Vol. 174, No. 20JOURNAL OF BACrERIOLOGY, Oct. 1992, p. 6424-64310021-9193/92/206424-08$02.00/0Copyright © 1992, American Society for Microbiology

Development of Thermus-Eschenichia Shuttle Vectors and TheirUse for Expression of the Clostridium thermocellum

celA Gene in Thermus thermophilusINIGO LASA, M. DE GRADO, M. A. DE PEDRO, AND JOSt BERENGUER*

Centro de Biologia Molecular, Universidad Aut6noma de Madrid-Consejo Superiorde Investigaciones Cientificas, 28049 Madrid Spain

Received 29 April 1992/Accepted 30 July 1992

We describe the self-selection of replication origins of undescribed cryptic plasmids from Thermus aquaticusY-VH-51B (ATCC 25105) and a Thermus sp. strain (ATCC 27737) by random insertion of a thermostablekanamycin adenyltransferase cartridge. Once selected, these autonomous replication origins were cloned intothe Escherichia coli vector pUC9 or pUC19. The bifunctional plasmids were analyzed for their sizes,relationships, and properties as shuttle vectors for Thermnus-Escherichia cloning. Seven different vectors withdiverse kanamycin resistance levels, stabilities, transformation efficiencies, and copy numbers were obtained.As a general rule, those from T. aquaticus (pLUl to pLU4) were more stable than those from the Thermus sp.(pMYl to pMY3). To probe their usefulness, we used one of the plasmids (pMYl) to clone in E. coli a modifiedform of the cellulase gene (ceL4) from Clostridium thermoceflum in which the native signal peptide was replacedin vitro by that from the S-layer gene of T. thermophilus HB8. The hybrid product was expressed and exportedbyE. coli. When the gene was transferred by transformation into T. thermophilus, the cellulase protein was Alsoexpressed and secreted at 70°C.

The production of thermostable enzymes and structuralproteins is an economically important field for investigationbecause of their potential industrial applications (10). Mostof these enzymes are presently produced by large-scalefermentation processes, the high cost of which can bejustified only for very specific applications (18).

For increasing the yield of these processes, thermostableenzymes have been cloned in mesophilic organisms, such asEscherichia coli. However, differences in codon usage orimproper folding of the proteins at low temperatures haveprevented the expected activities or the desired amounts ofthe proteins to be obtained. On the other hand, expression ofthe cloned proteins in thermophilic organisms has been madedifficult by the thermal instability of selectable plasmidmarkers, such as antibiotic resistance genes, at high growthtemperatures (7).

In 1986, Matsumura et al. demonstrated that point muta-tions in a kanamycin adenyltransferase gene (kat) increasedthe thermostability of the enzyme by several degrees (17),thus allowing the selection of cloning vectors for moderatethermophiles, such as Bacillus stearothermophilus. To applythis system to extreme thermophilic eubacteria, we recentlydescribed the construction of S-layer (slpA) mutants ofThernus thennophilus HB8 by insertion, through homolo-gous recombination, of a hybrid kat gene controlled by thepromoter and ribosome binding site signals of slpA (11). Asingle copy of the kat gene per chromosome resulted in anMIC of kanamycin of about 20 ,ug/ml in agar plate assays.Similarly, Mather and Fee used this same thermostable katgene to produce a selectable plasmid for T. thermophilusHB8 by random insertion into its cryptic plasmid, pTT8 (16).We describe the construction of a series of selectable

bifunctional E. coli-Thermus plasmids by use of previouslyundescribed replication origins from two Thermus isolates.

* Corresponding author.

We also demonstrate their usefulness as bifunctional cloningvectors by using one of the plasmids to express a thermo-stable cellulase gene (ceU) from Clostridium thermocellumin both E. coli and T. thennophilus.The cellulase gene was chosen for its potential properties

as a "reporter gene" because of the extracellular nature andthermostability of its product. As a parallel observation fromthis work, we also demonstrate that the amino-terminalsequence of the T. thennophilus HB8 S-layer gene (5, 6), thesequence of which will appear in the EMBL gene bank underaccession number X57333, contains a signal peptide that candirect the secretion of the CelA protein both in T. thermo-philus and in E. coli.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions. T. ther-mophilus HB8 (ATCC 27634), T. aquaticus Y-VII-51B(ATCC 25105), and a Thermus sp. (ATCC 27737) wereobtained from the American Type Culture Collection (Rock-ville, Md.). T. thernophilus HB27 was generously providedby Y. Koyama. E. coli TG1 [K-12 supE hsdA&5 thi A(lac-proAB) F' (traD36 proAB+ laqIq lacZAM15)] and DH5aF'[F' supE44 A(lacZYA-argF)U169 (4)80 lacZAM15) hsdR17recAl endA41 gyrA96 thi-1 relAl] (Bethesda Research Labo-ratories, Gaithersburg, Md.) were used as hosts for geneticmanipulations of plasmids.

Plasmids pUC9 (21) and pUC19 (22) were used for in vitrogenetic manipulations. Plasmid pKT1 was constructed in ourlaboratory (11) and contains a thermostable kanamycinadenyltransferase gene (17) controlled by the transcriptionalsignals of the slpA gene from T. thennophilus HB8. PlasmidpTC105 (4) was generously provided by P. Beguin andcontains the celA gene from C. thermocellum in a 3.2-kbpHindIII-HindIII DNA fragment. Plasmid pSla2.3 is a pUC9derivative in which the promoter and amino-terminal regionof the sipA gene were cloned (5).

6424

on Decem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

EXPRESSION OF celA IN T. THERMOPHILUS 6425

T. thernnophilus HB8 was grown at 70°C under strongaeration in a rich medium containing 8 g of Trypticase (BBLMicrobiology Systems, Cockeysville, Md.), 4 g of yeastextract (Okoid, Hampshire, England), and 3 g of NaCl perliter of tap water. The pH of the medium was adjusted to 7.5.For petri plates, 1.5% (wt/vol) agar was added to solidify themedium. When necessary, 100 ,ug of kanamycin per ml wasadded to plates for selection. Plates were incubated in aninverted position at 70°C in a water-saturated atmosphere.

E. coli strains were grown at 37°C in LB medium (12). If aplasmid was present, a selective antibiotic was added (100,ug/ml for ampicillin or 30 ,g/ml for kanamycin). Cells weremade competent as described previously (13).DNA analysis. Plasmid DNA was purified from E. coli by

the alkaline lysis method (2). The same method was used forThermus spp., but a shorter lysis time was used to limit thedegradation of DNA. Total DNA was purified by the methodof Marmur (15), but a higher salt concentration was used inthe precipitation steps.Most of the DNA techniques were carried out as described

by Maniatis et al. (14). Restriction enzyme digestions andligation reactions were performed as recommended by themanufacturer (Boehringer-Mannheim GmbH, Penzberg,Germany). DNA sequencing was performed by the methodof Sanger et al. (20) in accordance with the instructions ofthe DNA sequencing kit from Pharmacia (Uppsala, Swe-den). DNA amplification (19) was performed with the syn-thetic oligonucleotides described in Results and DNA poly-merase from T. aquaticus (Amplitaq; Perkin Elmer-Cetus,Emeryville, Calif.). The amplification reaction mixtureswere incubated in a PHC-1-Dry-Block (Techne, Duxford,Cambridge, England).

Southern analysis of the chromosomal and plasmid DNAswas performed as described previously (14) with about 5 ,ugof digested DNA per sample. DNA probes were labeled withthe Pharmacia kit for random primer labeling.

Transformation of T. thermophilus. Essentially the methodof Koyama et al. was used (9) for transformation. T. ther-mophilus HB8 or HB27 was grown at 70°C in the mediumdescribed above with 1 mM MgCl2 and 0.5 mM CaCl2. At anoptical density at 550 nm of 0.5, samples (0.5 ml) containingapproximately 108 CFU were transferred to 5-ml sterile glasstubes, and the desired amount of DNA was added. After 2 hof incubation at 70°C under strong aeration, cells weredirectly plated onto selective agar plates.

Cellulase activity. The production of extracellular cellulasewas assayed by a modification of a method described previ-ously (4) to increase the sensitivity. Colonies grown for 24 hon selective plates were transferred to nitrocellulose filters(MSI, Westboro, Mass.), and the filters, with the coloniesfacing up, were placed on a second set of sterile agar plates.The plates were overlaid with agar (0.7% [wt/vol]) containing0.5% (wt/vol) carboxymethyl cellulose in 50 mMK2HPO4-12 mM citric acid buffer (pH 6.3). Once the agarwas solidified, the plates were incubated at 65°C for 3 h, andthe activity was detected by staining of the remainingsubstrate with Congo red.

RESULTS

Cloning in E. coli of replication origins from a Thermus sp.and T. aquaticus. As a source of replication origins fromThermus spp., we used T. aquaticus ATCC 25105 andThermus sp. strain ATCC 27737 because of their biochemi-cal differences and because, to our knowledge, the presenceof plasmids within these strains has not been described. As

shown in Fig. 1, total plasmid preparations from 20-mlcultures of both strains were digested with restriction en-zyme BamHI or PstI and ligated by use of cohesive ends toa thermostable kanamycin adenyltransferase cartridge (kat).This cartridge was purified from E. coli plasmid pKT1, inwhich the kat gene is controlled by the promoter region ofthe S-layer gene (slpA) (11). Ligation mixtures were subse-quently used to transform the plasmid-free strain T. therno-philus HB27. After incubation at 70°C for 48 h, a largenumber of colonies resistant to 20 ,ug of kanamycin per mlwere obtained for the PstI-Thennus sp. and BamHI-T.aquaticus ligations. The BamHI-Thermus sp. and PstI-T.aquaticus ligations yielded very few colonies, so they werenot used.To clone the putative plasmids from the kanamycin-

resistant transformant colonies into E. coli, we pooledmixtures of such colonies from both transformations andpurified total plasmids. These mixtures were digested withEcoRI, an enzyme with few recognition sites in Thermusspp. and a single recognition site in the kat cartridge, andligated to the EcoRI site of pUC9 or pUC19. Upon transfor-mation of E. coli, colonies resistant to ampicillin (100 ,ug/ml)and kanamycin (30 pug/ml) were analyzed for the presence ofplasmids.

This method allowed us to obtain seven Thermus autono-mous replicons of different sizes: four from T. aquaticus(pTAQ1 to pTAQ4) and three from the Thermus sp. (pTSP1to pTSP3). When cloned into pUC vectors, the plasmidswere called pLUl to pLU4 and pMY1 to pMY3, respectively(Fig. 1). The plasmids were probed to determine whetherthey were bifunctional by testing their ability to transform T.thermophilus Hfl27 and to be recovered by transformationinto E. coli. The MIC of kanamycin for T. thermophilus cellscarrying any one of these plasmids was >200 ,ug/ml.

Analysis of pLU and pMY plasmids. The most importantproperties analyzed in plasmids pLUl to pLU4 and pMY1 topMY3 are summarized in Table 1. As expected, the restric-tion patterns of the plasmids revealed the presence of severalsites for enzymes that recognize sequences with a high G+Ccontent, such as XhoI. The plasmids either had very fewsites with a high A+T content (ClaI and EcoRI) or lackedthese sites altogether. Nevertheless, the differences in therestriction patterns between all plasmids but pMY1 andpMY2 suggest that we have cloned distinct replicationorigins (data not shown). To verify this suggestion, wepurified and labeled the corresponding Thermus DNA frag-ments from all the plasmids and assayed them independentlyfor the ability to hybridize to each other. As Table 1 shows,only plasmids pMY1 and pMY2 had common fragments.

Since the procedure for the purification of plasmids fromcultures of T. thermophilus HB27 did not yield DNA cleanenough to be quantified by spectrophotometry (especially forthe pLU plasmids), the transformation efficiencies wereassayed with plasmid DNA obtained from E. coli. Theefficiencies for the pMY plasmids (Table 2) were about 1/10those obtained when the same plasmids were purified fromT. thermophilus (data not shown).The stability of the plasmids was analyzed by probing the

kanamycin resistance of colonies plated after 32 h of growthin nonselective medium, with an intermediate reinoculationduring this period. The results (Table 2) demonstrate therelative instability of plasmids pMY1 and pMY2 comparedwith that of plasmids pMY3 or pLUl to pLU4. A possibleexplanation for this result could be that plasmids pMY1 andpMY2 have lower copy numbers. To test this possibility, wepurified total DNA of T. thermophilus HB27 cells trans-

VOL. 174, 1992

on Decem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

HCHPSBXN

Cryptic plasmids ofThermus aquaticus

.P X B Sct

B

B4

Ligation

Cryptic plasmids ofThermus sp

LILigation

Transformation of Thermus thermophilus HB27

Selection on Kanamycin (20 gg/mI)BX N E PXB PSBX N EP

pTSP1pTAQ1pTAQ2

pTAQ3pTAQ4IE

E4* E pUC9

Ligation

pTSP2

pTSP3

pUC19 E

Ligation

Transformation of E. coli DH5aF

Selection on Ampicillin (100,g/mI) and Kanamycin (30,g/mI)

P S B X

IH

pTSPpUC9

P E X B PEKXBSH

FIG. 1. Selection of bifunctional Thennus-E. coli plasmids. Low-scale preparations of plasmids from T. aquaticus Y-VII-51B and theThermus sp. were digested with the indicated restriction enzymes and ligated to the kat cartridge (KAT). Plasmid mixtures (pTAQ and pTSP)of kanamycin-resistant T. thermophilus HB27 colonies, obtained after transformation with the ligations mixtures, were digested with EcoRIand ligated to pUC9 or pUC19. Both ligation mixtures were then used to transform E. coli cells, and plasmids from individual colonies wereanalyzed as described in Results. Abbreviations for restriction enzymes: B, BamHI; C, ClaI; E, EcoRI; H, HindIII; K, KpnI; N, NdeI; P,PstI; S, SalI; Sc, SacI; X, XmaI.

6426

pT)

m

on Decem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

EXPRESSION OF celA IN T. THERMOPHILUS 6427

TABLE 1. Restriction analysis and relationships between plasmids

Plasmid Insert No. of the following sites: Cross-hybridization with:size (kbp) BamHI PstI HindIII KpnI XhoI pLUl pLU2 pLU3 pLU4 pMY1 pMY2 pMY3

pLUl 3.5 1 0 0 3 5 +pLU2 8.3 2 2 3 2 6 +pLU3 6.8 2 1 0 2 4 +pLU4 6.2 2 1 0 2 7 +pMYl 4.7 2 1 0 3 4 + +pMY2 4 1 1 0 1 4 +pMY3 9.4 9 1 3 1 7 +

formed with each plasmid from kanamycin-resistant, station-ary-phase cultures and digested it with EcoRI and BamHI.The DNA fragments were separated by electrophoresis onagarose gels, transferred to nitrocellulose filters, and hybrid-ized with the labeled promoter region of sipA. These diges-tions generated two labeled DNA fragments, of 1 and 5.8kbp, corresponding to the kat cartridge of the plasmid and tothe S-layer gene of this strain, respectively. The levels oflabeling of the DNA bands were related to the copy numbersof the corresponding plasmids (Table 2). Surprisingly, stableplasmids pLU1, pLU2, and pLU4 had copy numbers similarto those of unstable plasmids pMYl and pMY2, making thisexplanation for the instability improbable.A second explanation for the instability of plasmids pMY1

and pMY2 could be their origin. Southern blots demon-strated that both plasmids hybridized to a 16-kbp PstI bandfrom plasmid preparations of the Thermus sp. (data notshown). This result means that these plasmids were deletionproducts derived from a larger plasmid, an explanation thatcould justify their instability.

Construction of a fusion between the signal peptide region ofsipA and the celA gene of C. thermocelum. The celA gene ofC. thermocellum, with its own promoter and signal peptide,can be expressed and exported through the cell envelope ofE. coli (4). Preliminary attempts to clone and express thisgene in T. thennophilus HB27 by direct insertion of select-able replication origins into plasmid pMY1 were unsuccess-ful. Therefore, we decided to produce a fusion between thisgene and a segment containing the promoter and the amino-terminal region (33 amino acids) of the S-layer gene (slpA) ofT. thernophilus HB8 (5).The method used (Fig. 2) was developed in three steps.

First, we designed the oligonucleotide shown at the top leftof the figure to include a BamHI restriction site after thehypothetical signal peptide ofsipA. This oligonucleotide alsoincluded an extra adenine to adjust the reading frame. The

TABLE 2. Properties of pLU and pMY plasmids as vectors

TransformationPlasmid efficiency (no. Stability (%)b Copy no.C

of colonies/4Lg)'pLUl ix102 100 4pLU2 1x102 100 4pLU3 1 x 104 96 15pLU4 6 x 103 92 2pMY1 1 x 104 14 3pMY2 1 x 103 28 3pMY3 3 x 103 100 40a DNA was purified from m- r- E. coli.b Determined as described in the text.c DNA was purified from stationary-phase cells.

other oligonucleotide used for the amplification step wascomplementary to the pUC9 vector (reverse primer), intowhich a fragment of sipA containing the promoter and theamino-terminal region had been cloned. The amplified frag-ment was digested with XmaI and BamHI and cloned intopUC9 to produce plasmid pPSl. In this plasmid, the pro-moter and the amino-terminal region (amino acids 1 to 33) ofslpA are followed by a polylinker region containing restric-tion sites for BamHI, Sall, PstI, and HindIII.

In the second step, we amplified a celA fragment thatextended from its codon for amino acid 29 to the end of thegene. The synthetic oligonucleotide included a SalI restric-tion site to enable the production of a fusion with the sipAfragment that had been amplified and cloned. The secondoligonucleotide (direct primer) was complementary to theregion of the pUC9 vector in which the 3.2-kbp HindIII-HindIII fragment was cloned (5). Once amplified, the pro-jected fusion was obtained by cloning the celA fragmentbetween the Sall and HindIII sites of pPS1. Surprisingly, theresultant construction, pPSC1, was positive for cellulaseactivity in E. coli (Fig. 3C), thus demonstrating the function-ality of the signal peptide of the S-layer gene from T.thermophilus HB8 in this mesophile.

Finally, to express the fusion protein in T. thermophilus,we inserted into pPSC1 the 5.7-kbp EcoRI fragment frompMY1.1, a derivative of pMYl in which various restrictionsites were deleted from the polylinker regions (Fig. 2). Theresultant construction, called pTCM1, was then used totransform competent cells of T. thennophilus HB27. Trans-formants were positive for cellulase activity (Fig. 3B),whereas in a parallel experiment, colonies transformed withpTC105-pTSP1 were negative (Fig. 3A).

Figure 3 shows the results of an experiment in which E.coli (Fig. 3C) and T. thermophilus HB8 and HB27 (Fig. 3D)cells transformed with plasmid pMY1.1 (1), pTC105-pTSP1(2), or pTCM1 (3 and 3') were assayed for the production ofextracellular cellulase. When the transcription and secretionsignals of the S-layer gene of T. thermophilus HB8 werepresent, cellulase was synthesized and apparently exportedfrom both the mesophile and the thermophile. Furthermore,the usefulness of the novel bifunctional plasmids was clearlydemonstrated.

DISCUSSION

The use of positive criteria for the selection of plasmids inextreme thermophiles was limited until recently to genes thatcomplement mutants in the synthesis of amino acids (8).Although this system allowed the development of bifunc-tional Thermus-E. coli vectors (7), the high recombinationfrequency in the thermophile (9) produced, after transforma-

VOL. 174, 1992

on Decem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

6428 LASA ET AL.

36A G H W A K E

5'.GCC GGG CAC TGG GCC AAG GAG.3'3'CC QTQ ACC CGG TTC CT

A AGM 5'

Bam Hi

RP Bam HI

|X/B

pUC9 X/-B--*

H

Sal I

AC ACT GTM TCA Gac CC 3'5' .G GC AAC ACT GTG TCA GCO GCL GOT.3 '

A N T V S A27

B ~~~P

Sal I

S/H

EX N BS B P H

BamHI Sa

5 TGG 4i-Z CAA GGA TCC GTC GALC A? (GM TCZL...3

....ACC CdG GTT CCT AGG CAG CTG TGC CAC AaT....W V Q G S R Q T V S

pPSC1PE

E

E

CELA P

B ~~pLac

BS

PE pTCM1 EK-' 10.7 Kbp.

KAT

>.f.e PBX

FIG. 2. Construction of plasmids used for the expression of celA in T. thermophilus. A DNA fragment containing the promoter and thefirst coding region of slpA was amplified by the polymerase chain reaction with the oligonucleotides indicated. The fragment (including a newBamHI site) was then cloned into restriction sites XmaI and BamHI of pUC9, and plasmid pPS1 was obtained. The celA gene was amplifiedand cloned into pPS1 by use of SalI (included in the polymerase chain reaction) and HindIII, and plasmid pPSC1 was obtained. The EcoRIsite of this plasmid was further used to clone a replication origin from the Thennus sp. (pMY1.1) containing the selectable kat marker, andthe bifunctional plasmid pTCM1 was obtained. Symbols for restriction enzymes are identical to those in Fig. 1. DP, direct primer; RP, reverseprimer.

H

DP

I ....II.,.,.,.,.,.,.,.,.,.,"" :" e,:-,

.:-.: - - -

J. BACTERIOL.

0-

w-

S/H

on Decem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

EXPRESSION OF ceL4 IN T. THERMOPHILUS 6429

A

4

B

C D

a .3

2 2

FIG. 3. Expression of celA in E. coli and T. thermophilus. The assay for the detection of cellulase activity was done on kanamycin-resistant colonies of T. thermophilus HB27 transformed with pTC105-pTSP1 (A), kanamycin-resistant colonies of the same strain transformedwith pTCM1 (B), E. coli DH5aF' cells transformed with pMYl.l (1), pTC105-pTSP1 (2), pTCM1 (3), and pPSC1 (4) (C), and T. thernophilusHB27 cells transformed with pMY1.1 (1), pTC105-pTSP1 (2), and pTCM1 (3) and T. thermophilus HB8 cells transformed with pTCM1 (3')(D).

tion, a significant background of wild-type colonies withoutplasmids.For this reason, the recent adaptation to T. thernophilus

of a thermostable kanamycin adenyltransferase gene withvery low or no homology to its chromosomic DNA (11, 16)represents an important step in the development of cloningvectors for these extreme thermophiles. In this study, wetook advantage of the development (11) of a cartridge inwhich this selectable resistance marker (17), expressed from

a strong Thermus promoter, can be ligated to the restrictionsite for PstI, BamHI, or XmaI. The ligation of this cartridgeto BamHI- and PstI-digested samples of total plasmids fromT. aquaticus and the Thermus sp., respectively, and thesubsequent transformation of the plasmid-free strain T.thermophilus HB27 yielded a large number of kanamycin-resistant colonies that, hypothetically, should have con-tained plasmids. As the cartridge contains an internal EcoRIrestriction site, rarely found within the chromosome of

VOL. 174, 1992

on Decem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

6430 LASA ET AL.

Thermus spp., we used it to clone a mixture of the putativeplasmids into pUC9 and pUC19. Analysis of the E. colicolonies that were resistant to both ampicillin and kanamy-cin yielded seven different (by size) plasmids. These plas-mids were demonstrated to be bifunctional, as they couldtransform T. thennophilus HB27. Furthermore, they alsocould be purified from transformed cells of the thermophileand recovered by transformation in E. coli.The transformation efficiencies for T. thermophilus HB27

were not directly related to the size of the plasmid (Table 2),as large plasmids, such as pLU3 (10 kbp) and pMY3 (13kbp), produced more colonies per microgram than didsmaller plasmids (pLU1, pLU2, and pMY2). As expectedfrom the presence of a restriction system, transformationefficiencies for plasmids pMYl to pMY3 purified from Ther-mus spp. were at least 1 order of magnitude higher thanthose for the same plasmids purified from E. coli. The samecould be suggested for the pLU plasmids, although difficul-ties in the purification of these plasmids from Thermus spp.made it impossible to quantitate the amount of DNA used inthe transformation experiments. This effect seems to berelated to the intrinsic nature of the replication origins ofthese plasmids, because it was also extremely difficult todetect clear bands in plasmid preparations of T. aquaticus.The results from the Southern blotting are summarized in

Table 1 and clearly demonstrated that only pMY1 and pMY2had common fragments. This result was in good agreementwith those of other experiments, in which both plasmidshybridized to a PstI-generated band of 16 kbp that wasobserved in plasmids prepared from the Thermus sp. (datanot shown). Nevertheless, as minipreparations of plasmidDNA from Thermus spp. frequently appeared to be partiallydegraded, we believe that pMY1 and pMY2 could representthe replication origin and the PstI site within rearrangedfragments of this 16 kbp. The origin of pMY3 is much moreclear, as it hybridized to a second PstI band of 9.3 kbp inminipreparations of plasmid DNA from the Thermus sp.The presence of four different replication origins in T.

aquaticus was striking and much more difficult to explain,because minipreparations of plasmid DNA from this speciesoften revealed a degradation trail. However, the exochro-mosomal nature of the cloned fragments was demonstratedby Southern hybridization with a labeled mixture of plasmidspLUl to pLU4. This mixture recognized only the miniprep-aration fraction and did not hybridize to the chromosomalDNA.With the double intent of probing the functionality of these

plasmids and of developing secretion vectors for Thermusspp., we decided to prepare a fusion between a thermostableand easy-to-assay enzyme (CeUA) and the signal peptide ofthe S-layer gene of T. thermophilus HB8 (5). Since theS-layer is the outermost envelope of T. thermophilus (3), thisconstruction should have been exported and, because of theabsence of other hydrophobic sequences along the structuralcelA gene (1), secreted into the medium.The expression of this fusion in E. coli (Fig. 3) revealed an

important collateral observation: the amino-terminal se-quence of slpA acts as an effective export signal in E. coli(Fig. 3C). However, because of the low level of expressionof the fusion, we could not sequence the amino terminus.Therefore, we could not determine whether the signal pep-tide was actually processed in the mesophile.The insertion of a replication origin from the Thermus sp.

with the kat cartridge into the EcoRI site of pPSC1 allowedthe cloning of the fusion gene into T. thermophilus. Theexpression of cellulase was detected in the surrounding

medium of T. thermophilus cells transformed with plasmidpTCM1 (Fig. 3B and D), whereas in cells transformed withpTC105-pTSP1, it was not detected. This result stronglysuggests that the cellulase expressed from plasmid pTCM1 isactively exported in T. thermophilus although the possibilityexists that part of the observed activity was due to a smallamount of cell lysis. However, in experiments not describedhere, cellulase activity was detected in the supernatant ofliquid cultures, in which the protein concentration was toolow to be detected by Coomassie blue staining. Therefore,the bulk of the activity probably did not originate from celllysis.The cellulase activity observed in T. thermophilus was

apparently lower than that detected when the same plasmidwas cloned in E. coli. This effect could have been related inpart to the high temperature at which the thermophile wasgrown (70°C); such high-temperature growth makes theexpressed product unstable. To facilitate the use of the celAgene as a reporter in the future, investigators should isolateceLA mutants with increased thermal stability by directselection of colonies transformed with mutagenized pTCM1.The results presented in this paper clearly demonstrate

that genes cloned in E. coli can be expressed in T. thenno-philus by use of our bifunctional plasmids. Nevertheless,before these plasmids can be generally used, it will beimportant to analyze the regulatory mechanisms of promot-ers from Therinus spp. and to apply this information to thecontrol of the expression of cloned genes.

ACKNOWLEDGMENTS

We are very grateful to M. Matsumura for giving us the thermo-stable kanamycin adenyltransferase gene and to P. B6guin forproviding the celA gene. The comments and suggestions of A.Ottolenghi are also acknowledged.

This work was supported by grant BI091-0523 from the CICYTand by an institutional grant from the Fundaci6n Ram6n Areces. I.Lasa is the recipient of a fellowship from F.P.I.

REFERENCES1. Beguin, P., P. Cornet, and J.-P. Aubert. 1985. Sequence of a

cellulase gene of the thermophilic bacterium Clostridium ther-mocellum. J. Bacteriol. 162:102-105.

2. Birnboim, H. C., and J. Doly. 1979. A rapid alkaline extractionprocedure for screening recombinant plasmid DNA. NucleicAcids Res. 7:1513-1523.

3. Cast6n, J. R., J. L. Carrascosa, M. A. de Pedro, and J.Berenguer. 1988. Identification of a crystalline surface layer onthe cell envelope of the thermophilic eubacterium Thermusthermophilus. FEMS Microbiol. Lett. 51:225-230.

4. Cornet, P., J. Millet, P. BEguin, and J.-P. Aubert. 1983. Char-acterization of two cel (cellulose degradation) genes of Clostrid-ium thermocellum coding for endoglucanases. Bio/Technology1:589-594.

5. Faraldo, M. M. 1990. Ph.D. thesis. Universidad Aut6noma deMadrid, Madrid, Spain.

6. Faraldo, M. M., M. A. de Pedro, and J. Berenguer. 1991.Cloning and expression in E. coli of the structural gene codingfor the monomeric protein of the S-layer of Thermus thermo-philus HB8. J. Bacteriol. 173:5346-5351.

7. Koyama, Y., Y. Arikawa, and K. Furukawa. 1990. A plasmidvector for an extreme thermophile, Thermus thennophilus.FEMS Microbiol. Lett. 72:97-102.

8. Koyama, Y., and K. Furukawa. 1990. Cloning and sequenceanalysis of tryptophan synthetase genes of an extreme thermo-phile, Thermus thermophilus HB27: plasmid transfer from rep-lica-plated Escherichia coli recombinant colonies to competentT. thermophilus cells. J. Bacteriol. 172:3490-3495.

9. Koyama, Y., T. Hoshino, N. Tomizuka, and K. Furukawa. 1986.Genetic transformation of the extreme thermophile Thermus

J. BAcrERIOL.

on Decem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

EXPRESSION OF cel4 IN T. THERMOPHILUS 6431

thermophilus and other Thermus spp. J. Bacteriol. 166:338-340.10. Kristjansson, J. K. 1989. Thermophilic organisms as sources of

thermostable enzymes. Trends Biotechnol. 7:349-353.11. Lasa, I., J. R. Cast6n, L. A. Fernandez-Herrero, M. A. de Pedro,

and J. Berenguer. 1992. Insertional mutagenesis in the extremethermophilic eubacteria Thermus thermophilus HB8. Mol. Mi-crobiol. 6:1555-1564.

12. Lennox, E. X. 1955. Transduction of linked genetic characters ofthe host by bacteriophage P1. Virology 1:190-206.

13. Mandel, M., and A. Higa. 1970. Calcium dependent bacterio-phage DNA infection. J. Mol. Biol. 53:159-162.

14. Maniatis, T., E. F. Fritsch, and J. Sambrook 1982. Molecularcloning: a laboratory manual. Cold Spring Harbor Laboratory,Cold Spring Harbor, N.Y.

15. Marmur, J. 1961. A procedure for the isolation of deoxyribo-nucleic acid from microorganisms. J. Mol. Biol. 3:208-218.

16. Mather, M. W., and J. A. Fee. 1992. Development of plasmidcloning vectors for Thennus thermophilus HB8: expression of aheterologous, plasmid-borne kanamycin nucleotidyltransferasegene. Appl. Environ. Microbiol. 58:421-425.

17. Matsumura, M., S. Yasumura, and S. Aiba. 1986. Cumulativeeffect of intragenic amino acid replacement on the thermosta-bility of a protein. Nature (London) 323:356-358.

18. Ng, T. K., and W. Kenealy. 1986. Industrial applications ofthermostable enzymes, p. 197-215. In T. D. Brock (ed.), Ther-mophiles: general, molecular, and applied microbiology. JohnWiley & Sons, Inc., New York.

19. Saikl, T., S. Scharf, F. Faloona, K. B. Muilis, G. T. Horn, H. A.Erlich, and N. Arnheim. 1985. Enzymatic amplification of,B-globin genomic sequences and restriction site analysis fordiagnosis of sickle cell anemia. Science 230:1350-1354.

20. Sanger, F., S. Nicklen, and A. R. Coulson. 1977. DNA sequenc-ing with chain-terminating inhibitors. Proc. Natl. Acad. Sci.USA 74:5463-5467.

21. VMeira, J., and J. Messing. 1982. The pUC plasmids, anM13mp7-derived system for insertion mutagenesis and sequenc-ing with synthetic universal primers. Gene 19:259-268.

22. Yanisch-Perron, C., J. Vieira, and J. Messing. 1985. ImprovedM13 phage cloning vectors and host strains: nucleotide se-quences of the M13mpl8 and pUC19 vectors. Gene 33:103-119.

VOL. 174, 1992

on Decem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from