Comparison of cell wall proteins of Saccharomyces cerevisiae as anchors for cell surface expression of heterologous proteins

APPLIED AND ENVIRONMENTAL MICROBIOLOGY,0099-2240/97/$04.0010

Feb. 1997, p. 615–620 Vol. 63, No. 2

Copyright q 1997, American Society for Microbiology

Comparison of Cell Wall Proteins of Saccharomyces cerevisiae asAnchors for Cell Surface Expression of Heterologous Proteins

J. MARCEL VAN DER VAART,1,2* ROB TE BIESEBEKE,1 JOHN W. CHAPMAN,2

HOLGER Y. TOSCHKA,3 FRANS M. KLIS,4 AND C. THEO VERRIPS1,2

Department of Molecular Cell Biology, University of Utrecht, 3584 CH Utrecht,1 Unilever Research Laboratorium,3133 AT Vlaardingen,2 and Department of Molecular Cell Biology, Biotechnology Centre, University of Amsterdam,

1098 SM Amsterdam,4 The Netherlands, and Langnese Iglo GmbH, 31513 Wunsdorf, Germany3

Received 11 October 1996/Accepted 5 December 1996

The carboxyl-terminal regions of five cell wall proteins (Cwp1p, Cwp2p, Aga1p, Tip1p, and Flo1p) and threepotential cell wall proteins (Sed1p, YCR89w, and Tir1p) all proved capable of immobilizing a-galactosidase inthe cell wall of Saccharomyces cerevisiae. The fraction of the total amount of fusion protein that was localizedto the cell wall varied depending on the anchor domain used. The highest proportion of cell wall incorporationwas achieved with Cwp2p, Aga1p, or Sed1p as an anchor. Although 80% of these fusion proteins wereincorporated in the cell wall, the total production of a-galactosidase–Aga1p was sixfold lower than that ofa-galactosidase–Cwp2p and eightfold lower than that of a-galactosidase–Sed1p. Differences in mRNA levelswere not responsible for this discrepancy, nor was an intracellular accumulation of a-galactosidase–Aga1pdetectable. A lower translation efficiency of the a-galactosidase–AGa1 fusion construct is most likely to beresponsible for the low level of protein production. a-Galactosidase immobilized by the carboxyl-terminal 67amino acids of Cwp2p was most effective in the hydrolysis of the high-molecular-weight substrate guar gumfrom Cyamopsis tetragonoloba. This indicates that the use of a large anchoring domain does not necessarilyresult in a better exposure of the immobilized enzyme to the exterior of the yeast cell.

Expression of proteins on the cell surface of microorganismscan be used for production of live vaccines, whole cell adsor-bents, and display and selection of antibody libraries (6, 16).Surface display of heterologous proteins has been describedfor a large number of microorganisms. In gram-negative bac-teria, outer membrane proteins (1, 5), lipoproteins (11), fim-briae (12), and flagellar proteins (22) have all been used toimmobilize heterologous proteins on the cell surface. In gram-positive bacteria, cell surface localization of several heterolo-gous proteins was achieved by making use of protein A (7, 26),M6 protein (24), and the fibronectin-binding protein as an-chors (8). Enzymes immobilized on the surface of eukaryotessuch as yeast cells provide a way to develop a new generationof enzyme reactors. Such an enzymatic catalyst can be simplyremoved from the reaction medium by filtration or centrifuga-tion and, if necessary, can be regenerated easily. In fungi,immobilization of heterologous proteins is possible by makinguse of the C-terminal anchor domains of glucanase-extractablecell wall proteins of Saccharomyces cerevisiae (15, 28). Twotypes of mannoproteins are present in the cell wall of S. cer-evisiae (3, 14). The sodium dodecyl sulfate (SDS)-extractablemannoproteins are loosely associated with the cell wall, and atleast a proportion are likely to be plasma membrane proteins(33). Based on the thickness of the cell wall and the size of theimmobilized proteins, it is most likely that enzymes immobi-lized on the cell membrane will not be exposed on the outsideof the cell wall and therefore will not be available for a sub-strate which is not capable of penetrating the cell wall. The useof an SDS-extractable mannoprotein as an anchoring domainfor enzymes in the cell wall is therefore less attractive. Thesecond type of mannoprotein is the glucanase-extractable man-

noproteins, which are released after glucanase digestion of theglucan layer of the cell wall and not by SDS extraction. Theseproteins are generally rich in serine and/or threonine and theyall contain a putative glycosyl phosphatidylinositol (GPI) at-tachment signal. Schreuder et al. (27) showed that a fusionprotein consisting of the invertase signal sequence, the guara-galactosidase coding sequence, and the C-terminal half ofa-agglutinin was incorporated into the cell wall and was onlyreleased by glucanase extraction of isolated cell walls. In con-trast, a control protein lacking the a-agglutinin part was re-leased into the medium. a-Agglutinin also proved to be capa-ble of surface expression of lipase from Humicola lanuginosa,cutinase from Fusarium solani, and hepatitis B virus surfaceantigen (28).In this study, we have compared the incorporation capacity

of the C-terminal half of a-agglutinin with those of otherglucanase-extractable cell wall proteins and of proteins thatcontain all characteristics of cell wall proteins. Chimeric pro-teins consisting of a-galactosidase fused to the C-terminal re-gions of the cell wall proteins Flo1p (31), Cwp1p, Cwp2p, andTip1p (32) and the potential cell wall proteins Sed1p (9), Tir1p(Srp1p [19]) and YCR89w (23) were tested for their ability toanchor a-galactosidase in the cell wall of S. cerevisiae. TheC-terminal regions of all of these protein regions proved ca-pable of immobilizing a-galactosidase on cell walls. The largestfraction of the total amount of fusion protein produced wasfound localized to the cell wall when the C-terminal parts ofCwp2p, Sed1p, or Aga1p were used.

MATERIALS AND METHODS

Strains and media. The Escherichia coli strain used in this study was JM109[endA1 recA1 gyrA96 thi hsdR17 (rk2, mk1) relA1 supE44 D(lac-proAB) (F9traD36 proAB lacIqZDM15)] (36) and was grown in Luria broth (25) with 100 mgof ampicillin/ml when appropriate. The S. cerevisiae yeast strains used were SU50(YT6-2-1 L) MATa cir0 leu2-3,112 his4-519 can1 (4), SU51 (YT6-2-1 L) MATacir1 leu2-3,112 his4-519 can1, SU52 (YT6-2-1 L) MATa cir1 leu2-3,112 his4-519ura3 can1, and SU52cwp2::URA3 (YT6-2-1 L) MATa cir1 cwp2 leu2-3,112 his4-

* Corresponding author. Mailing address: Unilever Research Labo-ratorium, Olivier van Noortlaan 120, 3133 AT Vlaardingen, The Neth-erlands. Phone: 31-10 460 5263. Fax: 31-10 460 5383. E-mail: [email protected].

615

519 can1. Yeast strains were grown in YPD (1% [wt/vol] Difco yeast extract, 2%[wt/vol] Difco Bacto-Peptone, 2% [wt/vol] glucose) or, when induction of theGAL7 promoter was appropriate, in YPG (1% [wt/vol] Difco yeast extract, 2%[wt/vol] Difco Bacto-Peptone, 5% [wt/vol] galactose), or in synthetic minimalmedium (MM) consisting of 0.7% (wt/vol) Difco yeast nitrogen base (YNB), 2%(wt/vol) glucose, and amino acids as necessary (29). The E. coli strain and the S.cerevisiae strains were transformed as described by Van Der Vaart et al. (32).Reagents.DNA restriction and modification enzymes were from New England

Biolabs, Inc. (Beverly, Mass.), and Boehringer Mannheim Biochemicals (Mann-heim, Germany). Phenylmethylsulfonyl fluoride, the chromogenic substrate X-a-Gal (5-bromo-4-chloro-3-indolyl-a-D-galactopyranoside), b-galactose dehydro-genase, NAD, and pepstatin were purchased from Boehringer MannheimBiochemicals. Benzamidine-HCl was purchased from ICN Biochemicals (CostaMesa, Calif.), and laminarinase was purchased from Mollusk (L5144). Glassbeads with a diameter of 425 to 600 mm and guar gum from Cyamopsis tet-ragonoloba were obtained from Sigma Chemical Co. (St. Louis, Mo.). Polyvinyli-dene difluoride membranes (Immobilon P) were from Millipore Corp. (Bedford,Mass). The secondary goat anti-rabbit antiserum and nitrocellulose membranewere purchased from Bio-Rad Laboratories (Hercules, Calif.). The ECL en-hanced chemiluminescence kit and the Hybond-N membrane were obtainedfrom Amersham (Arlington Heights, Ill.). Zymolyase 100T was purchased fromSeikagaku Kogyo Co. (Tokyo, Japan).PCR amplification. PCR amplifications (Perkin-Elmer Cetus DNA Thermal

Cycler) were performed with 100 ml of 10 mM Tris-HCl (pH 8.3)–50 mMKCl–1.5 mM MgCl2–0.001% gelatin, with 0.2 mM (each) deoxynucleosidetriphosphate, 100 pmol of the DNA oligonucleotides, ;0.25 mg ofHindIII-digested chromosomal DNA of SU50 (or 2 ng of plasmid DNA) and 1U of Ampli-Taq polymerase. Incubation parameters were set as follows: 25 cyclesof 1 min at 958C, 1.5 min at 548C, and 2 min at 728C.Construction of fusion fragments. Fusions of the C-terminal ends of CWP1

(642 bp), CWP2 (204 bp), TIP1 (435 bp), TIR1 (639 bp), SED1 (690 bp),YCR89w (942 and 2,235 bp), or FLO1 (1,035 and 1,791 bp) with the codingregion of a-galactosidase were created (Fig. 1). The a-galactosidase–AGa1 fu-sion construct was created by Schreuder et al. (27). An NheI recognition se-quence was created at the C terminus of the a-galactosidase gene by insertion ofan oligonucleotide linker adapter (59 CAAGGAGCGCTAGCGGTACCGAAAGTTAACA 39 and 59 TCGAACAATTGAAAGCCATGGCGATCGCGAG 39)into the StyI-HindIII site of pUR2650 (DNA sequence of invertase signal se-quence and a-galactosidase coding region from pUR2740 [35]) to give pUR2984.At the 59 end of the C terminus of CWP1, an NheI recognition site was createdby PCR with the primer 59 GCGCGCTAGCCTGGTGAGTATCCGTTCC 39.The resulting PCR fragment (with primer GCGCAAGCTTAGGAGGACAATGTAAGA 39 at the 39 end) was inserted in frame at the NheI recognition site inpUR2984. The fusion constructs consisting of a-galactosidase and CWP2, TIP1,TIR1, SED1, or YCR89w were created by the same principle. The fusion con-structs described above were cloned into the SacI-HindIII-digested episomalexpression vector pSY16 (10). The plasmid containing the a-galactosidase–FLO1(1,035 bp) fusion construct (pUR2991) was created by insertion of the SnaBI-PvuII DNA fragment of pYY105 (39 end of the FLO1 open reading frame) (31)into HpaI-linearized pUR2984. Plasmid pUR2992, which contains the a-galac-tosidase–FLO1 (1,791 bp) fusion construct, was created by insertion of theKpnI-BglII fragment from pYY105 into KpnI (partial)-BglII-digested pUR2991.Both a-galactosidase–FLO1 fusion constructs were cloned into the SacI-HindIII-digested multicopy rDNA (gene coding for rRNA) integration plasmid

pUR2778, which is based on pMIRY2 (17). The final plasmids were namedpUR2994 and pUR2995, respectively.Galactose induction of fusion protein expression and the a-galactosidase

activity assay. Yeast transformants carrying the fusion protein genes were pre-grown in MM containing histidine and subsequently diluted 1:100 into YPG.These cultures were grown at 308C for 17 h, and approximately 10 3 108 cellswere harvested for a-galactosidase activity measurement. To analyze the incor-poration of the fusion protein into the cell wall, transformants carrying the fusionprotein gene were pregrown in 50 ml of MM containing histidine, after which thegrowth medium was replaced by YPG. The optical density at 660 nm (OD660) ofthe cultures and the a-galactosidase enzyme activities on cells and in the growthmedium were subsequently followed in time. a-Galactosidase activities weredetermined by the ability of a-galactosidase to convert the chromogenic sub-strate X-a-Gal. Fifty microliters of sample was mixed with 450 ml of this substrate(20 mM) in 0.1 M NaAc (pH 5) and incubated for 5 min at 378C. The reactionwas stopped by the addition of 1 ml of 10% (wt/vol) Na2CO3. The enzyme activitywas measured as the OD405. The enzyme unit is defined as the amount of enzymewhich hydrolyzes 1 mmol of substrate in 1 min at 378C at pH 5.Western and dot blot analysis. SDS- and laminarinase-extracted cell wall

mannoproteins, isolated as described by Schreuder et al. (27), and mediumproteins were separated by polyacrylamide gel electrophoresis and transferredonto Immobilon polyvinylidene difluoride membrane as described by Van DerVaart et al. (32). The efficiencies of the laminarinase extractions were monitoredby the amount of released Cwp1p that was recognized by an anti-b-1,6 glucanantiserum (data not shown). For dot blot analysis, protein samples were taken upin 0.5 M Tris-HCl (pH 6.8)–0.03% (wt/vol) SDS–40 mM b-mercaptoethanol.These samples (100 ml), which contained between 0.08 and 80 ng of the fusionproteins used (calculated from their corresponding a-galactosidase activities),were incubated at 1008C for 2 min and applied to pure nitrocellulose membrane(0.45-mm pore diameter) with a Bio-Rad Bio Dot apparatus. Blots were blockedin phosphate-buffered saline (PBS) containing 5% (wt/vol) milk powder for 30min. Subsequently, the blots were washed in PBS and incubated with a 1:3,000dilution of the anti-a-galactosidase antiserum in PBS containing 3% (wt/vol)bovine serum albumin for 2 h. Antiserum binding was visualized by ECL fluo-rescent labeling and chemiluminescent exposure of X-ray film. For quantificationof protein spots in the dot blot analysis, autoradiograms were scanned with theBio-Rad GS-700 Image Densitometer, and scanned images were analyzed withBio-Rad Molecular Analyst software (version 1.4).Northern analysis and quantification. RNA samples were isolated with

TRIzol Reagent from Gibco BRL. The isolation was performed as described bythe supplier after a cell disruption with glass beads (five times for 30 s each withglass beads 425 to 600 mm in diameter). Separation of RNA, transfer to aHybond-N membrane, and hybridization were performed according to themethod of Sierkstra et al. (30). Hybridization was performed with either a1,500-bp BamHI-HindIII-digested actin fragment or a 750-bp PstI-StuI-digesteda-galactosidase fragment. Both fragments were labeled with [a-32P]dCTP withthe High Prime labeling kit from Boehringer Mannheim Biochemicals. Quanti-fication of mRNA signals was performed with the Instant Imager of the PackardInstrument Company (Meriden, Conn.) with Instant Imager software. ThemRNA level of actin was used to correct for the amount of mRNA that wasblotted.Guar gum hydrolysis and galactose assays. Yeast cells expressing the fusion

proteins were incubated in H2O for 2.5 h at room temperature to allow for therelease of galactose contained in these galactose-grown cells. Subsequently, 2 3108 to 4 3 108 cells (0.5 U of a-galactosidase activity) were collected by centrif-ugation and resuspended in a 0.5% (wt/vol) guar gum solution in 0.1 M NaAc(pH 4.7) and incubated at 308C for 3 h. After centrifugation (14,0003 g, 15 min),the galactose concentrations in the supernatant were determined as described byBulpin et al. (2).

RESULTS

Comparison of 10 carboxyl termini of (potential) cell wallproteins as anchors for the immobilization of enzymes. Tentransformants expressing fusion constructs consisting of thecoding region of a-galactosidase (a-Gal) and AGa1 (27) orCWP1 or CWP2 or TIP1 or TIR1 or SED1 or YCR89w orFLO1, all under control of theGAL7 promoter, were analyzed.In a control strain, SU51-pSY16, which secretes a-galactosi-dase as a free enzyme, almost all enzyme activity was found inthe growth medium (Fig. 2). The addition of the C-terminalpart of any of the (potential) cell wall proteins used in thisstudy resulted in the detection of an increased enzyme activityon the cells (Fig. 2). The efficiency of immobilization of a-ga-lactosidase (immobilized enzyme activity/total enzyme activityratio) varied, depending on the anchor domain used. Fusionproteins consisting of a-galactosidase and Cwp1p or Tip1p are,for the greater part, found in the growth medium, but for the

FIG. 1. Schematic representation of fusion proteins constructed in this study.Plasmid numbers from which the corresponding fusion proteins are expressedare shown on the left. The number of C-terminal amino acids of each (potential)cell wall protein present in the corresponding fusion protein is indicated in theC-terminal anchoring part. The number of amino acids specified by the full-length coding sequence for each of the confirmed or potential cell wall proteinsis given in parentheses. All constructs contain the SUC2 signal sequence.

616 VAN DER VAART ET AL. APPL. ENVIRON. MICROBIOL.

a-galactosidase–Flo1p-344 (a-Gal–Flo1p-344) fusion protein,most of the enzyme activity was found in the intracellularfraction. The majority of the a-Gal–Tir1p and a-Gal–Flo1p-596 fusion proteins were found immobilized on cells, but rel-atively large amounts of the fusion proteins were also found inthe growth medium or in the intracellular fraction, respec-tively. The transformants expressing the a-Gal–YCR89-313and a-Gal–YCR89-744 fusion constructs immobilized much ofthe fusion protein produced, but total enzyme activity was low.The a-Gal–Aga1p, a-Gal–Cwp2p, and a-Gal–Sed1p fusionproteins were immobilized most efficiently. At least 74% of theenzyme activity was found on the cells, and for the a-Gal–Cwp2p fusion protein, an immobilization efficiency of up to80% was achieved.Cell wall localization of a-Gal–Cwp2p and a-Gal–Sed1p.

Cell walls of yeast cells containing a-Gal–CWP2 or a-Gal–SED1 were isolated and extracted with SDS and glucanase.Western analysis of these fractions and of the growth mediumshowed recognition of protein bands by an anti-a-galactosidaseantiserum in the first SDS extracts and in the glucanase ex-tracts (Fig. 3). Almost no recognition by the antiserum wasseen in the second SDS extraction, demonstrating that themajority of the plasma membrane and of the loosely associatedform of the fusion proteins was removed during the first SDSextraction. As expected from the enzyme activity measure-ments, there was also almost no recognition by the antiserumin the medium fractions. Western analysis of SU51-pSY16,which secretes a-galactosidase, showed a reactive protein bandin the growth medium and a faint reactive band in the first SDSextract.Incorporation levels of a-Gal–Aga1p, a-Gal–Cwp2p, and

a-Gal–Sed1p in the cell wall. The location and activity ofa-galactosidase were monitored after induction of fusion pro-tein expression. After induction of the transformants with ga-

lactose, the OD660 of 2 increased to an OD660 of 40 in 32 h. Foreach sample, the a-galactosidase enzyme activity on the cellswas divided by the OD660 of the culture. In this way, theincorporation level for each fusion protein per 107 cells wasdetermined. During the first 8 h, there was an increase in theincorporation level of the three fusion proteins. At this point,the transformant expressing the a-Gal–AGa1 fusion constructreached its maximal incorporation level of 50 mU/107 cells. Forthe other two transformants, the maximal incorporation levelwas seen after approximately 10 to 15 h. For the transformantexpressing the a-Gal–CWP2 fusion construct and for the trans-formant expressing the a-Gal–SED1 fusion construct, the max-imal incorporation level was approximately 120 mU/107 cells.This incorporation level was stably maintained during furthergrowth of the cultures (Fig. 4).It is possible that the incorporation capacity of the fusion

proteins is limited by the presence of other cell wall proteins.To investigate whether the depletion of other cell wall proteinswould provide more space for the fusion proteins expressed,we expressed a-Gal–Aga1p, a-Gal–Cwp2p, and a-Gal–Sed1pin a strain depleted of the cell wall protein Cwp2p, which isabundantly present in the yeast cell wall (32). Depletion ofCwp2p did not result in an increased incorporation capacity ofthe cell wall for the fusion proteins compared to that of theparent strain (data not shown).Specific activities of a-Gal–Aga1p, a-Gal–Sed1p, and

a-Gal–Cwp2p. Dot blot analysis with isolated fusion proteinswas performed to compare their specific activities (data notshown). To determine the linear area of the dot blot analysis,8 mU to 8 mU of the a-Gal–Cwp2 fusion protein was appliedto a nitrocellulose blot. After development of the blot, it wasexposed to an autoradiogram, which was analyzed with a den-sitometer. A linear increment of the density of the proteinspots was observed if between 50 mU and 1 mU of enzymeactivity of the fusion protein was applied to the blot. In four

FIG. 2. Comparison of immobilization efficiencies of C-terminal domains ofseveral (potential) cell wall proteins. Transformants expressing fusion proteinsconsisting of a-galactosidase and a (potential) cell wall protein were grown inYPG and harvested at the mid-exponential growth phase. After centrifugation,the pellet (cells) and supernatant (medium) were analyzed for a-galactosidaseactivity. Subsequently, the cells were broken with glass beads and separated intoa supernatant fraction (intracellular, including membranes) and cell wall fraction(enzyme activity determinations for these fractions were identical to the corre-sponding determinations for the intact cells). “Total” represents the sum of theenzyme activities found on cells, in the medium, and in the intracellular fraction.The percentage of enzyme activity on the cells is calculated as (enzyme activityon cells/total enzyme activity)3 100. All enzyme activities presented are averageenzyme activities determined from four independent experiments.

FIG. 3. a-Gal–Cwp2p and a-Gal–Sed1p are glucanase-extractable cell wallfusion proteins. Results of Western analysis of growth medium and cell wallfractions of transformants producing a-galactosidase–Cwp2p, a-galactosidase–Sed1p, or a-galactosidase as a free enzyme are shown. The blot was incubatedwith an anti-a-galactosidase antiserum. SDS1, first SDS extraction of isolated cellwalls; SDS2, second SDS extraction of isolated cell walls; laminarinase, extrac-tion of SDS-extracted cell walls with laminarinase. The applied cell wall extractsare equivalent to 2 mg of cell walls (fresh weight). The medium fractions cor-respond to the growth medium of 0.4 mg of cells, which corresponds to 0.2 mgof cell walls.

VOL. 63, 1997 CELL WALL IMMOBILIZATION OF a-GALACTOSIDASE IN YEAST 617

independent experiments, equal enzymatic activities of a-Gal–Aga1p, a-Gal–Cwp2p, and a-Gal–Sed1p were applied to theblotting membrane. The specific activities of a-Gal–Cwp2pand a-Gal–Sed1p were 2.756 0.21 and 3.436 0.18 times lowerthan the specific activity of a-Gal–Aga1p, respectively.Expression levels of a-Gal–AGa1, a-Gal–CWP2, and a-Gal–

SED1. The observed differences between production of a-Gal–Aga1p, a-Gal–Cwp2p, and a-Gal–Sed1p may be caused bydifferences in the transcription levels of these genes or differ-ences in mRNA stability, translation efficiency, or protein pro-cessing in the secretory route. To discriminate among thoseoptions, we performed an mRNA quantification for the threetransformants expressing these three fusion proteins. At thestart of the induction experiment (t 5 0), we detected nomRNA of the three fusion proteins. Three hours after induc-tion, mRNA of the fusion proteins was detectable, and themRNA level of a-Gal–CWP2 was twofold higher than thelevels of a-Gal–AGa1 and a-Gal–SED1. Nine hours after in-duction, the mRNA levels of both a-Gal–CWP2 and a-Gal–SED1 were slightly elevated (1.6 and 1.4 times, respectively)compared to the level of a-Gal–AGa1 (Fig. 5).Accessibility of guar gum for immobilized a-galactosidase.

Immobilized a-galactosidase is capable of converting the smallsubstrate p-nitrophenyl-b-D-glucoside (PNPG). The conver-sion of polymeric substrates, like guar gum, in enzyme reactorsbased on enzymes immobilized on the cell surface of yeasts hadnot been reported but can be very important (15). Therefore,we tested whether guar gum could be converted by immobi-lized a-galactosidase. Cells expressing a-Gal–AGa1, a-Gal–CWP2, or a-Gal–SED1 were incubated with guar gum, and theamount of galactose released from this substrate was deter-mined. To ensure that leakage of galactose from the galactose-grown cells did not interfere with this measurement, the cellswere preincubated in water for 2.5 h. The fusion proteins wereactive against the guar gum (Table 1), but the efficiency ofhydrolysis was lower than that of the native a-galactosidase.This lower efficiency was not the result of an uptake of galac-tose by the cells (data not shown). It is possible that the ob-served guar gum hydrolysis by the immobilized enzyme was

caused by the release of the enzyme from the cell wall into thesupernatant. To rule out this possibility, the cells were re-moved at the end of the assay, and the remaining supernatantwas assayed for guar hydrolyzing activity. The residual galac-tose concentration in the supernatant was subtracted from theamount of galactose measured after the guar gum incubationof the remaining supernatant (Table 1). Comparable amountsof wild-type yeast cells, grown in galactose, were not capable ofhydrolyzing guar gum.

DISCUSSION

Immobilization of a-galactosidase in the cell wall of yeastcells was performed by construction of fusions between thea-galactosidase gene and the C-terminal domains of AGa1(27), CWP1, CWP2, TIP1, TIR1, SED1, YCR89w, and FLO1.All anchoring domains proved capable of immobilizing a-ga-lactosidase, but large differences in immobilization efficiencybetween these C-terminal domains were observed. The lengthof the anchoring domain is not the dominant factor for thisvariation in immobilization efficiency. Shortening of theYCR89w anchoring domain from 744 amino acids to 313amino acids results in an increase in the immobilization effi-ciency for a-galactosidase, whereas the immobilization effi-

FIG. 4. Incorporation capacities for a-Gal–Aga1p, a-Gal–Cwp2p, anda-Gal–Sed1p. Transformants were induced with galactose and grown for 32 h.The a-galactosidase enzyme activities on the cells were monitored over time anddivided by the current OD660 of the culture. In this way, the incorporationcapacity (in milliunits) for each fusion protein per 107 cells was determined. AnOD660 of 1 5 107 cells.

FIG. 5. Northern analysis of SU51 expressing a-Gal–AGa1, a-Gal–CWP2, ora-Gal–SED1. Total RNA was isolated at times (T) 0, 3, and 9 h after inductionof the GAL7 promoter. The RNA samples were separated, transferred to a blot,and hybridized with a [a-32P]dCTP-labeled 750-bp PstI-StuI fragment of a-ga-lactosidase. Relative mRNA levels are shown below the corresponding lanes.The mRNA level of actin was used to correct for the amount of mRNA that wasblotted.

TABLE 1. Hydrolysis of guar gum by the action ofimmobilized a-galactosidase

EnzymeaAmt of galactose (mg) released

Cellsb Supernatantc

None 0 0a-Gald NAe 1,395a-Gal–Aga1p 390 180a-Gal–Cwp2p 605 30a-Gal–Sed1p 300 50

a A total of 0.5 U of a-galactosidase per ml was used.b The amount of galactose released by an incubation of yeast cells with guar

gum.c The amount of galactose released by a-galactosidase activity, which had been

released by the cells during the prior guar gum incubation.d Secreted a-galactosidase was added to guar gum.e NA, not applicable.

618 VAN DER VAART ET AL. APPL. ENVIRON. MICROBIOL.

ciency of the Flo1p anchoring domain decreased if the lengthof this domain decreased. The difference in cellular localiza-tion between the two Flo1 fusion proteins demonstrates thatthe position of the fusion junction can influence the immobi-lization efficiency. The location of the fusion junction may alsoinfluence the stability, specific activity, and posttranslationalmodifications of the corresponding fusion protein. Intracellu-lar accumulation or secretion of a particular fusion protein intothe growth medium does not necessarily indicate that the an-choring domain of this protein is unsuitable for the immobili-zation of proteins in the cell wall.The structure of particular fusion proteins may give rise to a

detrimental interaction with the mechanism responsible for theanchorage of cell wall proteins. It has been shown that glu-canase-extractable cell wall proteins carry a carbohydrate sidechain containing b-1,6-linked glucose residues (21, 34), and itwas shown that this chain, which is bound to the proteinthrough a phosphodiester linkage, is involved in cell wall an-chorage (13). Furthermore, the addition of this b-1,6-glucanchain to the cell-wall-bound form of a-agglutinin is dependenton the prior addition of a GPI anchor to the C terminus of thisprotein (18). The linkage of a cell wall protein to the cell wallby means of a b-1,6-glucan chain seems to be part of a generalanchoring mechanism. The cell wall proteins Cwp1p, Cwp2p,Tip1p (32), Sed1p, and Flo1p (33) all proved to contain ab-1,6-glucan chain in their cell-wall-bound forms. Further-more, it was shown that Cwp2p is GPI anchored in the intra-cellular form (33). Secretion of large amounts of a fusionprotein into the growth medium may be the result of the lackof the accessibility of the C terminus of a fusion protein forGPI addition. In analogy with a-agglutinin, this will result in animpairment of cell wall incorporation. It is also possible thatthe addition of the b-1,6-glucan chain to the acceptor site ofthe fusion protein, and not GPI anchoring, is disturbed.On the basis of total a-galactosidase enzyme activity, com-

bined with the percentage of the enzyme activity immobilizedon cells, three fusion proteins, a-Gal–Aga1p, a-Gal–Cwp2p,and a-Gal–Sed1p, were selected for further study. After im-mobilization in the cell wall, these fusion proteins could onlybe released by an enzymatic digestion of the glucan layer. Onlysmall amounts of these fusion proteins were released during anSDS extraction of the cell wall prior to the glucanase digestion.This shows that the fusion proteins become tightly linked to thecell wall. This has also been found for a-Gal–Aga1p by Schre-uder et al. (27). The amount of cell-wall-immobilized a-galac-tosidase was measured after induction of synthesis of the fu-sion proteins. The saturation level of the cell wall for a-Gal–Cwp2p and a-Gal–Sed1p is approximately 2.5 times higherthan that for a-Gal–Aga1p. Because the specific activities ofa-Gal–Cwp2p and a-Gal–Sed1p were lower than that ofa-Gal–Aga1p, the difference in protein level is even moredramatic. The total amount of a-Gal–Cwp2p and a-Gal–Sed1pis six to nine times higher than the total amount of a-Gal–Aga1p. This difference is not the result of a smaller amount ofmRNA of a-Gal-AGa1. No linear relationship between thedifference in protein levels of the fusion proteins and the dif-ference in the corresponding mRNA levels was observed. Ninehours after induction, protein levels of a-Gal–Cwp2p anda-Gal–Sed1p were approximately five times higher than theprotein level of a-Gal–Aga1p, whereas the mRNA levels ofa-Gal–CWP2 and a-Gal–SED1 were 1.6 and 1.4 times higher,respectively, than the mRNA level of a-Gal–AGa1. Becausethere was no intracellular accumulation of the fusion proteinsdetectable, the lower protein level of a-Gal–Aga1p is possiblycaused by a lower translation efficiency of the correspondingmRNA. This is supported by a comparison of codon bias fre-

quencies among the chimeric protein genes. The anchoringdomain of a-Gal–Aga1p has a codon bias frequency of 0.11,compared to codon bias frequencies of 0.77 and 0.55 for theanchoring domains of a-Gal–Cwp2p and a-Gal–Sed1p, respec-tively. Another possible cause for the lower protein level ofa-Gal–Aga1p is a lower protein stability of this fusion protein.Because of the lower protein concentration of a-Gal–Aga1p,this fusion protein constitutes a smaller fraction of the totalpool of proteins destined for the cell wall. However, competi-tion for cell wall localization between the different fusion pro-teins and the native cell wall proteins is unlikely to explain thelower incorporation capacity of the a-Gal–Aga1 fusion pro-tein, because expression of a-Gal–AGa1, a-Gal–CWP2, anda-Gal–SED1 in a strain depleted of Cwp2p did not result in anincrease of the incorporation capacity of these fusion proteins,although it is possible that the cell wall localization of Cwp2pis different from the localization of the fusion proteins tested.For many applications, the immobilized protein must be

expressed on the outside of the cell wall in such a way that it isaccessible for large substrates. Lipase from Humicola lanugi-nosa and cutinase from Fusarium solani subsp. pisi, immobi-lized by the anchoring domain of a-agglutinin, proved to beimmunologically detectable on the surface of intact cells, butdisplayed dramatically reduced enzymatic activity towards anemulsion of olive oil (28). These immobilized enzymes may beembedded in the glycoprotein outer layer, only accessible tosmaller substrates which are capable of penetrating the cellwall. To test the accessibility of immobilized a-galactosidase,we incubated cells carrying the enzyme immobilized on the cellwall, with guar gum from Cyamopsis tetragonoloba. The a-ga-lactosidase enzyme is capable of removing the side chain (1-6)-a-linked D-galactosyl residue from the (1-4)-b-D mannanbackbone (20). The a-galactosidase enzyme, immobilized bya-agglutinin, was capable of hydrolyzing guar gum. The ob-served largely reduced enzymatic activity towards an emulsionof olive oil by Humicola lipase and Fusarium cutinase, whichwere also immobilized by a-agglutinin, is probably caused byhindrance of the lipid-binding ability by the C-terminal exten-sion and low level of specific activity, respectively (28). Theimmobilized a-Gal–Cwp2p fusion protein was most efficient inhydrolyzing guar gum. Furthermore, after 3 h of incubation ofcells in guar gum, the amount of enzyme activity released intothe growth medium was small. However, the efficiency of guargum hydrolysis by the fusion proteins was much lower than theefficiency of guar gum hydrolysis by the free a-galactosidaseenzyme. It is possible that a fraction of the immobilized a-Gal–Cwp2p fusion protein is not exposed on the outside of the cellwall, but that it is located at the inner part of the protein layer.The enzyme activity of this fraction of the fusion protein ispresumably not available for the large guar gum substrate.Normalization of the enzyme activities of a-Gal–Cwp2p anda-galactosidase as a free enzyme was performed with the smallPNPG substrate, which is capable of penetrating the cell wall.Another possible explanation for the lower efficiency of guargum hydrolysis is repulsion of the guar gum substrate by thehydrophobicity of the cell wall.Although the anchoring part of the a-Gal–Cwp2p fusion

protein is only 67 amino acids long (including the GPI attach-ment signal), this fusion protein is very efficiently immobilizedand is most accessible to the large substrate guar gum. Aspacer region between the anchoring part and the enzymetherefore does not seem necessary to make the enzyme extra-cellularly available. The small a-Gal–Cwp2p fusion proteinmight penetrate the glucan layer of the cell wall more easily,whereas the larger fusion proteins may be entrapped in theinternal part of the glucan layer.

VOL. 63, 1997 CELL WALL IMMOBILIZATION OF a-GALACTOSIDASE IN YEAST 619

Here we have shown that it is possible to increase the effi-ciency of guar gum hydrolysis by immobilized a-galactosidaseby varying the anchoring domain of the fusion protein. The useof another host strain in order to increase the efficiency of guargum hydrolysis is also an option. Recently, it has been shownthat the expression of a-Gal–AGa1 in Hansenula polymorpha(31a) and Kluyveromyces lactis (28) resulted in the immobili-zation of this fusion protein in the cell walls of these yeasts.

ACKNOWLEDGMENTS

We thank Arno Mooren for the DNA construction work with theflocculin gene and Monique Plandsoen for creating a YCR89w-con-taining fusion construct.This project was supported by SENTER, a program of the Dutch

Ministry of Economical Affairs.

REFERENCES1. Agterberg, M., H. Adriaanse, H. Lankhof, R. Meloen, and J. Tommassen.1990. Outer membrane PhoE protein of Escherichia coli as a carrier forforeign antigenic determinants: immunogenicity of epitopes of foot-and-mouth disease virus. Vaccine 8:85–91.

2. Bulpin, P. V., M. J. Gidley, R. Jeffcoat, and D. R. Underwood. 1990. Devel-opment of a biotechnological process for the modification of galactomannanpolymers with plant a-galactosidase. Carbohydr. Polym. 12:155–168.

3. Cid, V. J., A. Duran, F. del Rey, M. P. Snyder, C. Nombela, and M. Sanchez.1995. Molecular basis of cell integrity and morphogenesis in Saccharomycescerevisiae. Microbiol. Rev. 59:345–386.

4. Erhart, E., and C. P. Hollenberg. 1983. The presence of a defective LEU2gene on 2m DNA recombinant plasmids of Saccharomyces cerevisiae is re-sponsible for curing and high copy number. J. Bacteriol. 156:625–633.

5. Francisco, J. A., C. F. Earhart, and G. Georgiou. 1992. Transport andanchoring of b-lactamase to the external surface of Escherichia coli. Proc.Natl. Acad. Sci. USA 89:2713–2717.

6. Georgiou, G., H. L. Poetschke, C. Stathopoulos, and J. A. Francisco. 1993.Practical applications of engineering gram-negative bacterial cell surfaces.Trends Biotechnol. 11:6–10.

7. Gunneriusson, E., P. Samuelson, M. Uhlen, P.-Å. Nygren, and S. Ståhl. 1996.Surface display of a functional single-chain Fv antibody on staphylococci. J.Bacteriol. 178:1341–1346.

8. Hanski, E., P. A. Horwitz, and M. G. Caparon. 1992. Expression of proteinF, the fibronectin-binding protein of Streptococcus pyogenes JRS4, in heter-ologous streptococcal and enterococcal strains promotes their adherence torespiratory epithelial cells. Infect. Immun. 60:5119–5125.

9. Hardwick, K. G., J. C. Boothroyd, A. D. Rudner, and R. B. Pelham. 1992.Genes that allow yeast cells to grow in the absence of the HDEL receptor.EMBO J. 11:4187–4195.

10. Harmsen, M. M., A. C. Langedijk, E. van Tuinen, R. H. Geerse, H. A. Raue,and J. Maat. 1993. Effect of a pmr1 disruption and different signal sequenceson the intracellular processing and secretion of Cyamopsis tetragonolobaa-galactosidase by Saccharomyces cerevisiae. Gene 125:115–123.

11. Harrison, J. L., I. M. Taylor, and C. D. O’Connor. 1990. Presentation offoreign antigenic determinants at the bacterial cell surface using the TraTlipoprotein. Res. Microbiol. 141:1009–1012.

12. Hedegaard, L., and P. Klemm. 1989. Type I fimbriae of Escherichia coli ascarriers of heterologous antigenic sequences. Gene 85:115–124.

13. Kapteyn, J. C., R. C. Montijn, E. Vink, J. de la Cruz, A. Llobell, J. E. Douwes,H. Shimoi, P. N. Lipke, and F. M. Klis. 1996. Retention of Saccharomycescerevisiae cell wall proteins through a phosphodiester-linked b-1,3-/b-1,6-glucan heteropolymer. Glycobiology 6:337–345.

14. Klis, F. M. 1994. Review: cell wall assembly in yeast. Yeast 10:851–869.15. Klis, F. M., M. P. Schreuder, H. Y. Toschka, and C. T. Verrips. 1994. Process

for immobilizing enzymes to the cell wall of a microbial cell by producing afusion protein. Patent wo 94/01567.

16. Little, M., P. Fuchs, F. Breitling, and S. Dubel. 1993. Bacterial surfacepresentation of proteins and peptides: an alternative to phage technology?Trends Biotechnol. 11:3–5.

17. Lopes, T. S., J. Klootwijk, A. E. Veenstra, P. C. van der Aar, H. van Heri-khuizen, H. A. Raue, and R. J. Planta. 1989. High-copy number integrationinto the ribosomal DNA of Saccharomyces cerevisiae: a new vector for high-level expression. Gene 79:199–206.

18. Lu, C.-F., R. C. Montijn, J. L. Brown, F. Klis, J. Kurjan, H. Bussey, and P. N.Lipke. 1995. Glycosyl phosphatidylinositol-dependent cross-linking of a-ag-glutinin and b1,6-glucan in the Saccharomyces cerevisiae cell wall. J. CellBiol. 128:333–340.

19. Marguet, D., X. J. Guo, and G. J. M. Lauquin. 1988. Intragenic repeatstructure and identification of a family of SRP1-related DNA sequences. J.Mol. Biol. 202:455–470.

20. Meier, H., and J. S. G. Reid. 1982. Reserve polysaccharides other than starchin higher plants, p. 418–471. In F. A. Loewus and W. Tanner (ed.), Ency-clopedia of plant physiology, new series 13A. Springer-Verlag, New York,N.Y.

21. Montijn, R. C., J. Van Rinsum, F. A. Van Schagen, and F. M. Klis. 1994.Glucomannoproteins in the cell wall of Saccharomyces cerevisiae contain anovel type of carbohydrate side-chain. J. Biol. Chem. 269:19338–19342.

22. Newton, S. M. C., C. O. Jacob, and B. A. D. Stocker. 1989. Immune responseto cholera toxin epitope inserted in Salmonella flagellin. Science 244:70–72.

23. Oliver, S. G., et al. 1992. The complete DNA sequence of yeast chromosomeIII. Nature 357:38–46.

24. Pozzi, G. M. R., R. Oggioni, R. Manganelli, D. Medaglini, V. A. Fischetti, D.Fenoglio, M. T. Valle, A. Kunkl, and F. Manca. 1994. Human T-helper cellrecognition of an immunodominant epitope of HIV-1 gp120 expressed onthe surface of Streptococcus gordonii. Vaccine 12:1071–1077.

25. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: alaboratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, ColdSpring Harbor, N.Y.

26. Schneewind, O., A. Fowler, and K. F. Faull. 1995. Structure of the cell wallanchor of surface proteins in Staphylococcus aureus. Science 268:103–106.

27. Schreuder, M. P., S. Brekelmans, H. Van Den Ende, and F. M. Klis. 1993.Targeting of a heterologous protein to the cell wall of Saccharomyces cer-evisiae. Yeast 9:399–409.

28. Schreuder, M. P., A. T. A. Mooren, H. Y. Toschka, C. T. Verrips, and F. M.Klis. 1996. Immobilizing proteins on the surface of yeast cells. Trends Bio-technol. 14:115–120.

29. Sherman, F., G. R. Fink, and C. W. Lawrence. 1979. Methods in yeastgenetics: a laboratory manual. Cold Spring Harbor Laboratory, Cold SpringHarbor, N.Y.

30. Sierkstra, L. N., J. M. A. Verbakel, and C. T. Verrips. 1992. Analysis oftranscription and translation of glycolytic enzymes in glucose limited contin-uous cultures of Saccharomyces cerevisiae. J. Gen. Microbiol. 138:2559–2566.

31. Teunissen, A. W. R. H., E. Holub, J. Van Der Hucht, J. A. Van Den Berg, andH. Y. Steensma. 1993. Sequence of the open reading frame of the FLO1 genefrom Saccharomyces cerevisiae. Yeast 9:423–427.

31a.van der Klei, I. Personal communication.32. Van Der Vaart, J. M., L. H. P. Caro, J. W. Chapman, F. M. Klis, and C. T.

Verrips. 1995. Identification of three mannoproteins in the cell wall ofSaccharomyces cerevisiae. J. Bacteriol. 177:3104–3110.

33. Van Der Vaart, J. M., F. S. van Schagen, A. T. A. Mooren, J. W. Chapman,F. M. Klis, and C. T. Verrips. 1996. The retention mechanism of cell wallproteins in Saccharomyces cerevisiae. Wall-bound Cwp2p is b-1,6-glucos-ylated. Biochim. Biophys. Acta 1291:206–214.

34. Van Rinsum, J., F. M. Klis, and H. Van Den Ende. 1991. Cell wall gluco-mannoproteins of Saccharomyces cerevisiae mnn9. Yeast 7:717–726.

35. Verbakel, J. M. A. 1991. Heterologous gene expression in the yeast Saccha-romyces cerevisiae. Ph.D. thesis. Rijksuniversiteit Utrecht, Utrecht, TheNetherlands.

36. Yanisch-Perron, C., J. Vieira, and J. Messing. 1985. Improved M13 phagecloning vectors and host strains: nucleotide sequences of the M13mp18 andpUC19 vectors. Gene 33:103–119.

620 VAN DER VAART ET AL. APPL. ENVIRON. MICROBIOL.