Protein Expression and PuriWcation 46 (2006) 85–91

www.elsevier.com/locate/yprep

Expression of glycosylated haloalkane dehalogenase LinB in Pichia pastoris

Takashi Nakamura a, Marcel Zámocký a,b, Zbyn5k Zdráhal c, Radka Chaloupková a, Marta Monincová a, Zbyn5k Prokop a, Yuji Nagata d, Jilí Damborský a,¤

a Loschmidt Laboratories, Masaryk University, Kamenice 5/A4, 625 00 Brno, Czech Republicb Institute of Molecular Biology, Slovak Academy of Sciences, Dubravska cesta 21, 84551 Bratislava, Slovakia

c Laboratory of Functional Genomics and Proteomics, Masaryk University, Kotlálská 2, 611 37 Brno, Czech Republicd Department of Environmental Life Sciences, Graduate School of Life Sciences, Tohoku University, 2-1-1 Katahira, Sendai, 980-8577, Japan

Received 31 May 2005, and in revised form 25 August 2005Available online 22 September 2005

Abstract

Heterologous expression of the bacterial enzyme haloalkane dehalogenase LinB from Sphingomonas paucimobilis UT26 in methylo-trophic yeast Pichia pastoris is reported. The haloalkane dehalogenase gene linB was subcloned into the pPICZ�A vector and integratedinto the genome of P. pastoris. The recombinant LinB secreted from the yeast was puriWed to homogeneity and biochemically character-ized. The deglycosylation experiment and mass spectrometry measurements showed that the recombinant LinB expressed in P. pastoris isglycosylated with a 2.8 kDa size of high mannose core. The speciWc activity of the glycosylated LinB was 15.6 § 3.7 �mol/min/mg of pro-tein with 1,2-dibromoethane and 1.86 § 0.36 �mol/min/mg of protein with 1-chlorobutane. Activity and solution structure of the proteinproduced in P. pastoris is comparable with that of recombinant LinB expressed in Escherichia coli. The melting temperature determinedby the circular dichroism (41.7 § 0.3 °C for LinB expressed in P. pastoris and 41.8 § 0.3 °C expressed in E. coli) and thermal stability mea-sured by speciWc activity to 1-chlorobutane were also similar for two enzymes. Our results show that LinB can be extracellularly expressedin eukaryotic cell and glycosylation had no eVect on activity, protein fold and thermal stability of LinB. 2005 Elsevier Inc. All rights reserved.

Keywords: Dehalogenation; Pichia pastoris; Heterologous expression; Glycosylation; Thermal stability; Catalytic activity

Microbial enzymes haloalkane dehalogenases (EC3.8.1.5) utilize water to transform haloalkanes into an inor-ganic halide and an alcohol. Large scale industrial produc-tion of halocarbons and the persistence of thesecompounds in the environment led to interest in haloalkanedehalogenases for bioremediation purposes [1–3]. Haloal-kane dehalogenase, LinB, is isolated from a �-hexachloro-cyclohexane degrading bacterial strain Sphingomonaspaucimobilis UT26 [4]. It is the second enzyme in the bio-chemical pathway enabling the bacterium to utilize �-hexa-chlorocyclohexane as its sole carbon and energy source. Inthis step, LinB catalyzes the conversion of 1,3,4,6-tetra-

* Corresponding author. Fax: +420 5 49494694.E-mail address: jiri@chemi.muni.cz (J. Damborský).

1046-5928/$ - see front matter 2005 Elsevier Inc. All rights reserved.doi:10.1016/j.pep.2005.08.022

chloro-1,4-cyclohexadiene to 2,5-dichloro-2,5-cyclohexadi-ene-1,4-diol via 2,4,5-trichloro-2,5-cyclohexadiene-1-ol. Inaddition to cyclic dienes, LinB converts a broad range ofhalogenated alkanes and alkenes to their correspondingalcohols [4]. Moreover, the structure of LinB has beensolved by X-ray crystallography [5,6], and the reactionmechanism was examined by transient kinetics [7] andquantum mechanical calculations [8]. On-going mechanisticstudies and intended use of LinB in practice requireseYcient production of homogenous protein.

Pichia pastoris is one of the most attractive systems toobtain recombinant protein by heterologous expression. P.pastoris is a methylotrophic yeast genetically engineered toexpress proteins for basic research and industrial use. Thesystem uses strong promoters such as alcohol oxidase

86 T. Nakamura et al. / Protein Expression and PuriWcation 46 (2006) 85–91

(AOX)1 promoter enabling production of large amount ofthe target protein with technical ease and at lower cost thanmost other eukaryotic systems. Furthermore, the yeast issuitable for large scale production in fermenters whereparameters such as pH, aeration and carbon source feedrate can be controlled. The fermenter enables to achieveultra-high cell densities (>100 g/L dry cell weight; >400 g/Lwet cell weight; >500 OD600 U/ml) and obtain largeramount of protein per volume of a culture [9–11].

In this study, we examined the possibility for expressionof haloalkane dehalogenase in the eukaryotic cell. LinBexpression system in P. pastoris was constructed, secretedenzyme was biochemically characterized and comparedwith the enzyme produced in Escherichia coli.

Materials and methods

Cloning and sequencing

The linB gene coding for haloalkane dehalogenase fromS. paucimobilis UT26 has been previously cloned into theE. coli expression vector pAQN [12]. The gene was sub-cloned in the pPICZ�A expression vector (Invitrogen,Carlsbad, CA) to obtain its integration into the genome ofthe methylotropic yeast P. pastoris. The linB gene was inte-grated in this shuttle vector through the restriction sitesEcoRI and KpnI. The His-tag on the C-terminus wasincluded because it was originally located before the stopcodon. This cloning allowed the in-frame translationalfusion of the dehalogenase gene (starting with completeamino acid sequence without N-terminal methionine) withthe �-factor signal sequence. First, this recombinant plas-mid, 4469 bp long was transformed into E. coli DH5�.Transformants were selected on LB plates containing 25 �g/ml Zeocin and the correct insertion of the gene was veriWedby sequencing of inserted DNA. Afterwards the plasmidDNA of three correct clones was transformed in the P. pas-toris strain GS115 with the lithium chloride method asdescribed in the manual for EasySelect Pichia ExpressionKit of Invitrogen. Suitable transformants were selected onplates containing 1% yeast extract, 2% peptone, 1% glucosewith 200–500 �g/ml of Zeocin.

Heterologous expression

The transformants were inoculated Wrst in 100 ml YPGmedium (1% (w/v) yeast extract, 2% (w/v) peptone, and 1%(v/v) glycerol) and grown until the stationary phase(approximately 18 h) at 28 °C in a shaking incubator at160 rpm for screening of expression levels. Aliquots weretransferred in 150 ml of induction medium containing 1%(w/v) yeast extract, 2% (w/v) peptone, 4 £ 10¡5% (w/v) Bio-tin when OD600 was approximately 0.2. The expression of

1 Abbreviations used: AOX, alcohol oxidase; CD, circular dichroism; 1-CB, 1-chlorobutane; 1,2-DBE, 1,2-dibromoethane; MALDI, matrix-assist-ed laser desorption/ionization, TOF, time of Xight.

LinB was originally induced according to the conditionsdescribed in the manual of Invitrogen, i.e., 0.5% (v/v) meth-anol was added for every 24 h to induce expression of LinB(total induction time 94 h) when the culture reachedOD600 D 2.0. These conditions provided more than one pro-tein variant possibly due to variable glycosylation of LinBproteins. To obtain single LinB protein, the culture solutionwas incubated under various conditions (Table 1) selectedand analyzed by using the software for experimental designMODDE 7.0 (Umetrics AB, Umeå, Sweden). In total sixfactors (OD600 before induction, induction temperature,methanol concentration, length of induction, addition ofcasamino acid and protease inhibitor) and one response(expression level estimated from the amount of protein inthe culture medium [13]) were used for the screeningdesigns. Important factors were identiWed by using factorialdesigns with linear and interaction model. Partial leastsquares projection to latent structures analysis was used forWtting. The model quality was estimated from the percentof variation explained by the model, the percent of varia-tion of the response predicted from the model and modelreproducibility, i.e. the variation of the response under thesame conditions, often at the center points, compared to thetotal variation of the response.

Protein puriWcation

Cells were harvested by centrifugation (2000g, 7 min at4 °C). Ammonium sulfate was added to the supernatant tobecome 75% saturated solution. The solution was stirredfurther for 30 min after the added ammonium sulfate wascompletely dissolved and then centrifuged at 11,000g for15 min. The pellet was resuspended in a equilibrating buVerfor puriWcation, which contained 20 mM potassium phos-phate buVer (pH 7.5), 0.5 M sodium chloride, and 10 mMimidazole. The His-tagged LinB was puriWed on the Ni–NTA and Sepharose column HR 16/10 (QIAGEN, Hilden,Germany) and dialyzed overnight against 50 mM potas-sium phosphate buVer, pH 7.5, as described previously [14].The dialyzed sample was concentrated to about 2 ml byAmicon stirred ultraWltration cell (model 8010, Millipore,Billerica, MA). To prepare a sample for circular dichroism(CD) spectra measurement, the concentrated enzyme solu-tion was loaded into a small column Wlled with 1.5 ml of

Table 1Selected conditions of LinB expression in P. pastoris

a No signiWcant improvement of protein expression level was observedwhen both casamino acid and PMSF were added to the culture.

Tested conditions Selected conditions

OD600 before induction (2, 25, and 40) OD600 D 2Induction temperature (25 °C and 28 °C) 28 °CMethanol concentration (v/v, 0.30, 0.50,

0.70, 1.0, 1.2, 1.5 and 2.0 %)0.70%

Lengh of induction (8, 9, 10, 12 and14 h) 10 hAddition of casamino acid 1% (w/v)a

or notAddition of 1% casamino acid

Addition of PMSF (0.5, 1 and 2 mM)a 1 mM

T. Nakamura et al. / Protein Expression and PuriWcation 46 (2006) 85–91 87

Ni–NTA–Sepharose solution to remove dark brown colourcoming from the culture medium. The colour was probablycaused by the presence of low molecular weight compoundbound to proteins nonspeciWcally.

Protein electrophoresis

SDS–PAGE was performed according to Laemmli [15].The gel was run in the Miniprotean II apparatus (BioRad,Hercules, CA), and then stained with 0.25% (w/v) Coo-massie brilliant blue R-250 in 30% methanol and 10% aceticacid. The estimation of protein molecular weight fromSDS–PAGE was performed by BioCapt software version99.04s for Windows (Vilbert Lourmat, Marne la Vallee,France).

Enzymatic deglycosylation

Deglycosylation of LinB secreted from P. pastoris(20 �g) was performed by using 3000 U of endoglycosidaseHf (Endo Hf, New England Biolabs, Beverly, USA) at 37 °Cfor 1 h according to the manufacturer instructions. Thedeglycosylated proteins were analyzed in a 15% SDS–PAGE as described above. To make a sample for CD spec-trum measurement of deglycosylated LinB, 350 �g of gly-cosylated protein was incubated with 10,000 U of Endo Hfat 28 °C for 2 h without denaturing step. The deglycosylatedenzyme solution was loaded into a small column stuVedwith 1.5 ml of Ni–NTA–Sepharose solution and puriWedusing the same procedure as described above.

SpeciWc activity assay

SpeciWc activities of LinB with two diVerent halogenatedsubstrates, 1-chlorobutane (1-CB) and 1,2-dibromoethane(1,2-DBE), were assessed by determination of the substrateand product concentrations using the gas chromatographTrace GC 2000 (Finnigan, San Jose, CA) equipped with aXame ionization detector and the capillary column (DB-FFAP 30m£0.25mm£0.25�m, J & W ScientiWc, Folsom,CA). The reaction was conducted in 25ml Reacti-Flasksclosed by Mininert Valves. The enzymatic reaction was initi-ated by adding 200�l (for 1,2-DBE) or 400�l (for 1-CB) of0.2mg/ml enzyme solution into 10ml of substrate solution(10 mM of 1,2-DBE and 10mM of 1-CB in a 0.1M glycinebuVer, pH 8.6, respectively). The mixture was incubated at37°C and progress of the reaction was monitored by with-drawing 0.5ml samples at 0, 5, 10, 20, and 40 min using asyringe needle to reduce evaporation of the substrate fromthe reaction mixture. The reaction mixture samples weremixed with 0.5 ml of methanol to terminate the reaction. Thereaction mixture without enzyme served as an abiotic control.

Thermal stability assay

A sample of enzyme dissolved in 50 mM potassiumphosphate buVer (pH 7.5) was incubated for 0, 5, 10, or

20 min at 50 °C. After cooling of the samples to 4 °C formore than 30 min, the measurement of remaining enzy-matic activity to 1-CB was performed under same condi-tions as speciWc activities. The enzymatic reaction wasinitiated by adding 400�l of 0.2 mg/ml enzyme solution into10 ml of substrate solution (10 mM of 1-CB in a 0.1 M gly-cine buVer, pH 8.6, respectively). The enzymatic activitywas determined by the method of Iwasaki et al. [16]. Inbrief, the amount of released halide ions was measuredspectrophotometrically at 460 nm with mercuric thiocya-nate and ferric ammonium sulfate. The progress of thereaction was monitored by withdrawing 1.0 ml samples at 0,5, 10, 20, and 40 min using a syringe needle. All activitieswere calculated relative to the samples without heating.

Circular dichroism

CD spectra were recorded at room temperature usingthe Jasco J-810 spectrometer (Jasco, Tokyo, Japan). Datawere collected from 185 to 260 nm, at 100 nm/min, 1 sresponse time and 2 nm bandwidth using a 0.1 cm quartzcuvette containing 0.15 mg/ml LinB enzyme in 50 mMpotassium phosphate buVer (pH 7.5). Each spectrum shownis the average of 10 individual scans corrected for absor-bance caused by the buVer. For thermal denaturation, theprotein solutions were heated from 22 to 72 °C at 1 °C/min.The changes in the ellipticity were monitored at 222 and221 nm for LinB expressed in P. pastoris and E. coli, respec-tively. Recorded thermal denaturation curves of both pro-teins were normalized to represent signal changes betweenapproximately 1 and 0 and Wtted to sigmoidal curves. Themelting temperatures were evaluated from the collecteddata as a midpoint of the normalized thermal transition.

Matrix-assisted laser desorption/ionization (MALDI)-time of Xight (TOF) spectrometry

Mass spectra of proteins were recorded on a ReXex IV(Bruker, Bremen, Germany) operated in the linear modewith detection of positive ions. Mixture of 2,5-dihydroxyben-zoic acid and 5-methoxysalicylic acid (super DHB, Bruker,Bremen, Germany) was used as a matrix. Solution of thematrix (20mg/ml) was prepared in 1% triXuoroacetic acid of20% acetonitrile solution. Samples were mixed with thematrix solution in a ratio 1:3. The XMASS 5.1.5 software(Bruker, Bremen, Germany) was used for data processing.

Results and discussion

Cloning of linB gene and its transformation into P. pastoris GS115

Cloning of linB gene in the pPICZ�A vector was per-formed through EcoRI and KpnI restriction sites. The entireconstruct is 4469 bp long, linB gene has 909 bp and is in-frame with the �-factor signal sequence. The translatedsequence of LinB starts with its second amino acid, because

88 T. Nakamura et al. / Protein Expression and PuriWcation 46 (2006) 85–91

the Wrst methionine was omitted to enable fusion with the �-peptide for eYcient secretion. All other amino acids are thesame as in the original protein as veriWed by sequencing ofthe corresponding plasmid DNA with 5�AOX1 and3�AOX1 primers. Fused linB gene is under the control of theinducible AOX1 promoter. After transformation of thisrecombinant plasmid named pPICZALinB in E. coli DH5�,the plasmid DNA of positive clones was isolated in suYcientamount. After its puriWcation and linearization with restric-tion endonuclease SacI, the plasmid was transformed in P.pastoris GS115 with the lithium chloride method accordingto Invitrogen manual (Invitrogen, Carlsbad, CA). Theexpected integration of linB gene in the genome was veriWedfor clones that were used in expression studies after the iso-lation of the total genomic DNA from the yeast cells withthe usage of lyticase and ethanol precipitation according tothe protocol provided in Invitrogen manual. The clonesselected for the heterologous expression of linB gene weretested for growth in a medium containing up to 500�g/ml ofZeocin and because the expression vectors pPICZ containno yeast origin of replication we concluded that linB gene ispresent in the genome of all transformants selected for theexpression of recombinant LinB.

ModiWcation of expression condition of LinB in P. pastoris

Expression procedure was initially carried out accordingto conditions described in the manual of Invitrogen (see

Fig. 1. Response surface plot obtained from optimization experiment test-ing the eVect of methanol concentration, temperature and OD600 beforeinduction (not plotted) on the protein concentration.

Materials and methods). After puriWcation, 2.4 mg/L of pro-tein was obtained and the speciWc activity to 1-CB was com-parable to enzyme from E. coli. Obtained LinB however wasnot a single protein as seen on Fig. 2A (lane 1). To obtainhomogenous LinB protein, the conditions for expressionwere selected based on the literature data ([17–22] and Sup-plementary material, Table S1) and adjusted to the speciWcproperties of LinB. First, LinB protein is deactivated at pHbelow 5.0, while pH in the culture medium without buVertends to get lower during methanol induction phase becauseof formation of formic acid by a methanol-utilizing pathway[9]. Second, LinB is losing its activity within 24 h at 28 °Cmaking longer cultivations at higher temperatures diYcult.

Taking into consideration the above limitations, theeVect of parameters such as methanol concentration, lengthof induction, and addition of casamino acid (Table 1) wasexamined at the same time in 150 ml-culture scale experi-ment. The expression level of LinB in P. pastoris was esti-mated from the amount of protein in the medium measuredby Bradford method because LinB activity could not bemeasured directly from the culture medium. The eVect offactors and their combinations on the level of proteinexpression was studied in a set of screening experiments(Fig. 1). Repeating this screening several times, conditionsof LinB expression were Wnally optimized (Table 1).Selected conditions provided 0.6 mg/L of pure and activeprotein (Table 2). The sample contained a single protein(Fig. 2) with the yield somewhat lower than obtained from94 h induction. The other protein species were avoidedunder optimized conditions by the shorter induction timeand the presence of casamino acids preventing partialdeglycosylation or proteolysis. The homogenous proteinsample was used for biochemical characterization.

Protein loss due to its production inside the yeast cell wasconsidered, but the amount of protein produced in the cellswas insigniWcant (Supplementary material, Fig. S1). Thereare several possibilities for further improvement of the yield.Some of the protein was lost during the concentration stepby ammonium sulfate precipitation that could be avoided bymodiWcation of protein puriWcation protocol. The number ofintegrated copies of the expression cassette can aVect theamount of expressed protein. Screening in higher concentra-tion of zeocin may be used to select transformant of multi-copy linB gene [23,24]. Other possibility is to use a fermenterinstead of shaken Xasks [25]. In our experience, the lowervolume of a culture resulted in the higher yield. This seemsto be related to the extent of aeration, which can be intensi-Wed and more easily controlled by fermenter. Optimizationof the codon usage for P. pastoris was also shown to providehigher levels of protein expression [26,27].

Table 2PuriWcation of LinB secreted from P. pastoris

Protein (mg)

Activity (nmol/min/mg of protein)

Total activity (nmol/min)

Yield (%)

PuriWcation factor

Culture supernatant 29.7 227 6742 100 1Ni–NTA puriWcation 0.6 1437 862 13 6.3

T. Nakamura et al. / Protein Expression and PuriWcation 46 (2006) 85–91 89

Characterization of secreted enzyme by SDS–PAGE

The LinB secreted from P. pastoris was analyzed by SDS–PAGE. It was found that there was a diVerence in band posi-tion between protein expressed in E. coli and P. pastoris(Fig. 2A, lanes 2 and 4). The protein was deglycosylated byEndo Hf to conWrm whether the band shift is derived fromits glycosylation. Band for deglycosylated enzyme expressed

Fig. 2. (A) Electrophoretic analysis of puriWed glycosylated LinB anddeglycosylated LinB by Endo Hf. Lanes M, pre-stained SDS–PAGE pro-tein standard marker; 1, puriWed LinB expressed in P. pastoris for 94 hunder recommended induction condition in the manual of Invitrogen (nobuVer, OD600 before induction 2.0, methanol concentration 0.5%, induc-tion time 94 h, and temperature 28 °C, 3.0 �g); 2, PuriWed LinB expressedin E. coli (2.4 �g); 3, Glycosylated LinB expressed in P. pastoris after enzy-matic deglycosylation with Endo Hf (4.5 �g), the band of Endo Hf isshown in 70 kDa; 4, PuriWed glycosylated LinB expressed in P. pastorisunder selected conditions from Table 1 (pH 6.5, OD600 before induction2.0, methanol concentration 0.7%, induction time 10 h, addition of 1%casamino acid, and temperature 28 °C, 2.0 �g). (B) MALDI-TOF MSspectra of recombinant LinB expressed in E. coli and P. pastoris. Thenumbers of the peaks and identity of samples correspond to (A).

A M 1 2 3 4

B

30000 40000 50000 m/z

20

40

60

80

100

120

140

160

180

a.i.2

3

4

0 0Mass (m/z)

Rel

ativ

e in

ten

sity

20.7

28.8

34.3

50.0

77.0 103

kDa

in P. pastoris shifted to the position of LinB expressed in E.coli (Fig. 2A, lanes 2 and 3). These results indicate glycosyla-tion of LinB expressed in P. pastoris. The glycosylation wasalso conWrmed by mass spectrometry (see below). Theenzyme puriWed from non-modiWed condition had threebands (Fig. 2A, lane 1), possibly corresponding to glycosyl-ated, deglycosylated or non-glycosylated proteins. The esti-mated molecular weight of the proteins in Fig. 2 wasapproximately 34.2, 32.1, 31.4 (lane 1), 31.7 (lane 2), 32.5(lane 3), and 34.8 kDa (lane 4), respectively.

Characterization of secreted enzyme by mass spectrometry

The molecular weight of the proteins determined from theMALDI-TOF mass spectrometry measurement was 33.8(peak 2), 34.8 (peak 3), and 36.6kDa (peak 4), respectively(Fig. 2B). Molecular weight of LinB with six histidinesexpressed in E. coli, corresponding to peak 2, calculatedusing the ExPASy ProtParm tool homepage (http://au.exp-asy.org/tools/protparam.html) was approximately 33.8 kDa.The glycosylated and non-glycosylated LinB (Fig. 2B, peaks2 and 4) showed 2.8kDa of mass diVerence. Potential N-linked glycosylation site [NXS/T motif (X, except P)] in theamino acid sequence of LinB (UniProt Accession No.P51698) was identiWed as a motif between positions 262 and265 “NQTE” by using search of PROSITE database (data-base of protein families and domains, http://www.expasy.org/prosite/). The diVerence in molecular weight (2.8kDa) is con-sistent with a glycosylation pattern in P. pastoris reportedpreviously, where the majority of high-mannose-type oligo-saccharides species are ranging from Man3GlcNAc2(0.9kDa) to Man17GlcNAc2 (3.2kDa) [11]. We speculate thata high-mannose-type oligosaccharides species is attached tothe fragments 262–265, that is located on the protein surfacefar from the entrance tunnel leading to the catalytic site [6].The mass diVerence between non-glycosylated and deglyco-sylated protein (peaks 2 and 3, 1kDa), however, cannot beexplained by the attachment of the innermost N-acetyl gluco-samine to asparagine residue at N-linked glycosylation site(approximately 0.2kDa). We speculate that additional massdiVerence could be due to the presence of second glycosyla-tion site on LinB, e.g., O-glycosylation on Ser or Thr, or dueto the incomplete cleavage of the signal peptide.

Characterization of secreted enzyme by circular dichroism

Far-UV CD spectra were used for assessment of the sec-ondary structure for glycosylated LinB expressed in P. pas-toris and non-glycosylated LinB expressed in E. coli. Bothenzymes showed CD spectra typical for predominantly �-helical conformation with two negative minima at approxi-mately 221 and 208 nm and a positive peak at approxi-mately 195 nm. No signiWcant diVerence between the CDspectra of glycosylated and non-glycosylated form ofenzyme was found (Fig. 3A). This indicates that glycosyla-tion does not have signiWcant eVect on the secondary struc-ture of LinB. The integrity of secondary structure of LinB

90 T. Nakamura et al. / Protein Expression and PuriWcation 46 (2006) 85–91

expressed in P. pastoris after its deglycosylation was con-Wrmed also by Far-UV CD spectroscopy. Obtained results(data not shown) were in good agreement with CD spectraof glycosylated and non-glycosylated LinB enzyme. Ther-mally induced denaturation was studied to detect eVect ofglycosylation on thermal stability of LinB enzyme. Bothtested LinB enzymes showed changes in ellipticity duringincreasing temperature (Fig. 3B). The intensity of the signaldecreased with temperature for both proteins following thesigmoidal curves. The melting temperature (Tm) estimatedfrom measured curves is 41.7 § 0.3 for LinB expressed in P.pastoris and 41.8 § 0.3 °C for LinB expressed in E. coli. Sim-ilarity of Tm values suggests that glycosylation does nothave any eVect on thermal stability of LinB.

Characterization of secreted enzyme by kinetic measurements

SpeciWc activity of glycosylated LinB was examined bygas chromatography analysis and determined rates werecompared with the values previously obtained for LinB

Fig. 3. Comparison of structure and conformational stability of LinB pro-duced in two diVerent expression systems under increasing temperature.(A) Far-UV CD spectra of LinB expressed in P. pastoris (closed squares)and LinB expressed in E. coli (open squares). (B) Thermal denaturationcurves of LinB expressed in P. pastoris (closed squares) and LinBexpressed in E.coli (open squares). The global Wt of sigmoidal model to thedata of protein denaturation curves is shown as solid line.

180 200 220 240 260-8

-4

0

4

8

12

16

CD

[mde

g]

wavelength [nm]

20 30 40 50 60 70-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

Rel

ativ

e sc

ale

T [˚C]

A

B

expressed in E. coli. 1,2-DBE and 1-CB were used as testingsubstrates. The activity was 15.6 § 3.7 �mol/min/mg of pro-tein for 1,2-DBE and 1.86 § 0.36�mol/min/mg of proteinfor 1-CB. These values are comparable to published data[14] for LinB expressed in E. coli, which are 13.2 and11.3 �mol/min/mg of protein for 1,2-DBE, and 2.04 and1.74 �mol/min/mg of protein for 1-CB. The eVect of glyco-sylation on thermal stability and enzymatic activity wasdescribed in the literature [26,28,29]. Therefore, the eVect oftemperature on protein deactivation was assessed by activ-ity measurements. Glycosylated LinB expressed in P. pasto-ris showed similar deactivation curves compared to LinBexpressed in E. coli (Fig. 4) conWrming conclusion madefrom CD experiments, that glycosylation had no eVect onthermal stability of LinB. Of note is the quantitative diVer-ence in thermostability assessed by CD-experiment andkinetic assay. In CD measurement, the sample is heated instepwise manner exposing the sample to increasing temper-ature for considerably longer time than in activity assaywhere an individual samples were incubated for 0, 5, 10, or20 min at 50 °C, cooled to 4 °C, and then the remainingenzymatic activity with 1-CB was measured. Furthermore,missing cool down step in CD experiment makes re-foldingof the protein sample at the temperatures close to Tm diY-cult.

In conclusion, we conWrmed that haloalkane dehalogen-ase LinB can be extracellularly expressed in eukaryotic cell.Protein secreted from P. pastoris is glycosylated. Glycosyla-tion of LinB has no eVect on protein activity, secondarystructure and thermal stability. Haloalkane dehalogenaseLinB suitable for fundamental studies and practical appli-cations can be produced by heterologous expression in P.pastoris.

Fig. 4. Comparison of thermal deactivation of LinB produced in twodiVerent expression systems. Data collected for LinB expressed in E. coli(closed circles) are compared with the data for glycosylated LinBexpressed in P. pastoris (open circles). A sample of enzyme dissolved in 50mM potassium phosphate buVer (pH 7.5) was incubated for 0, 5, 10, or20 min at 50 °C. After cooling of the samples, the remaining enzymaticactivity to 1-CB was determined. All activities were calculated relative tothe samples without heating. Results are expressed as the mean § stan-dard deviation from eight replicates.

0

20

40

60

80

100

0 5 10 15 20

Time (min)

Rel

ativ

e ac

tivi

ty

E. coliP. pastoris

T. Nakamura et al. / Protein Expression and PuriWcation 46 (2006) 85–91 91

Acknowledgments

This work was supported by postdoctoral fellowship forresearch abroad from Yamada Science Foundation (toT.N.). We thank Hana Konebná (Laboratory of FunctionalGenomics and Proteomics, Masaryk University, Brno,Czech Republic) for protein digestion for MS analysis. Thiswork was supported by the grants from the Czech Ministryof Education (No. MSM 0021622412 andMSM0021622415).

Appendix A. Supplementary data

Supplementary data associated with this article can befound, in the online version, at doi:10.1016/j.pep.2005.08.022.

References

[1] D.B. Janssen, F. Pries, J.R. VanderPloeg, Genetics and biochemistryof dehalogenating enzymes, Annu. Rev. Microbiol. 48 (1994) 163–191.

[2] S. Fetzner, Bacterial dehalogenation, Appl. Microbiol. Biotechnol. 50(1998) 633–657.

[3] S.D. Copley, Microbial dehalogenases: enzymes recruited to convertxenobiotic substrates, Curr. Opin. Chem. Biol. 2 (1998) 613–617.

[4] Y. Nagata, K. Miyauchi, J. Damborsky, K. Manova, A. Ansorgova,M. Takagi, PuriWcation and characterization of haloalkane dehalo-genase of a new substrate class from a �-hexachlorocyclohexane-degrading bacterium, Sphingomonas paucimobilis UT26, Appl. Envi-ron. Microbiol. 63 (1997) 3707–3710.

[5] J. Marek, J. Vevodova, I. Kuta-Smatanova, Y. Nagata, L.A. Svensson,J. Newman, M. Takagi, J. Damborsky, Crystal structure of the haloal-kane dehalogenase from Sphingomonas paucimobilis UT26, Biochem-istry 39 (2000) 14082–14086.

[6] A.J. Oakley, M. Klvana, M. Otyepka, Y. Nagata, M.C.J. Wilce, J.Damborsky, Crystal structure of haloalkane dehalogenase LinB fromSphingomonas paucimobilis UT26 at 0.96 A resolution: Dynamics ofcatalytic residues, Biochemistry 43 (2004) 870–878.

[7] Z. Prokop, M. Monincova, R. Chaloupkova, M. Klvana, Y. Nagata,D.B. Janssen, J. Damborsky, Catalytic mechamism of the haloalkanedehalogenase LinB from Sphingomonas paucimobilis UT26, J. Biol.Chem. 278 (2003) 45094–45100.

[8] M. Bohac, Y. Nagata, Z. Prokop, M. Prokop, M. Monincova, J. Koca,M. Tsuda, J. Damborsky, Halide-stabilizing residues of haloalkanedehalogenases studied by quantum mechanic calculations and site-directed mutagenesis, Biochemistry 41 (2002) 14272–14280.

[9] J.L. Cereghino, J.M. Cregg, Heterologous protein expression in themethylotrophic yeast Pichia pastoris, FEMS Microbiol. Rev. 24(2000) 45–66.

[10] G.P. Cereghino, J.L. Cereghino, C. Ilgen, J.M. Cregg, Production ofrecombinant proteins in fermenter cultures of the yeast Pichia pasto-ris, Curr. Opin. Biotechnol. 13 (2002) 329–332.

[11] R. Daly, M.T. Hearn, Expression of heterologous proteins in Pichiapastoris: a useful experimental tool in protein engineering and pro-duction, J. Mol. Recognit. 18 (2005) 119–138.

[12] Y. Nagata, K. Hynkova, J. Damborsky, M. Takagi, Construction andcharacterization of histidine-tagged haloalkane dehalogenase (LinB)of a new substrate class from a �-hexachlorocyclohexane-degrading

bacterium, Sphingomonas paucimobilis UT26, Protein Expr. Purif. 17(1999) 299–304.

[13] M.M. Bradford, A rapid and sensitive method for the quantiWcationof microgram quantities of protein utilizing the principle of protein-dye binding, Anal. Biochem. 72 (1976) 248.

[14] R. Chaloupkova, J. Sykorova, Z. Prokop, A. Jesenska, M. Monincova,M. Pavlova, Y. Nagata, J. Damborsky, ModiWcation of activity andspeciWcity of haloalkane dehalogenase from Sphingomonas paucimobi-lis UT26 by engineering of its entrance tunnel, J. Biol. Chem. 278(2003) 52622–52628.

[15] U.K. Laemmli, Cleavage of structural proteins during the assembly ofthe head of bacteriophage T4, Nature 227 (1970) 680–685.

[16] I. Iwasaki, S. Utsumi, T. Ozawa, New colorimetric determination ofchloride using mercuric thiocyanate and ferric ion, Bull. Chem. Soc.Japan 25 (1952) 226.

[17] E. Rodriguez, Y. Han, X.G. Lei, Cloning, sequencing, and expressionof an Escherichia coli acid phosphatase/phytase gene (appA2) isolatedfrom pig colon, Biochem. Biophys. Res. Commun. 257 (1999) 117–123.

[18] R.G. Miele, M. Prorok, V.A. Costa, F.J. Castellino, Glycosylation ofasparagine-28 of recombinant staphylokinase with high-mannose-type oligosaccharides results in a protein with highly attenuated plas-minogen activator activity, J. Biol. Chem. 274 (1999) 7769–7776.

[19] J.L. Adrio, J. Velasco, G. Soler, M. Rodriguez-Saiz, J.L. Barredo, M.A.Moreno, Extracellular production of biologically active deacetoxy-cephalosporin C synthase from Streptomyces clavuligerus in Pichiapastoris, Biotechnol. Bioeng. 75 (2001) 485–491.

[20] L. Thomas, D.L. Crawford, Cloning of clustered Streptomyces viri-dosporus T7A lignocellulose catabolism genes encoding peroxidaseand endoglucanase and their extracellular expression in Pichia pasto-ris, Can. J. Microbiol. 44 (1998) 364–372.

[21] K.J. Potter, M.A. Bevins, E.V. Vassilieva, V.R. Chiruvolu, T. Smith,L.A. Smith, M.M. Meagher, Production and puriWcation of the heavy-chain fragment C of botulinum neurotoxin, serotype B, expressed inthe methylotrophic yeast Pichia pastoris, Protein Expr. Purif. 13(1998) 357–365.

[22] M.P. Byrne, T.J. Smith, V.A. Montgomery, L.A. Smith, PuriWcation,potency, and eYcacy of the botulinum neurotoxin type A bindingdomain from Pichia pastoris as a recombinant vaccine candidate,Infect. Immun. 66 (1998) 4817–4822.

[23] A. Vassileva, D.A. Chugh, S. Swaminathan, N. Khanna, EVect of copynumber on the expression levels of hepatitis B surface antigen in themethylotrophic yeast Pichia pastoris, Protein Expr. Purif. 21 (2001) 71–80.

[24] V. Sarramegna, P. Demange, A. Milon, F. Talmont, Optimizing func-tional versus total expression of the human �-opioid receptor inPichia pastoris, Protein Expr. Purif. 24 (2002) 212–220.

[25] C.E. White, N.M. Kempi, E.A. Komives, Expression of highly disul-Wde-bonded proteins in Pichia pastoris, Structure 2 (1994) 1003–1005.

[26] D. Tull, T.E. Gottschalk, I. Svendsen, B. Kramhoft, B.A. Phillipson,H. Bisgard-Frantzen, O. Olsen, B. Svensson, Extensive N-glycosyla-tion reduces the thermal stability of a recombinant alkalophilic bacil-lus �-amylase produced in Pichia pastoris, Protein Expr. Purif. 21(2001) 13–23.

[27] G. Sinclair, F.Y. Choy, Synonymous codon usage bias and the expres-sion of human glucocerebrosidase in the methylotrophic yeast, Pichiapastoris, Protein Expr. Purif. 26 (2002) 96–105.

[28] K. Amano, M. Takase, A. Ando, Y. Nagata, Production of functionallectin in Pichia pastoris directed by cloned cDNA from Aleuria auran-tia, Biosci. Biotechnol. Biochem. 67 (2003) 2277–2279.

[29] B.M. Kim, H. Kim, R.T. Raines, Y. Lee, Glycosylation of onconaseincreases its conformational stability and toxicity for cancer cells,Biochem. Biophys. Res. Commun. 315 (2004) 976–983.