3D-CE unit (Agilent Technologies, Waldbronn, Germany)with a built-in diode array detector, using a procedure ofderivatization and separation modified from a method describedpreviously [19]. Five mL (100 nmol) saccharides were driedunder reduced pressure and derivatized with 20 mL aminopyri-dine solution (1 g 2-aminopyridin in 470 mL acetic acid and600 mL methanol) and incubated for 15 min at 90 8C. Afterevaporating the excess reagent under a stream of nitrogen(30 min, 60 8C), 25 mL borohydride solution (59 mg NaCNBH3

in 1 mL of 30% acetic acid) were added and the sample wasincubated for 30 min at 90 8C. The sample was dried underreduced pressure and redissolved in 500 mL water. The para-meters for capillary electrophoresis were as follows: fusedsilica capillary with an effective column length of 40 cm;running buffer, 100 mm phosphate pH 2.5; temperature, 30 8C;voltage, 22 kV; capillary preconditioning, 0.1 m NaOH for2 min, buffer for 3 min; pressure injection, 250 mbar s; detec-tion, 240 nm (10 nm bandwith). The system was calibratedusing pyridylaminated isomalto-oligosaccharides from partiallyhydrolysed dextran [20]. The results reported regarding theconversion of lactose were verified independently by using asecond method of separation and quantification [21]. Thismethod uses ethyl p-aminobenzoate for derivatization of thesaccharides and a running buffer of 0.2 m sodium borate,pH 10.5.

High-performance anion-exchange chromatography (HPAEC).This provides a better resolution of individual carbohydratecomponents than is achievable by CE. Therefore, it was used toobtain a complete, although semi-quantitative, picture on thespectrum of GalOS produced during lactose conversion. Thefiltered samples were analysed on a Dionex model AI 450system using a Carbopac PA-1 (250 � 4 mm i.d.) columnequipped with a PA-1 guard column (50 � 4 mm i.d.). Amodel PAD 2 pulsed amperometric detector was used. Thepulse potentials and durations were set at E1 � 0.05 V(t1 � 480 ms), E2 � 0.6 V (t2 � 300 ms) and E3 � 2 0.6 V(t3 � 240 ms). Samples were eluted in stepwise manner with aconstant flow rate of 1 mL´min21, by using isocratic elution for65 min with 10 mm sodium hydroxide, and gradient elution fora further 25 min with sodium hydroxide in a concentrationrange of 10±100 mm.

NMR-measurements. Spectra were recorded at 300 K in anonspinning 5 mm tube at 300.13 MHz for 1H and at

q FEBS 2000 Galactosyl transfer by thermostable b-glycosidases (Eur. J. Biochem. 267) 5057

75.47 MHz for 13C with a Bruker AVANCE 300 spectrometerequipped with a 5-mm QNP 2 probehead with z-gradientsusing Bruker standard software. Acetylated oligosaccharides(1±2 mg) were recorded as solutions in CDCl3, reducingoligosaccharides obtained upon Zemplen-deacetylation(0.5±1 mg) were measured in D2O. The 1H spectra werereferenced internally to tetramethylsilane (TMS) (d � 0 p.p.m.)or 3-(trimethylsilyl)-1-propanesulfonic acid (DSS), respec-tively. 13C spectra were referenced externally to 1,4-dioxane(d � 67.4 p.p.m.). Coupling constants were determined on afirst-order basis. Assignments were achieved using COSY andHMQC experiments.

Kinetic analyses

Hydrolysis. Initial velocities were measured at 80 8C in 50 mmsodium phosphate buffer pH 6.5. Kinetic parameters for theb-glycosidase catalysed hydrolysis of oNPGal, lactose andmethyl-b-d-galactoside were determined from fits of Eqn (1)to initial-velocity data by using the nonlinear least-squaresmethod with the programme sigmaplot (Jandel).

v � Vmax�S�/�Km 1 �S�� 1�and Vmax was converted into kcat by using a molecular mass of230 kDa for the functional tetramers of SsbGly and CelB. Thecatalytic efficiencies (kcat /Km) for the hydrolysis of sophorose,laminaribiose, cellobiose, and gentiobiose were derived fromthe part of the Michaelis±Menten curve when, at nonsaturatingconcentrations of the substrate, the initial velocity of the releaseof glucose is linearly dependent on the substrate concentration.

By using Scheme 2 the following relationships can be derived[22]:

kcat � k3kwater /�k3 1 kwater� �2�

kcat /Km � k3 /Kd �3�where k3 and kwater are the first-order rate constants for theglycosylation and deglycosylation steps, and Kd is the dissoci-ation constant of the substrate. If binding interactions betweenthe galactosylated enzyme and nucleophiles are neglected, theexpression for kcat in the presence of acceptor becomes [23]:

kcat � k3�kwater 1 kNu�Nu��/�k3 1 kwater 1 kNu�Nu�� 4�where [Nu] is the concentration of a nucleophilic acceptor, andkNu is a second-order rate constant (m21´s21) for reaction of thegalactosylated enzyme intermediate with Nu.

If, however, the reaction of E-Gal with nucleophiles is a two-step process that involves fast and reversible binding of thenucleophile at a saturatable enzyme site (Scheme 3), the expres-sion for kcat in the presence of the nucleophile becomes [23]:

kcat � k3�kwater 1 k4�Nu�/Kd0�/{k3 1 kwater 1 �k3 1 k4��Nu�/Kd

0}�5�

where k4 is the first-order rate constant for the reaction of thecomplex between E-Gal and bound nucleophile, and Kd

0 is thedissociation constant of that complex. Note that in Eqn (5)when k4 , kwater, added nucleophile will inhibit the rate ofrelease of the aglycon independently of the relative magnitudes

of kwater and k3, and this is a situation not accounted for by Eqn(4). When k4 . kwater, the effect of added nucleophiles on kcat

will mirror whether rate limitation is due to k3 (no change) orkwater (increase) [23].

The classical two-step reaction mechanism of b-glycosidase,shown in Schemes 2 and 3, requires that there be a fast loss ofleaving group from the complex with the galactosylatedenzyme. If, however, the rate of release of the leaving groupis slow and contributes to rate limitation, an expanded kineticscheme is needed to describe the reaction mechanism of theenzyme. This is shown in Scheme 4 from which in the absenceof nucleophiles the following relationship for kcat can bederived:

kcat � k3k5kwater /�k3k5 1 k3kwater 1 k5kwater� �6�where k5 is the first-order rate constant for the dissociationof the complex of the galactosylated enzyme and the leavinggroup. The expression relating the catalytic efficiency tomicroscopic rate constants remains unaltered because glycosidecleavage (k3) remains the first irreversible step in themechanism. For the reaction mechanism in Scheme 4, addednucleophile could affect an increase in the observed value ofkcat in two different ways, depending on whether kwater or k5 israte-limiting overall. By using Cleland's concept of net rateconstants [24] we can derive the following expression (Eqn 7)for the net rate constant (k 0water) for the breakdown of thecomplex of the galactosylated enzyme and the leaving group tofree enzyme:

k 0water � k5 1 k 0Nu/�1 1 k5/kwater 1 k 0Nu/k4� �7�whereby k 0Nu equals k4 [Nu]/Kd

0.The relation for the turnover number for the mechanism in

Scheme 4 is therefore given by Eqn (8):

kcat � k3k 0water /�k3 1 k 0water� �8�which can be rearranged by using Eqn (7) to give Eqn (9):

kcat � k3�k5 1 k4�Nu�/Kd0�/

{k3 1 k5 1 k3k5/kwater 1 �k3 1 k4��Nu�/Kd0} �9�

Added nucleophile will affect a decrease in kcat when k5 . k4

or kwater . k4. It will lead to an increase in kcat when k3 is notrate-limiting and k4 is greater than kwater or k5.

Scheme 2. Nucleophilic competition in the hydrolysis of Gal±OR by

b-glycosidases (E ).

Scheme 3. Nucleophilic competition in the hydrolysis of Gal±OR by

b-glycosidases (E), assuming binding of the nucleophile to a binding site

at the galactosylated enzyme intermediate. The binding of the

nucleophile can be described by a dissociation constant K 0d.

Scheme 4. Three-step mechanism of b-glycosidase-catalysed hydrolysis

of Gal±OR in the absence and presence of nucleophiles, assuming a

kinetically significant complex of galactosylated E and the leaving

group ±OR. Binding of nucleophiles affects the rate of dissociation of

±OR.


Partitioning analysis. The partitioning of the galactosylated (orglucosylated) enzyme (see Schemes 2±4) between reactionwith water and reaction with another nucleophile has beenanalysed according to Eqn (10) [13,25,26]:

voNP/vGly � 1 1 kNu�Nu�/kwater �10�where voNP and vGly are the initial velocities based on therelease of oNP and the release of the glycone moiety (d-Glc,d-Gal), respectively. Therefore, a plot of voNP /vGly against [Nu]should be linear with a slope of kNu /kwater. Note that kNu equalsk4 /Kd

0 , that is, the specificity constant for the reaction of E-Glywith Nu.

R E S U LT S A N D D I S C U S S I O N

Isolation and identification of major products of theenzymatic galactosyl transfer

As authentic standards were needed for CE and HPAEC, theGalOS produced during the hydrolysis of lactose catalysed byCelB were subjected to O-acetylation and purified by silica gelchromatography. This procedure allowed the separation ofmono- and disaccharides, the isolation of two trisaccharides,and the partial purification of one tetrasaccharide. 1H-NMRspectroscopy of the trisaccharide fractions enabled the unam-biguous assignment of the anomeric configuration and linkagesites of the constituent sugars. Furthermore, 13C-NMR data ofthe de-O-acetylated trisaccharides could be assigned usingHMQC and COSY spectra (Table 1), which compared favour-ably with reported data of galactosyl lactoses isolated from goator bovine colostrum, respectively [27,28]. Thus, the structuresof the trisaccharides were firmly established: b-d-Galp-(1!3)-b-d-Galp-(1!4)-d-Glc and b-d-Galp-(1!6)-b-d-Galp-(1!4)-

d-Glc. The 1H-NMR and COSY data of the partially purifiedO-acetylated tetrasaccharide were in accordance with the struc-ture b-d-Galp-(1!3)-b-d-Galp-(1!3)-b-d-Galp-(1!4)-d-Glc.The purified de-O-acetylated trisaccharides and other authenticstandards were used for the identification of the major GalOSproduced by SsbGly and CelB, using the standard-additiontechnique in CE (Fig. 1) and HPAEC (data not shown). Forthe standard-addition experiments, the crude sugar mixtureobtained upon lactose conversion was used. In the case of theSsbGly-catalysed reaction, partially purified oligosaccharidefractions were also used; this made it easier to achieve aclear-cut identification of the products.

Table 2 summarizes the major GalOS produced by SsbGlyand CelB during lactose hydrolysis. The overall productspectrum is similar for both enzymes and contains disaccha-rides and trisaccharides as the predominant components. Minorby-products were identified by CE and HPAEC. They includeup to three tetrasaccharides, and disaccharides such as b-d-Galp-(1!3)-b-d-Gal, b-d-Galp-(1!6)-b-d-Gal and b-d-Galp-(1!4)-b-d-Gal. However, these products taken togetheraccount for a maximum 20±25% of the total GalOS producedby the b-glycosidases. There is clear preference of bothenzymes for making new b(1!6) and b(1!3) glycosidicbonds. The synthesis of b(1!3) linked glycosides as the majorproducts of the b-glycosidase-catalysed glycosyl transfer hasbeen reported in relatively few instances thus far [4,30]. Onerecent example, however, is the thermostable b-glycosidasefrom Thermus thermophilus [31]. This enzyme produces b(1!3)linked glycosides with high regioselectivity upon action onnitrophenyl-b-d-glycosides and shares with SsbGly and CelBthe ability to efficiently transfer glycosyl residues to oNPGaland oNPGlc as acceptors [13]. During studies of an activesite mutant of SsbGly in which the catalytic nucleophile was

Table 1. 13C-NMR data of purified trisaccharides. Spectra (75.47 MHz) were recorded at 297 K and referenced to 1,4-dioxane (67.40), and assignments

were based on HMQC-spectra. ND, not determined.

Compound Residue C-1 C-2 C-3 C-4 C-5 C-6

3 0 galactosyl b-d-Galp-(1! 105�.36 72�.06 73�.55 69�.61a 76�.09b 62�.00

lactose 3)-b-d-Galp-(1! 103�.58 71�.22 82�.91 69�.45a 76�.09b 62�.00

4)-a-d-Glcp 93�.03 72�.25c 72�.44c 79�.31 ND ND

4)-b-d-Glcp 96�.80 74�.86 75�.39 79�.21 75�.82b 61�.13

6 0 galactosyl b-d-Galp-(1! 104�.36d 71�.83 73�.65e 69�.66 76�.15 62�.02

lactose 6)-b-d-Galp-(1! 104�.14d 71�.83 73�.47e 69�.50 76�.15 70�.06

4)-a-d-Glcp 93�.15 ND 72�.67 80�.23 70�.99 61�.20f

4)-b-d-Glcp 96�.72 74�.79 75�.57h 80�.48 75�.71h 61�.05f

b±h Assignments may be interchanged.

Table 2. Galacto-oligosaccharides formed as major by-products during the hydrolysis of lactose catalysed by SsbGly and CelB.

SsbGly CelB

b-d-Galp-(1!6)-b-d-Galp-(1!4)-d-Glca b-d-Galp-(1!3)-b-d-Galp-(1!4)-d-Glc

b-d-Galp-(1!3)-b-d-Galp-(1!4)-d-Glc b-d-Galp-(1!3)-b-d-Glc

b-d-Galp-(1!6)-b-d-Glc b-d-Galp-(1!6)-b-d-Galp-(1!4)-d-Glc

b-d-Galp-(1!3)-b-d-Glc b-d-Galp-(1!6)-b-d-Glc

b-d-Galp-(1!3)-b-d-Galp-(1!3)-b-d-Galp-

(1!4)-d-Glcb

a Ranked according to the relative proportions of the individual components in the oligosaccharide mixture produced by the enzyme. b Isolated, and identified

by NMR, but not quantitated by CE to show whether it is a major by-product.


replaced by site-directed mutagenesis, Moracci et al. [32] foundthe formation of new 3-O-b-linked glycosides by this mutant.

Formation and degradation of transfer products duringlactose hydrolysis

Figure 2 shows how the concentrations of the major GalOSchange with reaction time during lactose hydrolysis catalysedby SsbGly and CelB. CelB produced b(1!3) and b(1!6)linked trisaccharides in a molar ratio of < 2 : 1, and this ratiowas constant up to a 3-h incubation. The ratio of b-d-Galp-(1!3)-b-d-Galp-(1!4)-d-Glc to b-d-Galp-(1!6)-b-d-Galp-(1!4)-d-Glc was significantly smaller (1.3 : 1) for the reactioncatalysed by SsbGly. The distribution of glycosidic bonds inthe GalOS produced by SsbGly and CelB correlates with thespecificity of each enzyme for the hydrolysis of homodisac-charides of d-Glc, differing in the type of glycosidic linkage.According to catalytic efficiencies at 80 8C (Fig. 3), bothenzymes clearly prefer laminaribiose, b-d-Glcp-(1!3)-d-Glc.However, it is obvious from Fig. 3 that SsbGly and CelBdiffer with regard to specificity, and CelB shows a greaterpreference than SsbGly for hydrolysing b-(1!3)-linked gluco-sides.

The profiles of concentration vs. reaction time reveal aplateau for b-d-Galp-(1!6)-b-d-Galp-(1!4)-d-Glc which ismuch broader than that for b-d-Galp-(1!3)-b-d-Galp-(1!4)-d-Glc (Fig. 2). The greater kinetic stability of b-d-Galp-(1!6)-b-d-Galp-(1!4)-d-Glc than b-d-Galp-(1!3)-b-d-Galp-(1!4)-d-Glc is rationalized by considering the results reported inFig. 3. Not unexpectedly, the b-d-Galp-(1!6)-b-d-Galp-(1!4)-d-Glc is slowly degraded by both enzymes, especially by CelB.SsbGly and CelB differ in the extent to which they produce thedisaccharide, b-d-Galp-(1!3)-d-Glc, and this difference hasbeen clearly revealed by the results shown in Fig. 2. During theaction of CelB on lactose, formation of b-d-Galp-(1!3)-d-Glcis found at very early reaction times which is not the case forthe reaction catalysed by SsbGly. Therefore, it accounts chieflyfor the difference observed between CelB and SsbGly withregard to the prevailing ratio of disaccharides to trisaccharidesduring lactose conversion [13]. As b-d-Galp-(1!3)-d-Glc ismade by CelB even when the amount of free d-Glc is small, itsinitial production is very probably due to intramoleculartransgalactosylation (see Scheme 1). With both enzymes, the

Fig. 1. Typical elution profile upon CE of the pyridylaminated product

mixture obtained from the hydrolysis of lactose catalysed by SsbGly at

70 8C. The initial concentration of lactose was 0.8 m, and the substrate

conversion in the sample shown is < 70%. The peaks are: d-Gal and d-Glc

(1), lactose (2), b-d-Galp-(1!3)-b-d-Glc (3), b-d-Galp-(1!6)-b-d-Glc

(4), b-d-Galp-(1!6)-b-d-Gal (5), b-d-Galp-(1!6)-b-d-Galp-(1!4)-d-Glc

(6), b-d-Galp-(1!3)-b-d-Galp-(1!4)-d-Glc (7). The `x' refers to uniden-

tified minor products including at least two trisaccharides, and two or more

tetrasaccharides. Note that d-Gal and d-Glc are separated by the method

used and can be quantified (not visible in the figure).

Fig. 2. Synthesis and degradation of GalOS during the hydrolysis of

0.8 mm lactose at 70 8C catalysed by SsbGly (A) and CelB (B). b-d-Galp-

(1!3)-b-d-Glc, X,W; b-d-Galp-(1!6)-b-d-Glc, P,L b-d-Galp-(1!3)-b-

d-Galp-(1!4)-d-Glc, V,S b-d-Galp-(1!6)-b-d-Galp-(1!4)-d-Glc, B,A.

Fig. 3. Catalytic efficiencies of SsbGly (open bars) and CelB (filled

bars) for the hydrolysis of disaccharides composed of dd-glucose at 80 8C.


maximum concentration of disaccharides, b-d-Galp-(1!6)-d-Glc and b-d-Galp-(1!3)-d-Glc, occurred after a 1-day incu-bation when lactose had been depleted almost completely. Incontrast, the concentration of b-d-Galp-(1!3)-d-Glc wasmaximum at < 70% substrate conversion, corresponding to areaction time of < 3±5 h in Fig. 2. Therefore, b-d-Galp-(1!6)-d-Glc accumulated at the end of the hydrolysis reactionand accounted then for < 35±40% of the remaining transferproducts, as shown in Fig. 2. The b-d-Galp-(1!6)-d-Glcseemed to be hydrolysed poorly by SsbGly, and CelB did notexhibit a measurable net degradation of b-d-Galp-(1!6)-d-Glcwithin the reaction time studied. The information in Fig. 2would be essential if the goal of the enzymatic reaction was tooptimize specifically the production of trisaccharides or disac-charides, or transgalactosylation products containing b(1!6)or b(1!3) glycosidic linkages.

Effect of transgalactosylation on the rate of hydrolysis ofoNPGal

Figure 4 shows the effect of increasing the concentration ofadded nucleophile on the catalytic constant for the release ofaglycon from oNPGal, catalysed by CelB and SsbGly. Inthe case of SsbGly, kcat increased to a maximum value of2200±3000 s21 with increasing concentrations of nucleophilessuch as methanol, 1-propanol or d-Glc, for example. Theapparent activation of a retentive b-glycosidase by an exo-genous nucleophile is indicative of a reaction mechanism inwhich the rate of degalactosylation, the rate of loss of theleaving group or both are slow reaction steps in the hydrolysisof oNPGal [1±3,33]. In other words, k 0water is partly rate-limiting for the overall glycoside hydrolysis, and thereforek 0water increases in the presence of alcohols until another stepsuch as k3, for example, becomes rate-limiting. Partial ratelimitation by k 0water and the fact that galactosyl transfer takesplace to oNPGal [13], explains the observation of a rateenhancement over that expected from the value of kcat at highconcentration of oNPGal, observed in this work and by others[34]. With CelB, a maximum 1.2±1.4-fold enhancement of themaximum rate of oNP release from oNPGal was observedupon addition of methanol and 2-propanol (data not shown). A1.2-fold increase in kcat upon addition of 2-propanol was

observed for the SsbGly-catalysed hydrolysis of oNPGal. Aswe show below that 2-propanol does not react with galacto-sylated SsbGly or CelB to give transfer products in measurableconcentrations, the apparent activation of CelB by alcohols, andthat of SsbGly by 2-propanol could be brought about by`medium effects' of the nucleophile. Indeed, it has been shownby D'Auria et al. [35] that minor structural changes are inducedin SsbGly on addition of straight-chain primary alcohols. Analternative explanation would be that binding of the alcoholsremote from the site where cleavage of the glycosidic bondoccurs, helps to stabilize a more reactive conformation of theenzyme (CelB) and thereby increases the rate of the galacto-sylation step (k3). In addition, binding of the alcohol at anextended nucleophile-binding site at the galactosylated enzymecould occur in a way such as to enhance the rate of dissociationof the leaving group (k5) even though no galactosyl transferto the nucleophile takes place. The decrease in rate at highconcentration of nucleophile (Fig. 4) could reflect competitiveand other-type inhibitions of SsbGly and CelB which have notbeen investigated further. Note, however, that if: (a) the trans-galactosylation rate constant, k4, were small compared with k5

or kwater; and (b) medium effects were at the same timeresponsible for the increase of kcat in the presence of alcohols,the galactosyl transfer to exogenous nucleophiles would beexpected to produce inhibition of the enzyme, as it wasobserved for SsbGly and CelB at high concentrations ofalcohols.

d-Glc inhibited the rate of oNP release from oNPGal cata-lysed by CelB. However, galactosyl transfer to glucose tookplace during that reaction (see Table 4 below and [13]). Theinhibition by d-Glc could be overcome at high concentrationsof oNPGal (20±25 mm), apparently suggesting that it is due tocompetition of d-Glc with substrate for binding to the freeenzyme (see below). The fact that activation of CelB by d-Glcin the reaction with oNPGal is either completely absent or is atleast much smaller than that of SsbGly in the same reaction,can be rationalized in three different ways. Firstly, by assuminga reaction mechanism of CelB in which the rate constant k3

comprising the steps of cleavage of the glycosidic linkageof the substrate and aglycon release (Scheme 2), chieflycontributes to rate limitation when oNPGal is the substrate.Second, by considering that the true activation of CelB byadded nucleophile is masked completely because of thecompetitive inhibition by the nucleophile. A third and favouredinterpretation is, however, that kcat of CelB decreases in thepresence of d-Glc because the reaction of E-Gal with d-Glc is atwo-step process (Scheme 3) in which after binding of glucosehas occurred, the turnover of the complex between E-Gal andd-Glc is slower than the turnover of E-Gal with water. The rateconstant ratios for the partitioning of galactosylated CelBbetween reactions with oNPGal and water, and d-Glc and waterare 97 m21 and 17 m21, respectively [13]. As the rate constantratios can be interpreted as specificity constants for the reactionof E-Gal with nucleophiles, galactosyl transfer will occurpreferentially to substrate at high concentrations of oNPGal.If reaction of E-Gal with oNPGal (koNPGal, m21´s21) occurs ata faster rate than the reaction with d-Glc (kGlc, m21´s21), andgalactosylation (k3) is the chiefly rate-determining step in thereaction of CelB with oNPGal, oNPGal will overcome theinhibition observed in the presence of d-Glc even though thereis no or little competition of d-Glc with oNPGal for binding tothe free enzyme.

With oNPGlc as substrate an increase in kcat by added d-Galor d-fucose (< 1.7-fold) was observed for SsbGly whereas kcat

for the reaction catalysed by CelB was not affected by d-Gal for

Fig. 4. Rate acceleration and inhibition by added nucleophile in the

hydrolysis of oNPGal catalysed by SsbGly (filled symbols) and CelB

(open symbols). The substrate concentration was saturating and constant at

15 mm. d-Glc, B,A; methanol, X,W. The lines should illustrate the trend of

the data.


the concentration range 0±400 mm (Fig. 5). d-Fucose, how-ever, inhibited the rate of oNP release catalysed by CelB.Recently, Bauer and Kelly [36] have shown that the values ofkcat for reactions of CelB with a number of aryl-b-d-glucosidesare independent of the pKa of the aglycon between pKa valuesof 6 and approx. 9.5, suggesting that deglucosylation is rate-limiting for that pKa range. As a consequence, nucleophiliccompetition [33,37] is expected to give an increase in kcat forthe reaction of CelB with oNPGlc (pKa of oNP < 7.1) unlessthe complex of E-Glc with nucleophile turns over more slowlythan E-Glc with water in which case, d-Gal is expected toproduce inhibition of CelB. We did not observe significant rateenhancement by using d-Gal as added nucleophile althoughd-Gal is clearly shown to intercept the glucosyl residue (seeTable 4) and does not appear to inhibit the release of oNP athigh d-Gal concentrations. Furthermore, there was no increasein the kcat for the CelB-catalysed reaction with oNPGlc at highsubstrate concentrations. By contrast, plots of the rate of oNPrelease vs. concentration of oNPGlc (or the correspondingdouble-reciprocal plots) were biphasic for SsbGly, and athigh concentrations of oNPGlc ($ 15 mm) there was a linear

dependence of reaction rate on the substrate concentration.Therefore, deglucosylation may present a slow step but isprobably not entirely rate limiting for the CelB-catalysedreaction with oNPGlc.

Effect of transgalactosylation on the rate of lactosehydrolysis

When lactose was the substrate, and d-Gal, methanol or1-propanol were the intercepting nucleophiles, no significantincreases were observed for the turnover numbers of lactosehydrolysis by SsbGly and CelB based on the measurementsof initial velocities of d-Glc release. Therefore, the net rateconstant of degalactosylation does not appear to be a slow stepin the hydrolysis of lactose catalysed by SsbGly and CelB, sothat kcat approximates k3. Rate limitation brought about byk3 presumably reflects a larger contribution of acid catalysisto bond cleavage in lactose than oNPGal since glucose(pKa � 15.1) is a much more basic leaving group than oNP(pKa � 7.1). (The pKa of the 4 0-OH of d-glucose is, to the bestof our knowledge, not known. The pKa of glucose is < 12.4[38] but that value probably underestimates the pKa of thehydroxy group at C-4. Herschlag et al. [39] proposed that thepKa of ethylenglycol (15.1 [38]; 14.8 [39]) sets an upper limit tothe pKa of the 3 0-OH group of d-ribose. By analogy, a value of15.1 was assumed for the 4 0-OH group of d-glucose and usedhere for structure±reactivity correlations; however, it should benoted that this value contains considerable uncertainty.)

The catalytic efficiency of the reaction with lactose will bedetermined by contributions from electronic effects and bindingeffects. To partially distinguish between these contributions,hydrolysis of methyl-b-d-galactoside was used, for which therole of binding interactions with the aglycon is expected to besmall in the transition state for glycosidic-bond cleavage. ThepKa of methanol (15.1 [38]) is similar to that of the C-4hydroxy group of glucose (15.1), suggesting that in termsof structure±reactivity considerations [1±3], both should becomparable with regard to leaving group ability. In other words,unless there are significant binding effects resulting frominteractions between the enzyme and d-Glc, values of kcat < k3

should be similar for reactions of the b-glycosidases withmethyl-b-d-galactoside and lactose. Table 3 summarizes thekinetic parameters of CelB and SsbGly obtained from non-linear fits of Eqn (1) to initial rate data based on measuring therelease of the leaving group. By comparing the data for

Fig. 5. Rate acceleration and inhibition by added nucleophile in the

hydrolysis of oNPGlc catalysed by SsbGly (filled symbols) and CelB

(open symbols). The substrate concentration was saturating and constant at

3 mm (The Km values of SsbGly and CelB for the hydrolysis of oNPGlc are

0.8 mm and 0.2 mm, respectively). Added d-Gal, B,A; 6-deoxy-d-Gal,

X,W.

Table 3. Kinetic parameters for the hydrolysis of Gal±OR catalysed by SsbGly and CelB at 80 8C and pH 6.5, where ±OR is the leaving group. ND,

Not determined.

SsbGly CelB

Leaving

group (±OR) kcat (s21) Km (mm)

kcat /Km

(m21´s21)

± DDG³

(kJ´mol21) kcat (s21) Km (mm)

kcat /Km

(m21´s21)

± DDG³

(kJ´mol21)

Methanola 6�.6 192 �̂ 4 34 ± 8�.3 861 �^ 216 10 ±

kcat < k3 Km < Kd kcat < k3 Km < Kd

Glucoseb 1500 196 �^ 14 7700 16 3800 186 �̂ 19 20 400 22

kcat < k3 Km < Kd kcat < k3 Km < Kd

oNPc 1300 1.1 �^ 0.2 1 200 000 31 10 000 5.3 �^ 0.5 1 900 000 36

kcat < k5d (Km , Kd)e ND ND

a Determination of release of methanol. b Determination of release of glucose. c Determination of release of oNP. d See Scheme 4. e Kd � k2 /k1, and

Km � [(k2 1 k3)/k1] [k 0water /(k3 1 k 0water)] whereby k 0water is given by Eqn (7).


reactions with lactose and methyl-b-d-galactoside, the resultsshow clearly that there is a large positive deviation in the valuesof kcat and kcat /Km over those expected from the pKa differencebetween methanol and the 4 0-OH of d-Glc. Therefore, bindingenergy derived from the interactions with the glucose leavinggroup appear to be utilized by both enzymes to stabilize thetransition state of the reaction (kcat /Km) and to decrease theactivation energy of the rate-determining step (kcat). By usingthe relationship

DDG³ � 2RT ln ��kcat /Km�Gal2OR/�kcat /Km�methylgalactoside�where ±OR is the leaving group, we can calculate theincremental binding energy (Table 3) in the transition statefor the reaction with lactose or oNPGal, relative to that for thereaction with methyl-b-d-galactoside. The binding energy thatis made available to stabilize the transition state for thehydrolysis of lactose is almost fully expressed (80±100% logfraction) in the value of kcat (< k3). Therefore, this implies thatcompared with methanol, the extra binding energy resultingfrom the interactions with the glucose leaving group iscompletely utilized by both enzymes to achieve more efficientelectrophilic catalysis of glycosidic-bond cleavage (k3) and not`wasted' to generate tighter binding (Kd) (see [40] for thegeneral case; for studies on the role of acid/base catalysis in thehydrolysis of glycosides by b-glycosidases see [1±3,41±44]and by SsbGly see [45]).

Nature of the rate-limiting step in hydrolysis of oNPGal

The observed turnover numbers of SsbGly with oNPGal andlactose are almost identical. However, the results obtained innucleophilic-competition studies lend strong support to thesuggestion that different reaction steps contribute to ratelimitation in the reactions with these substrates. The obser-vation of a change in the rate-limiting step from kcat < k3

(hydrolysis of lactose) to kcat being partly rate-limited by k 0water

(hydrolysis of oNPGal) requires that the ratio k3 /k 0water forreaction with oNPGal increase significantly ($ 16-fold) com-pared with k3 /k 0water for reaction with lactose. (The estimate isbased on the experimental limit of detecting an increase in kcat

brought about by the addition of exogenous nucleophiles. Thislimit is < 3% of the value of kcat in the absence of nucleophiles.By assuming an average value, f, of < 1 for the increase ofk 0water in the presence of alcohols, we can calculate a lowerlimit of # 0.062 for the ratio of rate constants, k3 /k 0water, in thereaction of SsbGly with lactose, according to:

kcat;eff /kcat;0 � �k3/k 0water 1 1��1 1 f �/�k3/k 0water 1 1 1 f � � 1:03

where kcat,eff and kcat,0 are the turnover numbers in the presenceand absence of nucleophile, respectively. The above expressionis derived from Eqn (8), assuming that k 0water in the presence ofnucleophiles equals the expression k 0water (1 1 f ). Now, as therate constant of the degalactosylation step, kwater, contributes tothe magnitude of k 0water [see Eqn (7)] but is independent onthe nature of the leaving group, an increase in k3 /k 0water forhydrolysis of oNPGal, relative to the corresponding rate con-stant ratio for hydrolysis of lactose, could be affected by: (a) anincrease in k3, probably reflecting the weak basicity of the oNPleaving group; or (b) a decrease in k5. Considering the closelysimilar kcat values for the reaction with oNPGal and lactose, theobserved kcat for the reaction with oNPGal cannot, therefore,represent an estimate of the value of kwater and degalactosyl-ation is probably not a major rate-limiting step in the reactionof SsbGly with oNPGal. In other words, k 0water for this reaction

does not come close to kwater. However, if the reaction ofSsbGly with oNPGal is a true three-step mechanism (Scheme 4)in which the dissociation of the leaving group, represented byk5, is a kinetically significant reaction step, k 0water reflectscontributions from kwater and k5 [Eqn (7)], and k 0water can besmaller than both rate constants. By the above reasoning, wesuggest k5 rather than kwater to be partly rate-limiting for theoverall hydrolysis of oNPGal by SsbGly (cf. Table 3). Thepresence of alcohols probably affects k 0water by increasing therate of dissociation of the leaving group, and k 0water increasesuntil another reaction step (< 2200±3000 s21), possibly k3,becomes rate-limiting for kcat.

Intramolecular galactosyl transfer during lactose conversion

At early times during lactose hydrolysis when monosaccharideacceptors are present in small concentrations, the rate ratio offormation of d-Glc and disaccharides reflects the partitioning ofnoncovalently enzyme-bound glucose between release intosolution, and back-reaction with the galactosyl enzyme inter-mediate (Scheme 1, [4±6]). Because the rate of disaccharideformation was not available to our analysis, we extrapolatedback to zero lactose conversion the ratio of the concentration ofd-Glc to the concentration of the sum of the newly formeddisaccharides, as shown in Fig. 6. If we assume that the integraldata reflect the corresponding rate ratio, this analysis allowsus to obtain an estimate of the relative importance of intra-molecular transgalactosylation in the reactions of CelB andSsbGly with lactose. The initial ratio of [d-Glc]/S [disaccha-rides] thus found is 20 and 44 for lactose conversion by CelBand SsbGly, respectively. These figures indicate that with bothenzymes, intramolecular galactosyl transfer occurs at a slowrate, compared with that of release of d-Glc. However, theintramolecular transfer is 2.2-fold more important in thereaction of CelB with lactose, compared with the correspondingreaction of SsbGly. We must emphasize that the values givenare estimates for the upper limit of the contribution of intra-molecular galactosyl transfer to by-product formation in lactosehydrolysis. This is so because the intermolecular reaction of the

Fig. 6. Contribution of intramolecular transgalactosylation to the

formation of by-products during lactose hydrolysis catalysed by SsbGly

(X) and CelB (A). When substrate depletion is small, the ratio of d-Glc/S

(new disaccharides) on the y-axis provides an estimate of how, after the

glycosidic bond cleavage has occurred, the enzyme-bound d-Glc partitions

between release into solution, and intramolecular reaction to give new

disaccharides.


galactosylated enzymes with free d-Glc cannot be avoidedcompletely. However, with the methods used we were able todetect but could not quantify small concentrations of disaccha-rides (# 3 mm) in the presence of a large excess of lactose(0.8 m) when substrate conversion was less than 5%. Underthese conditions, essentially no transfer of d-Gal to mono-saccharides should occur by intermolecular transfer. The majorproduct of the intramolecular transfer reactions catalysed byCelB is b-d-Galp-(1!3)-d-Glc (< 60%), with the remainderbeing contributed by allolactose. In contrast, with SsbGly therelative proportion of these two transferase products is balanced.Production of b-d-Galp-(1!3)-d-Glc as a predominant trans-galactosylation product upon lactose conversion has beenreported with a b-galactosidase from Bacillus circulans [29]and cells of Bifidobacterium bifidum [30]. However, it is hasnot been studied by these authors whether disaccharideformation is due to intra- or intermolecular reactions.

Intermolecular galactosyl transfer

In general, plots of voNP /vGal against [Nu] were linear for acertain range of acceptor concentrations (cf. Fig. 7). Deviationsfrom linearity occurred mainly at high and low concentrationsof nucleophile. At low acceptor concentrations, the efficientgalactosyl transfer to substrate could be a reason for the upwardcurvature in Fig. 7. The observation that voNP /vNu levels off athigh concentration of nucleophile (data not shown) is difficultto rationalize because it requires that the galactosylated enzymein complex with the acceptor reacts to a certain extent withwater to give d-Gal.

Fits of the data (in the linear range) to Eqn (1) were used toobtain kNu /kwater, and the partitioning ratios are summarized inTable 4. As the true values of kwater are not known for bothenzymes, kNu cannot be obtained from these partitioning ratios.

Considering nucleophilic reactivity [46], primary alcoholsare expected to be more reactive than secondary alcohols towardsintercepting a glycosyl enzyme intermediate. 1-propanol and2-propanol have been used here as a model system to determinethe reaction of small primary and secondary alcohols withgalactosylated CelB and SsbGly. The data shown in Table 4suggest that there is a marked, at least 25-fold, selectivity ofboth enzymes for reaction with the primary alcohol, and little ifany galactosyl transfer to 2-propanol occurs.

Also shown in Table 4 are the partitioning ratios for thetransfer of a galactosyl residues from SsbGly and CelB tomethanol and d-Glc. The ratio of kGlc /kMeOH which is obtainedfrom the ratio of (kGlc /kwater) to (kMeOH /kwater), is approx. 45 forboth enzymes. This result shows that there is a 45-fold largerstabilization of the transition state (< 11.2 kJ´mol21) for thereaction of the galactosylated enzyme intermediates with d-Glcthan for reaction with methanol. The apparently much smallertransition state stabilization observed in the reaction of thegalactosylated enzyme with d-Glc, compared with the reactionof the enzyme with lactose (cf. Table 3), using the kineticparameters for the galactosyl transfer to methanol and for thehydrolysis of methyl-b-d-galactoside as the references, mustreflect inter alia the fact that the transfer of the galactosyl groupto d-Glc yields besides lactose, allolactose and b-d-Galp-(1!3)-d-Glc. However, there is a clear requirement of bindinginteractions with the nucleophile to obtain efficient base cata-lysis in the reaction of galactosylated SsbGly and CelB withalcohols.

If alcohol reactivity was the sole determinant for reaction ofthe glycosyl enzyme intermediate with nucleophiles, inter-molecular glycosyl transfer to sugars would be expected to: (a)yield exclusively b(1!6) linked products; and (b) be inhibitedwhen no primary hydroxy group was available for reaction.Using oNPGlc as substrate, the transfer of a glucosyl moietyto d-Gal and 6-deoxy-d-Gal was measured, and results areshown in Table 4. Both enzymes catalysed glucosyl transfer to6-deoxy-d-Gal, and in case of SsbGly, the transfer constant forreaction with 6-deoxy-d-Gal was even higher than that forreaction with d-Gal. Therefore, binding of the nucleophile tothe active site of SsbGly greatly increases, at least 50-fold, thereactivity of an otherwise unreactive secondary alcohol. Bind-ing effects make major contributions to the product spectrumobtained on lactose conversion, and that finding is clearlymanifested by the spectrum of new b-glycosides synthesized bySsbGly and CelB during lactose conversion.

A C K N O W L E D G E M E N T

Financial support from the European Commission is gratefully acknowledged

(grant EC FAIR CT 96-1048). The groups in Naples, Italy (M. Moracci,

Fig. 7. Partitioning of the galactosylated enzyme intermediate of

SsbGly (X) and CelB (A) between reaction with water and reaction

with added methanol. The data are for measurements at 80 8C using

7.5 mm oNPGal as the substrate.

Table 4. Partitioning ratios (kNu /kwater, mm21) for the reaction of

galactosylated (or glucosylated) SsbGly and CelB with exogenous

nucleophiles and with water, measured at 80 8C and pH 6.5 using

7.5 mmm oNPGal as the substrate. Values are ^ 10%.

Nucleophile SsbGly CelB

1-propanol 4�.5 2�.5

2-propanol , 0�.1 , 0�.1

Methanol 0�.85 0�.58

oNPa , 10 , 10

d-Galactoseb 5�.3 4�.1

d-Fucoseb 6�.1 2�.2

d-Glucosec 34 17

oNPGalc,d 125 97

oNPGlcd 122 85

a Measured with 180 mm lactose as the substrate and calculated from the

velocities of formation of glucose and galactose in the absence and presence

of 0.1±20 mm oNP. b Measured with 9 mm oNPGlc as the substrate and

calculated from the velocities of formation of glucose and oNP. c Data from

[13]. d Determined by measuring velocities of formation of oNP and

galactose (or glucose) at varying concentrations of oNPGal (or oNPGlc).


M. Ciaramella and M. Rossi) and Wageningen, the Netherlands (J. van der

Oost, T. Kapers and W. de Vos) are thanked for providing SsbGly and CelB,

respectively. S. Riva (CNR, Milano) is thanked for providing allolactose.

The help of M. Puchberger (Institute of Chemistry, BOKU, Vienna) with

NMR measurements is gratefully acknowledged. R. Zeleny and S. Baum-

gartner (IFA Tulln, BOKU, Vienna), and E. Staudacher (Institute of

Chemistry, BOKU, Vienna) kindly supported the measurements by CE and

HPAEC, respectively. K. D. Kulbe (Institute of Food Technology, BOKU,

Vienna) is thanked for encouragement and support. Regarding kinetic

analysis, the valuable comments of a reviewer are gratefully acknowledged.

R E F E R E N C E S

1. Davies, G., Sinnott, M.L. & Withers, S.G. (1998) Glycosyl transfer. In

Comprehensive Biological Catalysis (Sinnott, M.L., ed.), Vol. 1, pp.

119±208. Academic Press, San Diego.

2. Sinnott, M.L. (1990) Catalytic mechanisms of enzymic glycosyl

transfer. Chem. Rev. 90, 1171±1202.

3. Ly, H.D. & Withers, S.G. (1999) Mutagenesis of glycosidases. Annu.

Rev. Biochem. 68, 487±522.

4. Nakayama, T. & Amachi, T. (1999) b-Galactosidase. In Encyclopedia

of Bioprocess Technology: Fermentation, Biocatalysis, and Biosepa-

ration (Flickinger, M.C. & Drew, S.W., eds), Vol. 3, pp. 1291±1305.

John Wiley & Sons, New York.

5. Huber, R.E., Kurz, G. & Wallenfels, K. (1976) A quantitation of the

factors which affect the hydrolase and transgalactosylase activities of

b-galactosidase (E. coli) on lactose. Biochemistry 15, 1994±2001.

6. Huber, R.E., Gaunt, M.T. & Hurlburt, K.L. (1984) Binding and

reactivity at the `glucose' site of galactosyl-b-galactosidase (Escheri-

chia coli). Arch. Biochem. Biophys. 234, 151±160.

7. Aguilar, C.F., Sanderson, I., Moracci, M., Ciaramella, M., Nucci, R.,

Rossi, M. & Pearl, L.H. (1997) Crystal structure of the b-glycosidase

from the hyperthermophilic archaeon Sulfolobus solfataricus: Resili-

ence as a key factor in thermostability. J. Mol. Biol. 271, 789±802.

8. Pisani, F.M., Rella, R., Raia, C.A., Rozzo, C., Nucci, R., Gambacorta,

A., De Rosa, M. & Rossi, M. (1990) Thermostable b-galactosidase

from the archaebacterium Sulfolobus solfataricus. Purification and

properties. Eur. J. Biochem. 187, 321±328.

9. Kengen, S.W., Luesink, E.J., Stams, A.J. & Zehnder, A.J. (1993)

Purification and characterization of an extremely thermostable

b-glucosidase from hyperthermophilic archaeon Pyrococcus furiosus.

Eur. J. Biochem. 213, 305±312.

10. Kengen, S.W. & Stams, A.J.M. (1994) An extremely thermostable

b-glucosidase from the hyperthermophilic archaeon Pyrococcus

furiosus; a comparison with other glycosidases. Biocatalysis 11,

79±88.

11. Davies, G.J., Wilson, K.S. & Henrissat, B. (1997) Nomenclature for

sugar-binding subsites in glycosyl hydrolases. Biochem. J. 321,

557±559.

12. Nucci, R., Moracci, M., Vaccaro, C., Vespa, N. & Rossi, M. (1993)

Exo-glucosidase activity and substrate specificity of the b-glycosidase

isolated from the extreme thermophile Sulfolobus solfataricus.

Biotechnol. Appl. Biochem. 17, 239±250.

13. Petzelbauer, I., Nidetzky, B., Haltrich, D. & Kulbe, K.D. (1999)

Development of an ultra-high-temperature process for the

enzymatic hydrolysis of lactose. I. The properties of two

thermostable b-glycosidases. Biotechnol. Bioeng. 65, 322±332.

14. Murata, T. & Usui, T. (1997) Preparation of oligosaccharide units

library and its utilization. Biosci. Biotechnol. Biochem. 61, 1059±1066.

15. Nilsson, K.G.I. (1988) Enzymatic synthesis of oligosaccharides. Trends

Biotechnol. 6, 256±264.

16. Moracci, M., Nucci, R., Febbraio, F., Vaccaro, C., Vespa, N., La Cara,

F. & Rossi, M. (1995) Expression and extensive characterization of a

b-glycosidase from the extreme thermoacidophilic archaeon Sulfo-

lobus solfataricus in Escherichia coli: authenticity of the recombinant

enzyme. Enzyme Microb. Technol. 17, 992±997.

17. Voorhorst, W.G., Eggen, R.I., Luesink, E.J. & de Vos, W.M. (1995)

Characterization of the celB gene coding for b-glucosidase from the

hyperthermophilic archaeon Pyrococcus furiosus and its expression

and site-directed mutation in Escherichia coli. J. Bacteriol. 177,

7105±7111.

18. Bergmeyer, H.U., Bergmeyer, J. & Graûl, M., eds (1988) Methods of

Enzymatic Analysis, Vol. VI. VCH Publishers, Weinheim.

19. Suzuki, S., Kakehi, K. & Honda, S. (1992) Two-dimensional mapping

of N-glycosidically linked asialo-oligosaccharides from glycoproteins

as reductively pyridylaminated derivatives using dual separation

modes of high-performance capillary electrophoresis. Anal. Biochem.

205, 227±236.

20. Hase, S. (1996) Precolumn derivatization for chromatographic and

electrophoretic analyses of carbohydrates. J. Chromatogr. 720,

173±182.

21. Zeleny, R., Altmann, F. & Praznik, W. (1997) A capillary electro-

phoretic study on the specificity of b-galactosidases from Aspergillus

oryzae, Escherichia coli, Streptococcus pneumoniae, and Canavalia

ensiformis (jack bean). Anal. Biochem. 246, 96±101.

22. Richard, J.P., Westerfeld, J.G. & Lin, S. (1995) Structure±reactivity

relationships for b-galactosidase (Escherichia coli, lac Z). 1.

Brùnsted parameters for cleavage of alkyl b-d-galactopyranosides.

Biochemistry 34, 11703±11712.

23. Deschavanne, P.J., Viratelle, O.M. & Yon, J.M. (1978) Conformational

adaptability of the active site of b-galactosidase. J. Biol. Chem. 253,

833±837.

24. Cleland, W.W. (1975) Partition analysis and the concept of net rate

constants as tools in enzyme kinetics. Biochemistry 14, 3220±3224.

25. Richard, J.P., Westerfeld, J.G., Lin, S. & Beard, J. (1995) Structure±

reactivity relationships for b-galactosidase (Escherichia coli, lac Z).

2. Reactions of the galactosyl-enzyme intermediate with alcohols and

azide ion. Biochemistry 34, 11713±11724.

26. Richard, J.P., Huber, R.E., Heo, C., Amyes, T. & Lin, S. (1996)

Structure±reactivity relationships for b-galactosidase (Escherichia

coli, lac Z). 4. Mechanism for reaction of nucleophiles with the

galactosyl-enzyme intermediates of E461G and E461Q b-galactosi-

dases. Biochemistry 34, 11713±11724.

27. Urashima, T., Bubb, W.A., Messer, M., Tsuji, Y. & Taneda, Y. (1994)

Studies of the neutral trisaccharides of goat (Capra hircus) colostrum

and of the one- and two-dimensional 1H and 13C NMR spectra of

6 0-N-acetylglucosaminyllactose. Carbohydr. Res. 262, 173±184.

28. Saito, T., Itoh, T. & Adachi, S. (1987) Chemical structure of three

neutral trisaccharides isolated in free form from bovine colostrum.

Carbohydr. Res. 165, 43±51.

29. Yanahara, S., Kobayashi, T., Suguri, T., Nakakoshi, M., Miura, S.,

Ishikawa, H. & Nakajima, I. (1995) Formation of oligosaccharides

from lactose by Bacillus circulans b-galactosidase. Biosci. Biotech.

Biochem. 59, 1021±1026.

30. Dumortier, V., Montreuil, J. & Bouquelet, S. (1990) Formation of

oligosaccharides from lactose by Bacillus circulans b-galactosidase.

Carbohydr. Res. 201, 115±123.

31. Chiffoleau-Giraud, V., Spangenberg, P., Dion, M. & Rabiller, C. (1999)

Transferase activity of a b-glycosidase from Thermus thermophilus:

Specificities and limits ± Application to the synthesis of b-[1!3]-

disaccharides. Eur. J. Org. Chem. 1999, 757±763.

32. Moracci, M., Trincone, A., Perugino, G., Ciaramella, M. & Rossi, M.

(1998) Restoration of the activity of active-site mutants of the

hyperthermophilic b-glycosidase from Sulfolobus solfataricus:

dependence of the mechanism on the action of external nucleophiles.

Biochemistry 37, 17262±17270.

33. Sinnott, M.L. & Viratelle, O.M. (1973) The effect of methanol and

dioxan on the b-galactosidase-catalyzed hydrolyses of some b-d-

galactopyranosides: rate-limiting degalactosylation. Biochem. J. 133,

81±87.

34. Pulvin, S., Friboulet, A. & Thomas, D. (1990) Substrate inhibition and

activation kinetics of the b-galactosidase from the extreme thermo-

acidophile archaebacterium Caldariella acidophila. Biochim. Biophys.

Acta 1041, 97±100.

35. D'Auria, S., Nucci, R., Rossi, M., Bertoli, E., Tanfani, F., Gryczynski,

I., Malak, H. & Lakowicz, J.R. (1999) b-glycosidase from the

hyperthermophilic archaeon Sulfolobus solfataricus: structure and

activity in the presence of alcohols. J. Biochem. 126, 545±552.


36. Bauer, M.W. & Kelly, R.M. (1998) The family 1 b-glucosidases from

Pyrococcus furiosus and Agrobacterium faecalis share a common

catalytic mechanism. Biochemistry 37, 17170±17178.

37. Viratelle, O.M. & Yon, J.M. (1973) Nucleophilic competition in some

b-galactosidase-catalyzed reactions. Eur. J. Biochem. 33, 110±116.

38. Serjeant, E.P. & Dempsey, B. (1979) Ionisation constants of organic

acids in aqueous solution. In International Union of Pure and Applied

Chemistry. Commission on Equilibrium Data. Chemical Data Series

± No. 23. Pergamon Press. Oxford.

39. Herschlag, D., Eckstein, F. & Cech, T.R. (1993) The importance of

being ribose at the cleavage site in the Tetrahymena ribozyme

reaction. Biochemistry 32, 8312±8321.

40. Jencks, W.P. (1975) Binding energy, specificity, and enzymic catalysis:

the Circe effect. Adv. Enzymol. 43, 219±410.

41. Huber, R.E. & Chivers, P.T. (1993) b-Galactosidase of Escherichia coli

with substitutions for Glu-461 can be activated by nucleophiles and

can form b-galactosyl adducts. Carbohydr. Res. 250, 9±18.

42. Richard, J.P. (1998) The enhancement of enzymatic rate accelerations

by Brùnsted acid-base catalysis. Biochemistry 37, 4305±4309.

43. Lawson, S.L., Wakarchuk, W.W. & Withers, S.G. (1997) Positioning

the acid/base catalyst in a glycosidase: studies with bacillus circulans

xylanase. Biochemistry 36, 2257±2265.

44. Wang, Q., Trimbur, D., Graham, R., Warren, A.J.R. & Withers, S.G.

(1995) Identification of the acid/base catalyst in Agrobacterium

faecalis b-glucosidase by kinetic analysis of mutants. Biochemistry

34, 14554±14562.

45. Moracci, M., Capalbo, L., Ciaramella, M. & Rossi, M. (1996)

Identification of two glutamic acid residues essential for catalysis

in the b-glycosidase from the thermoacidophilic archaeon Sulfolobus

solfataricus. Protein Eng. 9, 1191±1195.

46. Ritchie, C.D. (1972) Nucleophilic reactivities towards cations. Acc.

Chem. Res. 5, 348±354.


Transgalactosylation by thermostable glycosidases … and 2-propanol, respectively. Therefore, the...

Documents

Transcript of Transgalactosylation by thermostable glycosidases … and 2-propanol, respectively. Therefore, the...