CA2070057A1 - Process for the production of glycosyltransferases - Google Patents

Abstract

Improved process for the production of glycosyltransferases Abstract of the disclosure The invention relates to the field of recombinant DNA technology and provides a novel method for the production of glycosyltransferases by use of transformed yeast strains.

Description

h~ ~ 7 ~ ~ ~ 7 Improved process for the production of glycosyltransferases The invention relates to the field of recombinant DNA technology and provides animproved method for the production of glycosyltransferases by use of transformed yeast strains.

Glycosyltransferases transfer sugar residues from an activated donor substrate, usually a nucleotide sugar, to a specific acceptor sugar thus forming a glycosidic linkage. Based on the type of sugar transferred, these enzymes are grouped into families, e.g. galactosyl-transferases, sialyltransferases and fucosyltransferases. Being resident membrane proteins primalily located in the Golgi appara~lls, the glycosyl~ransl`erases shate a common domain structure consisting of a short amino-terminal cytoplasmic tail, a signal-anchor domain, and an extended stem region which is followed by a large carboxy-terminal catalytic domain. The signal-anchor domain acts as both uncleavable signal peptide and as meMbrane spanning region and orients the catalytic domain of the glycosyltransferase within thc lumen of the Golgi apparatus. The luminal stem or spacer region is supposed to selve as a flexible tether, allowing the catalytic domain to glycosylate carbohydrate gro~lps of membrane-bound and soluble proteins of the secretory pathway enroute through the Golgi apparatus. Furthermore, the stem portion was discovered to function as retention signal to keep the enzyme bound to the Golgi membrane (PCT Application No. 91/0663~).
Soluble forms of glycosyltransferases are found in milk, serum and other body fluids.
These soluble glycosyltransferases are supposed to result from proteolytic release from the corresponding membrane-bound forms of the enzymes by endogenous proteases~
presumably by cleavage between the catalytic domain and the transmembrane domain.

Enzymatic synthesis Or carbohydrate structures has the advantage of a high stereo-selectivity and regioselectivity, rendering the glycosyltransferases a valuable tool for the modification or synthesis of glycoproteins, glycolipids and oligosaccharides. In contrast to chemical methods the time-consuming introduction of protective groups is superfluous.

As glycosyltransferases are naturally occurring in very low amounts, isolation from natural sources and subsequent purification are difficult. Therefore, production using 2 ~ 7 recombinant DNA technology has been worked on. For example, galactosyltransferases have been expressed in E. coli (PCT 90/07000) and Chinese hamster ovary (CHO) cells (~mith, D.F. e~ al. (1990) J. Biol. Chem. 265, 6225-34), sialyltransferases have been expressed in ~HO cells (Lee, E.U. (1990) Diss. Abstr; Int.B.5~), 3453-4) and COS-l cells (Paulsen, J.C. et al. (1988) J. Cell. Biol. 107, lOA), and fucosyltransferases have been produced in COS-1 cells (Goelz, S.E. et al. (1990) Cell 63, 1349-1356; Larsen R.D. et al.
(1990) Proc. Natl. Acad. Sci. USA 87, 6674-6678) and CHO cells (Potvin, B. (1990) J.
Biol. Chem. 265, 1615-1622). Considering the facts that heterologous expression in prokaryotes has the disadvantage of providing unglycosylated products, glycosyl-transferases, however, being glycoproteins, and that glycosyltransferase production by use of mammalian hosts is very expensive as well as complicated due to the presence of many endogenous glycosyltransferases which would contaminate the desired product, there is a need for improved methods which render possible the economic production of glycosyltransferases on a large scale.

It is an object of the present invention to provide such melllods.

The present invention provides a process for the production of biologically active glycosyltransferases by a recombinant DNA tcchnology usino a yeast vec~or expression system.

More specifically, the present invention provides a process for the production of a membrane-bound glycosyltransferase selected from the group consisting of a galactosyl-transferase, a sialyltransferase and a fucosyltransferase, or a variant thereof, respectively, said process comprising culturing a yeast strain which has been transformed with a hybrid vector comprising an expression cassette comprising a promoter and a DNA sequence coding for said glycosyltransferase or variant which DNAis controlled by said promoter, and recovering the enzyrnatic activity.

In a first embodiment, the invention relates to a process for the production of a membrane-bound glycosyltransferase selected from the grollp consisting of a galactosyl-transferase, a sialyltransferase and a fucosyltransferase, or a variant thereof, respectively, said process comprising culturing a yeast strain which has been transformed with a hybrid vector comprising an expression casset~e comprising a promoler, a DNAsequence codin~, for said glycosyltransferase or variant which DNA sequence is controlled by saidpromoter, and a DNA sequence containing yeast transcription termination signals, and - 2 ~ 7 recovering the enzymatic activity.

In a second embodiment, the invention relates to a process for the production of a membrane-bound glycosyltransferase selected from the group consisting of a galactosyl-transferase, a sialyltransferase and a fucosyltransferase, or a variant thereof, respectively, said process comprising culturing a yeast strain which has been transformed with a hybrid vector comprising an expression cassette comprising a promoter operably linked to a first DNA sequence encoding a signal peptide linked in the proper reading frame to a second DNA sequence coding for said glycosyltransferase or variant, and a DNA sequence containing yeast transcription termination signals, and recovering the enzymatic activity.

The term "glycosyltransferase" whenever used hereinbefore or hereinafter is intended to embrace the family of galactosyltransferases, the family of sialyltransferases and the family of fucosyltransferases. Said glycosyltransferases are naturally occurring enzymes of mammalian, e.g. bovine, murine, rat and human origin. Preferred are naturallyoccurring human full-lenglh glycosyltranst`erases including those enzymes identil`ied hereinafter by their EC-numbers.

~'he membrane-bound galactosyltransfcrases and their variants obtainable according to the inventive process catalyse the transfer ol` a gaLlctose resklue lrom an activated donor, usually a nucleotide activated donor such as uridine diphosphate galactose (UDP-Gal), to a carbohydrate group.

Examples of membrane-bound galactosyltransferases are UDP-Galactose: ~-galactoside c~(1-3)-galactosyltransferase (EC 2.4.1.151) which uses galactose as acceptor substrate forming an oc(1-3)-linkage-and UDP-Galactose: ~-N-acetylglucosamine ,B(1-4)-galactosyltransferase (EC 2.4.1.22) which transfers galactose to N-acetylglucosamine (GlcNAc) forming a ~(1-4)-linkage, including variants thereof, respectively. In the presence of c~-lactalbumin, said ,B(1-4)-galactosyltransferase also accepts glucose as an acceptor substrate, thus catalysing the synthesis of lactose.

The most preferred membrane-bound galactosyltransferase is the en~yme having theamino acid sequence depicted in the sequence listing with the SEQ ID NO. 1.

The mcmbrane-bound sialyltransferases and their variants obtainable according to the process of the invention catalyse lhe transfer of sialic acids (for example N-ace~yl 2 ~

neuraminic acid (NeuAc)) from an activated donor, usually a cytidine monophosphate sialic acid (CMP-SA) to a carbohydrate acceptor residue. An example of a membrane-bound sialylhransferase obtainable according to the inventive method is the CMP-NeuAc: ~-galactoside o~(2-6)-sialylhransferase (EC 2.4.99.1) which forms theNeuAc-o~(2-6)Gal-,~ 4)GlcNAc-sequence common to many N-linked carbohydrate groups.

The most preferred membrane-bound sialyltransferase is the enzyme having the amino acid sequence depicted in the sequence listing with the SEQ ID NO. 3.

The membrane-bound fucosylhransferases and their variants obtainable according to the process of the invention catalyse the transfer of a fucose residue from an activated donor, usually a nucleotide-activated donor, such as guanosine diphosphate fucose (GDP-Fuc), to a carbohydrate group. Examples of such fucosyltransferases are GDP-Fucose:~-galactoside o~(1-2)-fucosylhransferase (EC 2.4.1.69) and GDP-Fucose:N-acetylglucosamine cc(l-3/4)-fucosyltransferase (EC 2.4.1.65).

The most preferred membrane-bound fucosyltransferase is the enzyme having the amino acid sequence depicted in the sequence listing with SEQ ID NO. 5.

The term variants as used helein is intended to embrace both membrane-bound and soluble variants of the naturally occurring membrane-bound glycosylhansferases of mammalian origin with the provision that these variants are enzymatically active.
Preferred are variants of human origin.

For example, the term "variants" is intended to include naturally occurring membrane-bound variants of membrane-bound glycosylhransferases found within a particuLlr species, e.g. a variant of a galactosyltransferase which differs from the enzyme having the aMino acid sequence with the SEQ ID NO. I in that it lacks serine in position 11 and has the amino acids valine and tyrosine instead of alanine and leucine in positions 31 and 32, respectively. Such a variant may be encoded by a related gene of the same gene family or by an allelic variant of a particular gene. The term "variants" also embraces glycosyltransferases produced from a DNA which has been subjected to m vitro mutagenesis, with the provision that the protein encoded by said DNA has the en~ymatic activity of the nativc glycosyltransferase. Such modifications may consist in an addition, exchange and/or deletion of amino acids, the latter resulting in shortened variants.

- s -Preferred variants prepared according to the process of the invention are shortened variants, particularly soluble variants, i.e. variants which are not membrane-bound.
Shortened variants include for example soluble forms of membrane-bound glycosyltransferases and their membrane-bound variants, e.g. those variants mentioned above, which are secretable by a transformed yeast strain used in the process according to the invention. According to the present invention, these soluble enzymes are the preferred truncated variants.

The invention also relates to a process for the production of a soluble variant of a membrane-bound glycosyltransferase selected from the group consisting of a galactosyl-transferase, a sialyltransferase and a fucosyltransferase, said process comprising culturing a yeast strain which has been transformed with a hybrid vector comprising an expression cassette comprising a promoter operably linked to a first DNA sequence enco(lillg a signal peptide linked in the proper reading frame to a second ONA sequence coding for said variant, and a DNA sequence containing yeast transcription termination signals, and isolating said variant.

By definition the soluble form of a glycosyltransl`erase is a shortencd valiallt differing from the corresponding full-length, i.e. the membrane-bound forrn naturally located in the endoplasmic reticulum or the Golgi complex, by lack of the cytoplasmic tail, thesignal-anchor and, optionally, part of the stem region. The term "part of the stem" region as used herein is defined tO be a minor part of the N-terminal side of the stem region consisting of up to 12 amino acids. In other words, the soluble variants prepared according to the process of the present invention consist of essentially the whole stem region and the catalytic domain.

The soluble variants are enzymatically active enzymes differing from the corresponding full-length forms by the absence of an NH2-terminal peptide consisting of 26 to 67, particularly 26 to 61, amino acid residues, with the provision that those forms lacking part of the stem region only lack the above-defined minor part therof. Preferred are soluble variants obtainable from the membrane-bound glycosyltransferases identified hereinbefore by their EC-numbers and additionally soluble variants obtainable from the membrane-bound fucosyltransferase with S~Q. ID NO. 5.

Preferred soluble variants of galactosyltransferases are distinct from the corresponding full-length forms in that they lack an NH2-terrninal peplide consisting of 37 to 55, particularly 41 to 44, amino acids. The most preferred soluble variant prepared according to the inventive process is the enzyme having the amino acid sequence depicted in the sequence listing with the SEQ :[D NO. 2.

Preferred soluble variants of sialyltransferases miss an NH2-terminal peptide consisting of 26 to 38 amino acids compared to the full length form. The most preferred soluble variant prepared according to the inventive process is the enzyme having the amino acid sequence depicted in the sequence listing with the SEQ ID NO. 4. Likewise preferred is the soluble variant designated ST(Lys27-Cys406) consisting of the amino acids 27 to 406 of the amino acid sequence listed in SEQ ID NO. 3.

Preferred soluble variants of fucosylhransferases differ from the corresponding full-length enzymes in that they lack an NH2-terminal peptide consisting of 56 to 67, particularly 56 to 61, amino acids. Especially preferred is the soluble variant designated FT(Arg62-Arg405) consisting of the amino acids 62 to 405 of the amino acid sequence listed in SEQ ID NO. 5.

The yeast host strains and the constituents of the hybrid vectors are those specified below.

The hansformed yeast strains are cultured using methods known in thc art.

Thus, the transformed yeast strains according to the invention arc cultured in a liquid medium containing assimilable sources of carbon, nitrogen and inorganic salts.

Various carbon sources are usable. Examples of preferred carbon sources are assimilable carbohydrates, such as glucose, maltose, mannitol, fructose or lactose, or an acetate such as sodium acet~te, which can be used either alone or in suitable mixtures. Suitable nihro-gen sources include, for example, amino acids, such as casamino acids, peptides and pro-teins and their degradation products, such as tryptone, peptone or meat extracts, further-more yeast extract, malt extract, corn steep liquor, as well as ammonium salts, such as ammonium chloride, sulphate or nitrate which can be used either alone or in suitable mixtures. Inorganic salts which may be used include, for example, sulphates, chlorides, phosphates and carbonates of sodium, potassium, magncsium and calcium. Additionally, the nutrient medium may also contain growth promoting substances. Substances which promote growth include, for example, trace elemcnts, such as iron, zinc, manganese and 2 ~

the like, or individual amino acids.

Due to the incompat;bility between the endogenous two-micron DNA and hybrid vectors carrying its replicon, yeast cells transformed with such hybrid vectors tend to lose the latter. Such yeast cells have to be grown under selective conditions, i.e. conditions which require the expression of a plasmid-encoded gene for growth. Most selective markers cur-rently in use and present in the hybrid vectors according to the invention (infra) are genes coding for enzymes of amino acid or purine biosynthesis. This makes it necessary to use synthetic minimal media ieficient in the corresponding amino acid or purine base. How-ever, genes conferring resistance to an appropriate biocide may be used as well [e.g. a gene conferring resistance to the amino-glycoside G418]. Yeast cells transformed with vectors containing antibiotic resistance genes are grown in complex media containing the corresponding antibiotic whereby faster growth rates and higher cell densities are reached.

Hybrid vectors comprising the complete two-micron DNA (including a functional origin of replication) ~Lre stably maintained within strains of SaccharomYces cerevisi~c which are devoid of endogenous two-rnicron plasmids (so-called cir strains) so that the cultivation can be carried out under non-selective growth conditions, i.e. in a complex medium.

Yeast cells containing hybrid phlsmids with a constitutive promoter explcss thc DNA
encoding a glycosyltransferase, or a valiant thereof, controlled by said promoter without induction. I~owever, if said DNA is under the control of a regulated promoter the composition of the growth medium has to be adapted in order to obtain maximum levels of mRNA transcripts, e.g. when using the PH05 promoter the growth medium MUst contain a low concentration of inorganic phosphate for derepression of this promoter.

The cultivation is carried out by employing conventional techniques. The culturillg conditions, such as temperature, pH of the medium and fermentation time are selected in such a way that maximal levels of the heterologo~ls protein are produced. A chosen yeast strain is preferably grown under aerobic conditions in submerged culture with shaking or stirring at a temperature of about 25 to 35C, preferably at about 28C, at a pH value of from 4 to 7, for example at approximately pH 5, and t`or at least I to 3 days, preferably as long as satisfactory yielcls of protein are obtained.

After expression in yeast the glycosyltransferase, or its variant, is either accumulatcd inside the cells or secreted into the culture medium and is isolate(i hy conventional means.

For example, the first step usually consists in sèparating the cells from the culture fluid by centrifugation. In case the glycosyltransferase, or its variant, has accumulated within the cells, the protein has to be liberated from the cell interior by cell disruption. Yeast cells can be disrupted in various ways well-known in the art: e.g. by exerting mechanical forces such as shaking with glass beads, by ultrasonic vibration, osmotic shock and/or by enzymatic digestion of the cell wall. In case the glycosyltransferase, or its variant, to be isolated is associated with or bound to a membranous fraction, further enrichment may be achieved for example by differential centrifugation of the cell extract and, optionally, subsequent treatment of the particular fraction with a detergent, such as Triton. Methods suitable for the purification of the crude glycosyltransferase, or the variant thereof include standard chromatographic procedures such as afl`inity chromatography, for example with a suitable substrate, antibodies or Concanavalin A, ion exchange chromatography, gel filtra-tion, partition chromatography, HPLC, electrophoresis, precipitation steps sllch as ammonium sulfate precipitation and other processes, espccially those known from the literature.

In case the glycosyltransferase, or its variant, is secreted by the yeast cell into the periplasmic space, a simplified isolation protocol can be used: the protein is recovered without cell Iysis by enzymatic removal of the cell wall or by chemical agents, e.g. thiol reagents or EDTA, whicll gives rise to cell wall damages permitting the producedglycosyltransferase to be released. In case the glycosyltransferase, or its variant, is secrected into the culture broth, it can be recovered directly therefrom and be purified using the methods specified above.

In order to detect glycosyltransferase activity assays known from the literature can be used. For example, galactosyltransferase activity can be measured by determing the amount of radioacti~ely labelled galactose incorporated into a suitable acceptor molecule such as a glycoprotein or a free sugar residue. Analogously, sialyltransferase activity may be assayed e.g. by the incorporation of sialic acid into suitable substrates, and fucosyltransferase activity can be assayed by the transfer of fucose to a suitable acceptor.

The transformed yeast host cells according to the invention can be prepared by recombinant DNA techniques comprising the steps of:
- preparing a hybrid vector coMprising a yeast promoter and a DNA seqllence coding for a membrane-bound glycosyltransferase, or a variant thereof, which DNA scquence is 2 ~
g controlled by said promoter, - transforming a yeast host strain with said hybrid vector, - and selecting transformed yeast cells from untransformed yeast cells.

Expression vectors The yeast hybrid vector according to the invention comprises an expression cassette comprising a yeast promoter and a DNA sequence coding for a mernbrane-bound glycosyltransferase, or a variant thereof, which DNA sequence is controlled by said promoter.

In a first embodiment, the yeast hybrid vector according to the invention comprises an expression cassette comprising a yeast promoter, a DNA sequence coding for a membrane-bound glycosyltransferase, or a variant thereof, which DNA sequence is controlled by said promoter, and a DNA sequence containing yeast transcription termination signals.

In a second embodiment, the yeast hybrid vector accordillg to the invention comprises an expression cassette comprising a yeast promoter operably linked to a first DNA sequence encoding a signal peptide linked in the proper reading, frame to a second DNA sequence encoding a membrane-boulld glycosyltransferase, or a variant Iherof, and a DNA sequence containing yeast transcription termination signals.

The yeast promoter is a regulated or a constitutive promoter preferably derived from a highly expressed yeast gene, especially a SaccharomYces cerevisiae gene. Thus, the promoter of the TRP1 gene, the ADHI or Al~-III gene, the acid phosphatase (~) gene, a promoter of the yeast mating pheromone genes coding for the _- or c~-factor or a promoter derived from a gene encoding a glycolytic en~yme such as the promoter of the enolase, glyceraldehyde-3-phosphate dehydrogenase (GAP), 3-phosphoglycerate kinase (PGK), hexokinase, pyruvate decarboxylase, phosphofructokinase, glucose-6-phosphate isomerase, 3-phosphoglycerate mutase, pyruvate kinase, tliosephosphate isomerase, phosphoglucose isomerase or glucokinase genes can be used. Furthermore, it is possible to use hybrid promoters comprising upstream activation sequences (UAS) of one yeast gene and downstream promoter elements including a l`unctional TATA box of another yeast gene, for example a hybrid proMoter including the UAS(s) of the yeast PH05 gene and downstream promoter elements including a functional TATA box of the yeast GAP gene (PHO~ - GAP hyblid promoter). A preferred promoter is the promoter of the GAP gene, - 2 ~ 7 especially functional fragments thereof starting at nucleotides between positions -550 and -180, in particular at nucleolide -5~0, -263 or -198, and ending at nucleotide -5 of the GAP
gene. Anather preferred ?romoter of the regulated type is the PH05 promoter. As a constitutive promoter a shortened acid phosphatase PH05 promoter devoid of the upstream regulatory elements (UAS) is preferred as is the PH05 (-173) promoter element starting at nucleotide -173 and ending at nucleotide -9 of the PH05 gene.

The DNA sequence encoding a signal peptide ("signal sequence") is preferably derived from a yeast gene coding for a polypeptide which is ordinarily secreted. Other signal sequences of heterologous proteins, which are ordinarily secreted can also be chosen.
~east signal sequences are, for example, the signal and prepro sequences of the yeast invertase, oc-factor, pheromone peptidase (KEXl), "killer toxin" and repressible acid phosphatase (PH05) genes and the glucoamylase signal sequence from Aspergillus awa-mori. Alternatively, fllsed signal sequences may be constructed by ligating part of the signal sequence (if present) of the gene naturally linked to the promoter used (for exarnple PH05), with part of the signal sequence of another heterologolls protein. Those combina-tions are favoured which allow a precise cleavage between the signal sequence and the glycosyltransferase amino acid sequence. Additional sequences, such as pro- or spacer-sequences which may or may not carry specific processin~ signals cal1 also be included in the constructions to tacilitate accurate processing of precllrsor molecules. Alternatively, fused proteins can be generatcd containing internal processing signals which allow proper maturation in vivo or in vitro. For example, the processing signals contain Lys-Arg, which is recognized by a yeast endopeptidase located in the Golgi membranes. The preferred signal sequence according to the present invention is that of the yeast invertase gene.

If a t`ull-length glycosyltransferase, or a rnembrane-bound variant thereof, is expressed in yeast, the preferred yeast hybrid vector comprises an expression cassette comprising a yeast promoter, a DNA sequence encoding said glycosyltransferase or variant, which DNA sequence is controlled by said promoter, and a DNA sequence containing yeasttranscription terminatiorl signals. If the DNA encodes a membrane-bound enzyme there is no need for an additional signal sequence.

In case a soluble variant of a membrane-bound glycosyltransferase is expressed in yeast, the preferred yeast hybrid vector comprises an expression cassette comprising a yeast promoter operably linked to a first DNA sequence encoding a signal peptide linked in the proper reading frame to a second DNA sequence encoding said varial1t and a DNA

5 ~

sequence containing yeast transcription termination signals.

DNA encoding a membrane-bound glycosyltransferase, or a variant thereof, can be prepared by methods known in the art and comprises genomic DNA, e.g. isolated from a mammalian genomic DNA library, e.g. from rat, murine, bovine or human cells. If necessary, the introns occurring in genomic DNA encoding the enzyme are deleted.Furthermore, DNA encoding a membrane-bound glycosyltransferase, or a variant thereof, comprises cDNA which can be isolated from a mammalian cDNA library or produced from the corresponding mRNA. The cDNA library may be derived from cells from different tissues, e.g. placenta cells or liver cells. The preparation of cDNA via the mRNA
route is achieved using conventional methods such as the polymerase chain reaction (PCR).

For example, isolation of poly(A)+RNA from mammalian cells, e.g. HeLa cells, and sub-sequent first strand cDNA synthesis are performed following standard procedures known in the art. Starting from this synthesized DNA template, PCI~ can bc used to amplity the targeted sequence, i.e. the glycosyltransferase DNA or a fragMent thereof, while the amplification of the numerically overwhelming nontarget sequenccs is minimized. For this purpose, the sequence of a small stretch of nucleotides on each side of thc target sequence must be known. These flanking sequences are used to design two synthetic single-s~randed primer oligonucleotides the sequence of which is chosen so that each has basepair comple mentarity with its respective tlanking sequence. PCR starts by denaturing of the mRNA-DNA hybrid strand, followed by annealing the primers to the sequences flanking the target. Addition of a DNA polymerase and desoxynucleoside triphosphates causes two pieces of double-stranded DNA to form, each beginning at the primer and extending across the target seq-lence, thereby copying the latter. ~ach of the newly synthesized products can serve as templates for primer annealing and extensions (next cycle) thus leading to an exponential increase in double-stranded fragments of discrete length.

Furthermore, DNA encoding a membrane-bound glycosyltransferase, or a variant thereof, can be enzymatically or chemically synthesized. A variant of a membrane-bound glycosyltransferase having enzymatic activity and an amino acid sequence in which one or more amino acids are deleted (DNA fragments) and/or exchanged with one or more other amino acids, is encoded by a mutant DNA. Furthermore, a mutant DNA is intended to include a silent mutant wherein one or more nucleotides are replaced with other nucleotides with the ncw codons coding for the saMe amino acid(s). Such a mutant 5 ~

sequence is also a degenerated DNA sequence. Degenerated DNA sequences are degenerated within the meaning of the genetic code in that an unlimited number of nucleotides are replaced by other nucleotides without resulting in a change of the amino acid sequence originally encoded. Such degenerated DNA sequences may be useful due to their different restriction sites and/or frequency of particular codons which are preferred by the specific host to obtain optimal expression of a glycosyltransferase or a variant thereof. Preferably, such DNA sequences have the yeast preferred cods~n usage.

A mutant DNA can also be obtained by in vitro mutation of a naturally occurrtng genomic DNA or a cDNA according to methods Icnown in the art. For example, the partial DNA
coding for a soluble form of a glycosyltransferase may be excised from the full-length DNA coding for the corresponding membrane-bound glycosyrtransferase by using restriction enzymes. The availability of an appropriate restriction site is advantageous lherefor.

A ~NA sequence containing yeast transcriptiol1 termination signals is preferably the 3' flanking sequence of a yeast gene which contains proper signals for transcription termination and polyadenylation. Suilable 3' llanking sequences are for example those of the yeast gene naturally linked to the promoter used. The prel`erred flanking sequence is thllt of the yeast PHOS gene.

The yeast promoter, the optional DNA sequence coding for the signal peptide, the DNA
sequence coding for a membrane-bound giycosyltransfeMse, or a variant thereof, artd the DNA sequence containing yeast transcription termination signals are operably linked in a tandem array, i.e. they are ju~ctaposed in such a manner that their normal functions are maintained. The array is such that the promoter effects proper expression of the DNA
sequence encoding a meMbrane-bound glycosyltransferase, or a variant thereof, (optionally preceded by a signal sequence), the transcription termination signals effect proper termination of transcription and polyadenylation and the optional signal sequence is linked in the proper reading frame to the above-mentioned DNA sequence in such a manner that the last codon of the signal sequence is directly linked to the first codon of said DNA sequence and secretion of the protein occurs. If the promoter and the signal sequence are derived from different genes, the promoter ;s preferably joined to the signal sequence at a site between the major mRNA start and the ATG of the gene naturally linked to the promoter. The signal sequence should have its own ATG for translation initiation. The junction of lhese sequences may be effected by means of synthetic 2 ~ 7 oligodeoxynucleotide linkers carrying the recognition sequence of an endonuclease.

Vectors suitable for replication and expression in yeast contain a yeast replication origin.
~Iybrid vectors that contain a yeast replication origin, for example the chromosomal autonomously replicating segment (ars), are retained extrachromosomally within the yeast cell after transformation and are replicated autonomously during mitosis. Also, hybrid vectors that contain sequences homologous to the yeast 211 plasmid DNA can be used.
Such hybrid vectors are integrated by recombination in 2.u plasmids already present within the cell, or replicate autonomously.

Preferably, the hybrid vectors according to the invention include one or more, especially one or two, selective genetic markers for yeast and swch a marker and an origin of replication for a bacterial host, especially Escherichia coli.

As to the selective gene markels for yeast, any marker gene can be used which tacilitates the selection for transformants due to the phenotypic expression of the marker gene.
Suitable markers for yeast are, for example, those expressing antibiotic resistance or, in the case of auxotrophic yeast mutants, genes which complement host lesions.
Corresponding genes confer, t`or example, resistance to the antibiotics G418, hygromycin or bleomyein or provide for prototrophy in an auxoLropl1ic yeast mutant, for exasnple the URA3, LEU2, LYS2 or TRPI gene.

As the amplification of the hybrid vectors is conveniently done in E. coli, an E. coli genetic marker and an E. coli replication origin are included advantageously. These can be obtained from E. coli plasmids, such as pBR322 or a pUC plasmid, for example pUC18 or pUC19, which contain both E. coli replication origin and E. coli genetic marker conferring resistance to antibiotics, such as ampicillin.

The hybrid vectors according to the invention are prepared by methods known in the art, for example by linking the expression cassette comprising a yeast promoter and a DNA
sequence coding for a glycosyltransferase, or a variant thereof, which DNA sequence is controlled by said promoter, or the several constituents of the expression cassette, and the DNA lragments containing selective genetic markers for yeast and for a bacterial host and origins of replication for yeast and for a bacterial host in the predetermined order.

The hybrid vectors of the invention are used for the transformation of the yeast strains 6~

described below.

Yeast strains and transformation thereof Suitable yeast host organisms are strains of the genus Saccharomyces, especially strains of Saccharomyces cerevisiae. Said yeast strains include strains which, optionally, have been cured of endogenous two-micron plasmids and/or which optionally lack yeast peptidase activity(ies), e.g. peptidase yscc~, yscA,yscB,yscY and/or yscS activity.

The invention concerns furthermore a yeast strain which has been transformed with a hybrid vector comprising an expression cassette comprising a yeast promoter and a DNA
sequence coding for a membrane-bound glycosyltransferase, or a variant therof, which DNAis controlled by said promoter.

In a first embodiment, the yeast strain according to the invention has been transformed with a hybrid vector comprising an expression cassette comprising a yeast promoter, a DNA sequence coding for a membrane-bound glycosyltransferase, or a variant thereof.
which DNA sequence is controlled by said promoter, and a DNAsequence containing yeast transcliption termination signals.

In a second embodiment, the yeast strain according to the invention has been lransformed with a hybrid vector comprising an expression cassette consisting of a yeast promoter operably linked to a t`irst DNA sequence encoding a signal peptide linked in the proper reading frame to a second DNA sequence encoding a membrane-bound glycosyltransferase, or a variant thereof, and a DNA sequence containing yeast transcription termination signals.

The yeast strains of the invention are used for the preparation of a membrane-bound gly~osyltransferase or ~ variant thereof.

The transformation of yeast with the hybrid vectors according to the invention is accomplished by methods known in the art, for example according to the methods described by Hinnen et al. (Proc. Natl. Acad. Sci. USA (1978) 75, 1929) and Ito et al.
(J. Bact. (1983) 153, 163-168).

The membrane-bound glycosyltransferases, and the variants thereof, prepared by the process according to the invention can be used in a manner known per se, e.g. for the 2 ~

syn~hesis and/or modification of glycoproteins, oligosaccharides cand glycolipids (US Patent 4,925,796; EP Application 414 171).

The invention concerns especially the method for the production of membrane-bound glycosyltransferases, and variants therof, the hybrid vectors, the transformed yeast strains, and the glycosyltransfer~ses obtainable according to the inventive process, as described in the Examples.

In the Examples, the following abbreviations are used: GT " galactosyltransferase (EC 2.4.1.22), PCR '` polymerase chain reaction; ST A sialyltransferase (EC 2.4.99.1);
FT= fucosyltransferase.

E~xample 1: Cloning of the galactosyltransferase (CT) cONA from HeLa cells GT cDNA is isolated from HeLa cells (Wat~ele, G. and Berger, ~.G. (1990) Nucleic Acids Res. 18, 7174) by the polymerase chain reaction (PCR) method:

1.1 Preparation of poly(A)+RNA from HeLa cells For RNA preparation HeLa cells are grown in monolayer culture on 5 plates (23x23 cm).
The rapid and efficient isolation of RNA from cultured cells is performcd by extraction with guanidine-HCl as described by Mac Donald, R.J. et al (Meth. Enzymol. (1987) 152, 226-227). Generally, yields are about 0.6 - 1 mg total RNA per plate of confluent cells.
Enrichment of poly(A)+RNA is achieved by affinity chromatography on oligo(dT)-cellu- ~
lose according to the method described in the Maniatis manual (Sambrook, J., Fritsch, E.F.
and Maniatis, T. (1989) Molecular Cloning: A Laboratory Manual (2nd edition), Cold Spring Harbor Laboratory Press, Cold Spring Habor, USA), applying 4 mg of total RNA
on a 400 ,ul column. 3 % of the loaded RNA are recovered as enriched poly(A)+RNAwhich is stored in aliquots precipitated with 3 volumes of ethanol at -70C until it is used.

1.2 First strand cDNA synthesis for PCR
Poly(A)+RNA (mRNA) is reverse-transcribed into DNA by Moloney Murine Leukemia Virus RNase H- Reverse Transcriptase (M-MLV H- RT) (BRL). In setting up the 20 ,ul reaction mix, the protocol provided by BRL is followed with minor variations: l llg of HeLa cell poly(A)+RNA and 500 ng Oligo(dT)I2 l8 (Pharmacia) in 11.5 Ill sterile H2O are heated to 70C for lO min and then quickly chilled on ice. Then 4 ~I reaction buffer provided by BR~ (250 mM Tris-HCI pH 8.3~ 3?5 mM KCI, 15 mM MgC12), 2 1ll 0.1 M
dithiothreitol, 1 1ll mixed dNTP (lO mM each dATP, dCTP, dGTP, TTP, Pharmacia), 0.5 1ll (17.5 U) RNAguard (RNase Inhibitor of Pharmacia) and 1 111(200 U)M-MLVH- RT
are added. The reaction is carried out at 42C and stopped after l h by heating the tube to 95C for lO min.

In order to check the efficiency of the reaction an aliquot of the mixture (5 1ll) is incubated in thc presence of 2 I,lCi o~ 32p dCTP. By measuring the incorporated dCTP, the amount of cDNA synthesized is calculated. The yield of first strand synthesis is routinely between 5 and 15%.

1.3 Polymerase chain reaction The oligodeoxynucleoti(le primers used for PCR are synthesi~ed m vitro by the phosphor-amidite method (M.H. Carutllers, in Chemical and Enzymatic Synthesis of Gene Frag-ments, H.G. Gassen and A. Lang, eds., Verlag Chemie, Weinheim, FRG) on an Applied Biosystems Model 380B synthesizer. They are listed in Table 1.

Table 1: PCR-primers corresponding to primer sequence (5' to 3')1) bp in GT cDNA

P1up (KpnI) cgc~tACCCTTCTTAAAGCGGCGGCGGGAAGATG (-26)- 3 P1 (EcoRI) gcc_aattcATGAGGCTTCGGGAGCCGCTCCTGAGCG 1- 2~
P3 (SacI) CTGGAGCTCGTGGCAAAGCAGAACCC 448- 473 P2d (EcoRI) gcc~aaTTCAGTCTC~rATCCGTGTACCAAAACGC ~TA 1222 -1 192 P4 (HindIII) cccaa~ctTGGAATGATGATGGCCACCTTGTGAGG 546- 520 I) Capi~al Ictters represent sequences from GT. small letters are additional sequences, sites for restriction enzymes are underlined. Codons for 'start' and 'stop' of RNA translation are highligllted in boldface.
GT cDNA sequence from human placenta as published in GenBank (Accession Nr. M2292 1).

Standard PGR-conditions for a 30 Ill incubation mixture are 1 1ll of the Reverse Trans-criptase reaction (see Example 1.2), containing about 5 ng first strand cDNA, 15 pmol each of the relevant primers, 200 ~lmol each of the four deoxynucleoside triphosphates (dATP, dCTP, dGTP and TTP) in PCR-bufler (lO mM Tris-HCI pH ~.3 (at 23C), 50 mM

,~ 2 ~3 rS~ P~

KCI, 1.5 mM MgC12, 0.001 % gelatine) and 0.5 U AmpliTaq Polymerase (Perkin Elmer).
The amplification is performed in the Therrnocycler 60 (Biomed) using the following conditions: 0.5 min denaturing at 95C, 1 min annealing at 56C, and l min 15 sec exten-sion at 72C, for a total of 20 - 25 cycles. In the last cycle, primer extension at 72C is carried out for 5 min.

For sequencing and subcloning, the HeLa GT cDNA is amplified in two overlapping pieces, using different primer combinations:

(1) Fragment Pl -P4: Primers P1 and P4 are used to amplify a 0.55 kb DNA fragment covering nucleotide positions 7-556 in HeLa GT cDNA (SEQ ID N{). 1 ) (2) Fragment P3 - P2d: Primers P3 and P2d are used to amplify a 0.77 kb fragmentcovering nucleotide positions 457 - 1232 (SEQ ID NO. 1).

In order to avoid errors during amplification four independent PCRs are carried out for each fragment. Also primer Plup (KpnI) in combination with primer P4 is used to determine the DNA sequence followed by the 'start' codon.

After PCR amplification, tragment P1 - P4 is digested with the restriction enzymes EcoRI
and HindIII, analysed on a 1.2 % agarose gel, elutcd from the gel by GENECLEAN
(BIO lOl) and subcloned into the vector pUCl8 (Pharmacia), digested with the same enzymes. Fragment P3 - P2d is digested with SacI and EcoRI, analysed on a 1.2 % gel, eluted and subcloned inlo pUC18, digested with SacI and EcoRI. The resulting subclones are pUC18/Pl - P4 and pUC18/P3 - P2d, respectively. For subcloning, ligation andtransformation of E. coli strain DH5c~, standard protocols are followed as described in Example 2. Minipreparations of Plasmids pUC18/Pl - P4 and pUC18/P3 - P2d are used for dideoxy-sequencing of denatured double-stranded DNA with the T7 polymerase Sequencing kit (Pharmacia). M13/pUC sequencing primers and reverse sequencing primers (Pharmacia) are applied to sequence overlapping fragments produced rrom both DNA strands by digestion with various restriction enzymes. Further subcloning ofrestriction fragments of the GT gene is necessary for extensive sequencing of overlapping fragments of both strands. The sequence of fragments amplified by independent PCRs shows that the error of amplification is less than 1 in 3000 nucleotides. The complete nucleotide sequence of the HeLa cell GT cDNA which is presented in SEQ ID NO. 1 is 99.2 % homologous to that of human placenta (Genbank Accession No. M22921). Three differences are found:

(a) Three extra base pairs at nucleotide positions 37-39 (SEQ ID NO. 1) resul~ing in one extra amino acid (Ser) in the N-terminal region of lhe protein; (b) bp ~8 to lO1 are 'CTCT' instead of 'TCTG' in the sequence of human placenta, leading to two conservative amino acid substitutiolls (Ala Leu instead of ValTyr) at amino acidpositions 31 and 32 in the membrane spanning domain of GT; (c) the nucleotide atposition 1047 is changed from 'A' to 'G' without ensuing a change in amino acid sequence.

The two overlapping DNA-fragments P1 - P4 and P3 - P2d encoding the HeLa GT cDNAare joined via the NotI restriction site at nucleotide position 498 which is present in both fragments.

The complete HeLa cell GT cDNA (SEQ ID No. I) is cloned as a 1.2 kb EcoRI-EcoRI
restriction fragment in plasmid pIC-7, a derivative of pUC8 with additional restriction sites in the multicloning site (Marsh, J.L., Erfle, M. and Wykes, E.J. (1984) Gene 32, 481-485), resulting in vector p4AD113. For the pulpose of creating the GT expression cassette the EcoRI restriction site (bp 1227) at the 3' end of the cDNA sequence is deleted as follows: vector p4AD113 is first lineari~ed by digestion with EcoRV alld then treated with alkaline phosphatase. Furtherrllore, 1 ~Ig of the lme.lrised plasmi(l DNA is partially digested with 0.25 U EcoRI for 1 h at 37C. After agarose gel electrophoresis a fragment corresponding to the size of the lineari-~ed plasmid (3.~5 kb) is isolated from the gel by GENECLEAN (Bio 101). The protruding EcoRI end is filled in with Klenow polymerase as described in the Maniatis manual (supra). After phenolisation and ethanol precipitation the plasmid is religated and used to transform E. coli DH5cc (Gibco/BRL). Miniprepara-tion of plasmids are prepared from six transformants. The plasmids obtained are checked by restriction analysis for the absence of the EcoRI and EcoRV restriction sites at the 3' end of HeLa GT cDNA. The plasmid designated p4AE113 is chosen for the following experiments, its DNA sequence being identical to that of plasmid p4AD113, with the exception that bp 1232-1238 with the EcoRI-EcoRV restriction sites are deleted.

Example 2: Construction of expression cassettes for full length GT
For heterologous expression in Saccharomyces cerevisiae the full length HeLa GT cDNA
sequence (SEQ ID NO. I) is fused to transcriptional control signals of yeast for efficient initiation and termination of transcription. The promoter and terminator sequences originate from the yeast acid phosphatase gene (~) (EP 100561). The t`ull-length PH05 promoter is regulated by the supply of inorganic phosphate in the culture medium. High Pi 2 ~

concenlrations lead to promoter repression whereas low P' acts by induction.
Alternatively, a short, 173 bp PH05 promoter fragment is used, which is devoid of all regulatory elements and therefore behaves as a constitutive promoter.

2.1 Conshuction of a phosphate inducible expression cassette The GT cDNA sequence is combined with the yeast PH05 promotcr and transcription terminator sequences as follows:

(a) Full leng~h HeLa GT cDNA sequence:
Vector p4AE113 with the full length ~;T cDNA sequence is digested with the restriction enzymes EcoRI and BglII. The DNA fragments are elechrophoretically separaled on a 1 %
agarose gel. A 1.2 kb DNA fragment containing the complete cDNA sequence for HeLa GT is isolated from the gel by adsorption to glasmilk, using the GENECLEAN kit (BIO 101). On this fragment the 'ATG' start codon for protein synthesis of GT is located directly behind the restriction site for EcoRI, whereas the stop codon 'TAG' is followed by 32 bp contributed by the 3~unhranslated region of HeLa GT and the multiple cloning site of the vector with the BglII res~riction site.

(b) Vector for amplification in E. coli:
The vector for amplirication, plasmid p31R (cf. EP 100~561), a derivative of pBR322, is digested with the reshriction enzymes BamHI and Sall. 'I'he restriction lragments are separated on a 1 % agarose gel and a 3.5 kb vector fragment is isolated from the gel as described before. This DNA fragment contains the large SalI - HindI~I vector fragment of the pBR322 derivative as well as a 337 bp PH05 transcription terrninator sequence in place of the HindIII - BamHI sequence of pBR322.

(c) Sequence for the inducible PH05 promoter:
The PH05 promoter fragment containing the regulatory elements (UASp) for phosphate induction is isolated from plasmid p31R (cf. EP 10`0561) by digestion with the restriction enzymes SalI and EcoRI. The 0.8 kb SalI - EcoRI DNA fragment comprises the 276bpSalI - BamHI pBR322 sequence and the 534 bp BamHI-EcoRI PH05 promoter l`ragment with the EcoRI linker (5'-GAATTC-3') inhoduced at position -8 of the PH05 promoter sequence.

(d) Construction of plasmid pGTA 1132 The three DNA fragments (a) to (c) are ligated in a 121111igation mixture: 100 ng of DNA

2 ~ 7 fragment (a) and 30 ng each of fragments (b) and (c) are ligated using 0.3 U T4 DNA ligase (Boehringer) in the supplied ligase buffer (66 mM Tris-HCI pH 7.5, 1 mM
dithioerythritol, 5 mM MgCl2, 1 mM ATP) at 15C for 18 hours.
Half of the ligation mix is used to transform competent cells of E. coli strain DH5c~
(Gibco/BRL). For preparing competent cells and for transformation, the standard protocol as given in the Maniatis manual (supra) is followed. The cells are plated on selective LB-medium, supplemented with 75 ,Lg/ml ampicillin and incubated at 37C. About 120 transformants are obtained. Minipreparations of plasmid are pert`orrned from six inde-pendent transformants by using the modified alkaline Iysis protocol of Birnboim, H.C. and Doly, J. as described in the Maniatis manual (supra). The isolated plasmids are characterized by restriction analysis with four dil`ferent enzymes (EcoRI, PstI, HindIII, SalI, also in combination). All six plasmids show the expected restriction fragments. ~ne of the clones is chosen and referred to as pGTA 1132. Plasmid pGTA 1132 contains the expression cassette with the full-length HeLaGT cDNA under the control of the phosphate regulated PH05 promoter, and the PH05 transcriptional lerminator sequence. This expression cassette can be excised from pGTA 1132 as a 2.35 kb SalI - HindlII fragment, refelTed to as DNA fragment (IA).

2.2. Construction of a con~stitlltive expression cassette:
For the construction of an exprcssion c~sse~te with a consti~lltive, nonreglllated promoter, a 5' truncated PH05 promoter fragment witho-lt phosphate regulatory elements is used, which is isolated from plasmid p31/PH05(-173)RIT.

(a) Construction of plasmid p31/PH05(-173)RIT
Plasmid p31 RIT12 (EP 288435) comprises lhe full leng~th, regulated PH05 promoter (with an EcoRI site introduced at nucleotide position -8 on a 534bp BamHI - EcoRI fragment, followed by the coding sequence for lhe yeast invertase signal sequence (72bp EcoRI -~hol) and the PH05 transcription termination signal (135bp Xhol - HindIII) cloned in a tandem array between BamHI and HindIII of the pBR322 derived vector.

The constitutive PH05(-173) promoter element from plasmid pJDB207/PH05(-173)-YHIR
(EP 340170) comprises the nucleotide seqllence of the yeast PH05 promoter from nucleotide position -9 to -173 (BstEII restriction site), but has no upstream regulatory sequences (UASp). The PH05(-173) promoter, therefore, behaves like a constihltive promoter. This example describes the replacement of the regulated PH05 promoter in plasmid p31RIT12 by the short, constitutive PH05 (-173) promoter element in order to 2 ~ 7 obtain plasmid p31/PH05 (-173) RIT.

Plasmids p31RIT12 (EP 288435) and pJDB207/PH05(-173)-YHIR (EP 340170) are digested with restriction endonucleases SalI and EcoRI. The respective 3~6 kb and 0.4 kb SalI - EcoRI fragments are isolated on a 0.8 % agarose gel, eluted from the gel, ethanol precipitated and resuspended in H20 at a concentration of 0.1 pmoles/~,ll. Both DNA
fragments are ligated and 1 111 aliquots of the ligation mix are used to transform E coli HB101 (ATCC) competent cells. Ampicillin resistant colonies are grown individually iQ
LB medium supplemented with ampicillin (100 llg/ml). Plasmid DNA is isolated accord-ing to the method of Holmes, D.S. et al. (Anal. Biochem. (1981) 144, 193) and analysed by restriction digests with SalI and EcoRI. The plasmid of one clone with the correct restriction fragments is referred to as p31/PH05(-173)RIT.

(b) Constmction of plasmid pGTBl 135 Plasmid p31/PH05(-173)RIT is digested with the restriction enzyMes EcoRI and Sall.
After separation 011 a 1 % agarose gel, a 0.45 kb SalI - EcoRI lragment is isolated froM the gel by GENECLEAN (BIO 101). This fragment contains the 276 bp SalI-BamHI seqllence of pBR322 and the 173bp BamHI(BstEII)-EcoRI constitutive PH05 promoter l`ragment.
The 0.45 kb SalI-EcoRI fragment is ligated to the 1.2 kb EcoRI - Bglll GT cDNA
(fragment (a)) and the 3.5 kb BamHI-SalI vector part for ampliticalion in E. coli wilh the PH05 terminator (fragment (b)) described in Example 2.1. Ligation and transformation of E. coli strain DHSo~ are carried out as described above yielding 58 transformants. Plasmids are isolated from six independent colonies by minipreparations and characterize~l by restriction analysis. All six plasmids show the expected fragments. One correct clone is referred to as pGTB 1135 and used for further cloning experiments to provide theexpression cassette for HeLa GT under the control of the constitutive PH05 (- 173) promoter fragment. This expression cassette can be excised from the vector pGTB 1135 as a 2 kb SalI - HindIII fragment, referred to as DNA fragment (lB).

Example 3: Construction of the expression vectors pDPGTA8 and pDPGTB5 The yeast vector used for heterologous expression is the episomal vector pDP34 (11.8 kb) which is a yeast - E. coli shuttle vector with the ampicillin resistance marker for E. coli and the URA3 and dLEU2 yeast selective markers. Vector pDP34 (cf. EP 340170) is digested with the restriction enzyme BamHI. The linearized vector is isolated with &ENECLEAN and the protruding ends are filled in by Klenow polymerase treatment as described in the Maniatis mam~al (supra). The reaction is stopped after 30 min by heating to 65C for 20 min in the presence of 10 MM EDTA. After ethanol precipitation the plasmid is digested with Sall and subjected to gel electrophoresis on a 0.8 % agarose gel.
The (BamHI) blunl end-SalI cut vector pDP34 is isolated as an 11.8 kb DNA fragment from the gel with the GENEGLEAN kit.

In analogy to the Yector preparation plasmids pGTA 1132 and pGTB 1135 are each digested with HindIII. The protruding ends of the linealized plasmids are filled in by Klenow polymerase treatment and subsequently subjected to SalI digestion, resulting in (2A) a 2.35 kb (HindIII)blunt end - SalI fragment with the phosphate regulated expression cassette, or (2B) a 2.0 kb (HindIII)blllnt end - SalI fragment with the constitutive expression cassette.

Ligation of the blunt end-SalI pDP34 vector p~u t with fragment 2A or fragment 2B and transforMation of competen~ cells of E. coli strain DHSc~ is carried out as described in Example 2 using 80 ng of the vector part and 40 ng of fragment 2A or 2B, respectively. 58 and 24 transformants are obtained, respectively. From each transformation six plasmids are prepared and chauacterized by restriction analysis. For the constn~ction with the regulated expression cassette (fragment 2A), two plasmids show the e~pected restriction pattern. One of tlle clones is chosen and designate(l pDE'GTA8.
For the construction with the constitutive expression cassette (fragment 2B) one plasmid shows the expected restriction pattern, and is designated pDPGTB5.

Example 4: Transformation of S. cerevisiae strain BT 150 CsCI-purified DNA of the expression vectors pDPGTA8 and pDPGTB 5 is prepared following the protocol of R. Treisman in the Maniatis manual ~supra). The protease deficient S. cerevisiae strains BT 150 (MATo~, his4, leu2, ura3, pral, prbl, prcl, cpsl) and H 449 (MATa, prbl, cpsl, ura3~5, leu 2-3, 2-112, cir) are each transformed with S llg each of plasmids pDPGTA8, pDPGTB5 and pDP34 (control without expression cassette) according to the lithium-acetate transformation method (Ito, H. et al., supra).
Ura+-transformants are isolated and screened for GT activity (infra). Single transformed yeast cells are selected and referred to as SaccharomYces cerevisiae BT 150/pDPGTA8 Saccharomyces cerevisiae BT 150/pDPGTB5 Sa aromyces cerevisiae BT 150/pDP34 2 ~

~accharomyces cerevisiae H 449/pDPGTA8 Saccharomyces cerevisiae H 449/pDPGTB5 SaccharomYces cerevisiae H 449/pDP34 Example 5: Enzyme activitv of full-lenoth GT expressed in veast 5.1 Preparation of cell extracts Cells of transformed Saccharomyces cerevisiae strains BT 150 and H 449 are each grown under uracil selection in yeast minimal media (Difco) supplemented with histidine and leucine. The growth rate of the cells is not affected by the introduction of any of the expression vectors. Exponentially growing cells (at OD578 of 0.5) or stationary cells are collected by centrifugation, washed once with 50 mM Tris-HCI buffer pH 7.4 (buffer 1) and resuspended in buffcr 1 at a concentration corresponding to 0.1 - 0.2 OD578.Mechanical breakage of thc cells is efl`ec~ed by vigorous shaking on a vortcx mi~er with glass beads (0.45 - 0.5 mm diameter) for 4 min with intermittent cooling. Tlle crude extracts are used directly for determination of en~yme activity.

Fractionation of cellular components is achieved by differential centrifugation of the extract. First, the extract is centrifuged at 450 g for 5 mi~; secondly, the supernatant obtained is centrifuged at 20000 g for 45 min.

The supernatant is collected and the pellets are resuspended in buffer 1.

For phosphate induction of GT e~cpression under the control of the inducible PH05 promoter, cells of transfonnants BT 150/pDPGTA8 and H 449/pDPGTA8 are mass-shifted to low phosphate minimal medium, respectively, (Meyhack, B. et al. Embo J.
(1982) 1, 675-680). The cell extracts are prepared as described above.

5.2 Protein assay The protein concentration is determined by use oi` the BCA-Protein Assay Kit (Pierce).

5.3 Assay for GT activity GT activity can be measured with radiochemical methods using eiLher ovalbumin, aglycoprotein which solely exposes GlcNAc as acceptor site, or free GlcNAc as acceptor substrates. Cell extracts (of 1 - 2 ODs 578 Of cells) are assayed for 45 or 60 min at 37C in ~7~ 7 a lOO ,ul incubation mix~ure containing 100 ml~ Tris-HCI pH 7.4, 50 nCi IJDP-l4C-Gal (325 mCi/mmol), 80 nmol UDP-Gal, 1 llmol MnCI2,1 % Triton X-100 and 1 mg ovalbu-min or 211mol GlcNAc as acceptor. In case a glycoprotein acceptor substrate is used, the reaction is terminated by acid precipition of the protein and the amount of 14C galactose incorporated into ovalbumin is determined by liquid scintillation counting (Berger, E.G. et al. ~1978) Eur. J. Biochem. 90, 213-222). For GlcNAc as acceptor substrate, the reaction is terminated by the addition of 0.4 ml ice cold H2O and the unused UDP-l4C-galactose is separated from 14C products on an anion exchanged column (AG X1-8, BioRad) as des-cribed (Masibay, A.S. and Qasaba, P.K. (1989) Proc. Natl. Acad. Sci. USA 86, 5733-5737). Assays are performed with and without acceptor molecules to assess the extent of hydrolysis of UDP-Gal by nucleotide pyrophosphatases. The results are shown in Table 2.

Table 2: GT activity in S. cerevisiae strain BT 150 transformed with different plasmids.

GTspecit;cactivity (mU/mgprotein) plasmid PH05 promoterhigh Pi low Pi pDP34 -- <0.01 n.d. l) pDPGTA8 phosphate regulated 0.1 0.6-1 pDPGTB5 constitutive 0.6 n.d. l) 1) not deterrnined GT activity of cultures shifted to inducible conditions (lowPi minimal medium) is about the same as the activity of cultures in minimal media expressing GT constitutively (Table 2). As expected, no enzyme activity is found in cells trans~ormed with the vector only.

During fractionation, most of the GT activity (70 - 90 %, see Table 3) and highest specific activity is always found in the high speed pellet (20000 g). Enzyme activity can be increased by the addition of 1 - 2 % Triton X-100 to the enzyme assay. Both findings suggest that recombinant GT is membrane bound in yeast cells, as it is in HeLa cells.

2 ~ 7 Table 3: Distribution of GT activity during fractionation of BT 150/pDPGT5 cells GT activity cpm UDP-I4C-Gal incorporated Fraction in GlcNAc %
crude extract 19400 400 g pellet 1000 4.6 %
20000 g pellet 15800 72.5 %
20000 g supernatant S000 22.~ ~O
total:21800 100 %

xample 6: Purification of the recombinant GT
In order to liberate the Membrane-bound enzyme the 20000 g pellet fraction of BT150/pDPGTB5 cells is treated with 1 % (w/v) Triton X-100 for lO min at room tempera-ture and equilibrated with 50 mM Tris HCI pH 7.4, 25 mM MgCI2, O.S MM UMP and I %
(w/v) Triton X-100. The supernatant obtained at`ter centrifugation is 0.2 llm t`il~ered and passed over a GlcNAc-p-aminophenyl-Sepharose-column (Berger, E.G. et al. (1976) Experientia 32, 690-691) with a bed volume of 5 ml at a flow rate of 0.2 ml/min at 4C.
The enzyme is eluted with S0 mM Tris-HCI pH 7.4, 5 mM GlcNAc, 25 mM EDTA and 1 % Triton X-100. A single peak of en~yme activity is eluted within five fractions (size:
1 ml). These fractions are pooled and dialyzed against 2x 11 of 50 mM Tris-HCI pH 7.4, 0.1 % Triton X- 100. The purified GT is enzymatically active.

Example 7: Construction of an expression cassette for soluble GT
Galactosyltransferase expressed in yeast can be secreted into the culture medium if a yeast promoter is operably linked to a first DNA sequence encoding the signal sequence of the yeast invertase gen SUC2 linked in the proper reading frame to a cDNA encoding soluble GT;

(a) Partial HeLa GT cDNA sequence GT cDNA is excised from plasmid p4AD113 (Example 1) by digestion with EcoRI. The1.2 kb fragment containing the complete GT cDNA is isolated and partially digested with MvnI (Boehringer) cutting the GT sequence at positions 43, 55, 140 and 288 o~ the nucleotide sequence depicted in SEQ ID NO. 1. When digesting 0.2 llg of the EcoRI-EcoRI fragment with 0.75 11 MvnI for :I h the GT cDNA is cut only once at nucleotide position 134 yield;ng a l .1 kb MvnI-EcoRI fragment.

Q~

(b) Vector for amplification in E. eoli Plasmid pUCl8 (Pharmacia) is digested with BamHI and EeoRI in the multiple cloning site. Then the plasmid is treated with alkaline phosphatase as described in the Maniatis manual (supra), subjected to agarose gel electrophoresis and isolated from the gel as a 2.7 kb DNA fragment.

(c) Constitutive PH05 (-173) promoler and SUC2 signal sequence The veetor p31/PH05 (-173) RIT (Example 2) is digested with the restriction enzymes BamHI and XhoI. A 0.25 kb BamHI-XhoI fragment with the constitutive PH05 (-173) promoter and the adjacent coding sequence for the invertase signal sequence (inv ss) is isolated. Then the fragment is recut with HgaI (BioLabs). The HgaI recognition sequence is on the antisense DNA strand. The restriction enzyme euts upstream of the recognition sequence in sueh a way that the 5 staOgered end of the antisense strand eoincides with tlle end of the coding sequence o~ the inverlase signal sequence. The 0.24 kb BamHI and HgaI
cut DNA rragment contains the constitutive _H05 (- 173) promoter sequence linked to the yeast invertase signal sequence .

(d) Adaptor Fragment (a) is linked to fragmellt (c) by means of an adaptor sequence which is prepared from e~uimolar amounts of the synthetic oligonucleotides (Mierosynth) S'Cl' GCA CTG
GCT GGC CG 3' and S'CG GCC AGC CAG 3' for the complementary strand. The oligo-nucleotides are annealed to each other by first heating to 95C and then slowly cooling to 20C. The annealed adaptor is stored frozen.

(e) Construction of plasmid psGT
For ligation, linearized vector (b), the GTcDNA fragment (a), fragment (c) containing the promoter and the sequenee eneoding the si;,nal peptide and the adaptor (d) are used in a molar ratio of 1: 2: 2: 30-100. Ligation is calTied out in 12 111 of ligase bulfer (66 mM
Tris-HCl pH 7.5, 1 mM dithioerythritol, S mM MgCl2, 1 mM ATP) at 16C for 18 hours.
The ligation mix is used to transform competent cells ol` E. coii strain DHSc~ as described above. Minipreparations of plasmid are performed from 24 independent transformants.
The isolated plasmids are characterized by restriction analysis using four different enzymes (BamHI, PstI, EcoRI, XhoI, also in combination). A single clone with theexpected restriction pattern is referred to as psGT.

2 ~ ` 7 The correct sequence at the fusion site of the sequence encoding invertase signal peptide with the cDNA coding for soluble GT is contirmed for plasmid psGT by using the T7-Sequencing kit (Pharmacia) and primer 5'AGTCGAGGTTAGTATGGG 3' starLing at position -77 in the constitutive PH05 (-173) promoter.

Sequence for MvnI
DNAI): 5' AAA ATA Tct gca ctg gct ggc cgC GAC CTG AGC 3' 3' TTT TAT AGA CGT gac cga ccg gcG CAG GAC TCG 5' Protein: Lys Ile Ser Ala¦Leu Ala Gly Arg Asp Leu Ser inv ss ~ GT sequence cleavage site for signal endopeptidase 1) Small le~tcr~ represent Ibc ~ ptor seq~lcncc .

The expression cassette for secreted GT containing the constitutive P~05 (-173) promoter, the DNA sequcllcc encoding invertase signal peptide and thc partial GT cDNA l`rom plasmid psGT can be excised as a 1.35 kb Sall (BamHI~ - EcoRI lragmcnt. The expression cassette is still lacking the PH05 terminator sequences to be added in the ~ollowing cloning step.

Example 8: ConsLruction of the expression vector pDPGTS
For consLruction of the expression vector for soluble GT the following fragments are combined:

(a) Vector part Plasmid pDP34 is digested and pretreated as described in Example 3 resulting in a 11.8 kb blunt end - SalI vector fragment.

(b) Expression cassette Plasmid psGT is first linearized by digestion with SalI (in the multiple cloning site) and then partially digested with EcoRI A 1.35 kb DNA fragment is isolated containing Lhe constitutive PH05 (-173) promotor, a DNA sequence encoding the yeast inverlase signal peptide and the partial GT cDNA.

(c) PH05 terminator sequence The PH05 terrninator sequence is isolated from plasmid p31 which is constructed starting from plasmid p30 as described in EP 100561. After digestion with the restriction en~yme HindIII the protruding ends are filled up by Klenow polymerase treatment as described above. Then the plasmid is digested with EcoRI and a 0.39 kb blunt end - EcoRI fragment with the PH05 terminator sequences is isolated.

Ligation of fragments (a), (b) and (c) and transformation of competent cells of E. coli strain DH5c~ is carried out as described in Example 2. From the obtained transformants plasmids are isolated and characterized by restriction analysis. The plasmid of one clone with the correct restriction fragments is referred to as pDPGTS.

Example 9: Expression of soluble GT in Yeast In analogy to Example 4, CsCI-pllrified DNA of thc expression vector pDPGTS is used to transform S. cerevisiae strains BT 150 and H 449. Ura+-transformants are isolated and screened for GT activity. In each case one transformant is selected and referred to as Saccharomyces cerevisme BT 150/pDPGTS and Saccharomyces cerevisiae H 449/pDPGTS, respectively. Employing the assay described in Example 5 GT-activity is found in the culture broths of both transt`ol mants.

Example 10: Clonin.~ of the sialyltransferase (ST) cDNA t`rom human Hel7G2 cellsST cDNA is isolated from HepG2 cells by PCR in analogy to GT cDNA. Preparation of poly (A)+RNA and first strand cDNA synthesis are performed as described in Example 1.
The primers (Microsynth) listed in Table 5 are used for PCR.

~71~7 Table 5: PCR-primers corresponding to bp primer sequence (5' to 3')1) in ST cDNA
PstVEcoRI
SIA1 coct ca.gaattcaaaATGATTCACACCAACCTGAAGAAAAAGT 1 - 28 BamHI
SIA3 cC~atCCTGTGCTT~GCAGTGAATGGTCCGGAAGCC 1228 - 1198 1) Capit~l letters represent sequences from ST, small Ictters are ~dditional sequences ~vith sites for restriction enzymes (underlined). Codons for 'st~rl' and 'stop' for protein synlhesis a~e indicated in bol~face.
ST cDNA sequence f~om human placenta (27) as published in EMBL Data Bank (Accession Nr.X17247).

HepG2 ST cDNA can be amplified as one DNA fragmcnt of 1.2 kb using the primers SIAl and SIA3. PCR is performed as described for GT cDNA under slightly modifiedcycling conditions: 0.5 min della~uring at 95C, l min. 15 sec annealing a~ 56C, and 1 min 30 sec extension at 72C, for a total of 25-35 cycles. In the last cycle, primer extension at 72C is carried out for 5 min.

After PCR amplification, the 1.2 kb fragment is digested with the restriction enzymes BamHI and PstI, analysed on a l.2 % agarose gel, clu~e(l from the gel and subcloned into the vector pUC18. The resulting subclonc is designated pSIA2.

Example 11: Construction of the constitutive ST expression cassette For constitutive heterologous expression, ST cDNA is ligated to the constitutive PH05 (-173) promoter fragment and PH05 terminator sequences.

Plasrnid pSIA2 is first linearized by digestion with the restriction enzyme BamHI and sllbsequently partially digested with EcoRI. Since ST cDNA also contains an internal restriction site for the enzyme EcoRI (at bp 144), a 1.2 kb fragment with the complete ST
cDNA (SEQ. ID NO. 3) is created by partial digestion with EcoRI using 1 llg of DNA and 0.25 U EcoRI (1 h, 37C). After gel electrophoresis the 1.2 kb EcoRI-BamHI fragment comprising the complete ST cDNA (SEQ ID NO. 3) is isolated. On this DNA fragmentthe 'ATG' start codon for translation of ST is located close to the EcoRI restriction site.
Three adenosine phosphates (see PCR primer SIA 1) provide an 'A' at bp position 12, which is found in the consensus sequence around the 'ATG' from hi;,hly expressed genes in yeast (Hamilton, R. et al. (19~7) Nucleic Acids Res. 14, 5125-5149). The stop codon ~1~7~7 'TAA' and S bp of the 3' untranslated region of the gene are followed by the BamHI site.

~he 1.2 kb EcoRI-BamHI ST cDNA fragment is ligated to ~he 0.45 kb SalI-EcoRI frag-ment containing the constitutive PH05 (-173) promoter (Example 2.2(b)) and a 3.5 kb BamHI-SalI vector part for amplification in E. coli containing the PH05 terminator sequence (cf. Example 2.1, fragment (b)). Ligation and transformation of E. coIi strain DH5~ is performed as detailed in Example 2.1. One clone showing the expected restriction pattern is designated pST2.

Vector pST2 comprises the expression cassette for HepG2 ST under the control of the constitutive PH05 (-173) promotor as a 2.0 kb SalI-HindIII fragment, ret`erred to as DNA
fragment (IC).

Example 12: Expression of ST in yeast Vector pDP34 (cf. EP 340 170) is digested with ~he res~ric~ion enzyme BamHI. The linea-rized vector is isola~ed witll GENECLEAN and the protruding ends are filled in by Klenow polymerase treatment as described in the Maniatis manual (supra). The reaction is stopped after 30 min by heating to 65C for 20 min in the presence of 10 mM EDTA.
After ethanol precipitation the plasmid is digested wi~h '.all and subJected to gel electrophoresis on a 0.8 ~/O agnrose gel. The (BamHI) blllnt end-SalI cut vector pDP34 is isolated as an l 1.8 kb DNA fraoment with the GENECLEAN kit.

Plasmid p~T2 is digested with the restriction enzyme HindIII and in analogy to the preparation of the vector part filled in at the HindIII site by Klenow polymerase treatment.
The product is subjected to SalI digestion, resulting in a 2.0 kb (HindIII) blunt end - SalI
fragment comprising the constitutive ST expression cassette (2C).

Ligation of 80 ng of the pDP34 vector with 40 ng of fragment 2C and transformation of competent cells of E. coli strain DHSo is performed as described in Example 2. One clone showing the expected restriction pattern is chosen and referred to as pDPST5.

For transformation of yeast, CsCI purified DNA of ~he expression vector pDPST5 is prepared following the standard procedure given in lhe Maniatis manual (supra). S.
cerevisiae strains BT 150 and H 449 are each is transformed with 5 llg of plasmid DNA
according to the lithium-acetate transformation method (Ito, H. et al, supra).
Ura+-transformants are selected and screened for ST activity. In each case, one positive 2 ~ 7 transformant is selected and referred to as Saccharomyces cerevisiae BT 150/pDPST5 and Saccharomyces cerevisiae H 4491pDPST5.

Example 13: Enzyme activitY of full-len~th ST expressed in y~
13.1 Preparation of cell extracts Cells of Saccharomyces cerevisiae BT 150/pDPST5 are grown under uracil selection in yeast minimal media supplemented with histidine and leucine. Exponentionally growing cells (at OD578 of 0.5) or stationary cells are collected by centrifugation, washed once with 50 mM imida7.ole buffer, pH 7.0 (buffer 1) and resuspended in buffer 1 at a concentration corresponding to 0.1-0.2 OD578. Mechanical breakage of the cells is effected by vigorous shaking on a vortex mixer with glass beads (0.45-0.5 mm diameter) for 4 min withintermittent cooling.

ST-activity can be measured in the crude extracts employing the assay ùescribed below.

13.2 Assay for ST activity ST activity can be determined by measuring thc amount of radiolabeled sialic acid which is transferred from CMP-sialic acid to a glycoprotein acceptor. After tcrmination of the reaction by acid precipitation the precipitate is filtered using glass fiber filters (Whatman GFA), washed extensively with ice-cold ethalIol and assl ssed for radioactivily by liquid scintillation counting (Hest`ord et al: (1984), Glycoconjugate J. 1, 141-153). Cell ex~racts are assayed for 45 min in an ir.cubation mixture containing 37111 cell extract corresponding to approximately 0.5 mg protein, 3 Ill imida~ol buffer 50 mMol/1, pH 7.0;
50 nMol CMP-N-acetylneuraminic acid (Sigma) to which CMP-3H-N-acetylneuraminic acid (Amersham) is added to give a final specific activity of 7.3 Ci/mol, and 75 llg asialo-fetuin (prepared by acid hydrolysis using 0~1 M H2SO4 at 80C for 60 min,followed by neutralization, dialysis and Iyophili~ation).

ST-activity is found in the crude extracts prepared from S. cerevisiae BT 150/pDPST5 and H 449/pDPST5 cells.

Example 14: Construction of an expression cassette for soluble ST (~39-Cys,~062 The soluble ST designated ST(Lys39-Cys406) is an N-terminally truncated variant and consists of 368 amino acids ~SEQ ID NO. 4).

2 ~ 7 (a) Partial HepG2 ST cI)NA sequence Plasmid pSIA2 is digestcd with EcoRI and a 1.1 kb EcoRI-EcoR1 fragment is isolated.

(b) Vector for amplification in E. coli Plasmid pUC18 (Pharmacia) is digested with BamHI and EcoRI in the multiple cloning site. Then the plasmid is treated w;th alkaline phosphaLase as described in the Maniatis manual (supra), subjected to agarose gel electrophoresis and isolaled t`rom the gel as a 2.7 kb DNA fragment.

(c) Constitutive PH05 (-173) prornoter and SUC2 signal sequence The vector p31/PH05 (-173) RIT (E~xample 2) is digested with the restriction enzymes BamHI and XhoI. A 0.25 kb BamHI-XhoI fragment with the constitutive PH05 (-173) promoter and the ad jacent coding sequence for the invertase signal sequence (inv ss) is isolatecl. Then the fragment is recut with HgaI (BioLabs). The HgaI recognition sequence is on thc antisense DNA strand. The restriction enzyme cuts upstream of the recognition sequence in such a way that the S staggered end of thc antisense stralld coincides with th~
end of the coding sequence of the invertase signal sequence. The 0.24 kb BamHI and HgaI
cut DNA fragment contains the constitutive PH05 (-173) promoter sequence linked to the yeast invertase signal sequence .

(d) Adaptor Fragment (a) is linked to fragment (c) by means of an adaptor sequence which is prepared from equimolar amounts of the synthetic oligonucleotides 5'CTGCAAAATTGCAAACCAAGG 3' and 5'AATTCCTTGGF[TGCAAl~IT 3' for the complementary strand. The oligonucleotides are annealed to each other by first heating to 95C and then slowly cooling to 20C. The annealed adaptor is stored frozen.

(e) Construction of plasmid psST
For ligation, linearized vector (b), the STcDNA fragment (a), fragment (c) containing the promoter and Lhe sequence encoding the signal peptide and the adaptor (d) are used in a molar ratio of 1: 2: 2: 30-lOû. Ligation is carried out in 12 111 of ligase buffer (66 mM
Tris-HCI pH 7.5, 1 mM dithioerythritol, S mM MgCI2, 1 mM ATP) at 16C tor 18 hours.
The ligation mix is used to transform competent cells of E. coli strain DHSc~ as described above. Minipreparations of plasmid are performed from 24 indcpendent transformants. A
single clone showing the expected restriction pattern after characterisation by restriction analysis using four different enzymes (BamHI, Pstl, EcoRI, XhoI, also in combination) is 2`~

referred to as psST.

The correct sequence at the fusion site of the sequence encoding invertase signal peptide with the cDNA coding for soluble ST(Lys39-Cys406) is confirmed for plasmid psST by using the T7-Sequencing kit (Pharmacia) and primer 5'ACGAGGTTAATGGC 3' starting at position -77 in the constitutive PH05 (-173) promoter.

Se~uence for DNAI): 5' AAA ATA Tct gca aaa ttg caa acc aag gAA 3~
3~ TTT TAT AGA GTG ttt aac gtt tgg ttc ctt 5' Protein: Lys Ile Ser Ala¦ Lys Leu Gln Thr Lys Glu inv ss ¦ ST sequence cleavage site for signal endopeptidase 1) Sm~ll letters rcprescnl tbc ~(lnptor scqucl~cc .

The expression casse~te for secretcd ST contailling the constitll~ive P~lû5 (-173) promoter, the DNA sequence encoding invertase signal peplidc and the parlial ST cDNA from plasmid psST can be excised as a 1.35 kb SalI (BamHI) - EcoRI fragment. The expression cassette is still lacking the PH05 terminator sequences to be added in the following cloning step.

Example 15: Construction of the expression vector pDPSTS
For construction of the expression vector for soluble ST(Lys39-Cys406) the following fragments are combined.

(a) Vector part Plasmid pDP34 is digested and pretreated as described in Example 3 resulting in a 11.8 kb blunt end - SalI vector fragment.

(b) Expression casscttc Plasmid psST is first lineari~ed by digestion with SalI (in thc multiple cloning site) and then partially digestcd with EcoRI. A 1.35 kb DNA fragment is isolated containing the ~7~5~

constitutive PH05 (-173) promotor, a DNA sequence encoding the yeast invertase signal peptide and the partial ST cDNA.

(c) PH05 terrninator sequence The PH05 terminator sequence is isolated from plasmid p3 1 which is constructed starting from plasmid p30 as described in EP 100561. After digestion with the restriction enzyme HindIII the protruding ends are filled up by Klenow polymerase treatment as described above. Then the plasmid is digested with EcoRI and a 0.39 kb blunt end - EcoRI fragment wilh lhe PH05 terminator sequences is isolated.

Ligation of fragments (a), (b) and (c) and transformation of competent cells of E. coli strain DHSo~ is carried out as described in Example 2. From the obtained transformants plasmids are isolated and characterized by restriction analysis. The plasmid of one clone with the correct restriction t`ragments is referred to as pDPSTS.

Example 16: Expression of soluble ST in Yeast In analogy to Example 4, CsCI-puril`ied DNA of the exprcssion vector pDPSTS is used to transform S. cerevisiae strains BT 150 and H 449. Ura+-transformants are isolated and screened for ST activity. In each case one transformant is selected and referred to as SaccharomYces cerevisi~le ~3T 150/pDPSTS and Saccllaromyces cercvisine H 449/pDPSTS. Using the assay described in Example 13 ST-activity is found in the culture broths of both transformants.

Example 17: Construction o~ an expression cassette for soluble ST(Lys27-Cys,106) The soluble ST designated ST(Lys27-Cys406) is an N-terminally truncated variant containingthe catalytic domain and the entire stem region and consisting of 380 amino acids, i.e. amino acids 27 to 406 of the amino acid sequence listed in SEQ ID NO. 3.

(a) Partial HepG2 ST cDNA sequence Plasmid pSIA2 is digested with EcoRI and a 1.1 kb EcoRI-EcoRI fragment is isolated.

(b) Vector ~or amplification in E. coli Plasmid pUCl8 (Pharmacia) is digested with ~samHI and EcoRI in the multiple cloning site. Then the plasmid is treated with alkaline phosphatase as described in the Maniatis manual (supra), subjected to agarose gel electrophoresis and isolated from the gel as a 2~7~7 2.7 kb DNA fragment.

(c) Constitutive PH05 (-173) promoter and SUC2 signal sequence The vector p31/PH0~ ~-173) RIT (Example 2) is digested with the restriction enzymes BamHI and XhoI. A 0.25 kb BamHI-XhoI fragment with the constitutive PH05 (-173) promoter and the adjacent coding sequence for the invertase signal sequence (inv ss) is isolated. Then the fragment is recut with HgaI (BioLabs). The HgaI recognition sequence is on the antisense DNA strand. The restriction enzyme cuts upstream of the recognition sequence in such a way that the 5 staggered end of the antisense strand coincides with the end of the coding sequence of the invertase signal sequence. The 0.24 kb BamHI and Hg~l cut DNA fragment contains the constitudve PH05 (-173) promoter sequence linked to the yeast invertase signal sequence .

(d) Adaptor Fragmen~ (a) is linked to fragment (c) by mcans of an adaptor sequence which is prepared from equimolar aMounts of the synthe~ic oligonucleotides S'CTGCAAAGGAAAAGAAGAAAGGGAGTTAC-rATGATTCCTTTAAATTGCAAA
CCAAGG 3', and 5' AATTCCTl'GTTGCA~lTTAAAGGAATCATAGTAACTCCCTTTCTTCTTTT
CCTT 3' for the complementary strand. The oligonucleotides are annealed to each other by first heating to 95C and then slowly cooling to 20C. The annealed adaptor is stored frozen.

(e) Construction of plasmid psST1 For ligation, linearized vector (b), the STcDNA fragment (a), fragment (c) containing the promoter and the sequence encoding the signal peptide and the adaptor (d) are used in a molar ratio of 1: 2: 2: 30-100. Ligation is carried out in 12 1ll of ligase buffer ~66 mM
Tris-HCl pH 7.~, 1 mM dithioerythritol, 5 mM MgCI2, 1 mM ATP) at 16C for 1~ hours.
The ligation mix is used to transform competent cells of E. coli strain DH5cc as described above. Minipreparations of plasmid are performed from 24 independent transformants. A
single clone showing the expected restriction pattern after characterisation by restriction analysis using four different enzymes (BamHI, Pstl, EcoRI, XhoI, also in combination) is referred to as psST1.

The correct sequence at the fusion site of the sequence encoding invertase signal peptide with the cDNA coding for soluble ST(Lys27-Cys406) is confirmed for plasmid psST1 by using the T7-Sequencing kit (Pharmacia) and primer 5'AGTCGAGTAGTATGGC 3' starting at position -77 in the constitutive PH05 (-173) promoter.

Sequence for DNAI): 5' AAA ATA Tct gca aa~ gaa aag aag aaa ggg 3' 3' TTT TAT AGA GTG ttc ctt ttc ttc ttt ccc 5' Protein: Lys Ile Ser Ala¦Lys Glu Lys Lys Lys Gly inv ss ~ ST sequence cleavage site for signal endopeptidase 1) Sm~ll Ict~crs r~:plcscnt th~ ~d~ptor sequcncc .

The expression cassette ~or secreted ST(Lys27-Cys~06) containing the constitutive PH05 (-173) promoter, the DNA sequence encoding invertase signal peptide and the partial ST cDNA from plasmid psSTI can be excised as a 1.35 ~;b SalI (BamHI) - EcoRI
frag,lllent. The expression cassette is still lacking the P~1()5 terminator seq-lcncGs to be added in the following cloning step.

Example 18: Construction of the expression vector pDPSTS1 For construction of the e~pression vector for soluble ST(Lys27-Cys406) the following fragments are combined:

(a) Vector part Plasmid pDP34 is digested and pretreated as described in Example 3 resulting in a 11.8 kb blunt end - SalI vector fragment.

(b) Expression cassette Plasmid psST1 is ~irst linearized by digestion with SalI (in the multiple cloning sitc) and then partially digested with EcoRI. A 1.35 kb DNA fragment is isolatcd containing the constitutive PH05 (-173? promotor, a DNA sequence encoding the yeast invertase signal peptide and the partial ST cDNA.

-37- ~7~

(c) PH05 terminator sequence The PH05 terminator sequence is isolated from plasmid p31 which is constructed starting from plasmid p30 as described in EP 100561. After digestion with the restriction en~yme HindIII the protruding ends are filled up by Klenow polymerase treatment as described above. Then the plasmid is digested with EcoRI and a 0.39 kb blunt end - EcoRI fragment with the PH05 terminator sequences is isolated.

Ligation of fragments (a), (b) and (c) and transformation of competent cells of E. coli strain DH5c~ is carried out as described in Example 2. From the obtained transformants plasmids are isolated and characterized by restriction analysis. The plasmid of one clone with the correct restriction fragments is referred to as pDPSTS 1.

_xample 19: Expression of soluble ST(LYs27-cYs~06) in yeast In analogy to Example ~, CsCI-purified DNA of the expression vector pDPSTS1 is used to transform S. cerevisiae strains BT 150 and H 449. Ura ~-transforman~s are isolated and scrcened for ST activity. In each case one transforrnant ls selected and referred to as Saccharomyces cerevisiae BT 150/pDPSTS1 and SaccharomYces cerevisiae H 449/pDPSTS1. Using the assay described in Example 13, ST-activity is found in the cultllre broths of both transformants.

Example 20: Clonin~ of lhe FTcDNA from the human ~IL60 cell line On the basis of the published cDNA sequence (Goel~, S.E. et al. (1990) Cell 63, 1349-1356) for ELFT (ELAM-1 ligand fucosyltransferase) coding for oc(1-3)-fucosyltransferase the followirlg oligonucleotide primers are designed to amplify a DNA
fragment encompassing the open reading frame of the ELFT cDNA by PCR technology:5'-CAGCGCTGCCTGTTCGCGCCAT-3' (ELFT-lB) and 5'-GGAGATGCACAGTAGAGGATCA-3'(ELFT-2B). ELFT-lB primes at base pairs 38-59 of the published sequence, and ELFT-2B primes at bp 1347-1326 in the ~mtisense.
The primers are used lo amplify a 1.3 kb fragment using the Perkin-Elmer Cetus Taq polymerase kit. The fragment is amplified from fresh HL60 cDNA in the presence of 5%
DMSO and 2 mM MgCI2 with the following cycle:
95C 5 min, 25x (1 min 95C, 1 min 60C, 1.5 min 72C), 10x (1 min 95C, 1 min 60C, 1.5 min + 15 sec/cycle 72C).
Agarose gel electrophoresis reveals a prominent 1.3 kb band which when digested with Smal or ApaI gives the pattern predicted by the published sequencc. The 1.3 kb band is 2 ~

purif1ed using a Gene-Clean kit (Bio 101) and is subcloned into the pCR1000 vector (Invitrogen). A single clone with the correct 1.3 kb insert is selected and referred to as BRB.ELFT/pCR1000-13. The FTcDNA is inserted in the vector in a reverse orientation with respect to the T7 promoter. The open reading frame of the FTcDNA has been fully sequenced and is identical to the published sequence (SEQ.ID NO.~).

Example 21: Construction of plasmids for inducible and constitutive expression of soluble FT(Arg6~-Ar.g40sl in Yeast Soluble FT(Arg62-Arg405) is expressed from the FT cDNA skarting at nucleotide position 241 (NruI restriction site) omitting the N-terminal region coding for the cytoplasmic tail and the nnembrane spanning domain (see sequence ID NO. 5).

Plasmid BRB.ELFT/pCR1000-13 is digested with HindIlI, which cuts in the Multicloning region 3' of the FT cDNA insert. The sticky ends are converted to blunt ends in a reaction with Klenow DNA polymerase. Xhol linker (5' CCTCGAGG 3', Biolabs) cue kinased, annealed and ligated to the blunt en~s of the plasmid, using a 100-fold molar excess of linkers. Unreacted linkers are removed by isopropanol precipitation of the DNA, which is further digested with XhoI and Nrul (cleavage at nucleotide position 240 of the FT cDNA
according to Sequence ID NO. 5). The 1.1 kb Nrul-Xllo[ fragment (a) contains the FT
cDNA sequence lacking the region which codes for the cytoplasmic tail and the membrane-spanning domain up to amino acid 61.

Plasmids p31 RIT12 and p31/PHO5(-173)RIT (see Example ~) are each digested with SalI
and XhoI. The 0.9 kb and 0.5 kb fragments, respectively, are isolated and cut with HgaI.
The resulting sticky ends are filled by Klenow DNA polymerase. The created blunt ends coincide with the 3' end of the coding sequence for the yeast invertase signal sequence.
Subsequent BamHI cleavage releases a 596 bp BamHI-blunt end fragment (b) which comprises the inducible PHO5 promoter and the invertase signal sequence wilh its own ATG or a 234 bp BamHI-blunt end fragment (c) comprising the short, constituLive PHO5(-173) promoter and the inveltase signal sequence.

Plasmid p31RlT1~ is linearized with restriction endonuclease SalI. Partial HindIII
digestion in the presence of ethidiumbromide results in a 1 kb Sall-HindIII fragmcnt comprising the 276 bp SalI-BamHI pBR322 sequence, the 534 bp promoter of the yeast acid phosphatase PH{)5, the yeast invertase signal sequence (coding for 19 amino acids) 2 ~ 7 ~

and the PHO5 transcriptional terminator. The l l~b SalI-H~ndIII fragment of p31RIT12 is cloned into the yeast-E.coli shuttle vector pJDB207 (BegDs, J.D. in: ~Iolecular Genetics in yeast, ~Ifred Benzon Symposium 16, Copenhagen, 1981, pp. 383-389), which has been C~lt with SalI and HindIII. The resulting plasmid containing the 1 kb insert is refelTed to as pJDB207/PHO5-RIT12.

Plasmid pJDB207/PHO5-RIT12 is digested with BamHI and XhoI and the large, 6.8 Icb `
BamHI-XhoI fraoment (d) is isolated. This fragment contains all the pJDB207 vector sequences and the PHO5 transcriptional terminator.

The 596 bp BamHI-blunt end fragment (b), the 1.1 kb NruI-XhoI fagment (a) and the 6.8 kb XhoI-BamHI vector fragment (d) are ligated using standard conditions for blunt end ligation. Aliquots of the ligation mix are used to transform competent E. coli HB101 cells.
PlasMid DNA from ampicillin-resistant colonies is analysed by restriction digests. A
single clone with the correct expression plasmid is rc~errcd to as pJDB207/PHO5-I-FT.
LiOation of DNA fragments (c), (a) and (d) leads to expression plasmid pJDB~07/PHO5-(-173)-I-FT. The expression cassettes of these plasmids comprise the coding sequence of the invertase signal sequence fused in frame to that of the soluble o~(1-3)fucosyltransferase which is expressed under the control Or the inducible P~l[O5 or the constitutivePHO5(-173) promoter, respectively. The expression cassettes are clone(i into tlle yeast-E.
coli shuttle vector pJDB207 between the BamHI and HindIII restriction sites.
The nucleotide sequence at the site of the fusion between the invertase signal sequence and cDNA coding for soluble FT is confirmed by DNA sequencing on double strandedplasmid DNA using the primer 5' AGTCGAGGTTAGTATGGC 3' representing the nucleotide sequence at position -77 to -60 of the PHO5, as well as the PHO5(-173) promoter. The correct junction is:

5' AAA ATA TCT GCA CGA CCG GTG 3' 3' TTT TAT AGA CGT GCT GGC CAC 5' Lys Ile Ser Ala Arg Pro Val Inv.ss FT

cleavaDe site for signal peptidase - ~o -Example 22: Construction of plasmids for inducible and constitutive expression of membranc-bound FT in yeast These constructs use the coding sequence of the FT cDNA with its own ATG. The nucleotide sequence immediately upstream of the ATG, however, is rather unfavourable for expression in yeast due to its high G-C content. This region has been replaced by an A-T rich sequence using PCR methods. At the same time an EcoRI restriction site is introduced at the new nucleotide positions -4 to -9.

Table 6: PCR primers corresponding to primer sequence (5' to 3')1) bp in FT cDNA

FTl cgaqaattcataATGGGGGCACCGTGGGGC 58 to 75 FT2 ccgctcq~qGAGCGCGGCTTCACCGCTCG 1~85 to 1266 ., ~ C~pitnl l~ rs represcnt nucleoti(les from ~T, sm.lll Ic~tcrs ~rc ~l~lclilionnl 11~V s~qucnccs, reslriclion si~es ~rc un(lcrlillc~l, "st~rt" nlld "stop" eo~lons tlre hi~ llte~l.

Standard PCR conditions are used to amplify the FT cDNA with primers FT1 and FT2(see Table 6) in 30 cycles of DNA synthesis (Taq DNA polymerase, Perkin-Elmer, S U/
1 min at 72C), denaturation (10 sec at 93C) and annealing (40 sec at 60C). The resulting 1.25 kb DNA fragment is purified by phenol extraction and ethanol precipitation, then digested with EcoRI, XhoI and Nrul. The 191 bp EcoRI-NruI fragment (e) is isolated on a prepauative 4% Nusieve 3:1 agarose (FMC BioProducts, Rockland, ME, USA) gel in tris-borate buffer pH 8.3, gel-eluted and ethanol prècipitated. Fragment (e) comprises the S' part of the FT gene coding for amino acids 1 to ~1. The DNA has a 5' extension with the EcoRI site as in PCR primer FT1.

Plasmids p31RIT12 and p31/PHO5(-173)RIT are each digested with EcoRI and XhoI. The large vector fragments (f and g, respectively) are isolated on a preparative 0.8% agarose gel, eluted and purified. The 4.1 kb XhoI-EcoRI ~ragment (f) comprises the 2~70~7 pBR322-derived vector, the 534 bp PHO5 promoter (3' EeoRI sile) and the 131 bp PHO5 transcriptional terminator (5' XhoI site). The 3.7 kb XhoI-EcoRI fragment (g) only differs by the short, constitutive, 172 bp PHO5(-173) promoter (3' EcoRI site) instead of the full length PHO5 promoter.

The 191 bp EcoRI-NruI fragment (e), the 1.1 kb NruI-XhoI fragment (a) and the ~.1 kb XhoI-Ecol~I fragment (f) are ligated. A 1111 aliquot of the ligation mix is used to transforrn competent E. coli HB101 cells. Plasmid DNA from ampicillin resistant colonies is analysed. Plasmid DNA from a single clone is referred to as p31R/PHO5-ssFT.

Ligation of DNA fragments (e), (a) and (g) leads to plasmid p31R/PHO5(-173)-ssFT.
These plasmids comprise the coding sequence of the membrane-bound FT under the control of the inducible PHO5 or the constitutive PHO5(-173) promoter, respectively.

Example 23: Cloning of the FT expression cassettes into pDP34:

Plasmids p31RlPHO5-ssFT and p31RlPHO5(-173)-ssF'I' are digested with HindllI, which cuts 3' of the PHO5 transcriptional terminator. After a reaction wi~h Klenow DNApolymerase, the DNA is digested with Sall. The 2.3 kb and 1.9 kb SalI-blunt end l`ra"ments, respectively, are isolated.

Plasmids pJDB207/PHO5-I-E~T and pJDB207/PHO5(-173)-I-FT are p~tially digested with HindIlI in the presence of 0.1 mg/ml of ethidium bromide (to avoid cleavage at an additional HindIII site in the invertase signal sequence) and then treated with Klenow DNA polymerase and SalI as above. The 2.1 kb and 1.8 kb fragments, respectively, are isolated.

The four DNA fràgments are each ligated to the 11.8 kb SalI-blunt end vector fragment of pDP34 (see Example 3). Upon transformation of competent E. coli HB101 cells and analysis of plasmid DNA of individual transformants, four correct expression plasmids are referred to as pDP34/PHO5-I-FT; pOP34/PHO5(-173)-I-FT; pDP34R/PHO5-ssFT and pDP34R/PHO5(- 173)-ssFT.

S. cerevisiae strains BT150 and H449 are transl`ormed with 5 llg each of the four expression plasmids (above) according to Example 4. Single transformcd yeast colonies are selected and referred to as -42- 2~7~

Saccharomyces cerevisiae BTl50/pDP34/PHO5-I-FT;
" . " BT150/pDP34/PHO5(-173)-I-FT;
" " BT150/pDP34R/PHO5-ssFT;
" " BT150/pDP34RlPHO5(-173)-ssFT;
" " H449/pDP34/PHO5-T-FT;
" " H449/pDP34/PHO5(-173)-I-FT;
" " H449/pDP34RlPHO5-ssFT;
" " H449/pDP34R/PHO5(- 173)-ssFT

Fermentation and preparation of the cell extracts is performed according to Example 5.
Using an assay analogous to that described by Goelz et al. (supra) FT-activity is found in the crude extracts prepared from strains BT150/pDP341VPHO5-ssFT, BT150/pDP34R/PHO5(- 173)-ssFT, H449/pDP34R/PHO5-ssFT and H449/pDP34R/PH05(-173)-ssFr, and in the culture broth of strains H449/pDP34/PHO5-I-FT, H449/pDP34/PHO5(-173)-I-FT, BTl50/pDP34/PHO5-1-FT and BTl50/pDP34/PHO5(-173)-I-FT, eposition of microoronnisms The following microorgnnism strains were (leposited with the Deutsche Sammlllng von Mikroorganismen (DSM), Mascheroder We;,16, D-3300 Braullschweig (deposi~ion dates and accession numbers given):
Escherichia coli JM109/pDP34: March 14, 1988; DSM 4473 _cherichia coli HB101/p30: October 23,1987; DSM 4297 Escherichia coli HB101/p31R: December 19, 1988; DSM 5116 Saccharomyces cerevisiae H 449: Febn~ary 18, 1988; DSM 4413 Saccharomyces cerevisiae BT 150: May 23, 1991; DSM 6530 2 ~

Segllence listino SEQ ID I~O. 1 SEQUENCE TYPE: Nucleotide with corresponding protein SEQUENCE LENGTH: 1265 bp STRANDEDNESS: double TOPOLOGY: linear MOLECULE TYPE: recombinant IMMEDIATE EXPERIMENTAL SOURCE: Plasmid p4AD113 from E. coli DH5(x/p4AD 113 FEATURES: from 6 to 1200 bp cDNA sequence coding for HeLa cell galac tosyltransferase from 1 to 6 bp EcoRI site trom 497 to 504 bp NotI site from 1227 to 1232 bp EcoRI site from 1236 to 1241 bp EcoRV site from 1243 to 1248 bp BglIl site ROPERTIES: EcoRI-HindIII fragment t`roM plasmicl p4ADl l3 comprising HeLa cell cDNA coding for full-length galactosyllransfcrase (EC 2.4.1.22) Met Arg Leu Arg Glu Pro Leu Leu Ser Gly Ser Ala Ala Met Pro Gly Ala Ser Leu Gln Arg Ala Cys Arg Leu Leu Val Ala Val Cys Ala Leu His Leu Gly Val Thr Leu Val Tyr Tyr Leu Ala Gly Arg Asp Leu Ser Arg Leu 4~ 2~70~7 Pro Gln Leu Val Gly Val Ser Thr Pro Leu Gln Gly Gly Ser Asn Ser Ala Ala Ala Ile Gly Gln Ser Ser Gly Glu Leu Arg Thr Gly Gly Ala Arg Pro Pro Pro Pro Leu Gly Ala Ser Ser Gln Pro Ar~ Pro Gly Gly Asp Ser Ser Pro Val Val Asp Ser Gly Pro Gly Pro Ala Ser Asn Leu Thr Ser Val Pro Val Pro His Thr Thr Ala Leu Ser Leu Pro GCC TGC CCT GAG GAG TCC CCG CTG CTT GTG GGC CCC ATG 42g Ala Cys Pro Glu Glu Ser Pro Leu Leu Val Gly Pro Met Leu Ile Glu Phe Asn Met Pro Val Asp Leu Glu Leu Val Ala Lys Gln Asn Pro Asn Val Lys Met Gly Gly Arg Tyr 2 ~ 7 ~

Ala Pro Arg Asp Cys Val Ser Pro His Lys Val Ala Ile Ile Ile Pro Phe Arg Asn Arg Gln Glu His Leu Lys Tyr Trp Leu Tyr Tyr Leu His Pro Val Leu Gln Arg Gln Gln Leu Asp Tyr Gly Ile Tyr Val Ile Asn Gln Ala Gly Asp Thr Ile Phe Asn Arg Ala Lys Leu Leu Asn Val Gly Phe Gln Glu Ala Leu Lys Asp Tyr Asp Tyr Thr Cys Phe Val Phe Ser Asp Val Asp Leu Ile Pro Met Asn Asp His Asn Ala Tyr Arg Cys Phe Ser Gln Pro Arg His Ile Ser Val Ala Met Asp Lys Phe Gly Phe Ser Leu Pro Tyr Val Gln 2 ~ Pl Tyr Phe Gly Gly Val Ser Ala Leu Ser Lys Gln Gln Phe Leu Thr Ile Asn Gly Phe Pro Asn Asn Tyr Trp Gly Trp Gly Gly Glu Asp Asp Asp Ile Phe Asn Arg Leu Val Phe Arg Gly Met Ser Ile Ser Arg Pro Asn A:la Val Val G:Ly Arg Cys Arg Met Ile Arg His Ser Arg A~,p Lys Lys Asn Glu Pro Asn Pro Gln Arg Phe Asp Arg Ile Ala His Thr Lys Glu Thr Met Leu Ser Asp Gly Leu Asn Ser Leu Thr Tyr Gln Val Leu Asp Val Gln Arg Tyr Pro L.eu Tyr Thr Gln Ile Thr Val Asp Ile Gly Thr Pro Ser 2 ~

SEQ ID NO. 2 SEQUENCE TYPE: Protein SEQUENCE LENH: 357 amino acids MOLECULE TYPE: C-terminal fragment of full-lenoth HeLa cell galactosyl-lransferase PROPERTIES: soluble galactosyltransferase (EC 2.4.1.22) from HeLa cells Leu Ala Gly Arg Asp Leu Ser Arg Leu Pro Gln Leu Val Gly Val Ser Thr Pro Leu Gln Gly Gly Ser Asn Ser Ala Ala Ala Ile Gly Gln Ser Ser Gly Glu eu Arg Thr Gly Gly Ala Arg Pro Pro Pro Pro Leu Gly Ala Ser Ser Gln Pro Arg Pro Gly Gly Asp Ser Ser Pro Val Val Asp Ser Gly Pro Gly Pro Ala Ser Asn Leu Thr Ser Val Pro Val Pro His Thr Thr Ala Leu Ser Leu Pro Ala Cys Pro Glu Glu Ser Pro Leu Lèu Val Gly Pro Met go 95 100 eu Ile Glu Phe Asn Met Pro Val Asp Leu Glu Leu Val Ala Lys Gln Asn Pro Asn Val Lys Met Gly Gly Arg Tyr 2~7~7 Ala Pro Arg Asp Cys Val Ser Pro His Lys Val Ala Ile . 130 135 Ile Ile Pro Phe Arg Asn Arg Gln Glu His Leu Lys Tyr Trp Leu Tyr Tyr Leu His Pro Val Leu Gln Arg Gln Gln Leu Asp Tyr Gly Ile Tyr Val Ile Asn Gln Ala Gly Asp ~'hr Ile Phe Asn Arg Ala Lys Leu Leu Asn Val Gly Phe Gln Glu Ala I,eu Lys Asp Tyr Asp Tyr Thr Cys Phe Val Phe Ser Asp Val Asp Leu Ile Pro Met Asn Asp His Asn Ala Tyr Arg Cys Phe Ser Gln Pro Arg His Ile Ser Val Ala Met Asp Lys Phe Gly Phe Ser Leu Pro Tyr Val Gln Tyr Phe Gly Gly Val Ser Ala Leu Ser Lys Gln Gln Phe Leu Thr Ile Asn Gly Phe Pro Asn Asn Tyr Trp Gly Trp Gly Gly Glu Asp Asp Asp Ile Phe Asn Arg Leu Val Phe 2~7~7 - so -Arg Gly Met Ser Ile Ser Arg Pro Asn Ala Val Val Gly . 285 290 295 Arg Cys Arg Met Ile Arg His Ser Arg Asp Lys Lys Asn Glu Pro Asn Pro Gln Arg Phe Asp Arg Ile Ala His Thr Lys Glu Thr Met Leu Ser Asp Gly Leu Asn Ser Leu Thr Tyr Gln Val Leu Asp Val Gln Arg Tyr Pro Leu Tyr Thr Gln Ile Thr Val Asp Ile Gly Thr Pro Ser 2~7~7 5, SEQ ID NO. 3 SEQUENCE TYPE: Nucleotide with corresponding protein SEQUENCE LENH: 1246 bp STRANDEDNESS: double TOPOLOGY: linear MOLECULE TYPE: recombinant IMMEDIATE EXPERIMENTAL SOURCE: Plasmid pSIA2 from E. coli DHS xlpSIA2 FEATURES: from lS to 1232 bp cDNA sequence coding for HepG2 cell sialyltransferase from l to 6 bp PstI site from 6 to 1 l bp EcoRI site from l44 to 149 ~p EcoRI site from l 241 to l 246 bp BamHI site ROPERTIES: Pstl-BamHI fragmcnt from plasmid pSlA2 comprising HepG2 cDNA
coding for t`ull-length sialyltransferase (EC 2.~1.99.l) TGCAGAATT CAAA ATG ATT CAC ACC AAC C1'G AAG AAA 38 Met Ile His Thr Asn Leu Lys Lys Lys Phe Ser Cys Cys Val Leu Val Phe Leu Leu Phe Ala Val Ile Cys Val Trp Lys Glu Lys Lys Lys Gly Ser Tyr Tyr Asp Ser Phe Lys Leu Gln Thr Lys Glu Phe Gln Val ~7~7 Leu Lys Ser Leu Gly Lys Leu Ala Met Gly Ser Asp Ser Gln Ser Val Ser Ser Ser Ser Thr Gln Asp Pro His Arg Gly Arg Gln Thr Leu Gly Ser Leu Arg Gly Leu Ala Lys Ala Lys Pro Glu Ala Ser Phe Gln Val Trp Asn Lys Asp Ser Ser Ser Lys Asn Leu Ile Pro Arg Leu Gln Lys Ile Trp Lys Asn Tyr Leu Ser Met Asn Lys Tyr Lys Val Ser Tyr Lys Gly Pro Gly Pro Gly Ile Lys Phe Ser Ala Glu Ala Leu Arg Cys His Leu Arg Asp His Val Asn Val Ser Met Val Glu Val Thr Asp Phe Pro Phe Asn Thr Ser Glu ~ 3 Trp Glu Gly Tyr Leu Pro Lys Glu Ser Ile Arg Thr Lys 165 . 170 175 Ala Gly Pro Trp Gly Arg Cys Ala Val Val Ser Ser Ala Gly Ser Leu Lys Ser Ser Gln Leu Gly Arg Glu Ile Asp Asp His Asp Ala Val Leu Arg Phe Asn Gly Ala Pro Thr Ala Asn Phe Gln Gln Asp Val Gly Thr Lys Thr Thr Ile CGC CTG ATG AAC TCT CAG TTG GTT ACC AC`A GAG AAG CGC 740 Arg Leu Met Asn Ser Gln Leu Val Thr Thr Glu Lys Arg Phe Leu Lys Asp Ser Leu Tyr Asn Glu Gly Ile Leu Ile Val Trp Asp Pro Ser Val Tyr His Ser Asp Ile Pro Lys Trp Tyr Gln Asn Pro Asp Tyr Asn Phe Phe Asn Asn Tyr 2~7~7 AAG ACT TAT CGT AAG CTG CAC CCC AAT CAG CCC TTT TAC 8 g 6 Lys Thr Tyr Arg Lys Leu His Pro Asn Gln Pro Phe Tyr Ile Leu Lys Pro Gln Met Pro Trp Glu Leu Trp Asp Ile Leu Gln Glu Ile Ser Pro Glu Glu Ile Gln Pro Asn Pro Pro Ser Ser Gly Met Leu Gly Ile Ile Ile Met Met Thr Leu Cys Asp Gln Val Asp Ile Tyr Glu Phe Leu Pro Ser Lys Arg Lys Thr Asp Val Cys Tyr ~yr Tyr Gln Lys Phe Phe Asp Ser Ala Cys Thr Met Gly Ala Tyr His Pro Leu Leu Tyr Glu Lys Asn Leu Val Lys His Leu Asn Gln Gly Thr Asp Glu Asp lle Tyr Leu Leu Gly Lys Ala Thr Leu 2 ~

CCT GGC TTC CGG ACC ATT CAC TGC~TAAGCACAGG ATCC 1246 Pro Gly Phe Arg Thr Ile His Cys 2~7~7 SEQ ID NO. 4 SEQUENCE TYPE: Protein SEQUENCE LENH: 368 amino acids MOLECULE TYPE: C-terminal fragment of full-length sialyltransferase PROPERTIES: soluhle sialyltransferase (EC 2.4.99.1) from human HepG2 cells Lys Leu Gln Thr Lys Glu Phe Gln Val Leu Lys Ser Leu Gly Lys Leu Ala Met Gly Ser Asp Ser ln Ser Val Ser Ser Ser Ser Thr Gln Asp Pro His Arg ' 25 30 35 ly Arg Gln Thr Leu Gly Ser Leu Arg G:ly Leu Ala Lys ~15 Ala Lys Pro Glu Ala Ser Phe Gln Val Trp Asn Lys Asp Ser Ser Ser Lys Asn Leu Ile Pro Arg Leu Gln Lys Ile Trp Lys Asn Tyr Leu Ser Met Asn Lys Tyr Lys Val Ser yr Lys Gly Pro Gly Pro Gly Ile Lys Phe Ser Ala Glu la Leu Arg Cys His Leu Arg Asp His Val Asn Val Ser Met Val Glu Val Thr Asp Phe Pro Phe Asn Thr Ser Glu Trp Glu Gly Tyr Leu Pro Lys Glu Ser Ile Arg Thr Lys Ala Gly Pro Trp Gly Arg Cys Ala Val Val Ser Ser Ala Gly Ser Leu Lys Ser Ser Gln Leu Gly Arg Glu Ile Asp sp His Asp Ala Val Leu Arg Phe Asn Gly Ala Pro Thr Ala Asn Phe Gln Gln Asp Val Gly Thr Lys Thr Thr Ile Arg Leu Met Asn Ser Gln Leu Val Thr Thr Glu Lys Arg 195 ~00 Phe Leu Lys Asp Ser Leu Tyr Asn Glu Gly Ile Leu Ile Val Trp Asp Pro Ser Val Tyr His Ser Asp Ile Pro Lys rp Tyr Gln Asn Pro Asp ~yr Asn Phe Phe Asn Asn Tyr Lys Thr Tyr Arg Lys Leu His Pro Asn Gln Pro Phe Tyr Ile Leu Lys Pro Gln Met Pro Trp Glu Leu Trp Asp Ile eu Gln Glu Ile Ser Pro Glu Glu Ile Gln Pro Asn Pro 20700~7 Pro Ser Ser Gly Met Leu Gly Ile Ile Ile Met Met Thr Leu Cys Asp Gln Val Asp Ile Tyr Glu Phe Leu Pro Ser Lys Arg Lys Thr Asp Val Cys Tyr Tyr Tyr Gln Lys Phe Phe Asp Ser Ala Cys Thr Met Gly Ala Tyr His Pro Leu Leu Tyr Glu Lys Asn Leu Val Lys His Leu Asn Gln Gly Thr Asp Glu Asp Ile Tyr Leu Leu Gly Lys Ala Thr Leu Pro Gly Phe Arg Thr Ile His Cys , .

2 ~ 7 SEQ ID NO.5 SEQUENCE TYPE:Nucleotide sequence with corresponding protein SEQUENCE LENGTH:1400 bp STRANDEDNESS:double TOPOLOGY:linear IMMEDIATE EXPERIMENTAL SOURCE:BRB.ELFT/pCR1000-13 FEATURES:
from 58 to 1272 bp cDNA sequence coding forhuman~(l-3) fucosyltransferase from 238 to 243 bp NruI site PROPERTIES:HL60 cDNA coding forfull-length~(1-3) fucosyl-transferase Met Gly Ala Pro Trp Gly Ser Pro Thr Ala Ala Ala Gly Gly Arg Arg Gly Trp Arg Arg Gly Arg Gly Leu Pro Trp Thr Val Cys Val Leu Ala Ala Ala Gly Leu Thr , Cys Thr Ala Leu Ile Thr Tyr Ala Cys Trp Gly Gln Leu Pro Pro Leu Pro Trp Ala Ser Pro Thr Pro Ser Arg Pro ~`7~0~

Val Gly Val Leu Leu Trp Trp Glu Pro Phe Gly Gly Arg Asp Ser Ala Pro Arg Pro Pro Pro Asp Cys Arg Leu Arg Phe Asn Ile Ser Gly Cys Arg Leu Leu Thr Asp Arg Ala go 95 100 Ser Tyr Gly Glu Ala Gln Ala Val Leu Phe His His Arg Asp Leu Val Lys Gly Pro Pro Asp Trp Pro Pro Pro Trp 120 1;25 Gly Ile Gln Ala His Thr Ala Glu Glu Val Asp Leu Arg Val Leu Asp Tyr Glu Glu Ala Ala Ala Ala Ala Glu Ala ~45 150 Leu Ala Thr Ser Ser Pro Arg Pro Pro Gly Gln Arg Trp Val Trp Met Asn Phe Glu Ser Pro Ser His Ser Pro Gly ~2~7~

Leu Arg Ser Leu Ala Ser Asn Leu Phe Asn Trp Thr Leu Ser Tyr Arg Ala Asp Ser Asp Val Phe Val Pro Tyr Gly Tyr Leu Tyr Pro Arg Ser His Pro Gly Asp Pro Pro Ser Gly Leu Ala Pro Pro Leu Ser Arg Lys Gln Gly Leu Val Ala Trp Val Val Ser His Trp Asp Glu Arg Gln Ala Arg Val Arg Tyr Tyr His Gln Leu Ser Gln His Val Thr Val Asp Val Phe Gly Arg Gly Gly Pro Gly Gln Pro Val Pro Glu Ile Gly Leu Leu His Thr Val Ala Arg Tyr Lys Phe Tyr Leu Ala Phe Glu Asn Ser Gln His Leu Asp Tyr Ile Thr Glu Lys Leu Trp Arg Asn Ala Leu Leu Ala Gly Ala Val Pro Val Val Leu Gly Pro Asp Arg Ala Asn Tyr Glu Arg Phe Val Pro Arg Gly Ala Phe Ile His Val Asp Asp I'TC CCA AGT GCC TCC TCC CTG GCC TCG TAC CTG CTT TTC 1104 Phe Pro Ser Ala Ser Ser Leu Ala Ser Tyr Leu Leu Phe Leu Asp Arg Asn Pro Ala Val Tyr Arg Arg Tyr Phe His Trp Arg Arg Ser Tyr Ala Val His Ile Thr Ser Phe Trp Asp Glu Pro Trp Cys Arg Val Cys Gln Ala Val Gln Arg Ala Gly Asp Arg Pro Lys Ser Ile Arg Asn Leu Ala Ser Trp Phe Glu Arg 2~7~7 TGCCGCATCA TGGGAGTAAG TTCTTCAAAC ACCCATTTTT GCTCTATG 1~00

Claims (19)

1. Process for the production of a membrane-bound mammalian glycosyltransferase selected from the group consisting of a galactosyltransferase, a sialyltransferase and a fucosyltransferase, or a variant thereof, respectively, comprising culturing a yeast strain which has been transformed with a hybrid vector comprising an expression cassette comprising a promoter and a DNA sequence coding for said glycosyltransferase or variant which DNA is controlled by said promoter, and recovering the enzymatic activity.

2. Process according to claim 1, wherein the glycosyltransferase is of human origin.

3. Process for the production of a variant according to claim 1, wherein the variant differs from the corresponding full-length glycosyltransferase by lack of the cytoplasmic tail, the signal anchor and, optionally, a minor part of the stem region.

4. Process for the production of a variant according to claim 3 comprising culturing a yeast strain comprising an expression cassette comprising a promoter operably linked to a first DNA sequence encoding a signal peptide linked in the proper reading frame to a second DNA sequence coding for said variant which DNA sequence is controlled by said promoter, and recovering the enzymatic activity

5. Process according to claim 1, wherein the glycosyltransferase is a galactosyltransferase.

6. Process according to claim 5, wherein the galactosyltransferase is selected from the group consisting of UDP-Galactose: .beta.-galactoside .alpha.(1-3)-galactosyltransferase (EC 2.4.1.151) and UDP-Galactose: ,.beta.-N-acetylglucosamine ,.beta.(1-4)-galactosyltransferase (EC 2.4.1.22).

7. Process according to claim 5, wherein the galactosyltransferase has the amino acid sequence depicted in SEQ ID NO. 1.

8. Process according to claim 5, wherein the galactosyltransferase has the amino acid sequence depicted in SEQ ID NO. 2.

9. Process according to claim 1, wherein the glycosyltransferase is a sialyltransferase.

10. Process according to cla;m 9, wherein the sialyltransferase is CMP-NeuAc .beta.-galactoside .alpha.(2-6)-sialyltransferase (EC 2.4.99.1).

11. Process according to claim 9, wherein the sialyltransferase has the amino acid depicted in SFQ ID NO. 3.

12. Process according to claim 9, wherein the sialyltransferase is designated ST(Lys27-Cys406) and consists of amino acids 27 lo 406 of the amino acid sequence listed in SEQ. ID NO.3.

13. Process according to claim 9, wherein the sialyltransferase has the amino acid depicted in SEQ ID NO.4.

14. Process according to claim 1, wherein the glycosyltransferase is a fucosyltransferase.

15. Process according to claim 14, wherein the fucosyltransferase is selected from the group consisting of GDP-Fucose:.beta.-galactoside .alpha.(1-2)-fucosyltransferase(EC 2.4.1.69) and GDP-Fucose:N-acetylglucosamine .alpha.(1-3/4)-fucosyltransferase(EC 2.4.1.65).

16. Process according to claim 14, wherein the fucosyltransferase has the amino acid sequence depicted in the sequence listing with SEQ ID NO. 5.

17. Process according to claim 14, wherein the fucosyltransferase is designated FT(Arg62-Arg405) and consists of amino acids 62 to 405 of the amino acid sequence depicted in the sequence listing with SEQ ID. NO. 5.

18. A yeast hybrid vector comprising an expression cassette comprising a yeast promoter and a DNA sequence coding for a membrane-bound mammalian glycosyltransferase selected from the group consisting of a galactosyltransferase, a sialyltransferase and a fucosyltransferase, or a variant thereof, respectively, which DNA sequence is controlled by said promoter.

19. A yeast strain which has been transformed with a hybrid vector according to claim 18.

FD 4.4/DVC