Current G e n e t i c s

Curr Genet(1991) 19:81-87

9 Springer-Verlag 1991

Optimisation of a host/vector system for heterologous gene expression by Hansenula polymorpha Laurens N. Sierkstra, John M. A. Verbakel, and C. Theo Verrips Unilever Research Laboratorium Vlaardingen, Olivier van Noortlaan 120, NL-3133 AT Vlaardingen, The Netherlands Received June 21/October 8, 1990

Summary. For the methylotrophic yeast, Hansenula polymorpha, expression vectors with different origins of replication have been constructed in order to analyse their influence on transformation and integration efficiency, The constructed plasmids are identical except for their origin of replication, which involve, respectively, that of the Saccharomyces cerevisiae 2%tm plasmid and a H. polymorpha ARS sequence (HARS2). A plasmid with no origin of replication served as a control. The plasmids also contained the ~-galactosidase expression cassette, consisting of the Cyamopsis tetragonoloba c~-galactosidase gene, the H. polymorpha methanol oxidase promoter and terminator, and the S. cerevisiae invertase signal sequence. The transformation frequencies of the expression vectors containing the 2-~tm and the HARS2 origins of replication, and no origin of replication, were 2, 50 and 15 per ~tg of DNA respectively, which demonstrates the negative effect of the 2-gin sequence on the transformation frequency. Autonomously replicating plasmids could be isolated from the transformants obtained with the plasmid containing either the 2-gin or the HARS2 sequence. Integration of the 2-~tm based plasmid into the H. polymorpha genome could not be established using a standard procedure. This is in contrast with transformants containing a plasmid bearing the HARS2 sequence or else with no origin of replication, which shows that the 2-gin sequence negatively influences the integration of the expression vector into the 11. poIymolTha genome. Integration of expression plasmids occurred in 50% of the analysed integrants on the homologous methanol oxidase locus, and tandem integration was favoured. The level of specific mRNA, and the expression of the c~-galactosidase protein by these integrants, was proportional to the number of integrated copies of the expression plasmid in the H. potymorpha genome. Key words: Methylotrophic yeast - Integration - Origin of replication - Plant ~-galactosidase

Off)3rint requests to. J. M. A. Verbakel

Introduction

Hansenulapolymorpha is one of the four yeast genera that can use methanol as a sole carbon and energy source. The mechanism by which H. polymorpha uses methanol has been studied extensively (Veenhuis et al. 1983). The methanol utilization pathway, localised in the peroxisomes, begins with the oxidation of methanol to formaldehyde, a reaction catalysed by the enzyme methanol oxidase. The methanol oxidase gene can be induced by methanol and is repressed by glucose. Under fully induced conditions the cellular protein content consists of at least 40% methanol oxidase, so the use of this strong methanol oxidase promoter, in combination with the efficient growth and induction on the relatively inexpensive methanol, makes H. polymorpha a good option for the production of heterologous proteins (Giuseppin et al. 1988; Giuseppin 1989; Ledeboer et al. 1986). The gene coding for the enzyme methanol oxidase has been cloned and characterised (Ledeboer etal. 1985). Transformation procedures have been described and a host/vector system has been developed (Gleeson and Sudbery 1988). The production of heterologous proteins under the alcohol oxidase promoter of the related yeast Pichia pastoris has been reported; these include the S. cerevisiae invertase (Tschopp et al. 1987) and the hepatitis B surface antigen (Cregg et al. 1987). In H. polymorpha, we have expressed, amongst others, a plant enzyme, a-galactosidase, using the methanol oxidase promoter. This enzyme, normally produced by the plant C. tetragonoloba during germination, is responsible for the removal of the 1-6 linkedc~-D-galactopyranosyl units from galactomannan, a C. tetragonoloba reserve polysaccharide (Meier and Reid 1982). The cloning and characterisation of the C. tetragonoloba a-galactosidase gene has been described by Overbeeke et al. (1989) and Hughes et al. (1988). The c~-galactosidase has been expressed under the control of the H. polymorpha methanol oxidase promoter with an expression vector based on the S. cerevisiae 2-gin origin of replication sequence (Fellinger et al., submitted for publication). However, the

82

transformation frequency of this expression vector was low and its integration into the H. polymorpha genome was very difficult to accomplish. Only one stable integrant was obtained after laborious screening of transformants. Integration of the e-galactosidase expression vector into the H. polymorpha genome is desirable because of the high genetic stability of this situation, which is a prerequisite for the large-scale production of c~-galactosidase. In S. cerevisiae, plasmids based on the 2-gin sequence can be transformed with a high frequency and they combine a high copy number with a high stability (Futcher 1988). Recent reports from the literature, however, state that plasmids bearing the 2-gin sequence of S. cerevisiae cannot be transformed to H. polymorpha, indicating that the 2-gin sequence does not function as an origin of replication in this yeast (Roggenkamp et al. 1986). This is in contrast with results from Gleeson et al. (1986) who reported transformation and autonomous replication of YEp13, a plasmid containing the S. cerevisiae 2-1xm origin of replication (Broach et al. 1979), in H. polymorpha. Roggenkamp et al. (1986) found that an autonomously replicating sequence (ARS) isolated from H. polymorpha chromosomal DNA (HARS2) improved the transformation frequence and allowed autonomous replication of plasmids in H. polymorpha (Roggenkamp et al. 1986). As stable expression systems are important for industrial processes we decided to compare these different origins of replication, as well as the absence of an origin of replication, in identical vectors derived from YEpl 3 and in an identical genetic background, namely the leucine-deficient H. polymorpha strain A16 (Veale 1989). We, therefore, constructed three vectors, which were identical except for their origin of replication. These three different plasmids contain, respectively, the S. cerevisiae 2-gm sequence, the H. polymorpha HARS2 sequence and no origin of replication. The transformation frequencies of these different expression plasmids were determined and the integration pattern and copy number of the integrated expression plasmids were analysed by Southern hybridisation. The effect of the integration pattern, and the number of copies integrated into the H. polymorpha genome, on the expression level of the plant enzyme c~galactosidase was then analysed. Materials and methods Strains and media. E. coli JM109 (Yanisch-Perron et al. 1985) was grown in Luria broth medium and used for the propagation and cloning of recombinant plasmids. When needed, ampicillin was added to a final concentration of 100 lag/ml. H. polymorpha strain A16 ( L E U 2 - ) was grown in defined medium containing: 0.68% Yeast Nitrogen Base (YNB) without amino acids supplemented with 20 lag/ml leucine and 2% glucose, or in 1% yeast extract, 2% bactopeptone and 2% glucose (YEPD). Transformants were grown in defined media without leucine or in YEPD. Induction media consisted of 0.68% YNB (without amino acids) with 0.5% methanol or I % yeast extract, 2% bactopeptone and 0.5% methanol (YEPM).

technique of Hattori and Sakaki (1986). The foilowing D N A fragments were used as a probe in the Southern and Northern blot analyses: (1) The ~-galactosidase probe consists of a 800 bp PvuII/ HindIII fragment of the structural e-galactosidase gene from pUR2303 (Overbeeke et al. 1987), (2) The methanol oxidase coding sequence probe consists of an internal AeeI 744 bp fragment (position 151 to position 895) of the methanol oxidase gene (Ledeboer et al. 1985) and (3) The methanol oxidase promoter probe consists of a XhoI 878 bp fragment (position -1313 to position -435) of the methanol oxidase gene (Ledeboer et al. 1985). All D N A fragments were labelled using the hexamere labelling kit purchased from Amersham, UK.

Plasmids. The plasmid pUR3513 was obtained by deleting the SalI fragment, containing the 2-gin origin of replication of S. cerevisiae, of YEp13 (Fig. 2). In pUR3513, the 2.0 kb SalI fragment from pHARS2 (Roggenkamp et al. 1986), containing an autonomously replicating sequence for H. polymorpha, was inserted in the SalI site resulting in plasmid pUR3514 (Fig. 2). Plasmid pUR3501 (Fellinger et al., submitted for publication) contains the methanol oxidase promoter, the invertase signal sequence and the c~-galactosidase gene on a pEMBL9 vector. The methanol oxidase terminator located on a BglII (position 1905) - EcoRV (position 2269) 364 bp fragment (Ledeboer et al. 1985) was cloned in the BamHI/HincII sites of pEMBL9, resulting in plasmid pUR3511. Transformation and stable integration of plasmid vectors. The transformation procedure used is that of Klebe et al. (1983) with a few modifications as described by Roggenkamp et al. (1986). In order to obtain stable integration of the plasmids, the transformants were grown in YEPD agar in nunc tubes. This was repeated successively at least four times for each transformant. Following this procedure the presence of the plasmid (replicating/integrated) was checked by growing the transformants on defined medium without leucine. The presence of an integrated plasmid was checked by Southern hybridisation. Isolation of DNA and mRNA from H. polymorpha. Plasmid D N A from H. polymorpha was isolated using the method originally developed for S. cerevisiae by Hoffman and Winston (1987). For the isolation of chromosomal D N A from H. polymorpha we used a modification of the procedure described by Janowicz et al. (1985) with a lysis solution of 0.5 ml 50 mM EDTA (pH 8.5) containing 0.3% SDS. For m R N A isolation from H. polymorpha the cells were grown overnight in YEPD and diluted (1 : 100) in 20 ml YEPM for induction of the methanol oxidase promoter. After an induction period of 6 h the cells were collected and from this stage on the procedure described by Zitomer et al. (1979) was followed.

Southern blot analysis. The chromosomal D N A was digested and the resulting fragments were separated on a 0.7% agarose gel to perform a Southern blot procedure as described by Southern (1975). The D N A was blotted from the gel on Genescreen (Boston, Massachusetts) plus paper using a Pharmacia (Uppsala, Sweden) vacugene blot apparatus. Blots were prehybridised overnight in a hybridisation mixture (50 mM Tris-HC1 pH 7.5, 10 m M EDTA, 1 M NaC1, 0.1% SDS, 0.1% Na-pyrophosphate, 0.2% Ficoll, 0.2% BSA, 0.2% PVP and a denatured mixture of 0.1 mg/ml ss-DNA and 0.01 rag/m1 poly-rA) at 68 ~ Hybridisation was performed overnight at 68 ~ in the hybridisation mixture with the denaturated (5 min 100~ labelled fragment. After hybridisation the filters were washed twice with 2 x SSC, 0.1% SDS, 0.1% Na-pyrophosphate for 20 rain at 68~ and twice with 0.2% SSC, 0.1% SDS, 0.1% Na-pyrophosphate for 20 rain at 68 ~ The filters were air-dried and exposed to X-ray films. Northern blot analysis. The R N A samples were separated on a

DNA techniques. Restriction enzymes, T4 D N A ligase and Klenow enzyme were used as recommended by the manufacturer. All recombinant D N A manipulations were performed using standard methods (Maniatis et al. 1982). Double-strand sequencing followed the

denaturating formamide/formaldehyde gel. The R N A was blotted on Genescreen plus paper using the vacugene system of Pharmacia. The blots were heated at 80 ~ for 2 h. Hybridisation was performed as described for Southern blot analysis.

83

Analysis of e-gatactosidase expression. Extracellular c~-galactosidase was isolated by collecting the medium of an induced culture by centrifugation. Intracellular ~-galactosidase was isolated by collecting a 3 ml sample of a YEPM culture. The cells were pelleted, washed in PBS (170 mM NaC1, 10 mM Na2HPO~'2H20, 2 mM NaH2PO4-H20) and resuspended in 100 gl of this buffer. An equivalent volume of glassbeads was added to the suspension. To obtain lysis the suspension was shaken three times at maximum speed on a vortex mixer for 30 sec with 1 min intervals on ice. The ~-galactosidase was assayed as described by Overbeeke et al. (1987) using p-nitrophenyl e-D-galactopyranoside (Sigma, St. Louis, MO, USA) as a substrate. Results

Vector constructions When we succeeded in synthesizing the plant enzyme egalactosidase by H. polymorpha under control of the methanol oxidase promoter, the expression plasmid (pUR3510, Fig. 1 B) contained the e-galactosidase gene sandwiched between the methanol oxidase promoter and methanol oxidase terminator. The methanol oxidase terminator consisted of 1 kb of the methanol oxidase terminator sequence and 1.5 kb of the coding methanol oxidase sequence, which was not in frame with the preceding c~-galactosidase gene. The S. eerevisiae invertase signal sequence (Sarokin and Carlson 1984) was used to ensure secretion of the c~-galactosidase into the medium. The expression plasmid pUR3510, a derivative of YEp13 (an Escherichia coli-S, cerevisiae shuttle vector), also contained the fl-lactamase gene for selection in E. coli, the 2-gin sequence as the origin o f replication, and the fl-isopropylmalate dehydrogenase gene (LEU2, Andreadis et al. 1982) as a selection marker for transformation to the LEU2-deficient H. polymorpha strain A16. This vector, however, could not be transformed with high frequency and was very difficult to integrate into the H. polymorpha genome. Only one stable integrant was obtained. Therefore, we decided to investigate the influence of different origins of replication on transformation

0 EcoRk 6000 60~i00amH1 Sall

4000

IOOO

0

v

2OO0

-

3000

Transformation and stable integration of expression plasmids The plasmid without an origin of replication (pUR3515) could be transformed with an efficiency of 1 3 - 2 0 transformants per tag of DNA. When, subsequently, the plasmid containing the 2-1am sequence (pUR3516) was transformed only 1 - 5 transformants per gg of D N A were obtained. Transformation of the plasmid containing the H A R S 2 sequence resulted in a three-fold increase of the transformation fi'equency ( 4 0 - 6 0 transformants per lag DNA) compared to that of plasmid pUR3515. Either pUR3516 or pUR3517 could be isolated from H. poly-

500

35OO~

I NpUR3501 I ~I~ 5OOO

frequency and, more importantly, on integration. For this purpose the ~-galactosidase expression cassette from pUR3510 was cloned into three plasmids containing three different origins of replication. This expression cassette is identical to that o f p U R 3 5 1 0 except that 1.5 kb of the coding methanol oxidase sequence has been deleted. This sequence was deleted because transformants of pUR3510 produce two distinct c~-galactosidase mRNAs, probably due to the expected transcription termination at the methanol oxidase terminator and an incorrect transcription termination between the 3' part of the ~-galactosidase gene and the methanol oxidase terminator. The latter is possibly caused by this methanol oxidase coding sequence (Fellinger et al., submitted for publication). Deletion of the 1.5 kb coding methanol oxidase could result in a more efficient transcription termination, in its turn resulting in an increased production of one distinct m R N A of the correct length and, consequently, in a higher ~-galactosidase expression. The cloning of this ~galactosidase expression cassette (pUR3512) is shown in Fig. 1 A. The c~-galactosidase expression cassette of pUR3512 was cloned into the vectors pUR3513, Y E p l 3 and pUR3514, which is shown in Fig. 2; the resulting expression plasmids are pUR3515, pUR3516 and pUR3517, respectively.

loo~ 1500 EcoRI

Sel~

ma

Sal I digesk

isolate 2.6 kb fragment

fitted in with Ktenow' ~ 7000

/~

a I digest,

ate vector

Indil

0EcoRI

B000

OEcoRh

1000 IN~VSS 2000

l~ooo

sa

Sail

E R

iN~ss

12000

LOOO

5000

Sall IOOOO

fl[~

20~

4000

H;ndI

B

Sall

HindtLI BamHI

Fig. 1A, B. Construction of vector pUR3512 (A) containing the c~-galactosidase gene fused between the methanol oxidase promoter-invertase signal sequence and the methanol oxidase terminator. An EcoRI/HindIII digest gives the ~-galactosidase expression cassette. pUR3510 (B) is identical to pUR3516 (Fig. 2) except for the methanol oxidase terminator which consists of 1 kb of the methanol oxidase terminator and 1.5 kb of the methanol oxidase coding sequence

84 7OOO

6

0

0

0 EcoRI

0

IOOO

~

In~ss 5000

2000

40

H~ndlll

| fcoRIIHindlll digest,

| isolate 3kb fragment. filled in with Klenow IOOOO

Salt BamH[ Hmdill

Pvu~l

~ ~

Sol r

Pvull digest, E~oRc isolate vector

EcoR,

gcoRI0

BO HI HuldH 12000

2000

i

2000

eco

8000

.....

....

Xhol

Sail

4000 phi

Sallldigest, isolate vector OPvull

0 EcoRI .

.

.

.

.

.

IOOO

6000

Pvull digest, isolate vector

so.

5000 ECORI

sooe

2000

E,oFI~

oUR3515 1o000

2000 EcoRI

OOO E0o0

4000

Sall digest, isolate vector 0s

0

HindI[I Sat I

Pvul[

aooo Sail

100o

S~I

1000

10O00 ECORI

5000

7000

Pvull 2o00

5000

Sell

2000 ECoRI I

Sall digest,

isolate 2.0kb fragment

3000

2OO0

mill

partial Pvull digest,6000Sall

NvS5

isolate vector

so.

gOcOoOR i

4000

4000

Fig. 2. Constructionof vectorspUR3515, pUR3516 and pUR3517, containingthe ~-galactosidaseexpressioncassetteclonedinto the vectors with different origin of replications

morpha, transformed with the plasmids pUR3516 and pUR3517, and analysis revealed no rearrangements and no multimeric forms, which is in contrast to what Tikhomirova et al. reported (1986). The transformants obtained were stabilised by having the expression plasmids integrated into the H. polymorpha genome. Transformants of plasmid pUR3515 are integrants due to the absence of an origin of replication; pUR3515 can be integrated very easily as is shown by the transformation frequency. Plasmid pUR3517 could be integrated into the H. polymorpha genome by using the standard stabilisation procedure. Plasmid pUR3516, the expression plasmid with the 2-gm sequence, could not be stabilised by integration into the genome using this standard procedure. Integration pattern of expression plasmids Hybridisation of the parent strain A16 with the methanol oxidase promoter probe clearly shows one band of ap-

proximately 3.0 kb (Fig. 3A, lane 4). This band corresponds with an EcoRI fragment 3.0 kb upstream of the start codon to 15 bp downstream of the start codon of the methanol oxidase gene. When integration occurs of either of the constructed vectors into the homologous chromosomal methanol oxidase promoter sequence, this band should be disrupted. This is the case with the integrants 26, 41 and 27 (Fig. 3 A, lanes 1-3). Because integration of the plasmid into the methanol oxidase promoter introduces an EcoRI-site at the beginning of this promoter the band which appears is approximately 1.5 kb (the length of the methanol oxidase promoter in the expression cassette). Hybridisation of the chromosomal DNAs from these integrants with the methanol oxidase coding-sequence probe (Fig. 3 B, lanes 1 - 3) slaows that the methanol oxidase structural gene and the methanol oxidase terminator region are intact as compared with the parent strain A16 (Fig. 3 B, lane 4). The copy number of the integrated plasmids can be estimated from the results of

85

Fig. 3A-C. Southern blots of genomic H. polymorpha DNA digested to completion with EcoRI. In each panel lanes 1 to 4 correspond to genomic DNA of respectively 2% 41, 2v and parent strain A16. Panel A shows the hybridisation with the methanol oxidase promoter probe, panel B with the methanol oxidase coding sequence probe and panel C with the ~-galactosidase probe

the hybridisation experiment using the e-galactosidase probe. The parent strain AI6, as expected, gives no signal with this probe (Fig. 3 C, lane 4). If one copy of the expression plasmid is integrated into the methanol oxidase promoter it should give one band of approximately 6.3 kb ranging from the EcoRI-site, which is normally 3.0 kb upstream of the start codon of the methanol oxidase gene, to the EcoRI-site near the/Mactamase gene of the plasmid, e.g., 26 (Fig. 3 C, lane 1). Two integrated plasmids should give two separate bands of 6.3 (see above) and 4.8 kb ranging from the EcoRI-site 1.5 kb upstream of the ATG start codon of the e-galactosidase gene in the expression cassette to the EcoRI-site near the /Llactamase gene of the expression plasmid (the plasmid band), e.g., 41 (Fig. 3 C, lane 2). A third (or more) integrated copy should give a more intense blackening on the Southern blot of the 4.8 kb band (the plasmid band), e.g., 27 (Fig. 3 C, lane 3). Integration schemes of integrants consisting of one (e.g. 26) o r more (e.g., 41 or 27) copies are given in Fig. 4. In other cases, (50%), integration had not occurred on the homologous methanol oxidase locus as could be concluded from the Southern analysis (data not shown), because these integrants show the same pattern for the methanol oxidase promoter and methanol oxidase probe as the parent strain A16. Hybridisation of the chromosomal D N A isolated from these integrants with the e-galactosidase probe, showed that all of the integrated expression plasmids at sites other than the methanol oxidase promoter are single-copy integrants whereas homologous integrants are mostly tandem/triplet integrants. No autonomously replicating plasmids could be isolated from the integrants. Analysis of the c~-galactosidase expression

The enzyme c~-galactosidase was secreted by both pUR3516 and pUR3517 transformants in concentrations between 10 and 20 mg/1. These concentrations are comparable with those obtained for the e-galactosidase pro-

duction of a single copy integrant (see results below). No large differences could be detected in the expression level of e-galactosidase by pUR3516 or pUR3517 transformants. After the integration of the expression plasmids pUR3515 and pUR3517 into the H. polymorpha genome the ~-galactosidase production and secretion of the above-described integrants was measured. Because of the integrants high stability the induction experiments were performed by pre-culturing the cells non-selectively in YEPD and inducing them in YEPM. In all cases the OD66 o of the cultures (i.e., when the cells were harvested for the e-galactosidase assay) was 7.0 and the level of e-galactosidase as determined intracellularly was _+10% of the extracellular e-galactosidase level. A one-copy integrant of pUR3515, such as 26, secretes 14 _+0.8 mg/1 of e-galactosidase into the medium, whereas a two-copy integrant of pUR3517, such as 41, secretes twice as much e-galactosidase into the medium (30 _+2.3 mg/1). A threecopy integrant, such as 27, secretes 40+2.9 mg/1 of c~galactosidase into the medium. This gene-dose effect was also observed for other integrants. To obtain information about the transcription efficiency, m R N A was isolated from the one, two and threecopy integrants 26, 41 and 27, respectively. The corresponding Northern blot is shown in Fig. 5. The m R N A isolated from a pUR3510 integrant (82), containing the 1.5 kb methanol oxidase coding sequence, and hybridised with the ~-galactosidase probe, is shown in lane 4 of Fig. 5. The difference between this m R N A and that isolated from the integrants described in this study is obvious and shows that the deletion of the 1.5 kb of coding methanol sequence in pUR3510 has resulted in only one defined transcription stop for the ~-galactosidase mRNA, This deletion, however, has not resulted in a higher ~-galactosidase production. The experiment also clearly demonstrates that an increased copy number results in a proportionally increased level of specific c~galactosidase mRNA. An overview of the described results is given in Table 1.

86 ATG EcoRI

3008bp

I

\EcoRI

EcoRl

6OO0bp

pMOX

[

tMOX

pMOX-probe cod MOX-probe

o(-gal-probe pMOX

c~- gal

LEU2

r

]

pMOX o(- gal

LEU2

x pMOX

cod MOX

499 bp 6000 bp C =

6300 bp

4800 bp

Fig. 4. Schematic presentation of the integration of the c~-galactosidase expression plasmids into the methanol oxidase locus of H. polymorphaA16. Bands which should hybridise with each probe, and length of these bands, are given in base pairs. A, bands which should hybridise with the methanol oxidase probe; B, bands which should hybridise with the methanol oxidase coding sequence probe and C, bands which should hybridise with the e-galactosidase probe. A, B and C correspond with panel A, B and C in Fig. 3

Fig. 5. Northern blot of mRNA isolated from integrants after induction (6 h) in YEP with 0.5% methanol. The Northern blot was hybridised with the c~-galactosidase probe. Lanes 1 to 5 are mRNA isolated from respectively 26, 41, 27, 82 and parent strain. AI6 Table 1. An overview of the expression results of the integrants. The standard deviation is a resultant of four independent induction experiments Integrant

Integrated plasmid

Copy number

c~-galactosidaseexpression (mg/l)

~-galactosidase mRNA

26 41 2v

pUR3515 pUR3517 pUR3515

1 2 3

14_+0.8 30_+2.3 40_+2.9

+ ++ +++

Discussion In a previous study (Fellinger et al., submitted for publication) we encountered two problems. Firstly, the expression plasmid used could not be transformed to H. polymorpha with a high frequency and secondly, integration of the expression cassette into the H. polymorpha genome, essential for a stable expression, was very difficult to accomplish, resulting in only one stable integrant. Recent reports from the literature questioned the functioning of the 2-gm sequence as an origin of replication in

H. polymorpha (Roggenkamp et al. 1986; Gleeson et al. 1986) and mentioned the isolation of an autonomously replicating sequence from H. polymorpha (HARS2, Roggenkamp et al. 1986). Results presented by the latter authors, obtained with the uracil-deficient H. polymorpha LR9, showed less transformants using the 2-1am sequence-containing plasmid pRB58 (Carlson and Botstein 1982) than using YIp5 (Stinchcomb et al. 1980), a plasmid containing no origin of replication and approximately 8 kb smaller than pRB58. From these results Roggenkamp et al. concluded that the 2-gin sequence does not function as an origin of replication in H. polymorpha. Gleeson et al. (1986) transformed YEpl 3 to a leucine-deficient H. polymorpha with reasonable frequencies and reported autonomous replication of this plasmid, but no comparison was made with a plasmid with no origin of replication. Gleeson et al. concluded that if the 2-gin sequence does not act as an origin of replication there should be another sequence present on YEpl 3 that does. Because of these recent results the 2-gin sequence and the HARS2 sequence were compared with a vector containing no origin of replication in an identical e-galactosidase expression vector and using an identical genetic background, namely the leucine-deficient H. polymorpha A16. Therefore, three c~-galactosidase expression plasmids have been constructed which contain respectively the S. cerevisiae 2-gin sequence (pUR3516), the H. potymorpha HARS2 sequence (pUR3517) and no origin of replication (pUR3515). From our transformation data it can be concluded that the HARS2 sequence indeed functions as an origin of replication in H. polymorpha A16 because the transformation frequency of pUR3517 is three times higher than that in the control experiment (pUR3515). However, the increase in transformation frequency (of approximately 150 x ) which Roggenkamp et al. (1986) reported could not be obtained; this may be explained by the different strains used in the two sets of experiments. The u-galactosidase expression of transformants containing plasmid pUR3517, which could be isolated, was low. So the expected high copy number of this HARS2 plasmid, in combination with the low e-galactosidase expression, leads to the assumption that this expression plasmid is very unstable in H. polymorpha A16. This implies that, although autonomously replicating sequences have been isolated for H. polymorpha, integration of the expression vector in the H. polymorpha genome is still a prerequisite for the production of heterologous proteins by this organism. The transformation frequency of pUR3516 (2-gin sequence) was low, even lower than frequencies obtained with pUR3515 (no origin of replication). Plasmids could be isolated from pUR3516 transformants whereas this was not the case with pUR3515 transformants. By allowing autonomous replication, the 2-1am sequence functions as an origin of replication, though the transformation frequency is not increased by this sequence. The assumption that another sequence present on YEpl 3 may act as an origin of replication (Gleeson et al. 1986) is in disagreement with our data which demonstrate that it is the 2-1am sequence which is responsible for the autonomous replication of plasmids. From our data it cannot be con-

87 cluded that the decrease in transformation frequency of 2 lam-based plasmids is caused by inefficient partitioning of plasmids containing the 2-gm sequence or by inefficient replication. The transformation protocol and the use of a plasmid containing no origin of replication, as described in this paper, combined with a high carrier D N A concentration, resulted in a transformation frequency of 104 transformants per gg of plasmid D N A (P. Haima, personal communication). To obtain a stable situation for the production of e-galactosidase by H.polymorpha, integration of the e-galactosidase expression cassette into the H. polymorpha genome had to be established. Using the standard procedure the expression plasmids pUR3517 and pUR3515 could be successfully integrated into the H. polymorpha genome, whereas plasmid pUR3516, containing the 2-gm sequence, could not. The only difference between pUR3516 and the other two expression plasmids is in the origin of replication. The results described show that the 2-txm sequence of pUR3516, or the ARS activity of the 2-gm sequence, prevents integration of the e-galactosidase expression plasmid, or else that integration of the plasmid containing the 2-gm sequence results in an unstable situation leading to a rapid loss of the integrated plasmid. However, integration of the plasmid pUR3517, containing the H A R S 2 sequence, into the H. polymorpha genome is easy and stable which seems contradictory because both sequences are ARS sequences. The different behaviour of these origins of replication may be explained by the different types of sequence: the H A R S 2 sequence is genomic whereas the 2-1xm sequence is plasmid-originated. In S. cerevisiae 2-1am based expression vectors normally do not integrate; not even when homologous sequences are present. Integration of the expression plasmids pUR3515 and pUR3517 into the H. polymorpha genome occurred in 50% of the analysed integrants on the homologous methanol oxidase promoter of the methanol oxidase chromosomal locus. Tandem integration by pUR3515 and pUR3517 on the homologous methanol oxidase promoter was favoured above single-copy integration at other sites in the H. polymorpha genome. Tandem integration o f expression plasmids on homologous loci has also been described for several other host organisms, e.g. S. cerevisiae (Orr-Weaver and Szostak 1983), Aspergillus niger (van Hartingsveldt et al. 1987) and Trichoderma reesei (Penttil/i et al. 1987). The expression of the ~-galactosidase by integrants with different copy numbers showed that at low copy numbers the amount of enzyme produced is proportional to the number of integrated copies of the expression plasmid into the H. polymorpha genome. The amount of specific ct-galactosidase m R N A also increased with increasing copy number which implies that, in batch culture, transcription is limiting in :~-galactosidase expression. The c~-galactosidase produced by H. polymorpha is efficiently secreted and correctly processed. Expression levels of 100 mg/l c~-galactosidase at low cell densities have been obtained in intial fermentation studies. The host/vector system described in this study, in combination with the efficient and relatively cheap continuous fermentation process, makes H. poly-

rnorpha an excellent host for the production and secretion of functional heterologous proteins.

Acknowledgements. We are grateful to A.L Fetlinger, Dr. M.L.F. Giuseppin and Dr. A.M. Ledeboer for their stimulating discussions and for critically reading the manuscript. We thank P. Haima for providing experimental data of transformation frequencies. References Andreadis A, Hsu J, Kohlhaw GB, Schimmel P (1982) Cell 31:319325 Broach J, Strathern J, Hicks J (1979) Gene 8:121-133 Carlson M, Botstein D (1982) Cell 28:145-154 Cregg JM, Tschopp JF, Stillman C, Siegel R, Akong M, Craig WS, Buckholz RG, Madden KR, Kellaris PA, Davis GR, Smiley BE, Cruze J, Torregrossa R, VelicelebiG, Thill GP (1987) Bio/ Technology 5:479-485 Futcher B (1988) Yeast 4:27-40 Gleeson MA, Sudbery PE (1988) Yeast 4:1-15 Gleeson MA, Ortori GS, Sudbery PE (1986) J Gen Microbiol 132:3459 3465 Giuseppin MLF (1989) Biotechnol Bioeng 33:524-535 Giuseppin MLF, Eijk van HMJ, Bes BCM (1988) Biotechnol Bioeng 32:577-583 Hartingsveldt van W, Mattern IE, Zeijl CMJ, Pouwels PHi Hondel van den CAMJJ (1987) Mol Gen Genet 206:71-75 Hattori M, Sakaki Y (1986) Anal Biochem 152:232-238 Hoffhaan CS, Winston F (1987) Gene 57:267-277 Hughes SG, Overbeeke N, Robinson S, Pollock K, Smeets FLM (1988) Plant Mol Biol 11:783-789 Janowicz ZA, Eckart MR, Drewke C, Roggenkamp RO, Hollenberg CP (1985) Nucleic Acids Res 13:3043 3062 Klebe RJ; Harries JV, Sharp ZD, Douglas MG (1983) Gene 25: 333341 Ledeboer AM, Edens L, Maat J, Visser C, Bos JW, Verrips CT (1985) Nucleic Acids Res 13:3063-3082 Ledeboer AM, Verrips CT, Visser C, Janowicz ZA, Hollenberg CP (1986) patent EP-A O 173 378 Maniatis T, Fritsch El=, Sambrook J (1982) in Molecular cloning: a laboratory manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York Meier H, Reid JSG (1982) In: Loewus FA, Tanner W (eds) Encyclopedia of plant physiology, new series 13A. Springer, Berlin, p. 418-471 Orr-Weaver TC, Szostak JW (1983) Mol Cell Biol 3:747-749 Overbeeke N, Fellinger AJ, Hughes SG (1987) patent WO 87/07641 Overbeeke N, Fellinger AJ, Toonen MY, Wassenaar van PD, Verrips CT (1989) Plant Mol Biol 13:541-550 Penttilfi M, Nevalainen H, RS.tt6 M, Salminen E, Knowles J (1987) Gene 61:155-164 Roggenkamp R, Hansen H, Eckhart M, Janowicz Z, Hollenberg CP (1986) Mol Gen Genet 202:302 308 Sarokin L, Carlson M (1984) Mol Cell Biol 4:2750 2757 Southern EM (1975)J Mol Biol 98:503-517 Stinchcomb DT, Thomas M, Kelly I, Selker E, Davies RW (1980) Proc Natl Acad Sci USA 77:4559-4563 Tikhomirova LP, Ikonomova RN, Kuznetsova EN 0986) Curt Genet 10:741-747 Tschopp JF, Sverlow G, Kosson R, Craig W, Grinna L (1987) Bio/Technology 5:1305- 1308 Veale R (1989) PhD thesis, University of Sheffield, UK Veenhuis M, Van Dijken JP, Harder W (1983) Adv Microbiol Physiol 24:2-76 Yanisch-Perron C, Vieira J, Messing J (1985) Gene 33:103-119 Zitomer RS, Montgomery DL, Nichols DL, Hall BD (1979) Proc Natl Acad Sci USA 76:3627-3631 Communicated by L. A. Grivell