Structure and properties of the extracellular inulinase of Kluyveromyces marxianus CBS 6556

Structure

and

Properties of

the

Extracellular Inulinase

of

Kluyveromyces

marxianus

CBS 6556

ROBERT J. ROUWENHORST,1MARCO HENSING,1 JOHN VERBAKEL,2 W. ALEXANDERSCHEFFERS, ANDJOHANNES P. VAN DIJKEN1*

DepartmentofMicrobiology andEnzymology, Kluyver Laboratory of Biotechnology, Delft University of Technology,

Julianalaan 67, 2628BCDelft,' and Unilever Research Laboratory, Departmentof Genetics, 3133ATVlaardingen,2 The Netherlands

Received9 May 1990/Accepted 14 August1990

In theyeastKluyveromycesmarxianustwoforms of inulinasewerepresent,namely,aninulinasesecretedinto

the culturefluidandaninulinase retained in the cell wall. Both formswerepurifiedandanalyzedbydenaturing and nondenaturingpolyacrylamide gel electrophoresis. With theuseof

endo-ID-N-acetyl-glucosaminidase

H,it

was establishedthat theenzyme retained in the cell wall and theenzymesecretedinto the culturefluidhave

similar subunitsconsisting ofa64-kDa polypeptide with varyingamountsof carbohydrate (26to37% of the molecularmass).Thetwoformsof inulinase differed in size because of their differences in subunit aggregation. Theenzymepresentinthe culture fluidwasadimer, and theenzymeretained in the cell wallwasatetramer.

Thedifferences inoligomerization didnotaffect theapparentKm values towardsthesubstrates sucrose and

raffinose.Thesefindingssupportthehypothesis that the retentionof glycoproteins in theyeastcellwallmaybe

caused by a permeability barrier towards larger glycoproteins. The amino-terminal end of inulinase was

determined andcomparedwiththeaminoterminus of the closely related invertase. The kinetic and structural evidenceindicatesthatinyeasts twodistinct ,3-fructosidasesexist, namely, invertase and inulinase.

Inyeasts, twoextracellularglycoproteinsareknowntobe associatedwith growth onsucrose: invertase (EC3.2.1.26), e.g., in Saccharomyces cerevisiae, and inulinase (EC 3.2.1.7), e.g.,inKluyveromycesmarxianus. Theseenzymes

exhibitcorresponding hydrolytic activities towards sucrose

but differ in their specificities for higher-molecular-weight oligosaccharides and fructans of the inulintype(17, 33). The S/I ratio (relative activities with sucrose and inulin) is generally employed to discriminate between invertase and inulinase(33, 37).Alow S/Iratio (high activity withinulin) is taken toindicateinulinase.

Originally,theclassification ofinulinaseswasbasedonthe occurrence of specific enzymes in bacteria, molds, and plants. These inulinases rarely show activity with sucrose

and split fructans of the inulin type either endo-wise or exo-wise, producing a series ofoligofructans or only

fruc-tose, respectively (11, 15, 16, 33, 37). In contrast, yeasts

produce inulinases capableofhydrolysis of inulin and levan-type fructans exo-wise as well as of sucrose hydrolysis. Differences in S/I ratios and in apparent kineticconstantsare

consideredby some authors tobe insufficientfor a

distinc-tion between yeast inulinase and yeast invertase. In the view of theseauthors, inulinase isaspecialkind of invertase and

shouldbeclassifiedas such(2, 21, 32).

A considerable amountofresearch has been devoted to

the invertase ofS. cerevisiae because it appearedto be an

attractive model for studiesonprotein synthesisand excre-tion of glycoproteins (6, 7, 12, 13, 18, 24, 29, 30, 35). Secreted invertase resides mainly in the cell wall as an

octamer(1, 13).Thesmallamountofinvertasepresentin the

culture fluid, as well asthe fractionthat is removable from

thecellsbytreatmentwiththiols, wasfoundtobecomposed of dimers. It has been suggested thatoligomerization helps

to retain theenzyme within the cellwall (13).

*Correspondingauthor.

Similarly, theinulinaseof yeastisinpartassociated with the cellwall, but, in contrast to the invertase of Saccharo-myces, much moreofthe enzyme is actually secreted into

theculturefluid. Whengrownunderconditions which dere-press enzyme synthesis, the yeast K. marxianus secretes

over50%ofitsenzymeintotheculture fluidand35%canbe released from the cell wall by sulfhydryls (23, 32). In

contrast to the biochemistry of the invertase of Saccharo-myces, very little is known about thebiochemistry of inu-linase. Although inulinases of different yeasts have been partially purified (28, 37, 39)andcharacterizedwith respect

to theirkinetic properties, no information is available con-cerning their oligomeric structures and molecular weights. Duringthepurification of inulinase,the sucrose- and inulin-hydrolyzingactivities are never separated. This has led to

the conclusion that one enzyme is responsible for both activities and to the assumption that cell wall-associated activity and activity secreted into the culture fluid both representinulinase, withidentical kineticproperties (28, 39). However, as to the nature of the inulinase, two questions remain tobe answered. (i) Do the cell-wall-associated inu-linase and the inuinu-linase present in the culture fluid differin

oligomeric structure, like the invertase ofSaccharomyces? (ii) Does inulinase, purified to homogeneity, indeed not

hydrolyzesucrose?

In order to answer these questions we have not only purifiedtheinulinase associated with the cell wall(cellwall

inulinase), but also that present in the culture fluid

(super-natantinulinase)of K. marxianus,paying specialattentionto

theiroligomericstructures. In this reportwealsopresentthe

amino acid sequenceoftheamino-terminal end of inulinase.

MATERIALS AND METHODS

Organismandgrowthconditions.K. marxianusvar.

marx-ianus CBS 6556wasobtainedfrom the Yeast Divisionof the

Centraalbureau voor Schimmelcultures, Delft, The

Nether-lands, and maintained on YEPD agar slopes. YEPD

con-3337

0099-2240/90/113337-09$02.00/0

tained

(per

liter of distilled water) 10 gofyeastextract(Difco

Laboratories, Detroit,

Mich.), 10 g ofBacto-Peptone

(Dif-co),

and 20 g of

glucose.

The

organism

was grown under carbon and energy limitation in a laboratory fermentor

(Applikon, Schiedam,

The Netherlands), with a working

volume of 1

liter,

on amineral salts medium supplemented

with 5 gofsucrose perliter and 2.5 gofinulin per liter at a dilution rate of0.15

h-'

at

40°C

(pH 4.5), with an oxygen

concentrationthatwas50to70%ofair saturation. Dissolved oxygenwasmeasuredwitha

polarographic

oxygenelectrode

(Ingold, Urdorf,

Switzerland),and pHwascontrolledbythe automatic addition of1 M KOH. The mineral salts medium

was

prepared

by the method of Rouwenhorst et al. (32).

General conditions during purification. All

purification

steps were

performed

at room temperature, unless stated

otherwise.

Enzyme

from each stepofthe

purification

pro-cedure could be stored at either 4 or

-20°C

without any

significant

loss of

activity (either

withsucrose orinulinasthe

substrate).

Toremoveany

remaining

wholecells,the crude

inulinase

preparations

were membrane filtered (pore size,

0.22

pim;

Schleicher&

Schuell, Dassel,

Federal

Republic

of

Germany)

under mild pressure. Membranefiltration didnot alter the enzyme

activity

northequaternarystructure ofthe enzyme. When storage was

required,

the enzyme

prepara-tionwas

kept

frozen.

Preparationof inulinase. (i) Supernatantinulinase. During

steady-state cultivation,

15 liters of culture effluent was collected at

4°C.

Cells and culture fluid were separated by continuous-flow

centrifugation

(10,000 x g) with a Sorvall TZ-28

density gradient

rotor(DuPontInstruments, Biomed-ical

Division,

Newtown, Conn.). Culture fluidwas concen-trated to 150 ml with a Nephross Andante hollow-fiber system

(Organon Teknika,

Oss, TheNetherlands)followed

by

removal of

low-molecular-weight

contaminating protein with Centricon-30 microconcentrators (30kDacut-off; Am-icon

Corp.,

Scientific

Systems

Division,

Danvers, Mass.).

The concentrated

preparation

of extracellularinulinase from the culture fluid was taken as a source of supernatant

inulinase.

(ii)

Cell wall inulinase. Release of cell wall-associated

inulinase was induced by suspendingthe cells (0.2 g ofcell paste per

ml)

in 50 mM

potassium

phosphate (pH 6)

contain-ing

10mM2-mercaptoethanol and 10 mMdithiothreitol and

incubating

for2hat

30°C

withgentle agitation. The

suspen-sion was then centrifuged at 4°C (4,000 x g), and the

inulinase-containing

supernatant was exhaustivelydialyzed

against

50 mM

potassium phosphate

(pH 6.5) and concen-trated with Centricon-30 microconcentrators. The concen-trated

preparation

of this inulinase formerly trapped in the cell wallwill be referred to ascell wallinulinase.

To solubilize

activity

that was still associated with the cells after

primarily

release of cell wall inulinase, the cells

(0.2

g of cell paste per ml) were suspended in 50 mM

potassium

phosphate

(pH 6) containing10mM

2-mercapto-ethanol,

0.1 M

KCl,

and0.1% Triton X-100, and incubated for4h at

30°C.

Aftercentrifugation (4,000 x g) at 4°C, the supernatant contained thenewly released inulinase.

(iii)

Cell-bound inulinase. After the removal of cell wall

inulinase,

the cells still contained inulinase activity. This

cell-bound

activity

could only be solubilized by complete

breakage

ofthe cells. Ultrasonic disintegration, performed

asdescribed

previously

(32),wasusedtocollect this enzyme

fraction.

Purification of inulinase. (i) Anion-exchange

chromatogra-phy.

Supernatant

inulinase (4 ml) and cell wall inulinase (4

ml)

wereloadedontoa60-mlDEAE-SephadexA-50column

(2.5by 50 cm)equilibratedwith 20mMpotassium

phosphate

(pH 6.5). The column waswashed with the samebuffer and eluted with a100-ml linear

gradient

of 0to 500 mMsodium chloride in 20 mM potassium phosphate (pH 6.5). The flow rate throughout the experiment was 24 ml

h-1,

and effluent was collected in fractions of 2.3 ml. The column fractions containing inulinase activity were collected asthree

pools,

and the protein in the pools was concentrated with Centri-con-30 microconcentrators.

(ii) FPLC. Furtherpurification was performed on asmall scaleby means ofhigh-performance gel filtration

chromatog-raphy on a Superose 12 HR 10/30 prepacked column witha fast-proteinliquidchromatography(FPLC)system

(Pharma-cia, Uppsala, Sweden). Elution was carried out at room temperature with 500 mM potassium phosphate (pH 6)at a flowrate of 24 ml

h-1.

Determination ofmolecular mass. To estimate the molec-ular mass of the inulinases, the Superose 12 column was calibrated with standard proteins: carbonic anhydrase

(Sig-ma Chemical Co., St. Louis, Mo.), 29 kDa; ovalbumin (Sigma), 43 kDa; phosphorylase (Pharmacia), 94 kDa; bo-vine albumin (Pharmacia), 67 kDa; lactate dehydrogenase

(Sigma), 140 kDa; aldolase (Pharmacia), 158 kDa; catalase (Pharmacia), 232 kDa; ferritin (Pharmacia), 440kDa;

thyro-globulin (Pharmacia), 660kDa.

Removal of carbohydrate from inulinase. Depletion of oligosaccharide chains from inulinase wasaccomplished by

endo-p-N-acetyl-glucosaminidase

H (Endo-H; EC 3.2.1.96) from Streptococcus plicatus (Boehringer GmbH, Mann-heim, Federal Republic of Germany). Endo-H removes carbohydrate chains of the high-mannose type by cleaving

the di-N-acetylchitobiose unit linked to asparagine (34). Purified inulinases (6.5 ,ug) were suspended in 100 ,ul of 50 mM sodium citrate (pH

5.5)-0.1%

sodium dodecyl sulfate

(SDS)-0.02% sodium azide and heated for 5 min at 100°C. Endo-H (0.02 U) was added to the inulinases, and the solutions were incubated for 24 h at30°C. Released oligosac-charides and Endo-H were removed by filtration of the reaction mixture with Centricon-30 microconcentrators (cut-off, 30 kDa).

SDS-polyacrylamide gel electrophoresis (SDS-PAGE). Slab gel electrophoresis on a discontinuous polyacrylamide gel (12 cm high, 12 cm wide, 1.5 mm thick) was performedby the method of Laemmli (22) in a 3-cm 3.25% stacking gel and a 12.5% running gel. Samples were mixed in a 1:1 ratio (vol/vol) with a buffer containing 0.17 M Tris hydrochloride (pH 6.8), 3% SDS, 30% glycerol, and 10% 2-mercaptoetha-nol and heated (10

min,

100°C)before application. Gels were stained for protein with Coomassie brilliant blue and for carbohydrate by using the periodic acid-Schiff reagent as described by Beeley (4). Apparent molecular masses were determined by using commercially available marker proteins asstandards (Pharmacia).

Nondenaturing PAGE. For nondenaturing PAGE, essen-tially the same method was used as for SDS-PAGEexcept that SDS was omitted from the running bufferand from the sample solution and samples were not boiled before appli-cation. Inulinase activity in the gels was detected by using a modification of the method described by Grossmann and Zimmermann (18). Gels (7%) were incubated at room tem-peraturein asolution of 0.1 M sodium acetate (pH 4.5) with

4% sucrose. After washing (three times, 5 min each) with distilled water, the gels were transferred to 100 ml of 0.5 M NaOH-1 mg of 2,3,5-triphenyl tetrazolium chloride ml-1, and heated over an open flame until coloration occurred.

Staining was stopped by adding excess distilled water, followed by complete fixation of thegelin 10% aceticacid. Amino acid sequencing. Purified protein was dialyzed against double-distilled water, precipitated with 9 volumes of acetone, andresolubilized in water for amino acid sequenc-ing. Sequence analysis was performed with a gas-phase sequenator (Applied Biosystems model470A) by using25% trifluoroacetic acid as the conversion reagent. The resulting phenylthiohydantoin amino acids were analyzed on-line by reversed-phase high pressure liquid chromatography with a phenylthiohydantoin C18 column (2.1 by 220 mm) and a phenylthiohydantoin analyzer (Applied Biosystems model 120A).

Protein determination. Protein was determined by the method of Lowry et al. (25), by theCoomassie brilliant blue method of Bradford (5), and by total carbon analysis (TOC). From 23 proteins with known amino acid composition (19),

both the nitrogen content (16.12 + 0.79) and the carbon content(53.27 + 1.06)appeared very constant. Analogous to the representation of the protein concentration by the nitro-gen content, the protein concentration of globular proteins can be obtained by determination of the carbon content. When the degree of glucosylation (G; fraction of molecular weight accounted for by carbohydrate) is known and assum-ing a carbon content of carbohydrates of 44%, the concen-tration of glycosylated protein (PG [in parts per million]) can becalculated from the carbon content by: [PGJ - ppmn C x

[(1/0.53) x (1-G) x (1/0.44) x Gl.

A Beckman model 915B Total Organic Carbon Analyzer (Beckman Instruments, Inc., Fullerton, Calif.) was used to

determinethecarbon content of inulinase preparations. As a

standard, a2.137-gliter-1 solution of anhydrous potassium

biphthalate(1,000 mgof carbon

liter-')

was used.

Inulinase assay. Inulinase activity was measured as de-scribed previously (32). One unit of inulinase activity was definedasthe amountofenzymecatalyzing the liberation of 1

p.mol

of fructose

min-'

atpH 4.5 and50°C.

Chemicals. Fructose, sucrose, and 2-mercaptoethanol

were from Baker Chemicals BV, Deventer, The Nether-lands. Acrylamide, SDS, and N,N'-methylbisacrylamide

were from Merck, Darmstadt, Federal Republic of Ger-many. Dithiothreitol and inulin (chicory root) were from

Sigma. Otherchemicals were reagentgrade.

RESULTS

Purificationof supernatantinulinaseandcell wall inulinase. Preliminary experiments indicated that the inulinase ofK. marxianuscouldnotbeprecipitated conveniently by ammo-nium sulfate. Inulinase present in the culture fluid was thereforeconcentratedby usinga hollow-fiber device.

(i) Anion-exchange chromatography. In two successive

runs,performedunderexactlythesame conditions, 9,400U of the crudepreparation ofsupernatantinulinaseand10,350

Uof the crudepreparation ofcell wall inulinasewereloaded onto a

DEAE-Sephadex

A-50

column,

equilibrated

with 20 mMpotassium phosphate (pH6.5), and eluted withalinear NaCl gradient. Supernatant inulinase elutedat0.2 M NaCl

(Fig. 1A), but the peak coincided with a second,

contami-nating protein peak. In contrast, Sephadex A-50

gradient

elution of cell wall inulinase

provided

a

rapid

purification.

Mostof the

contaminating

protein

didnotbindtothecolumn and was eluted before NaCl

gradient

elution was initiated. The protein peak containing inulinase

activity

eluted atan

NaClmolarityof 0.15 M(Fig. 1B).

Thetotal enzyme activitiesrecoveredafter

Sephadex

A-50

c 0 0 c a 6 4 0 30 60 90 120 150 Elutlon volume (ml) 1.00 ,1

E

' C a 0 Cl 00,50 a 4 I 0.2.5 I0 x 6 0 E 0 z , II .I E IO Dx _. N >, 750 = 6 _ X-C 0 00 soo o I 0C 5 z0 250 - I . 0 30 60 90 120 150 Elutlon volume (ml)

FIG. 1. Chromatography of inulinase preparations on

DEAE-SephadexA-50.Crudepreparations ofsupernatantinulinase (A) and cell wallinulinase (B) wereappliedto thecolumnequilibratedwith 20 mM potassium phosphatebuffer(pH6.5) andelutedwith a0 to 0.5 M lineargradient ofNaCl. Fractions (2.3 ml) were collected,

scanned for absorbance at 280 nm, and assayed for

sucrose-hydrolyzing activity (0). Active peak fractions of supernatant inulinase and cell wall inulinase were separated into six pools as

indicated.

gradient elution,both ofsupernatantinulinase andcell wall

inulinase,werehigher(107 and129%recovery,respectively)

than the total activity originally loaded on the column. Apparently, the anion-exchange chromatography led to re-movalof constituents ofeither the mediumorthe cells that were inhibitory to inulinase. The S/I ratios of the crude

preparations of both supernatant inulinase and cell wall inulinase were 13. Inulinase obtained from Sephadex A-50 fractions had anS/Iratio of 13 + 2, and thisvalueappeared constant throughout the Sephadex A-50 inulinase

peaks,

althoughatthebeginning and the end of thesepeakstheS/I ratio showed greater deviations. Thiswasprobablybecause ofalowinulinaseconcentration

resulting

in

inulin-hydrolyz-ingactivities whichwere at the limits of the assay(datanot shown).

(ii) FPLC. The Sephadex A-50 inulinase activities were collected as six

pools,

marked I, II, and III (supernatant inulinase), and IV, V, and VI (cell wall inulinase). These pooled activities were concentrated by Amicon-30 ultrafil-trators,andaportionof eachpoolwas

applied

separately

to

anFPLCgel filtration system. The FPLCelution

profiles

of

pools

I,

II, and III of the supernatant inulinase all showed two major protein

peaks,

of which

only

one

corresponded

withaninulinaseactivitypeak (Fig. 2A).Theinulinase

pools

IV, V, and VI of cell wall inulinase

appeared

to be almost pure. The FPLC elution

profiles

all showed one

major

protein peak,

corresponding

with inulinase

activity

(Fig.

0,015I I I E0,010 E 0 -M _ 0,005 I op-0,01S ~0 E0 !op 0 0 "*-r III

A

10 20 30 / 5010 20 30 Go E 2 40₄₀

t,

a

c1

20 C 0 0 40 20 30

Time of ter Injection (min)

60= E a 40 ' 20; a 10 0 0 20 30 40V " 10 20 30 40V 50 10 20 30 40V 50

Time after Injection (min)

FIG. 2. FPLC gel filtration of the pooled DEAE-Sephadex A-50 chromatography fractions ofsupernatant inulinase (A)and cellwall

inulinase(B). The pooledinulinase activities I, II,III(supernatantinulinase), andIV,V,VI(cellwallinulinase)wereappliedtoaSuperose 12 HR10/30 gel filtration column,equilibratedwith 0.5 Mpotassium phosphatebuffer(pH 6). Chromatographywasperformedataflow rate

of 24 mlh-1. Inulinase (0)isgivenas sucrose-hydrolyzing activity.

2B). The inulinases purified by FPLC gel filtration were activetowardssucrose aswellasinulin,inan S/Iratio of15

+ 2 (results not shown). When compared with protein markersof known molecularweight,thesupernatant

inulin-aseelutedas apeakat approximately 180 kDa and the cell wall inulinase eluted as a peak at approximately 450 kDa (Fig. 3). However, glycoproteins tend to elute earlier from gel filtration columns than globular marker proteins, and

hence these apparent molecularmasses will be

overestima-tions ofthereal values (4).

Analytical gel electrophoresis. Theisolated peakfractions from FPLC gel filtration of supernatant inulinase and cell wall inulinase were concentrated and analyzed by both SDS-PAGE andnondenaturingPAGE before and after treat-mentwith Endo-H.Theproteininthe FPLCfractions ofcell wall inulinase and the protein in the FPLC fractions of

supernatant inulinase both migrated on SDS-PAGE as a polydisperse band between87 and 102 kDa(Fig. 4, lanes 1 and5). Noother protein bands were detectedon the SDS-polyacrylamide gels,either withCoomassiebrilliant blueor with the more sensitive silver staining, indicating that the enzyme preparations contained only inulinase protein. The polydispersityof theinulinase bandsapparentlyiscausedby heterogeneity in the size of the polysaccharide chains

at-tachedtothe inulinasepolypeptide.Toremovethese

carbo-> 0,4 0,3-0,2 _

0,1

40

105

106

Molecular weight

FIG. 3. Estimationof the molecular masses ofpurified superna-tantinulinaseandcellwall inulinasebyFPLCon aSuperose12gel

filtration column. Open circles represent the positions ofpurified supernatant inulinase (S) and cell wall inulinase (C). Standard proteins: 1, carbonic anhydrase (29 kDa); 2, ovalbumin (43 kDa); 3, bovine albumin (67 kDa); 4, phosphorylase (94 kDa); 5, lactate

dehydrogenase (140 kDa); 6, aldolase (158 kDa); 7, catalase (232 kDa); 8,ferritin(440kDa); 9,thyroglobulin (660kDa).

7

I

I

1

2

3

4

494

467

443

l430 FIG. 4. SDS-PAGE of native and carbohydrate-depleted

inuli-nases.Supernatant inulinase(128,ug ml-'), cellwall inulinase (135 ,ug ml-'), Endo-H-treated supernatantinulinase (65 ,ug ml-'), and cellwall inulinase (65 ,ug ml-')weremixed withanequal volume of sampling buffer and heatedat100°C forSmin. Samples(50,ul)were

then applied to a 12.5% discontinuous SDS-polyacrylamide gel

system. Lanes: 1, 2, and 4, cell wall inulinase; 3, 5, and 6,

supernatantinulinase. Treatment with Endo-Hwasperformed with

nativeinulinase (lanes 2 and3)anddenatured inulinase (lanes4and 6). Molecular mass standards (top to bottom): phosphorylase, bovineserumalbumin, ovalbumin, carbonic anhydrase.

hydrate chains, intact supernatant inulinase and cell wall inulinase were treated with Endo-H. SDS-PAGE of the Endo-H-treated inulinasesgavemoredistinct protein bands, with apparent molecular masses of about 72 kDa (Fig. 4,

lanes 2and 3). Deglycosylation was verified by stainingthe gel for carbohydrate. TheEndo-H-treated inulinase stained with periodic acid-Schiff reagent, indicating that it still contained carbohydrate. Deglycosylation experiments with Saccharomyces invertase have shown that the susceptibility ofoligosaccharidesinthenativeglycoprotein is enhanced by its denaturation (35). Inulinase samples were therefore

boiled in0.1% SDS before the addition of Endo-H.

Deple-tion ofthe oligosaccharide chains after denaturation of the inulinasesresulted in asharperresolution of the inulinases.

Thedeglycosylated protein backbones of cell wall inulinase and supernatant inulinase nowmigrated as oneband with a

uniform size of 64 kDa(Fig. 4, lanes4and6). The denatured and Endo-H-treated inulinase failed to stain with periodic acid-Schiff reagent. Apparently, all outer-chain carbohy-drate hadnowbeenremoved from the inulinases. The results of the SDS-PAGE with nontreated and Endo-H-treated inulinase clearly show that intact inulinase contains 26 to

37% of its mass as high-mannose type oligosaccharides linked toasparagine residueson a64-kDapolypeptide. The native enzymesfrom culture fluid and cell wall should then

representglycosylatedinulinase multimers. To establish the relationship between supernatant inulinase and cell wall inulinase, Ferguson analysis of the purified enzymes was

carried out by using nondenaturing gels ofdifferent poly-acrylamide concentrations and determining the migration distances of the inulinases relativetothose ofproteinswith

known molecularweight, asdescribedby Esmonetal.(13). It appeared that the supernatant inulinase corresponded to

an inulinase dimer ofapproximately 165 kDa and the cell

wall inulinase correspondedtoa tetramerofapproximately 335 kDa.

Release of cell wall inulinase by different treatments. A

comparison of the initial inulinase preparations with the

purified inulinases, with respecttotheirbehavioron

nonde-naturing PAGE withstainingforsucrose-hydrolyzing

activ-ity, showed that the purification procedure had not altered

FIG. 5. Nondenaturing PAGE of crude and purified inulinase preparations. Samplesweresubjectedtonondenaturing PAGEon a 10% polyacrylamide gel, and the gel was stained for

sucrose-hydrolyzing activity. Lanes: 1, crude cell wall inulinase; 2, purified cell wallinulinase; 3, crudesupernatantinulinase; 4, purified

super-natantinulinase.

the composition of either the supernatant inulinase orthe

cell wall inulinase (Fig. 5, lanes 1 and 2 and lanes 3 and4, respectively). Inulinase treated with SDS showed no su-crose-hydrolyzing activityuponnative PAGE.

To substantiate that theuseof thiolsasinulinase-releasing

agentsdidnotaffectthemultimericstructureof the cell wall inulinase, inulinase released by incubating cells for 24 h at

4°C in 50 mM potassium phosphate (pH 6) was purified. After Sephadex A-50 elution and subsequent FPLC gel filtration, the preparation of cell wall inulinase that was

obtained showed profiles and molecular weights (as deter-mined by SDS-PAGE) which did not differ from those observed for the cell wall inulinase released by thiols. Nondenaturing PAGE of crude preparations of cell wall inulinasefurther confirmed that themode ofenzymerelease did notaffect the structure of nativeinulinases (Fig. 6, lane 2).

Theuse ofenzyme release buffer did not result in com-pleteremoval of inulinase from the cells:approximately15% of the total inulinase produced by K. marxianus still

re-mained attachedtothe cells. Incubation of these cells in a

solution containing 2-mercaptoethanol, Triton X-100, and KCl resulted inafurther releaseof inulinase.However,after

thistreatment,cells still contained 3% of the total inulinase produced. This residual enzyme was solubilized by sonic disintegration of the cells. On nondenaturing PAGE, the

inulinase in thesepreparationsshowed the samemobilityas the cell wall inulinasethatwasremovedfromthecellwallby the enzyme release buffer (Fig. 6, lanes 3 through 5). The

sucrose-hydrolyzing activity was verified by running com-merciallyavailable(Boehringer)extracellular invertase ofS.

cerevisiae (molecular mass, ca. 800 kDa) on the same gel (Fig. 6, lane 1).

Kineticconstantsofpurifiedsupernatantinulinases andcell

wall inulinases. The apparent Km values of inulinase for

sucrose and raffinose appeared to be independent of the

degree ofoligomerization. Both the supernatant (dimeric) inulinase and the cell wall (tetrameric) inulinase showed

apparent Km values of14.6 and 5.5 mM with sucrose and

raffinose, respectively (Table 1).Noapparent Kmvalue with

1 2 3 4 5

FIG. 6. Nondenaturing PAGE of inulinase solubilized from the cellsby variousconsecutivetreatments.Cellsweresuspended in 50

mMpotassium phosphate buffer (pH 6) for 24 hat4°C (lane 2)orin

enzyme release buffer for 2 h at 30°C (lane 3). Inulinase still associated with thecells aftertreatmentwithenzymerelease buffer

wasinpartsolubilized by incubating the cells in buffer containing 0.1% TritonX-100, 10 mM2-mercaptoethanol,and0.1MKCl for4 hat 30°C (lane 4) and in partby sonication (lane 5) of the cells. Commercially available S. cerevisiae invertase (lane 1) and

inuli-nasesreleased by eachtreatmentwereappliedtoa7%

nondenatur-ingpolyacrylamidegel.

chicory inulin could becalculated,ashad alsobeenobserved for crude inulinase preparations (32). Because ofthe low solubility of this fructosepolymer, the maximalinulin con-centration that can be achieved iswell below theapparent

Kmvalues.

Amajorproblem in determining specificenzymeactivities can be the establishment of protein concentrations. This is especially true in the case of chromophoric proteins and glycoproteins. As a result, large differences between

ob-served and true protein concentrations may occur. Colori-metricprotein determinations include peptide-specific reac-tions, as in the reduction of Folin-Ciocalteu reagent by tryptophanandtyrosine residues in the Lowrymethod (25) or dye-binding properties in the Bradford method (5). The responseofthesemethodsdepends bothonthenatureof the

polypeptide and on the presence of either chromophoric groups or carbohydrate chains and may vary with the amountofprotein. Whenusingthesemethods, standardiza-tion with proteins ofcomparable properties is a necessity. Onthe other hand, thedetermination ofprotein concentra-tionsonthe basis of the carbon content isindependentofthe

peptide composition or the presence of chromophoric groups, andthe method need notbestandardized on other proteins. To convert the measured carbon content of a purified glycoprotein preparation into the protein content, only the degree of glycosylation should be known (see Materials and Methods).

Asexpected,the valuesobtainedfor thespecificinulinase activities strongly depended on the method employed to

determineprotein concentrationsintheinulinase solutions. Significantdifferences in theobtainedspecificactivitieswere observedbetween theLowrymethod,the Bradfordmethod, and the TOC method. The Lowry method tended to give lower values than TOC-calculated activities, whereas the Bradford method gave higher values. With all three meth-ods,thespecificactivitiesof the inulinase dimerwerehigher

than those ofthe inulinase tetramer. Assuminga degree of glycosylationof26%,thespecificsucrose-hydrolyzing activ-itiesof dimeric and tetrameric inulinasewereapproximately 1,840 and 610Umgof inulinaseprotein-1and1,370and460 Umgofglycosylated protein-', respectively (Table 1). The

maximal concentration of chicory inulin that conveniently could be used formeasuringthevelocityof inulinhydrolysis was 2% (wt/vol). At this concentration, the activities of

dimeric inulinase and tetrameric inulinasewere119 and 55 U

mgof inulinase

protein-1

or88 and40 Umgofinulinase-1.

The ratioofthe activitiesofpureinulinasewithsucroseand chicory inulin, determined at pH 4.5 and 50°C and at a saturating sucrose concentration and at 2% inulin, respec-tively,wasabout16. This value for theS/Iratio isconsistent withthat ofcrudeinulinase preparations (32).

The inulinases had an absorption maximum at 280 nm. Specific absorption coefficients were calculated from the

inulinaseconcentrations determined by TOCand the absorp-tion of inulinase diluabsorp-tionsat 280nm. These coefficients for solutionsof 1mg/ml

(A2O-')

were1.03 +0.01and1.00 0.01 for the inulinase dimer andtetramer, respectively.

N-terminal amino acid sequence ofinulinase. In order to

clarify the relationship between the inulinase of K. marx-ianusCBS 6556 and the invertaseof S.cerevisiae, theamino acid sequence ofthe N-terminal end of inulinase was

ana-TABLE 1. ApparentKmvalues andspecific activitiesofsupernatant(dimer) inulinase and cell wall(tetramer)inulinase of K. marxianus CBS6556forthree substrateswith different degrees ofpolymerizationa

Spactof inulinase (Umgofprotein-')

Substrate Inulinase Km Cb(UMg

(mM) Lowry' Bradford' TOCb TOfnUlne

ofinulinase- ) Sucrose Dimer 14.6 923 3,331 1,875 1,388 Tetramer 14.7 806 2,069 856 634 Raffinose Dimer 5.6 312 1,126 634 469 Tetramer 5.3 261 670 278 205 Inulind Dimer 59 212 119 88 Tetramer 51 132 55 40

a Protein concentrationswere determinedby theFolin Ciocalteu methodofLowry et al. (25), theBradford Coomassie brilliant blue method (5), and by

measuring the carbon contentin theinulinase solutionsby TOC.

bForcalculation of theconcentrationsofinulinase proteinand of nativeglycosylated inulinase,aglycosylationof 26%was used.

Bovineserumalbuminwasusedas astandard.

I

WkijHa

itvertase

WLmru8o

10 Ser- Gy-Asp- Ser - Lys-Ala- lie- TIr- ? -

Th-Ser - Mst - MTy - AM - GIu - TN' - Smr - Amp - Arg - Pro

-11 20

f -Phe- ? - Leu-An-Arg- Pro-Ser-Val-Tyr

-Invmrtam Lou -Vai - Hs - Phe - Ttr - Pro - An - Lys- Gy - Trp

-FIG. 7. Amino acid sequence of the amino-terminal end of the inulinase polypeptide of K. marxianus CBS 6556 and the invertase

polypeptide ofS. cerevisiae. Theamino-terminal sequence of invertase was obtained fromCarlson etal. (7).

lyzed and compared with the amino-terminal sequence of invertase as obtained by Carlson et al. (7) (Fig. 7). Apart from the serine residue at position 1, of the 20 amino acids determined no two amino acids aligned between the amino-terminal sequence of inulinase and invertase. Thus, there was no homology in the N-terminal sequences. Partial ami-no-terminal sequence analysis of both the purified superna-tant inulinase and the purified cell wall inulinase was con-ducted. Theamino acid sequences of the first 20 amino acid residuesof supernatant inulinase and cell wall inulinase were identical.

DISCUSSION

Structure oftheextracellular inulinase ofK. marxianus. In the present paper we have presented evidence showing that theinulinase secreted into the culture fluid and the inulinase retained in the cell wall of K. marxianus CBS 6556 have

similar subunits, consisting of a 64-kDa polypeptide that contains 26 to 37%of the molecular mass as carbohydrate. The two inulinase forms differ in size because of their differences in subunit aggregation.

Inulinase monomer may contain varying amounts of

car-bohydrate, as suggested by the diffuse protein bands that ranged from 87 to 102 kDa on SDS-PAGE (Fig. 4). This

outer-chain carbohydrate can be removed by treatment of denaturedinulinase withEndo-H, indicatingthatit is present ashigh-mannoseoligosaccharide chains linkedtoasparagine

residuesofthepolypeptide chain.Thepolydisperse behavior ofglycoproteinsonSDS-PAGEhasalso been observed with yeast invertase (9, 34). This dispersity appeared to be an effectofdifferentlengthsof thecarbohydratechains andnot of differencesin thenumberofcarbohydrate chainsattached tothe invertase polypeptide (9).

Duringthefirst stepofpurificationof the intact inulinases with anion-exchange chromatography, it was obvious that the supernatantinulinase and thecellwallinulinasediffered

(Fig.1). Cell wall inulinase elutedfromthecolumnat alower

sodium chloride concentration than supernatant inulinase.

Such

chromatographic

behavior can be caused

by

a differ-encein theamount ofcarbohydrateattachedor

by

a differ-ence in the quaternary structure that masks the

DEAE-binding properties of the

polypeptide

moieties (35). NondenaturingPAGE and FPLCgelfiltrationdemonstrated that theinulinases differ insize,but theanomalousbehavior

ofglycoproteinsin

electrophoresis

and in

gel

filtration makes it difficult to obtain accurate molecular weights. Ferguson analysisandanalytical gelfiltration gave estimated values of 165 to 180 kDa and 335 to 450 kDa for the supernatant inulinase and the cell wall inulinase,

respectively

(Fig. 3).

These valuescoincide with a dimer structure of the purified supernatant enzyme and atetrameric structure of the puri-fied cell wall enzyme. During purification, the structure of the inulinasesdid not alter. In addition, the use of 2-mercap-toethanol, in theinitialstepof inulinase release from the cell wall, did notinfluence the oligomeric structure of cell wall inulinase. The release of cell wall inulinase by potassium phosphate alone led to the solubilization of tetrameric inu-linase. Moreover, inulinase that could only be solubilized by incubating the cells in a buffer containingKCl,Triton X-100, and 2-mercaptoethanol or by cell disruption showed the sameelectrophoretic behavioronnondenaturing PAGE that

purifiedcell wall enzyme did(Fig. 5and 6). Apparently, the inulinase tetramer is the largest inulinase aggregate present in K. marxianus CBS 6556. The absence of extracellular inulinase monomer as well as activity with any peak smaller than a dimer, asdetermined by gel filtration, precludes the

recognitionofactivity with the monomeric form of inulinase. Indeed, during purification of supernatant inulinase with FPLC gel filtration, we observed a contaminating protein

with an apparentmolecularweight of 72 kDa. Although no

activity could be measured, this protein could represent inulinase monomerformed by the dissociation of multimer inulinase resulting from shear force and heat formation duringconcentration with the hollow-fiber device. The oc-currence of dissociation ofoligomers by heatformation or shear force was recentlyobserved with multimer invertase

(9, 13).

Ofthedifferentyeastinulinases reported, onlyafew have beenpurifiedandnoinformationis availableontheirsubunit structure.WorkmanandDay (39)

purified

cell wall inulinase

ofK.fragilis ATCC12424andstatedthatthis enzymewas a glycoprotein containing 66% carbohydrate, a percentage

exceeding by far the value reported here.

Unfortunately,

Workman andDay(39)didnotmention the molecular

weight

oftheir

purified

inulinase in which they estimated the per-centageofcarbohydrate.Amolecular

weight

of250kDawas

cursorily

mentioned forK. marxianus inulinase

by

Uhm et al. (36). More information is available on the

nonspecific

exoinulinases of

fungi.

The molecular masses of inulinase subunitsdeterminedby SDS-PAGEwerefoundtobe 74 kDa for

Aspergillusficuum

(14),81 kDafor

Aspergillus

sp.

(3),

85 kDa for Aspergillus niger(36), 70 and 84 kDafor

Chryso-sporium pannorum (40), and60 and 80 kDafor Penicillium sp. 1 (27), and all contained 20 to

40%

carbohydrate.

Moreover, native inulinase of A. niger

appeared

to be a

tetramerwith a molecular mass of

approximately

315 kDa

(36),asdetermined

by

analytical gel

filtration. Themolecular masses and the degrees of

glycosylation

of these

fungal

inulinasesare in the same range as those of theextracellular inulinases of K. marxianus.

Kinetic properties of the extracellular inulinases of K. marxianus. Inulinase dimer and tetramer had identical appar-ent Km values and specificabsorption coefficients at 280 nm, but the specific enzyme activities were different. Recently, wereported on the kinetic parameters of nonpurified inulin-ase preparations (32). The apparent Km values of inulinase

dimer and inulinase tetramer with sucrose (both 14.6 mM) had not altered during purification. However, an apparent Km of 5.6 mM with raffinose of thepurified inulinase differs

from the 8.2 mM observed for nonpurified inulinase. The reason for these slightly different Km values is not known (32).

The specificenzyme activity ofinulinase dimer was

con-sistently higherthan thespecificenzymeactivityof inulinase tetramer. Aggregation apparently leads to negative

cooper-ativity forenzyme activity. For the S. cerevisiae invertase, no such difference between specific enzyme activity of invertase dimer and specific enzyme activity of invertase octamerhas been observed(13). The specific enzyme activ-ities of purified inulinases were markedly lower than those reported for crude inulinases (32). The latter activities were calculated withprotein concentrations determined with the

Bradford protein assay. The Bradford protein assay

under-estimates the actual protein content 2.5- to 3-fold as

com-pared withvalues obtainedbytheTOC method. As aresult,

the specific inulinase activitieswill be too high.

Invertase versus inulinase. Inadditiontodifferences in the

S/I ratio and the apparent affinity constants, two further differences between yeast invertase and yeast inulinase are encountered when enzyme structures are compared. (i)The

62-kDapolypeptide backbone observed for Endo-H-treated

extracellularinvertase (9, 35) isslightly smaller thanthat of

theinulinasereported here. (ii) Thecarbohydrateattached to thepolypeptide ofinvertase accountsfor50% of the molec-ular mass ofinvertase (35), while in inulinasethe

carbohy-drate content is only around 30%. This difference could

reflecta smaller number of glycosylation sites in the inulin-ase polypeptide and thus a dissimilarity in polypeptide structuresof the twoproteins. Anotherfindingthatsuggests

different polypeptide structures is the low homology in

amino acid sequences at the amino-terminal ends of inulin-aseand invertase. Preliminary experiments with oligonucle-otides derived fromthe amino acid sequence of the amino-terminal end of K. marxianus CBS 6556 inulinase did not result in any detectable hybridization with the genome of various Saccharomyces yeasts (J. Verbakel [Unilever

Re-search, Vlaardingen, TheNetherlands], personal

communi-cation). The SUC gene family has been reported to be a

highly conserved sequence in Saccharomyces yeast (6).

Strong homology has been demonstrated evenbetweenthe SUC2 gene of S. cerevisiae and the invertase genes of the mold Neurospora crassa

(8)

and the bacterium Bacillus

subtilis(26). Remarkably, the S. cerevisiae SUC2 sequence hardly hybridized

with

thegenome of the yeast S. kluyveri UCD 51-242 (6). Further research into the nature of the sucrose-hydrolyzing activity of S. kluyveri showed that this yeast possesses an inulinase, as defined by S/I ratio and

behavioron native gelelectrophoresis (31).

Retentionof glycoproteins within the periplasmic space of thecell wall ofyeasts is thought to be caused by a

perme-ability barrier in the outer regions of the cell wall (13, 20).

Oligomerization ofglycoproteins may then play a role in this cell wallretention. Almost all of theinvertase produced by S. cerevisiae is ultimately retained in the cell wall as an

octamericcomplex of four invertase dimers. Invertase that is completely secreted into the culture fluid and invertase that can be released by the use of sulfhydryls are invertase dimers (13). Incontrast tothe S. cerevisiae invertase, both the cell wall-retained inulinase and the inulinase released from the cell wall ofK. marxianus by thiol treatment are tetramers. This probably is a consequence of differences between thetwoyeastsinsusceptibility of their cell walls to sulfhydryls. The cellwallofK. marxianus maybelessrigid than the cell wall of S. cerevisiae, and hence treatment with sulfhydryls might resultin a morepronounced elimination of thepermeability barrier (10, 38).

Various treatments such as shear force, heat, low-ionic-strength buffer, and sonic disintegration of the cells have

been reported to cause disintegration of the S. cerevisiae invertase octamer, resulting in the appearance ofdimeric invertase without any changeinkinetic parameters (9, 13). We did not observeadecayof inulinasetetramerinto dimers duringpurification. Onthecontrary, wecouldinduce tetra-merdecayonly by the use of enzyme-denaturing agents such asSDS orguanidine hydrochloride (results notshown). The tetramer inulinase thus appears to be a very stable form. However, we cannotexcludethepossibilitythatsupernatant inulinase is actually a tetramer that is dissociated during

concentration of culture fluid by the use of a hollow-fiber device.

ACKNOWLEDGMENTS

This work was supported by Unilever Research, Vlaardingen,

TheNetherlands, and theDutchMinistry of Economic Affairs.

LITERATURECITED

1. Arnold,W.N.1981. Enzymes, p. 1-46. In W. N.Arnold(ed.),

Yeastcellenvelopes: biochemistry, biophysics and ultrastruc-ture,vol. 2.CRC Press,BocaRaton, Fla.

2. Arnold, W. N. 1987.Hydrolyticenzymes, p.369-391. In D. R.

Berry,I.Russell, andG.G. Stewart (ed.), Yeastbiotechnology, 2nd ed. Allen&Unwin, London.

3. Azhari, R., A. M. Szlak, E. Elan, S. Sideman, and N. Lotaii. 1989.Purification andcharacterization of endo- and

exo-inulin-ase.Biotechnol.Appl. Biochem.11:105-117.

4. Beeley, J. G. 1985.Glycoprotein and proteoglycan techniques.

InR. H.Burden andP. H. vanKnippenberg(ed.), Laboratory techniques in biochemistry and molecular biology, vol. 16.

ElsevierBiomedical Press, Amsterdam.

5. Bradford, M. M. 1976. A rapid and sensitive method forthe quantitation ofmicrogram of protein utilizing the principle of protein dye-binding.Anal.Biochem. 72:248-254.

6. Carlson, M., and D. Botstein. 1983. Organization of the SUC

genefamily in Saccharomyces. Mol. Cell. Biol.3:351-359. 7. Carlson, M.,R. Taussig, S. Kustu, and D. Botstein. 1983.The

secreted form of invertase in Saccharomyces cerevisiae is

synthesizedfrommRNA encodingasignalsequence. Mol.Cell. Biol.3:439-447.

8. Casanova,M.,J. P. Martinez, M. L. Gil,R.Sentandreu, and J.

Ruiz-Herrera. 1987. Different molecular forms of invertase in

the slime variant ofNeurospora crassa: comparison with the

wild-type strain.J.Gen. Microbiol. 133:2447-2459.

9. Chu,F.K.,W.Watorek,and F.Maley. 1983.Factors affecting

the oligomeric structure of yeast external invertase. Arch.

Biochem. Biophys.223:543-565.

10. Davies, R., and F. J. Wayman. 1975. The effect ofthiols on

Saccharomyces fragilis. Antonievan Leeuwenhoek J. Micro-biol. Serol. 41:33-58.

11. Edelman, J., and T. G. Jefford. 1964. The metabolism of fructosepolymers in plants.4.j3-Fructofuranosidasesof tubers

12. Esmon, B., P. C. Esmon, and R.Schekman. 1984. Early steps in theprocessing of yeastglycoproteins. J. Biol. Chem. 259:10322-10327.

13. Esmon, P. C., B. Esmon, I. E. Schauer, A. Taylor, and R.

Schekman. 1987. Structure, assembly, and secretion of oc-tameric invertase. J. Biol. Chem. 262:4387-4394.

14. Ettalibi, M., and J. C. Baratti. 1987.Purification, properties and comparison of invertase, exoinulinases and endoinulinases of Aspergillus ficuum. Appl. Microbiol. Biotechnol. 26:13-20. 15. Flood, A. E., P. P.Rutherford, and E. W. Weston. 1967. Effects

of2:4-dichlorophenoxyacetic acid on enzyme systems in Jerus-alem artichoke tubers and chicory roots. Nature (London) 214:1049-1050.

16. Fuchs, A., J. M. deBruin, and C. J.Niedeveld. 1985. Bacteria and yeasts as possible candidates for the production of inuli-nases and levainuli-nases. Antonie van Leeuwenhoek J. Microbiol. Serol. 51:333-351.

17. GrootWassink, J. W. D., and S. E. Fleming. 1980. Non-specific P-fructofuranosidase (inulase) from Kluyveromyces fragilis: batch and continuous fermentation, simple recovery method, andsomeindustrialproperties. Enzyme Microb. Technol. 2:45-53.

18. Grossmann, M. K., and F. K. Zimmermann. 1979. The struc-tural genesofinternal invertases in Saccharomyces cerevisiae. Mol.Gen. Genet. 175:223-229.

19. Haschemeyer, R. H., and A. E. V. Haschemeyer. 1973. Proteins.

Aguidetostudy byphysicalandchemical methods. John Wiley &Sons, Inc., New York.

20. Kidby, D. K., and R. Davies. 1970. Thiol induced release of invertasefrom cell walls of Kluyveromycesfragilis. Biochim. Biophys. Acta 201:261-266.

21. Kovaljova, N. S., andV. V.Yurkevich. 1978.Ontheexistence of 1-fructofuranosidases of inulinasetypeinyeast, p. SV3. InP.

Galzy, A.Arnaud, C. Bizeau, and G. Moulin (ed.), Proceedings of the VIth International Specialized Symposium on Yeasts:

Metabolism and regulation of cellular processes. ENSAM-CRAM, Montpellier, France.

22. Laemmli, U.K.1970.Cleavage of structuralproteins during the assembly of the head of bacteriophage T4. Nature (London) 227:680-685.

23. Lam, K. S., andJ. W. D.GrootWassink. 1983. Efficient

non-killing extraction of ,-D-fructofuranosidase (an exo-inulase) from Kluyveromyces fragilis athigh cell-density. Enzyme Mi-crob.Technol. 7:239-242.

24. Lehle,L., R. E. Cohen, andC. E.Baliou. 1979. Carbohydrate

structureofyeast invertase: demonstration ofaformwith only

core oligosaccharides and aformwith completed polysaccha-ridechains.J. Biol. Chem. 254:12209-12218.

25. Lowry,0.H., N.J.Rosebrough,A. L.Farr,and R.J.Randall.

1951. Protein measurement with the Folin phenol reagent. J.

Biol. Chem. 193:265-275.

26. Martin, J., M. DebarbouilHe, E. Ferrari, A. Klier, and G.

Rapoport. 1987.Characterization of the levanase gene of Bacil-lus subtilis which showshomology to yeast invertase. Mol.Gen. Genet. 208:177-184.

27. Nakamura, T., T. Kurokawa, S. Nakatsu, and S. Ueda. 1978.

Crystallization and general properties ofanextracellular

inulin-asefromAspergillus sp. Nippon Nogei Kagakukoishi

52:159-166.

28. Negoro, H., and K. Kito. 1973. ,-Fructofuranosidase from Candidakefyr. J. Ferment.Technol. 51:96-102.

29. Neumann, N. P., and J. 0. Lampen. 1969. The glycoprotein structureof yeast invertase. Biochemistry 8:3552-3556.

30. Novick, P., S. Ferro,andR.Schekman. 1981. Order of eventsin

theyeast secretory pathway. Cell 25:461-469.

31. Rouwenhorst, R. J., W. S. Ritmeester, W. A. Scheffers, and J. P.

van Dijken. 1990. Localization of inulinase and invertase in

Kluyveromyces species. Appl. Environ. Microbiol.

56:3329-3336.

32. Rouwenhorst, R. J., L. E. Visser, A. A. van der Baan, W. A.

Scheffers,andJ.P. vanDiken.1988. Production, distribution, and kinetic properties of inulinase in continuous cultures of Kluyveromyces marxianus CBS 6556. Appl. Environ. Micro-biol. 54:1131-1137.

33. Snyder, H. E., and H. J. Phaff. 1960. Studies on a beta-fructosidase (inulinase) produced by Saccharomyces fragilis. AntonievanLeeuwenhoek J. Microbiol. Serol. 26:443-452.

34. Tarentino, A. L., T. H. Plummer, and F. Maley. 1974. The releaseofintact oligosaccharides from specificglycoproteins by endo-3-N-acetylglucosaminidase.J. Biol. Chem. 249:818-824. 35. Trimble, R. B., and F. Maley. 1977. Subunit structure of

external invertase from Saccharomyces serevisiae. J. Biol. Chem. 252:4409-4412.

36. Uhm, T. B.,D.Y.Jeon, S. M. Buyn, J. S. Hong, and J. W. D.

Grootwassink. 1987. Purification andpropertiesof ,B-fructofura-nosidase fromAspergillus niger. Biochim. Biophys. Acta 926:

119-126.

37. Vandamme, E. J., and D. G. Derycke. 1983. Microbial

inuli-nases:fermentation process, properties andapplications. Adv. Appl. Microbiol. 29:139-176.

38. Weimberg, R.,and W. L. Orton. 1966. Elution of exocellular enzymes fromSaccharomycesfragilisandSaccharomyces ce-revisiae. J. Bacteriol.91:1-13.

39. Workman,W.E., and D. F. Day. 1983. Purification and prop-ertiesofthe 3-fructofuranosidasefromKluyveromycesfragilis. FEBS Lett. 160:16-20.

40. Xiao, R., M. Tanida, and S. Takao. 1989. Purification and

characteristics oftwoexoinulinases from Chrysosporium pan-norum.J. Ferment. Bioeng.67:331-334.