WO2001012657A2 - Methods and means for selenoprotein expression - Google Patents

Abstract

In bacterial host cells a mammalian or other heterologous selenoprotein is produced by recombinant expression, employing nucleic acid which comprises a coding sequence for the selenoprotein, a bacterial selenocysteine insertion sequence (SECIS) for insertion of selenocysteine into the protein on production of the protein by expression from the nucleic acid in bacterial host cells and regulatory sequences for expression of the encoded selenoprotein in bacterial host cells. The host cells are preferably cultured under conditions for over-expression of selA, selB, and selC so that the selenoprotein is produced.

Description

METHODS AND MEANS FOR SELENOPROTEIN EXPRESSION

The present invention relates to the expression of selenoprotems. In particular, it relates to expression in bacteria, especially E . coli , of heterologous selenoprotems. The heterologous expression involves insertion of selenocysteine by means of manipulation of the bacterial insertion machinery and engmeered selenocysteine insertion sequence (SECIS) elements.

Selenoprotems carrying a selenocysteine residue are found in bacteria, archaea as well as eu arya and often have oxidoreductase activity. In bacteria and archaea these mclude formate dehydrogenases, hydrogenases or glycine reductase whereas in mammals the glutathione peroxidase family and the thyroid hormone deiodmases are selenoprotems, in addition to mammalian selenoprotems with unknown function such as selenoprotem P or (Bock et al . , (1991) Mol. Microbiol. , 5, 515-520; Low and Berry (1996) TIBS, 21, 203-208; Stadtman (1996) Annu . Rev. Biochemistry, 65, 83-100) . The selenocysteine residue is in all organisms inserted at the position of an opal (UGA) codon in the mRNA, normally conferring termination of translation. In selenoprotem mRNA's the UGA codon is alternatively encoded as selenocysteine by a complex machinery, best characterized in E . coli with formate dehydrogenase H as a model selenoprotem - for reviews, see ref. (Bock et al . , (1991) Mol. Microbiol., 5, 515-520; Huttenhofer and Bock (1998) Cold Spring Harbor Laboratory Press, New York; Stadtman (1996) Annu. Rev. Biochemistry, 65, 83-100).

In short, mRNA for selenoprotems m E . col i and other bacteria contain an about 40 nucleotides long selenocysteine insertion sequence (SECIS) positioned immediately at the 3 '-side of the UGA codon (Heider et al . , (1992) EMBO J., 11, 3759-3766; Zmoni et al . , (1990) Proc. Natl. Acad. Sci., 87, 4660-4664). The codons in the SECIS have dual functions: they provide the genetic code for translation of the ammo acids following the selenocysteine residue, and they fold mto a stem-loop type secondary structure. This SECIS element possibly inhibits bmdmg of RF2 but is also, most importantly, binding the 68 kDa SELB protem, the selB gene product. SELB is homologous to the 43 kDa elongation factor EF-Tu but addition carries a carboxytermmal elongation responsible for recognizing and binding the loop region of the SECIS element (Huttenhofer et al . , (1996) RNA, 2, 345-366; Kromayer et al . , (1996) J. Mol. Biol., 262, 413-420) . In contrast to EF-Tu, SELB does not bmd to many different tRNA species, but only to a selenocysteine specific tRNA (tRNA^Sec) , the selC gene product, in its selenocystemylated form (Baron and Bock (1991) J. Biol. Chem., 266, 20375-20379; Forchham er et al . , (1989) Nature, 342, 453-456) . Thereafter, SELB catalyzes selenocysteine insertion at the ribosome, at the specific position of the selenocysteine UGA codon.

After elongation with selenocysteine, EF-Tu continues catalyzing elongation of the growing peptide chain, with ammo acids now encoded by the same nucleotides that previously made up the SECIS element. The selC gene product, i.e. the selenocysteine specific tRNASec (Leinfelder et al . , (1988) Nature, 331, 723-725), is originally charged with a seryl-residue which by utilization of selenophosphate is converted to selenocystemyl by the pyrιdoxal-5 ' -phosphate- containing selenocysteine synthase, an oligomer of the 50 kDa selA gene product (Forchhammer and Bock (1991) J. Biol. Chem., 266, 6324-6328; Tormay et al . , (1998) Eur. J. Biochem. , 254, 655-661). The selenophosphate, m turn, is provided by selenophosphate synthetase, the selD gene product (Ehrenreich et a l . , (1992) Eur . J. Biochem. , 206, 767-773; Lemfelder et al . , (1990) Proc. Natl. Acad. Sci., 87, 543-547) .

Taken together, selenocysteine insertion during selenoprotem translation in E. colx involves the following factors: An E . coli-type SECIS element just following the UGA codon in the selenoprotem mRNA, and the selA, selB, selC and selD gene products .

Although not yet as fully characterized, it is clear that mammalian cells show significant qualitative differences in selenoprotem synthesis from the system found in E . coli . A SECIS element is present also m the mRNA of mammalian selenoprotems but displays other secondary structures and conserved features than those seen in E. coli . Moreover, in mammalian selenoprotem mRNA's the SECIS elements are situated in the 3 ' -untranslated region, several hundred nucleotides downstream of the UGA codon (Low and Berry (1996) TIBS, 21, 203-208; alczak et al . , (1996) RNA, 2, 367-379). It is therefore believed that the 3 ' -untranslated region

"loops back" for selenocysteine insertion, bmdmg a putative mammalian SELB homologue (Berry et al . , (1991) Nature, 353, 273-276; Low and Berry (1996) TIBS, 21, 203-208).

It should be emphasized that the structures and specific nucleotides of the SECIS elements of both E . coli and mammalian selenoprotem mRNA's are highly species specific, determining the bmdmg properties to the homologous SELB protems. This explains why mammalian selenoprotem genes cannot directly be expressed as recombinant protems E . coli . If a technique would exist, however, to by-pass barriers to heterologous expression of selenoprotems in E . col i (Tormay and Bock (1997) J. Bacteriol., 179, 576-582), this should be of significant value from both a basic and applied view. It was previously shown that both the SECIS element and the SELB protem interacting with this SECIS element and with the ribosome must be from the homologous organisms to achieve the production of recombmant selenoprotems m E. coli (Heider et al . , (1992) EMBO J. , 11, 3759-3766; Tormay and Bock (1997) J. Bacteriol . , 179, 576-582) .

The present inventors have now unexpectedly shown that mammalian proteins (exemplified with rat thioredoxm reductase) can be produced as a recombinant selenoprotem in E . coli , in a system conforming to the specificities of the bacterial selenocysteine translation machinery.

Thioredoxm reductase (TrxR) is a flavoprotem which catalyzes NADPH-dependent reduction of the active site disulfide in oxidized thioredoxm, an ubiquitous 12 kDa prote with a large number of activities (Holmgren (1989) J. Biol. Chem., 264, 13963-6; Holmgren and Bjornstedt (1995) Meth. Enzymol., 252, 199-208; Williams (1992) CRC Press, Boca Raton, FL, Vol. 3, pp. 121-211; Yodoi and Tursz (1991) Adv. Cancer Res., 57, 381-411). Thioredoxm reductase from Escheπ chia coli has been crystallized (Williams (1992) CRC Press, Boca Raton, FL, Vol. 3, pp. 121-211) and a high resolution X-ray structure shows surprisingly large differences to other members of the pyridme nucleotide-disulfide oxidoreductase family, notably glutathione reductase which is the best characterized (Kuπyan et a l . , (1991) Nature, 352, 172-4; Waksman et al . , (1994) J. Mol. Biol., 236, 800-16). The two subumts of 35 kDa are smaller than the 50 kDa subumts of glutathione reductase from all species and different domain organization suggests functional convergent evolution of the two enzymes (Kuriyan et a l . , (1991) Nature, 352, 172-4). The structural features of TrxR from E . coli and a narrow substrate specificity for its homologous Trx are typical also for TrxR from other prokaryotes, lower eukaryotes like yeast, or cytosol from plants (Dai et al . , (1996) J. Mol. Biol., 264, 1044-1057; Kuπyan et al . , (1991) Nature, 352, 172-4; Waksman 1" et al . , (1994) J. Mol. Biol., 236, 800-16; Williams (1992) CRC Press, Boca Raton, FL, Vol. 3, pp. 121-211). Mammalian TrxR, on the other hand, is strikingly different from that of lower organisms.

l TrxR from calf liver and thymus and rat liver were first purified to homogeneity and showed subumts with a molecular mass of 58 kDa (Holmgren (1977) J. Biol. Chem., 252, 4600-6; Luthman and Holmgren (1982) Biochemistry, 21, 6628-33), thereby being larger than the E . coli enzyme. The substrate

2C specificity of the mammalian enzyme is qualitatively different from the highly specific prokaryote type of TrxR and is surprisingly wide. Mammalian TrxR reduces not only thioredoxm from all species, but also disulfides in other protems like prote disulfide isomerase or NK-lysm, low 5 molecular weight disulfides like 5, 5 ' -dithiobis (2- trobenzoic acid) (DTNB, Ellman's reagent) or lipoic acid, low molecular weight non-disulfide substrates like selemte or alloxan, or, addition, lipid hydroperoxides (for references, see (Arner et al . , (1998) Meth. Enzymol., 300: 0 226-239) . Recently, it was found that mammalian TrxR is a selenoprotem, with the selenocysteine residue smuated in the carboxytermmal motif -Gly-Cys-Sec-Gly (Gladyshev et al . , (1996) Proc. Natl. Acad. Sci . (USA), 93, 6146-6151; Tamura and Stadtman (1996) Proc. Natl. Acad. Sci. (USA), 93, 5 1006-1011; Zhong et al . , (1998) J. Biol. Chem., 273,

8581-8591) . The selenocysteine residue in this motif is essential for enzymatic activity (Nordberg et al . , (1998) J. Biol. Chem., 273, 10835-10842; Zhong et a l . , (1998) J. Biol. Chem., 273, 8581-8591) and this presumably easily accessible carboxytermmal redox active motif most likely explains the wide substrate specificity of the enzyme. Except for a carboxytermmal elongation of 16 ammo acids, carrying this motif, mammalian TrxR is highly homologous to the other pyridme nucleotide-disulfide oxidoreductases glutathione reductase, lipoamide dehydrogenase or mercuric ion reductase whereas the homology to the prokaryotic TrxR is vague (Zhong et al . , (1998) J. Biol. Chem., 273, 8581-8591).

Brief Descripti on of the Fi gures

Figure 1 Structure of SECIS elements.

Figure 1A depicts, for comparison, the structure and key functional features of the SECIS element of formate dehydrogenase H from Escheπ chia coli .

Figure IB shows the SECIS cassettes used for construction of the pET-TRS and pET-TRSTER plasmids, encompassing the 3 ' -end of the TrxR open reading frame, the segment for the SECIS element (nucleotides in capital letters) and the BseAI and Sail sticky ends used for ligation with the TrxR open reading frame and the pET24(d)+ vector, as indicated. The ammo acid sequence resulting from this region of the inserts is given above in three-letter codes, with the selenocysteine codon and residue in bold letters and the introduced stop codons of pET-TRSTER being indicated by a box. Shown are also the resulting SECIS elements of the mRNA, manually folded and with circles indicating nucleotides that have been changed from those of the native SECIS element.

Figure 2 shows variants and ammo acid constraints of SECIS elements for targeted insertion of selenocysteine to the internal part of recombinant selenoprotems to be produced in E . coli . Variant SECIS elements that may be used for selenocysteine insertion carry constraints to the possible codon usage. As further discussed elsewhere herein, and also indicated in Figure 1A, the loop region encompassing 17 nucleotides binds the SELB elongation factor and should be positioned on the 3 ' -side 11 nucleotides from the UGA codon. However, the stem can be lengthened with at least 3 bp still maintaining 50% efficiency of selenocysteine insertion and the actual nucleotide bases of the stem are not of functional importance. The figure shows variants of SECIS elements, conforming to these functional restrains of the SECIS element in E . coli , that may be used in targeted insertion of selenocysteine to the internal part of recombinant selenoprotems with the alternative ammo acid sequences resulting from these variants given in three-letter codes. Additional changes of the SECIS element may be employed to further increase the ammo acid sequences translated carboxyterminally of the selenocysteine residue. Further discussion is included below.

Figure 3 shows a variant SECIS positioned withm the coding sequence for human GPxl. The nucleotides corresponding to the loop region of the SECIS are shown underlined in the upper nucleotide sequence, with the resulting ammo acids given underlined below the native sequence, shown in one- letter ammo acid codes.

One aspect of the present invention provides a nucleic acid comprising a coding sequence for a mammalian or other eukaryotic prote , or other heterologous prote whether low eukaryal, archeal or non-coli bacterial, and a bacterial selenocysteine insertion sequence (SECIS) for insertion of selenocysteine mto the protein on production of the protem by expression from the nucleic acid in a bacterial host cell. A further aspect of the invention provides a nucleic acid comprising or consisting of a bacterial selenocysteine insertion sequence (SECIS) including a TGA codon followed by one or more codons modified to encode one or more ammo acids which follow a selenocysteine residue in a mammalian or other heterologous protem. A mammalian or other heterologous selenoprotem may be identified, chosen or selected and ammo acids noted which follow a selenocysteine in the selenoprotem. Appropriate codons may then be chosen for inclusion in the modified SECIS following the TGA codon. The number of codons following the TGA that encode an ammo acid may be from none (by following the TGA with a termination codon such as TAA) , 1, 2, 3, 4 or more, 5, 6, 7, 8, 9, 10, 11 or more, limited by the size of the encoding sequence to be used. The first three to four codons following the selenocysteine-encoding TGA may be designed to be any of choice, whereas the codons positioned four or five to ten or eleven may be restricted to a choice between certain predetermined codons, as discussed further below with reference to Figure 2. A SECIS according to the present invention may be provided at the 3 ' -end (sense strand) or with the coding sequence of nucleic acid encoding a mammalian polypeptide, generally a selenoprotem.

Examples of selenoprotems include any mammalian selenoprotem, such as cellular glutathione peroxidase (Rotruck et al. (1973) Science 179, 588-590; Shen et al. (1993) J. Biol. Chem. 268, 11463-11469), plasma glutathione peroxidase (Takahashi et al. (1987) Arch. Biochem. Biophys. 256, 677-686), gastrointestinal glutathione peroxidase (Chu et al. (1993) J. Biol. Chem. 268, 2571-2576), phospholipid hydroperoxide glutathione peroxidase (Ursini et al. (1985) Biochim. Biophys. Acta 839, 62-70), lodothyromne deiodmase of type I (Berry et al. (1991) Nature 349, 438-440), type II (Davey et al. (1995) J. Biol. Chem. 270, 26786-26789) or type III (Croteau et al. (1995) J. Biol. Chem. 270, 16569-16575), selenoprotem W (Vendeland et al. (1993) J. Biol. Chem. 268, 17103-17107), thioredoxm reductase (Gladyshev et al. (1996) Proc. Natl. Acad. Sci. (USA) 93, 6146-6151; Zhong et al. (1998) J. Biol. Chem. 273, 8581-8591) or its mitochondπal isoenzyme (Lee et al. (1999) J. Biol. Chem. 274, 4722-4734; Miranda-Vizuete et al. (1999) Eur. J. Biochem. 261, 405-412), selenophosphate synthetase (Guimaraes et al. (1996) Proc. Natl. Acad. Sci. (USA) 93, 15086-15091), selenoprotem P (Read et al. (1990) J. Biol. Chem. 265, 17899-17905) or the recently discovered 15 kDa selenoprotem (Gladyshev et al. (1998) J. Biol. Chem. 273, 8910-8915) with unknown function. Other heterologous selenoprotems may include those found in archea, such as Methanococcus j annaschn , including the archeal formate dehydrogenase, heterosulphide reductase, formyl-methanofuran-dehydrogenase, F₄₂₀-reducιng hydrogenase, selenophosphate synthetase and the two selenoprotem subumts of the F₄₂₀-non-reducmg hydrogenase (Wilting et al. (1997) J. Mol. Biol. 266, 637-641), and thioredoxm reductase of Caenorhabdi tis elegans (Buettner et al . (1999) J. Biol. Chem. 274, 21598-21602), selenoprotem A of Clostridiae (Cone et al. (1976) Proc. Natl. Acad. Sci. (USA) 73, 2659-2663 or the peπplasmic hydrogenase large subumt of Desulfomicrobi um bacula tum (Tormay & Bock (1997) J. Bacteriol. 179, 576-582).

One preferred selenoprotem employed m embodiments of the present mvention is thioredoxm reductase. As explained below, this protem is in demand and its production m useful quantities in accordance with the present invention overcomes difficulties associated with current techniques for purification (of small amounts) of the prote from animal tissue .

A SECIS at the 3'-end, for instance one such as used experimentally below to provide -Gly-Cys-Sec-Gly at the C- terminus of the encoded polypeptide, may be used to introduce a selenocysteine residue mto a non-selenoprotem . This provides a valuable way of labelling a non-selenoprotem with selenium. Zhong et al . , (1998) J. Biol. Chem., 273, 8581-8591 showed that the oxidized carboxyterm us of thioredoxm reductase is very stable. Introducing a selenocysteine and/or replacing a cysteme with selenocysteine provides for enhanced resolution in X-ray crystallography, for PET studies, for probing redox activities and for introduction of highly energetic selenium isotopes (e.g. ⁷⁵Se) to enable radiochemical methods for the protem of interest. The use of a SECIS in accordance with the present invention for providing a selenocysteine residue in a non-selenoprotem, especially for labelling or tagging the protem at or towards the C-termmus, is provided as a further aspect of the present invention.

Methods for production of a mammalian selenopolypeptide from encoding nucleic acid in bacterial host cells, production and use of suitable host cells, and other manipulations of nucleic acid are provided by further aspects of the present invention, discussed further below.

A SECIS according to the present invention may include codons for desired ammo acids to follow a selenocysteine m a chosen mammalian protem. These need not be the precise ammo acids that occur naturally in the relevant mammalian protem following the selenocysteine.

In designing a SECIS in which ammo acids are to be encoded through the SECIS (because the selenated residue is withm the polypeptide rather than C-termmal), account may be taken not only of the degeneracy of the genetic code but also possibilities for substitution of one or more ammo acids without compromising, or unduly affecting, activity of the relevant polypeptide it is desired to produce, especially by employing "conservative variation", i.e. substitution of one hydrophobic residue such as lsoleucme, valme, leucine or methionme for another, or the substitution of one polar residue for another, such as argmme for lysme, glutamic for aspartic acid, or glutamme for asparagme. Note that it is not an absolute requirement of the present invention that the selenoprotem is functional following its production, nor that function is the same or to the same degree as corresponding naturally occurring selenoprotem (i.e. not recombmantly produced) . However, the experimental results herein show that it is possible to produce an enzymaticallv functional mammalian selenoprotem m bacteria, and for the selenoprotem product to show activity to some degree is preferred in various embodiments of the present invention.

In addition to including a selenocysteme-encodmg TGA codon, one or more codons encoding ammo acids to follow a selenocystein residue in a desired mammalian selenoprotem to be produced, a SECIS according to the present invention must include nucleotides providing for interaction with the host cell selenocysteine insertion machinery, specifically SELB protem, and preferably includes nucleotides encoding ammo acids which follow the selenocysteine in the relevant product protem, which ammo acids preferably are the same as in native protem or correspond to conservative substitution.

Two natural E . coli SECIS elements are available, that of formate dehydrogenase 0 and N and that of formate dehydrogenase H. The loop segment of 17 nucleotides binds

SELB protem, naturally positioned 11 nucleotides downstream of the UGA codon but adjustment to remove at least one, and maybe two or three nucleotides, may be made without loss of function. Also, this region between the UGA and the loop can be extended by at least one, two or three nucleotides. The loop regions of formate dehydrogenases N and 0 on the one hand and formate dehydrogenase H on the other are different, thereby providing for encoding different ammo acid sequences. Further modification may be made to the loop region to encode one or more different ammo acids provided a corresponding mutation is made in the SELB protein to retain functional interaction.

Exemplary embodiments of SECIS according to the present invention are shown m Figure IB, Figure 2 and Figure 3.

Figure 3 shows a variant SECIS positioned within the coding sequence for human GPxl. The underlined ammo acids are encoded by the loop region (also underlined) of the SECIS element. This region of the protem is variable between species (Ursim et al, 1995, Methods Enzymol 252, 38-53) , not involved in the active site and shown by crystal structure to correspond to a structural α-helix (Epp et al., 1983, Eur. J. Biochem. 133: 51-69). The mutant produced using the variant SECIS is predicted to have hydrophobicity and secondary structure not significantly different from the native enzyme. This illustrates use of a SECIS to introduce a selenocysteine into a selenoprotem at a site far from the carboxytermmal end.

In a further embodiment of the invention, a SECIS element is used to provide a portion of a selenoprotem, with the selenocysteine being mcorporated at or adjacent to the C- termmus . Self-splicmg elements such as interns (Mathys et al. (1999) Gene, 231: 1-13) or other techniques may be used to join the portion to the remaining carboxytermmal portion of the protem to provide a product incorporating the selenocysteine residue internally. Generally whether providing just a SECIS or a SECIS withm a construct, such as a construct encodmg a mammalian selenoprotem (such as thioredoxm reductase) in accordance with the present invention, nucleic acid or a polynucleotide is provided as an isolate, in isolated and/or purified form. Nucleic acid encoding a mammalian selenoprotem may be provided free or substantially free of material with which it is naturally associated, such as free or substantially free of nucleic acid flanking the gene in the relevant mammalian genome, except possibly one or more regulatory sequence (s) for expression. Note that nucleic acid encodmg a mammalian selenoprotem and a SECIS according to the present invention does not exist m nature. Nucleic acid may be wholly or partially synthetic and may include genomic DNA, cDNA or RNA. Where nucleic acid according to the invention includes RNA, reference to the sequence shown should be construed as encompassing reference to the RNA equivalent, with U substituted for T, except where context demands otherwise.

Nucleic acid sequences in accordance with the present invention can be readily prepared by the skilled person using the information and references contained herein and techniques known in the art (for example, see Sambrook, Fπtsch and Maniatis, "Molecular Clonmg, A Laboratory Manual, Cold Spring Harbor Laboratory Press (1989), and Ausubel et al . , Current Protocols in Molecular Biology, John Wiley and Sons, (1992)). These techniques include (l) the use of the polymerase chain reaction (PCR) to amplify samples of nucleic acid, (n) chemical synthesis, or (m) preparing cDNA sequences. DNA encoding a protem of interest or a portion thereof may be generated and used in any suitable way known to those of skill in the art, including by taking encoding DNA, identifying suitable restriction enzyme recognition sites either side of the portion to be expressed, and cutting out said portion from the DNA. The portion may then be operably linked to a suitable promoter in a standard commercially available expression system. Another recombinant approach is to amplify the relevant portion of the DNA with suitable PCR primers. Modifications to the relevant sequence may be made, e.g. using site directed mutagenesis, to lead to the expression of modified peptide or to take account of codon preference in the host cells used to express the nucleic acid.

In order to obtain expression of the nucleic acid sequences, the sequences may be incorporated in a vector having one or more control sequences operably linked to the nucleic acid to control its expression. The vectors may include other sequences such as promoters to drive the expression of the inserted nucleic acid, nucleic acid sequences so that the polypeptide or peptide is produced as a fusion and/or nucleic acid encoding secretion signals so that the polypeptide produced in the host cell is secreted from the cell. Polypeptide can then be obtained by transforming the vectors mto host cells in which the vector is functional, culturing the host cells so that the polypeptide is produced and recovering the polypeptide from the host cells or the surrounding medium. Prokaryotic cells useful in embodiments of the present invention include E . coli , and strains such as B121 (DE3) which have T7 polymerase may be preferred for ease of overexpression . Cells may be employed in which reside mutant SELB, extending possibilities for SECIS design, given the need for interaction between the SECIS and SELB.

The present invention further provides a method of making a mammalian selenopolypeptide or peptide (as disclosed) , the method including expression from nucleic acid encoding the polypeptide or peptide withm bacterial cells. This may conveniently be achieved by growing host cells m culture, containmg such a vector, under appropriate conditions which cause or allow expression of the polypeptide. For production of a selenoprotem selemte and cysteme are included in the culture medium.

A further aspect of the present invention provides a bacterial host cell containing nucleic acid according to the present invention as disclosed herein.

A still further aspect provides a method which includes introducing the nucleic acid mto a host cell. The introduction, which may be generally referred to without limitation as "transformation", may employ any available technique. Suitable techniques may include calcium chloride transformation, electroporation and transfection using bacteπophage .

Marker genes such as antibiotic resistance or sensitivity genes may be used in identifying clones containing nucleic acid of interest, as is well known in the art.

In most preferred embodiments of the invention the host cell overexpresses selA, selB and selC sequences, and this may be as the result of transformation with selA, selB and selC genes, either prior to transformation with the nucleic acid encoding the mammalian protem and SECIS or by cotransformation . SelD may also be overexpressed.

The introduction of nucleic acid including a SECIS and sequence encoding a mammalian protem may be followed by causmg or allowing expression from the nucleic acid, e.g. by culturing host cells (which may mclude cells actually transformed although more likely the cells will be descendants of the transformed cells) under conditions for expression of the gene, so that the encoded polypeptide (or peptide) is produced. If the polypeptide is expressed coupled to an appropriate signal leader peptide it may be secreted from the cell mto the culture medium. Following production by expression, a polypeptide or peptide may be isolated and/or purified from the host cell and/or culture medium, as the case may be, and subsequently used as desired, e.g. in the formulation of a composition which may include one or more additional components, such as a pharmaceutical composition which includes one or more pharmaceutically acceptable excipients, vehicles or carriers.

In embodiments wherem thioredoxm reductase is produced, this may be employed as desired, for instance in structural studies, activity-based assays, screening for novel substrates or inhibitors, raising antibodies, kits for assays for detection of Trx, and so on.

Recently it was shown that C. elegans contains a thioredoxm reductase with the same selenocyste e-contammg carboxytermmus as the rat thioredoxm reductase utilised herein (Buettner et al . , J. Biol. Chem. (1999) 274(31): 21598-602) and a second mammalian mitochondrial thioredoxm reductase has been found, containing the same carboxytermmus (Lee et al . , J. Biol. Chem. (1999) 274(8): 4722-34). These enzymes may be produced and studied using the technique of the present invention with significantly higher efficiency than purification of the endogenous nematode or mitochondrial enzymes. Moreover, using the invention it is now possible to produce mutant enzymes still carrying the selenocysteine residue but bemg mutated in other key residues. Without the mvention such production of mutants would require purification from transgemc animals or from transfected mammalian cell lines and these alternatives are both more expensive and time-consuming and are much less effective in terms of yield. A polypeptide produced in accordance with the present invention, or a nucleic acid molecule as disclosed, e.g. encoding a mammalian selenoprotem, may be provided in a kit, e.g. sealed in a suitable container which protects its contents from the external environment. Such a kit may mclude instructions for use.

Various further aspects and embodiments of the present invention will be apparent to those skilled in the art m view of the present disclosure. Certain aspects and embodiments of the invention will now be illustrated by way of example and with reference to the sequences discussed already above.

EXPERIMENTAL RESUL TS

The work of the present inventors involved design of novel variants of the bacterial SECIS element, their hope being production of a mammalian selenoprotem (exemplified with rat thioredoxm reductase) , using the bacterial selenocysteine insertion machinery.

The present inventors fused the open reading frame of rat selenoprotem thioredoxm reductase (TrxR) with the selenocysteine msertion sequence (SECIS element) of formate dehydrogenase H from Escherichia coli , demonstrating for the first time that heterologous expression of mammalian selenoprotems is possible in bacteria. A variant of the SECIS element, encoding the carboxytermmal end of TrxR (-Sec-Gly-COOH) and positioning the E . coli SELB binding motif in the 3 ' -untranslated region, did not impair selenium incorporation and thereby enabled production of enzymatically active enzyme. The selenocysteine-contammg recombinant proteins were produced at dramatically higher levels than endogenous formate dehydrogenase 0, demonstrating a high reserve capacity of selenoprotem production m E. coli . Co-transformation with the selA, selB and selC genes further increased the activity of the recombinant rat TrxR, to a specific activity about 25% of the native enzyme, with 20 mg recombinant mammalian TrxR produced per liter of culture. The results also reveal that the bacterial SECIS element can direct selenocysteine incorporation when the element is situated in the 3 ' -untranslated region, in proper position to the UGA codon.

The carboxytermmal end of mammalian thioredoxm reductase is -Gly-Cys-Sec-Gly-COOH. A variant of the bacterial SECIS element was changed to encode Gly after the selenocysteine, followed by a tandem UAA for translational termination. The conserved loop region of the SECIS element was maintained, to enable bmdmg of the SELB protem (Huttenhofer et al . , (1996) RNA, 2, 345-366; Kromayer et al . , (1996) J. Mol. Biol., 262, 413-420; Liu et al . , (1998) Nucleic Acids Res., 26, 904-910) .

Plasmids were constructed using the pET-24d(+) vector

(Novagen) , with the pET-TR plasmid containing only the rat TrxR open reading frame, pET-TRS containing the fusion to a native formate dehydrogenase H SECIS element and pET-TRSTER carrying the fusion to a mutated SECIS element, as shown in Figure 1.

Overexpression and selenium incorpora tion The inventors examined whether the gene fusions of the mammalian TrxR open reading frame with the variants of the SECIS element of formate dehydrogenase H fused to the 3 ' -end would direct selenocysteine insertion mto the resulting recombinant proteins.

It was found that this was the case. The pET-seπes of vectors with the T71ac promoter are highly stringent, with very little basal expression in the absence of IPTG but with remarkably high overexpression when induced (Dubendorff and Studier (1991) J. Mol. Biol., 219, 45-59). Accordingly, with 1 μM IPTG in the medium no recombinant protem could be seen in whole cell extracts, whereas cells induced with 50 μM or 1 mM IPTG for 1 - 4 h at 37°C contained high amounts of recombinant protem produced from the pET-TR, pET-TRS and pET-TRSTER plasmids, at about the same level for all three plasmids. The selenocysteine incorporation was also analyzed using ⁷⁵Se-selemum labelling.

The BL21(DE3) strain of Escheri chia coli was transformed with the plasmids and the transformants were grown aerobically at 37°C in the presence of ⁷⁵Selemte, induced with various concentrations of IPTG and harvested after 1, 2, 3 or 4 hours. Comassie-stamed 10% SDS-PAGE gels and autoradiograms of the same gels were examined.

Based upon the SDS-PAGE gels and autoradiograms, several conclusions could be drawn regarding the selenium incorporation mto the recombinant proteins:

(l) Gene fusions with the variants of the SECIS element were absolute requirements for selenium incorporation to occur, whereas there was no apparent difference in selenium incorporation between the two SECIS variants.

(n) The selenium incorporation machinery could efficiently be utilized in this system also under aerobic conditions.

(m) The selenium incorporation into the over-produced recombinant proteins greatly exceeded the basal selenium incorporation into formate dehydrogenase 0, the only endogenous E . coli selenoprotem produced under the aerobic conditions utilized.

(iv) There seemed to be a competition between selenium incorporation mto formate dehydrogenase 0 and the recombmant protem products of pET-TRS and pET-TRSTER. (v) More selenium incorporation was seen using 50 μM IPTG as compared to induction with 1 mM IPTG. (vi) There was a tendency to a decrease in selenium incorporation mto selenylated tRNAGlu and tRNALys (Lemfelder et al . , (1990) Proc. Natl. Acad. Sci., 87,

543-547; Wittwer and Stadtman (1986) Arch. Biochem. Biophys., 248, 540-550) with increasing induction of the recombinant protems of pET-TRS or pET-TRSTER.

Most of the highly overexpressed protem was found the insoluble pellet indicating the formation of inclusion bodies, with no difference in relative selenium content between insoluble and soluble protem. Induction of expression at 16°C, however, resulted in full solubility of the over-produced protein. Analysis of human Trx-dependent reduction of insulin disulfides - a highly specific assay for mammalian TrxR - revealed that only pET-TRSTER and not pET-TR or pET-TRS produced enzymatically active protem (Table I), although all three plasmids gave rise to the equal amount of produced recombinant protem, as judged by SDS-PAGE.

The inventors then co-transformed the pET-TRSTER plasmid with either a plasmid (pSUABC) carrying the selA, selB and selC genes under the control of their endogenous promoters, or a plasmid (pMN302) carrying the selD gene under control of its endogenous promoter. Time-dependent expression at different temperatures was then analyzed via the formation of soluble enzymatically active protem.

GenBank Accession numbers:

selA - M64177 selB - U00039 sel C - Y00299 selD - M30184

Co-transformation with the pSUABC plasmid was found to increase the production of enzymatically active protem about 5- to 6-fold, whereas the pMN302 plasmid had no apparent effect (Table II) . There was no apparent difference in the total amount of recombinant protem produced from the pET-TRSTER plasmid with or without co-transformation with any of the other two plasmids, when cells from the same growth conditions were analyzed using SDS-PAGE. This indicated that not the amount but rather the specific activity of the recombinant enzyme was increased by the co-transformation with the pSUABC plasmid, most likely due to a more efficient selenocysteine incorporation. To analyze if this was the case, the recombinant protem products of the pET-TRSTER plasmid induced with or without co-transformation with the pSUABC plasmid were then purified and the properties of this protem compared with that produced from the pET-TR and pET-TRS plasmids.

Enzyme purifi ca tion and a ctivi ty

Since the recombinant protem was highly over-produced, S100 extracts were directly used in 2 ' , 5 ' -ADP-Sepharose purification, which is a specific affinity purification for NADP (H) -binding protems.

This yielded >98% pure recombinant protem for both the products of pET-TRSTER, named rTrxR, pET-TRSTER co-transformed with pSUABC, named rTrxR(ABC), as well as the products of pET-TR (named rTR) and pET-TRS (named rTRS) . The latter showed that although rTR and rTRS lacked enzymatic activity, the soluble protems still had NADP (H) -bmdmg capacity indicating rightly folded subumts.

The recombinant protems purified also contained FAD at about 1 mol FAD per mol subumt and showed absorption spectra typical for TrxR (Arscott et al . , (1997) Proc. Natl. Acad. Sci. (USA), 94, 3621-3626; Luthman and Holmgren (1982) Biochemistry, 21, 6628-33) .

As expected from analysis of whole cell extracts (Table I), the purified rTR or rTRS had very low enzymatic activity using hTrx or selemte as substrates. However, both proteins showed some NADPH dependent reduction of DTNB (0.35 U/mg for rTR and 0.31 U/mg for TRS) . Both the purified rTrxR and rTrxR(ABC), however, carried significant enzymatic activity and were analyzed in further detail for the reduction of a number of known substrates of mammalian TrxR.

In reduction of DTNB, hTrx, selemte or lipoic acid, rTrx(ABC) showed 5.9- to 8.8-fold higher activity than rTrxR under the same assay conditions (Table III). The 7.7 U/mg found for rTrxR(ABC) in the model DTNB assay (Table III) was 22% of the previously highest reported specific activity of 35 U/mg for the purified mammalian enzyme (Gromer et al . , (1998) J. Biol. Chem., 273, 20096-20101; Luthman and Holmgren (1982) Biochemistry, 21, 6628-33). When kinetic parameters of rTrxR(ABC) were compared with those of native purified rat liver TrxR, rTrxR(ABC) showed correspondingly lower k_cat values for hTrx, DTNB or lipoic acid whereas K_^ values were in the same range (Table IV) .

These findings indicated that the rTrxR and rTrxR(ABC) products fact contained both inactive and active forms of the enzyme, with rTrxR(ABC) containing more of the active form than rTrxR. The difference between inactive and active forms of the enzyme might be due to different efficiency of selenocysteine insertion and/or termination of translation at the UGA codon. In an attempt to further clarify this question the different recombinant proteins were subjected to peptide analysis using mass spectrometry .

Mass spectrometry

The peptide analysis of rTR, rTrxR(ABC) and rTRS confirmed the high purity of all three purified protems, with about 25 out of 27 to 30 masses detected corresponding to peptides derived from the protems.

For both rTrxR(ABC) and rTRS peptide masses were found suggesting the presence of both full length protem and protem truncated just before the selenocysteine residue.

Interestingly, a peptide with a mass of 1197.7 that corresponded to the carboxytermmal peptide truncated just before the position of the selenocysteine residue was found in the peptide digest of all three recombinant protems. This showed that the presence of the SECIS element m junction with the UGA codon, did not totally inhibit its function as a termmation codon. The presence of the truncated protem in the rTrxR(ABC) sample, being identical to the enzymatically inactive rTR protem, should explain the lower specific activity of rTrxR(ABC) as compared to native purified enzyme (Table IV) .

In addition, a peptide with a mass of 1510.8 was also detected in the peptide digest of rTrxR(ABC), corresponding to the full length protem, which should be the enzymatically active form. This also demonstrated that the tandem UAA termination codons in the stem of the SECIS element of pET-TRSTER (Figure IB) were fully functional.

Ammo acids withm the consensus sequence were found to be different from the previously reported rat TrxR sequence (Zhong et al . , (1998) J. Biol. Chem., 273, 8581-8591) as explained below. Few compounds with masses above 500 were detected that could not be assigned to individual peptides of the recombinant protems.

No peptide with a mass corresponding to non-selenocysteine mediated UGA suppression was found in any of the samples.

Although the rTRS protem lacked enzymatic activity, it showed selenocysteine incorporation and the peptide with a mass of 1409.9 was derived from the 6xHιs-tagged carboxytermmus of the full length rTRS, encoded by the region of the pET-vector flanking the 3 '-side of the insert, illustrating the presence of full length protem also in this sample .

DISCUSSION The present work shows success in overproduction of mammalian selenoprotem in E . col i . The results unequivocally show that the presence of a bacterial-type SECIS element ust following the selenocysteine-encodmg UGA in the mRNA is the only prerequisite for selenocysteine msertion in E . coli , and that this can be utilized for recombinant selenoprotem synthesis. Interestingly, the results also show that the bacterial SECIS element does not need to be translated for selenocysteine insertion to occur and that a UAA termination codon is not suppressed but operative also when positioned in the stem part of the SECIS element.

It is not clear how the UGA codon at the base of a SECIS element is suppressed, i.e. how release factor 2 (RF2) is prevented to terminate translation at the position of the selenocysteine residue. Also, details of the general mechanism of translation termination are not fully understood although newly developed m vi tro studies have given rise to new msigths, eg. how RF3 accelerates the release of RF1 or RF2 bound at the πbosome ( Freistroffer et a l . , (1997) EMBO J., 16, 4126-4133). The context of stop codons affect the efficiency of termination and using these newly developed m vi tro studies, it was confirmed in line with previous data that the most efficient context of a UAA codon for RF1 function is UAA(U) (Pavlov et al . , (1997) J. Mol. Biol., 273, 389-401) . Here it should be noted that both of the tandem UAA codons of the SECIS variant in the pET-TRSTER construct were in a UAA(U) context (Figure IB) . The mass spectrometry analysis showed that the presence of the SECIS element did not totally prevent termination of translation at the selenocysteme-encoding UGA codon. This was most likely due to the over-expressing system utilized and agrees well with the previous finding that when the gene for the 80 kDa formate dehydrogenase N was over-expressed in E . coli , the major part of the protem was produced as a 15 kDa polypeptide, corresponding to termination of translation at the UGA selenocysteine codon (Leinfelder et al . , (1988a) J. Bact . , 170, 540-546). The increase by cotransformation with the SelA, SelB and SelC genes of the specific activity of over-produced rTrxR points to importance of the stoichiometry between selenoprotem mRNA and the Sel family of gene products m selenoprotem synthesis in E coli .

The inventors were surprised by the initial yield of about 20 mg/L of rTrxR(ABC), with about 5 mg/L being enzyme in an active form (as judged from the specific activity), although an earlier study had indicated that E . coli cells have a high capacity of endogenous selenoprotem synthesis (Chen et al . , (1992) Mol. Microbiol., 6, 781-785). In that study, formate dehydrogenase H was overproduced but was found to be synthesized only under anaerobic conditions. The reason for this must have been that the expression of the recombinant formate dehydrogenase H was under control of its anaerobically active endogenous promoter (Chen et al . , (1992) Mol. Microbiol., 6, 781-785), since clearly the results here show there is no need for anaerobiosis for the function of the selenocysteine translation machinery per se .

The yield of about 5 mg active rTrxR per liter culture should be compared to the more laborious purification of about 1-2 mg of enzyme (albeit without the presence of inactive forms) per kg animal tissue, the only source of mammalian thioredoxm reductase to date. This makes the method presented here worthwhile as an alternative for production of mammalian TrxR. More important, however, is that utilizing the technique presented here enables constructing and study of mammalian TrxR with point mutations at other residues than the selenocysteine, such as its neighbouring cysteme, the thiols in the N-termmal redox active site, or the histidme proposed to act as a base in the active site (Zhong et al . , (1998) J. Biol. Chem., 273, 8581-8591). These studies should certainly aid in our further understanding of this enzyme, that plays a central role in the cellular defence agamst oxidative stress and in redox regulation, and also is a target of several drugs in clinical use (Arner et al., 1995, J. Biol. Chem. 270: 3479-3482; Gromer et al . , (1998) J. Biol. Chem., 273, 20096-20101; Gromer et al . , (1997) FEBS Lett., 412, 318-320; Mau and Powis (1992) Biochem. Pharm. , 43, 1613-20; Powis et al . , (1995) Pharmac. Ther., 68, 149-173; Schallreuter et al . , (1990) Biochem. Biophys. Acta, 1054, 14-20; Schallreuter and Wood (1989) Biochem. Biophys. Res. Commun. , 160, 573-9) .

In thioredoxm reductase the selenocysteine residue is necessary for the catalytic properties of the enzyme and is positioned at the carboxytermmus. This gave no other restraint to the design of the SECIS element following the open reading frame, in terms of encoded ammo acids, than introduction of a glyc e codon followed by termination codons. A similar strategy as utilized here may be used for production of other mammalian selenoprotems, which contain the selenocysteine residue withm the open reading frame. The only constraint in this case would be the length of the stem and preservation of the conserved loop nucleotides in the SECIS element (see Figure 1A) . Moreover, the inventors have found that a -1 frame shift due to a one nucleotide shorter stem does not alter the efficiency of selenocysteine incorporation, and a stem lengthened by three base pairs still results in selenium incorporation with 50% capacity compared to the normal SECIS element (Heider et al . , (1992) EMBO J., 11, 3759-3766). Also, the nucleotides in the lower part of the SECIS element must not necessarily form a true base-pairing stem for its function, as long as the loop is positioned at the right distance from the UGA codon (Liu et al . , (1998) Nucleic Acids Res., 26, 904-910). This confers a flexibility for the design of SECIS elements to be used for selenocysteine insertion withm the open reading frame of a recombinant selenoprotem, although the nucleotides encoding five to six ammo acids positioned three to four residues carboxytermmally of the selenocysteine residue have to be conserved (Figure 2) .

A SECIS element positioned in an untranslated region of the selenoprotem mRNA, as in the mammalian (Low and Berry (1996) TIBS, 21, 203-208) and probably archaeal (Wilting et al . , (1997) J. Mol. Biol., 266, 637-641) organisms, eliminates the restriction on the structure imposed by the fact that the nucleotides of an m-frame SECIS element, as in bacteria, also need to encode ammo acids. Interestingly, the variant of the bacterial SECIS element in the pET-TRSTER plasmid, with the SELB-bmdmg loop in the untranslated region, was here found to be functional. This revealed that the bacterial SECIS element does not need to be translated for selenocysteine insertion to occur, which in turn could give a clue to the evolution of the SECIS elements of the mammalian or archaeal type. Possibly, a bacterial SECIS element containing a termination codon in the stem evolved from an m-frame SECIS element, thereby producing a protem with a selenocysteine residue close to the carboxytermmal end. If this protem carried some oxidoreductase activity and was of advantage for growth, the structure would be preserved. Once formed, the SECIS element with a termination codon in the stem would be relieved of the evolutionary constraint of preserving particular ammo acid-encodmg nucleotides. This could then be envisioned to evolve into a SECIS element positioned further away in the untranslated region, with concurrent evolution of the SECIS-bind g protem (s) needed for its function. This hypothesis could of course be strengthened by any future finding of a naturally occurring bacterial SECIS element with a termination codon in its stem part.

MATERIALS AND METHODS

Ma terials and general methods Human wild-type and mutant C61S/C72S thioredoxm and placental TrxR were prepared as described by Ren et al . ,

(1993) Biochemistry, 32, 9701-8). ⁷⁵Se-selemte (1.85 mCi/ml, 119 μg selemum/ml) came from Amersham international pic.

(Buckinghamshire) and bovine pancreas insulin was bought from Sigma Chemical Co. (St Louis, MO.) and prepared as described

(Arner et al . , (1998) Meth. Enzymol., in press). Oligonucleotides were ordered from GibcoBRL/Life Technologies or MWG-BIOTECH, with standard purity for PCR or sequence reactions and cartridge purified for construction of SECIS cassettes. The pET-2 (d) + vector came from Novagen Inc.

(Madison, WI . ) and the pGEM-T vector was from Promega Corp.

(Madison, WI . ) . Enzymes were bougnt from Boehringer Mannheim GmbH (Mannheim), Promega Corp. (Madison, WI . ) , Pharmacia Biotech (Uppsala), MBI Fermentas (Vilnius), New England BioLabs Inc. (Beverly, MA.) or GibcoBRL/Life Technologies Inc. (Ga thersburg, MD. ) and were used according to manufacturers' instructions. Sequencing reactions were performed on the A.L.F. Sequencer (Pharmacia Biotech, Uppsala) or an ABI Automated Sequencer (Applied Biosystems/Perkm Elmer, Warrington) , according to protocols supplied by the manufacturers. Standard DNA clonmg and expression techniques were performed as described by Sambrook et al . (Sambrook et al . , (1989) Molecular clonmg, a laboratory manual. Cold Spring Harbour Laboratory Press, New York) or in the protocols provided with the vectors utilized.

Ba cterial s trains and growth condi tions

E . coli strains DH5a or JM109 were used for clon g and propagation of plasmids whereas expression studies were performed in the BL21(DE3) strain. All cultures were grown under aerobic conditions m Luria-Bertam medium (Sambrook et al . , (1989) Molecular clonmg, a laboratory manual. Cold Spring Harbour Laboratory Press, New York) with additions as given in the text. The medium was supplemented with ampicillm (100 μg/ml) for cells carrying the pGEM-T derived plasmids, with kanamycin (30 μg/ml) for cells transformed with the pET-24d( +) derived plasmids and with chloramphemcol (34 μg/ml) for cells containing the pSU- or pACYC184-derιved plasmids .

Construction of plasmids

The open reading frame of the rat TrxR cDNA clone previously described (Zhong et al . , (1998) J. Biol. Chem., 273, 8581-8591) was PCR amplified using the primers EASTU (5'-CTGTCAACcATGgATGACTCTAAAGATGCC-3' ) and EASTL

(5'-CTGGGGCTTAACCTCAGCAtCCgGACTGGAGG-3' ) . The nucleotides given in lower case letters conferred site directed mutagenesis to enable cleavage with appropriate restriction enzymes. EASTU introduced an Ncol cleavage site at the initiating ATG codon, which also changed the translated penultimate N-termmal ammo acid from Asn to Asp. EASTL introduced a BseAI cleavage site at the 3 ' -end of the open reading frame, being a silent mutation regarding the encoded ammo acids. The resulting 1.5 kb product of this PCR reaction was cloned mto a pGEM-T vector to yield the pTa plasmid, which was purified using the Plasmid Midi Kit from QIAGEN GmbH (Hilden) and the sequence of the whole insert was determined several times. The mutations introduced by the EASTU and EASTL primers were thereby confirmed. It should also be noted that compared to the previously reported sequence (Zhong et al . , (1998) J. Biol. Chem., 273, 8581-8591) T365 was found to be C (a silent change) and 1479-GCTTTGCAGCCA-1490 was instead found to be GGCTTTGCAGCCGCA (nucleotide numbering as in the previous report (Zhong et al . , (1998) J. Biol. Chem., 273,

8581-8591)), yielding the corresponding one ammo acid longer sequence GFAAA, instead of ALQP. Since both these differences are homologous to the human sequence (Zhong et al . , (1998) J. Biol. Chem., 273, 8581-8591), they most likely do not represent PCR amplification errors, but should be regarded as corrections of the sequence reported in the previous study (Zhong et al . , (1998) J. Biol. Chem., 273, 8581-8591).

For subsequent ligation with the insert of the pTa plasmid, two different SECIS cassettes were made using the following oligonucleotides :

S1A, 5 ' -CCGGATGCTGACACGGCCCATCGGTTGCAGGTCTGCA-3 ' ;

S2A, 5 ' -CCAATCGGTCGGTAATGCGGCCG-3 ' ;

SIB, 5 ' -CCTGCAACCGATGGGCCOTGTCAGCAT-3 ' ; S2B, 5'-TCGACGGCCGCATTACCGACCGATTGGTGCAGA-3' ;

SttlA, 5 ' -CCGGATGCTGAGGCTAATAATCGGTTGCAGGTCTGCA-3 ' ;

Stt2A, 5 ' -CCAATCGTTAGCCTATGCGGCCG-3 ' ;

SttlB, 5 ' -CCTGCAACCGATTATTAGCCTCAGCAT-3 ' ;

Stt2B, 5 ' -TCGACGGCCGCATAGGCTAACGATTGGTGCAGA-3 ' . In one tube S2A and SIB and in another tube Stt2A and SttlB (333 pmol each) were incubated with 10 units of T4 polynucleotide kmase m a total volume of 15 μl containing 50 mM Tris-Cl, pH 8.0, 10 mM MgC12, 5 mM DTT and 1 mM ATP for 30 mm at 37°C, to phosphorylate the 5 ' -ends . Then the polynucleotide kmase was inactivated by 20 mm incubation at 65°C whereafter S1A and S2B (333 pmol each, less than 2 μl of each) were added to the tube containing S2A and SIB (to make the "S" SECIS cassette) whereas SttlA and Stt2B (333 pmol each, less than 2 μl of each) were added to the tube containing Stt2A and SttlB (to make the "STER" SECIS cassette) . The tubes were then set in a 95°C water bath, whereupon the thermostate was changed to 37°C and the water was allowed to slowly attain 37°C. Then an additional 1 mM ATP and 6 mM DTT were added to the same tubes together with 2 Weiss units of T4 DNA ligase and ligation was allowed to occur at 37°C for 40 mm. The reactions were stopped by incubation at 65°C for 10 mm and the resulting SECIS cassettes were confirmed on 12% polyacrylamide gels and stored at -20°C until further use.

The SECIS cassettes hereby constructed carried the genes for the desired variants of the SECIS element from formate dehydrogenase H with compatible sticky ends for subsequent ligation with the insert in the pTa plasmid and transfer to the pET-24(d)+ vector, as described below and shown in Figure 1.

Prior to ligation with the SECIS cassettes, pTa was cleaved with BseAI and Sail, removing a small (30 bp) fragment between the BseAI site in the 3 ' -end of the insert and the Sail site of the pGEM-T vector, with the small fragment carrymg a unique Pstl cleavaqe site. In separate tubes, the two SECIS cassettes made above were at a 20 molar excess mixed with the cut plasmid (0.1 μg SECIS cassette to 0.5 μg cut plasmid) and ligation was carried out at 4°C for 4h using 2 Weiss units of T4 DNA ligase. The ligation products were then precipitated with ethanol and taken up in buffer for cleavage with Pstl, carried out using 5 units of Pstl at 37°C for lh. This counterselection thereby linearized all ligation products containing the small fragment and the whole mix was used for transformation of JM109 cells with electroporation usmg a Gene Pulser from BioRad (Hercules, CA. ) , which works efficiently only with circular plasmids.

Potentially positive clones were identified using resistance of plasmids to Pstl cleavage and were subsequently isolated and sequenced over the region with the SECIS cassettes, usmg a vector-based reverse primer ( 5 ' -AGG GGA TAA CAA TTT CAC ACA GGA-3'). The inserts of plasmids with a confirmed correct sequence as well as the insert of the original pTa plasmid were recovered by cleavage with Ncol and Sail with subsequent purification of the about 1.5 kb inserts from preparative 1% agarose gels. These fragments were directionally ligated mto the pET-24d(+) vector cleaved at the Ncol and Sail restriction sites.

The ligation products were used for transformation of JM109 cells by electroporation and positive clones were identified by the correct restriction pattern after simultaneous cleavage with Sail and Ncol. The plasmids of single colonies carrying the correct constructs were purified and were then designated pET-TR (containmg only the insert from pTa, i.e. the open reading frame of TrxR), pET-TRS (insert in fusion with the "S" SECIS cassette) and pET-TRSTER (insert in fusion with the "STER" SECIS cassette) . These plasmids were used for transformation of the expression host BL21(DE3) by electroporation. For co-expression with the sel genes, BL21(DE3) clones already carrying the pET-derived plasmids were transformed using TSS transformation (Chung et a l . , (1989) Proc. Natl. Acad. Sci., 86, 2172-2175) with either of the two compatible plasmids pSUABC or pMN302.

The pSUABC plasmid is pSU-derived (Martinez et al . , (1988) Gene, 68, 159-162) with the 5.5 kb chromosomal Sau3A fragment carrymg the selA and selB genes under control of their natural promotors in conjunction with a 430 bp EcoRI/Hindlll chromosomal fragment carrying the selC gene under control of its natural promoter (Muller (1997) Design neuer Selenoproteme . Thesis, Ludwig-Maximilians-Umversitat ) .

The pMN302 plasmid is pACYC184-deπved (Chang and Cohen (1978) J. Bacteπol., 134, 1141-1156) with a 3.8 kb chromosomal fragment carrying the selD gene under control of its natural promoter (Le felder et al . , (1990) Proc. Natl. Acad. Sci., 87, 543-547).

The glycerol stock of BL21(DE3) transformed with a pET-28b(+) plasmid containing β-galactosidase, included with the pET-24d(+) vector as the "E" control from Novagen, was used as a positive induction control and a negative control for mammalian TrxR activity and was called pET-E. All plasmids utilized in the present study were found to be stable and single colonies either from agar plates stored at 4°C or from glycerol stocks stored at -70°C were utilized for expression experiments.

Expression of recombinan t protein and ^7:,Selenι um labelling For overproduction of protem, transformed BL21(DE3) cells were grown at 37°C to an OD600 of about 0.5, whereafter IPTG was added at concentrations given in the text. If induction was performed at temperatures other than 37°C, cultures were shifted from 37°C to the indicated temperature when the OD600 had reached about 0.5 and IPTG was added 30 mm to 1 h later. In the case of "Selenium labelling, the medium was supplemented with L-cysteine from the beginning of the cultures (100 μg/ml) to inhibit unspecific incorporation of selenium mto non-selenoprotems (Muller et al . , (1997) Arch. Microbiol., 168, 421-427) and ⁷⁵Se-selemte (1 μM) was added at the time of IPTG addition. For production of non-isotope labelled protem to be analyzed for enzyme activity, supplementation of the medium with L-cysteme (100 μg/ml) was also done and non-isotope labelled selemte (5 μM) was added lh prior to addition of IPTG. Cells were collected by centrifugation after periods of incubation given in the text and protem produced was then extracted or analyzed according to any of the following methods.

For SDS-PAGE analysis of whole cell lysates, SDS sample buffer (100 mM Tris-Cl pH 6.8, 200 mM DTT, 4% SDS, 0.2% bromphenol blue and 20% glycerol) was added directly to a cell pellet at a volume of 100 μl sample buffer per ml original culture per absorbance unit at 600 nm. For small scale analysis of aggregation m inclusion bodies, cells from 1 ml of culture were resuspended in 500 μl of 50 mM Tris-Cl pH 8.0, 2 mM EDTA (TE buffer) and were lyzed by somcation. The sonicated cells were centrifuged at 13000 g 20 mm at 4°C, from where supernatants (soluble fraction) were recovered and the pellet (insoluble fraction) resuspended in 500 μl TE buffer. Both were then subjected to analysis on SDS-PAGE or assays of enzyme activity. In cases of "Selenium labelling, gels were first stained, fixed and dried, and radioactive bands were then visualized and quantified using a Phosphorlmager with the Image Quant software from Molecular Dynamics (Sunnyvale, CA. ) .

Enzyme purifica tion

Cells from 50-200 ml cultures induced with IPTG (50 μM) at indicated temperatures were collected by centrifugation, taken up in 5-10 ml TE buffer per 100 ml culture and lysed using two consecutive passages at 1000 psi in a French Pressure Cell Press (American Instrument Co, ML) . The lysates were centrifuged at 30,000g for 30 mm at 4°C and the S30 supernatants were then centrifuged at 100,000g for 1.5 h at 4°C. The S100 supernatant was subsequently applied to a 2.5 ml 2 ' , 5 ' -ADP-Sepharose (Pharmacia Biotech, Uppsala) column equilibrated with TE buffer. After collection of the flow-through, the column was washed with TE buffer until A_280r™ reached baseline (at least 10 ml buffer) . Then enzyme was eluted with increasing concentrations of NaCl.

The major peak of hTrx-dependent insulin reduction (see below) was found in fractions eluted between 150 mM and 500 mM NaCl. These fractions were all >95% pure as judged by SDS-PAGE and were therefore pooled and concentrated using a Model 12 Ultraflltration cell (Amicon Corp., Lexington, MA.) with an Omega 10 K-cutoff membrane (Filtron GmbH, Karlstem). There was no apparent differences between rTR, rTRS, rTrxR and rTrxR(ABC) in bmdmg to the 2 ', 5 ' -ADP-Sepharose .

Enzyme concentrations were determined with the BioRad

(Hercules, CA. ) Protem assay, using bovine serum albumin as standard. Amount of bound FAD was calculated by the absorbance of the oxidized enzymes at 463 nm, using an extinction coefficient of 11300 M-lcm-1

Enzyme activi ty determina tion

Activity of recombinant mammalian TrxR in S100 supernatants was measured usmg the insulin assay as described (Arner et a l . , (1998) Meth. Enzymol., in press) . In short, soluble protein (10-20 ul) was incubated at 20°C in a total volume of 100 μl containing 1 mM NADPH, 1 mg/ml insulin and 6 μM hTrx (C63S/C72S) in TE buffer. After 30 mm, 1 ml of 1 mM DTNB, 6 M guamdine-HCl in 200 mM Tris-Cl, pH 8.0 was added. Derivatized thiols were measured by absorbance at 412 nm and the extrapolated amount of active enzyme was estimated by comparison to a standard-curve of purified native human placenta TrxR (0-300 pmol) added to a cell-extract of E. coli transformed with the pET-E plasmid and thereby lacking expressed recombinant mammalian TrxR, assayed in parallel to the other samples .

Reduction of DTNB, hTrx, selemte or lipoic acid by the purified recombinant enzymes was analyzed as described (Arner et al . , (1998) Meth. Enzymol., press), using substrate and enzyme concentrations as given in the text.

Peptide genera tion and mass spectrometry

Purified rTR, rTRS or rTrxR(ABC) was dissolved in 200 μl 0.1 M Tris-Cl, pH 8.5, 6 M guanidine hydrochloride (total protem 20 - 100 μg) and 5 μl β-mercaptoethanol was added. Samples were then incubated two hours at 50°C for full reduction. Thiol (and selenol) groups were subsequently alkylated by addition of 5 μl 4-vmylpyrιdιne and incubation one hour at 20°C. Alkylated protem was then recovered by an HPLC desalting step and peptides were generated with sequencing grade endoprotemase Lys-C from Boehrmger Mannheim GmbH (Mannheim) according to the procedure given by the manufacturer.

Peptides were then analysed using both HPLC-coupled mass spectrometry (LC-MS) or matrix-assisted laser desorption lomzation mass spectrometry (MALDI) . In the LC-MC, an ABI 140D HPLC-pump was coupled directly to an API 365 triple quadrupole instrument (Sciex, Thornhill, Ontario, Canada) equipped with the standard electrospray lomzation source. The HPLC flow rate was 5 μl/min and the gradient utilised was 12%B for 10 mm, linear increase to 50%B at 70 mm, to 70%B at 80 mm, to 99%B and 90 mm, then continued at 99%B until 110 mm and subsequently lmearily reduced to 12%B at 120 mm, where buffer A was 0.1% trifluoroacetic acid (TFA) in water whereas buffer B was 0.1% TFA in 80% acetomtrile . Peptide masses were recorded throughout the HPLC run by scanning the first quadrupole over the mass range 300-2000 in 6 seconds with the instrument calibrated with polypropylene glycol .

For MALDI, a Bruker Relfex III time-of-flight mass spectrometer (Bruker-Franzen, Bremen, Germany) was utilised with either alpha-matrix (alpha-cyano-4-hydroxy-cmnamιc acid) at 5 mg/ml in 50% acetomtrile, 0.1% TFA or 2,5- dihydroxybenzoic acid at 20 mg/ml in 30% acetomtrile, 0.1% TFA, with in both cases 0.7 μl sample and 0.7 μl matrix solution, dried at 20°C. The MALDI measurements were performed with a 337 nm nitrogen laser in positive reflector mode .

The results of both the LC-MS and the MALDI were combined, with a majority of the detected mass species found in both measurements .

Table I. Mammalian TrxR activity in soluble fractions of E. coh transformed \N i(lι the p ET-TR, pET-TRS, p ET-TRS_{1 1 R} or pET-E plasmids

isinid Insv.il in disul fide tcduction Enzymatically active

(A_{4 I 2} units)' mammalian TrxR

hTrx + hTrx mg/L culture²

pET-TR 0 052 0 067 < 0 01 pET-TRS 0 048 0 058 < 0 01 pET-TRS π-,, 0 05 1 1 564 1 2 pET-E 0 050 0 063 < 0 01

1 ) Insulin disulfide ι eduction w as measured by incubation foi Hi al 37°C of 30 μl of the soluble [laclioii ol L coh lysates (coi i cspondmy 10 30 μl culluic) in a total of i<10 μl TE buffer containing 1 inM NΛDPH, 1 mg/ml insulin with or without 6 μM hTι x(C 1 S/C72S) with subsequent dctci i nation of absorbance at 4 12 nm upon addition of 900 μl 1 M DTNB and 6 M gua dinc HC1 in 200 mM Tπs Cl pH 8 0

2) Extrapolaled fio a standard curve using purficd human placenta! TrxR (0 - 300 fnnol) added lo 30 μl of the soluble fraction of C coli iransfoi incd with the pET E plasmid and assayed simultaneously with the other samples

The soluble fraction of E coh transformed with the different plasmids and induced at 20°C over night with 50 μM IPTG was analyzed for insulin disulfide reduction with or without addition of human Trx, which is a highly specific assay for mammalian TrxR activity, as detailed in the text

Table II. Effect of co-transformation with the pSUABC or pMN302 plasmids on production of activ e rat TrxR produced from the pET- TRS_TER plasmid

Plasmids used for transformation

pET-TRS_T pET-TRS_TCR + pMN302 pET-TRS_TER + pSUABC

Amount of soluble active enzyme (mg L culture)¹

Temperature for induction Temperature for induction ¹ Temperature for induction ^*

Time for harvest 16°C 20°C 25°C 16°C 20°C 25°C 16°C 20°C 25°C after induction ^:

3 hours 0 05 0 13 0 50 0 03 0 13 0 45 0 08 0 17 2 79 8 hours 0 17 0 16 0.96 0.07 0 16 0 57 1 19 2 09 5 40 19 hours 0 37 0 64 1.42 0.38 0 64 0 62 2 81 3.37 5 14

¹ The soluble fraction of harvested cells was analyzed using hTrx-dependent insulin reduction with the amount of enzyme estimated from a standard curve using native purified TrxR, as described in Table I

^: Cells were induced and grown with 50 μM IPTG at the indicated temperatures and then harvested after 3, 8 or 19 hours, as shown in the table

Table III. Enzymatic activity of rTrxR and rTrxR(ABC)

Substrate Assay conditions Activity Ratio of activity

rTrxR rTrxR(ABC) ιTrxR(ABC) s rTrxR

DTNB 100 nM enzyme 1 3 U/mg 7 7 U/me 5 9 fold 300 μM NADPH 5 mM DTNB hTrx 15 nM enzyme 64 mm ' 513 mm 0-fold 150 μM NADPH 3 μM hTrx

Selemte ' 50 nM enzyme 12 mm 106 mm 8 8-fold 200 μM NADPH 100 μM selemte

Lipoic acid ^b 40 nM enzyme 9 8 ιn 70 mm 7 1 fold 200 μM NADPH 1 mM lipoic acid

1 U is defined as 2 μmol TNB formed per μmol dimeric enzyme per minute, determined by A _12nm, and activity is given as U/mg enzyme, being the standardized unit of specific activity for mammalian thioredoxm reductase

Activity given as μmol NADPH oxidized per μmol dimeric enzyme per minute, determined by A 340nιτ

Table IV. Kinetic parameters of rTrxR(ABC) compared to native rat TrxR

Substrate Assay conditions rTrxR(ABC) Native rat liver TrxR for rTrxR(ABC) l cal (Ref )

(mm ') (μM) (mm ') (μM)

DTNB^a 50 nM enzyme 863 124 4000 660 (Luthman and 300 μM NADPH Holmgren, 1982)

0 025 - 5 mM DTNB hTrx^b 15 nM enzyme 850 3 3 3000 2 5 (Luthman and 150 μM NADPH Holmgren 1982)

1 - 7 μM hTrx

Lipoic acιd^b 100 nM enzyme 263 679 368 710 (Arner et al 1996) 150 μM NADPH 0 25 - 2 mM lipoic acid

Reaction rates determined following formation of TNB by increase in A_412nm R Reeaaccttiioonn rraatteess ddeetteerrmmiinneed following oxidation of NADPH by decrease m A 340nrτ Using rat or bovine Trx

Claims

1. A method of producing in bacterial host cells a mammalian or other heterologous selenoprotem, the method comprising: providing nucleic acid which comprises a coding sequence for the selenoprotem, a bacterial selenocysteine insertion sequence (SECIS) for insertion of selenocysteine mto the protem on production of the protem by expression from the nucleic acid in a bacterial host cell and regulatory sequences for expression of the encoded selenoprotem in bacterial host cells, transforming bacterial host cells with the nucleic acid, and culturing the host cells under conditions for over- expression of selA, selB and selC so that the selenoprotem is produced.

2. A method according to claim 1 wherem the host cells are transformed with selA, selB and selC encoding sequences.

3. A method according to claim 1 or claim 2 wherem the host cells over-express selD.

4. A method according to claim 3 wherem the host cells are transformed with selD encoding sequences.

5. A method according to any one of claims 1 to 4 wherem the bacterial SECIS includes a TGA codon followed by one or more codons modified to encode one or more ammo acids which follow a selenocysteine residue in the heterologous selenoprotem .

6. A method according to claim 5 wherein the SECIS provides the ammo acid sequence -Gly-Cys-Sec-Gly at the C-terminus of the encoded protem.

7. A method according to claim 6 where the SECIS is as shown in Figure IB modified to encode said ammo acid sequence .

8. A method according to any one of claims 1 to 5 where the SECIS includes a nucleotide sequence shown in Figure 2.

9. A method according to any one of claims 1 to 5 wherem the SECIS includes the SECIS nucleotide sequence shown in

Figure 3.

10. A method according to any one of claims 1 to 9 further comprising recovering the selenoprotem from the host cells or surrounding medium.

11. A method according to claim 10 wherem the polypeptide is provided in isolated and/or purified form.

12. A method according to claim 10 or claim 11 wherein the polypeptide is formulated mto a composition comprising at least one additional component.

13. A method according to any one of claims 1 to 12 wherem the selenoprotem is thioredoxm reductase.

14. A method according to any one of claims 1 to 13 wherem the bacterial host cells are E . coli .

15. A selenoprotem produced by a method according to any one of claims 1 to 14.

16. A bacterial host cell which contains nucleic acid which comprises a coding sequence for the selenoprotem, a bacterial selenocysteine insertion sequence (SECIS) for insertion of selenocysteine mto the protem on production of the protem by expression from the nucleic acid in a bacterial host cell and regulatory sequences for expression of the encoded selenoprotem in bacterial host cells, and which contains nucleic acid for over-expression of selA, selB and selC.

17. A bacterial host cell according to claim 16 containing nucleic acid for over-expression of selD.

18. A bacterial host cell according to claim 16 which is over-expressing selA, selB and selC.

19. A bacterial host cell according to claim 18 which is over-expressing selD.

20. A bacterial host cell according to claim 18 or claim 19 which is producing the selenoprotem.

21. A nucleic acid construct for producing in bacterial host cells a mammalian or other heterologous selenoprotem, which construct comprises a coding sequence for the selenoprotem, a bacterial selenocysteine insertion sequence (SECIS) for insertion of selenocysteine mto the protem on production of the protem by expression from the nucleic acid in a bacterial host cell and regulatory sequences for expression of the encoded selenoprotem in bacterial host cells.

22. A nucleic acid construct according to claim 21 wherem the bacterial SECIS includes a TGA codon followed by one or more codons modified to encode one or more ammo acids which follow a selenocysteine residue in the heterologous selenoprotem.

23. A nucleic acid construct according to claim 22 wherem the SECIS provides the ammo acid sequence -Gly-Cys-Sec-Gly at the C-termmus of the encoded protem.

24. A nucleic acid construct according to claim 23 wherem the SECIS is as shown m Figure IB modified to encode said ammo acid sequence.

25. A nucleic acid construct according to any one of claims 21 to 24 wherem the SECIS includes a nucleotide sequence shown in Figure 2.

26. A nucleic acid construct according to any one of claims 21 to 24 wherem the SECIS includes the SECIS nucleotide sequence shown in Figure 3.