Although protein glycosylation is widespread in eukaryotes, reports of this post-translational modification in prokaryotes has been rare. On the basis of Con-A lectin binding it has been proposed that several antigens from both Mycobacterium tuberculosis and M. bovis are glycosylated including the M. bovis antigens MPB70 and MPB83 (T. Fifis et al., Infect. Immun.

59 (1991) 800-7) which are serodominant during M. bovis infection in badgers and cattle. These antigens are cross- reactive proteins showing 61 similarity in their amino acid sequence and have identical homologues in M. tuberculosis, namely MPT70 and MPT83.

W097/083222 describes and claims a gene mpt83 which encodes the glycosylated protein MPT83 of M. tuberculosis. It has previously been reported that MPT83 is post-translationally modified by the fast growing mycobacterium M. smegmatis (R. G. Hewinson et al., Scand. J. Immunol. 43 (1996) pp 490-499).

The complete gene sequence and the corresponding protein sequence is given in WO 97/083222 which is incorporated herein by reference. The numbering of amino acids and bases used herein corresponds to that given in W097/083222. In this reference, the glycosylation signal was identified as a 23 amino acid sequence comprising amino acids 35-57 in the protein sequence.

The applicants have found that a smaller motif will act as a glycosylation motif. In addition, a further glycosylation site

within the MPT83 protein has been identified. Such sites may advantageously be introduced into other proteins particularly when these are intended to be used as vaccines. Glycosylation may improve the immunogenicity of the protein. Furthermore, it has been found that in some instances, glycosylation leads to improved stability of the protein, leading for example to a longer and stronger immune response.

Thus the invention provides a recombinant DNA which encodes a polypeptide which is other than a wild-type or native polypeptide and which is glycosylated when expressed in an actinomycete host, said DNA comprising a sequence which encodes the amino acid sequence PAAPVTTAA (SEQ ID NO 1) or a fragment or variant thereof; said fragment or variant including at least both the threonine residues and being able to act as a glycosylation motif; and/or a sequence which encodes a glycosylation motif found in the region of the MPT83 protein including asparagine at position 192 in the sequence.

As used herein, the term"fragment"includes small peptide units for example of 5 amino acids or more. The term "variant"refers to peptide units wherein one or more amino acids, for example up to 4 amino acids, have been changed whilst not altering the qualitative function of the peptide. The changes may be what are regarded as"conservative substitutions"where amino acids are replaced by amino acids of an essentially similar nature (e. g. basicity etc.) in which case rather more subtitutions may be tolerated. Alternatively a small number of non-conservative substitutions may be introduced

without altering the function of the peptide, as would be determinable by the skilled person.

Suitably, the recombinant DNA of the invention comprises a sequence encoding SEQ ID No 1 as defined above. Such a sequence may be the sequence found in the mpt83 gene, i. e.

CCG GCG GCG CCC GTT ACC ACG GCG GCA (SEQ ID No 2) A futher glycosylation motif of MPT83 involves the asparagine at position 192 in the sequence. It is probable that this residue forms part of a glycosylation motif located between 179-196 of MPT83. The motif is of at least 5 amino acids, and preferably at least 10 amino acids in length. The entire 20 amino acid sequence is of sequence DLTVIGARDDLMVNNAGL (SEQ ID No. 3) The critical asparagine residue is indicated in bold type.

Thus, recombinant DNA of the invention may encode said sequence (SEQ ID No. 3).

For example, the DNA may comprise the sequence as found in mpt83, i. e.

GAC CTG ACG GTG ATA GGC GCC CGC GAC GAC GAC CTC ATG GTC AAC AAC GCC GGT TTG GTA (SEQ ID NO. 4) Polypeptides obtainable by expression of a recombinant DNA as described above form a further aspect of the invention. Such polypeptides may be obtained by incorporating recombinant DNA as described above into a replication vector, transforming a suitable host cell with said vector and culturing the cell.

Methods of preparing polypeptides as well as vectors and

transformed cells used in the method form further aspects of the invention.

Suitable host cells would be determinable by the skilled person using routine methods. However, preferably the host cell is an actinomycete such as a mycobacterial cell.

Applications of modified polypeptides of the invention and especially the DNA encoding said polypeptides are explained in W097/083222. In particular, these modified polypeptides would be useful in vaccines, especially those protective against mycobacterial infection such a tuberculosis, since they may improve immunogenicity and stability as discussed above.

Furthermore, they would provide a serological marker allowing the vaccine strain to be distinguished from infection by a native strain.

Methods of treatment using the DNA or polypeptides of the invention as well as vaccines containing them form yet further aspects of the invention. Vaccines may be in the form of live or "killed"vaccines as is understood in the art. In particular, the"naked DNA"approach to vaccines would be amenable to the use of the present invention.

Vaccines may be in the form of pharmaceutical compositions in combination with a pharmaceutically acceptable carrier.

The invention will now be particularly described by way of example with reference to the accompanying drawings in which: Figure 1 illustrates the E. coli alkaline phosphatse (PhoA) reporter system;

Figure 2 shows the alignment of amino acid sequences of MPT83 and MPT70; Figure 3 illustrates the modification of MPT70 to introduce a glycosylation site; Figure 4 shows a range of constructs tested for recognition by ConA; and Figure 5 illustrates a proposed model for glycosylation of the antigen MPT83.

Example 1 Idenfication of a peptide region required for Glycosylation The system used to identify the glycosylation region of MPT83 was similar to that used previously to identify the glycosylation site ot hte l9kDa lipoprotein from M. tuberculosis (J. L. Herrmann et al., EMBO J, 15 (1996) 3547-54). A mycobacterial shuttle vector was used to express MPT83 and MPT70 fused to E. coli alkaline phosphatase (PhoA) lacking its own signal sequence in M. smegmatis mu2155 as shown in Figure 1. The proteins were expressed under the control of the promoter region from the l9kDa antigen of M. tuberculosis. PhoA was used as a hybrid partner to allow monitoring of the level of expression of the different constructs. Glycosylation was determined by ConA recognition. When expressed in E. coli the fusion proteins did not bind ConA.

Sonicated extracts of recombinant M. smegmatis expressing the above constructs were tested for ConA binding after fractionation by SDS-PAGE and transfer to nitrocellulose membranes. Levels of protein expression were determined by anti-PhoA immunoblotting.

Equal expression of the MPT70 and MPT83-PhoA fusions were confirmed by the immunoblotting with anti-PhoA. The MPT83-PhoA fusion was recognised by ConA whereas the MPT70-PhoA fusion was not. Therefore, it was concluded that the signal motif for glycosylation of MPT83 lies within the amino acid residues that differ from those of MPT70 (See Figure 2).

Example2 Identification of a glycosylation motif within the amino terminus of MPT83.

Using the PhoA reporter system described above in Example 1, the regions of MPT83 involved in ConA binding were identified. One such region was identifed at the amino terminus of the molecule and was mapped as shown below in Table 1.

Table 1 Amino acids of MPT 83 ecognition by Con A 1 20<BR> MINVQAKPAAAASLAAIAIAFLAGCSST-PhoA-ve 35 1-20FLAGCSSTKPVSQDT-PhoA-ve 56 1-2OFLAGCSSTKPVSQDTSPKPATSPAAPVTTAAMADPA-PhoA +ve 1-20-63 FLAGCSSTKPVSQDTSPKPATSPAAPVTTAAMADPAADLIGRG-PhoA +ve A regions between amino acid residues 35 and 56 of MPT83, which is absent from MPT70 (see Figure 2) was found to be responsible for ConA binding. Within this ConA binding region is an amino

acid sequence PAAPVTTAA (SEQ ID No 1) (residues 43-51) that was thought to be similar to the sequence of a glycopeptide, PAPPVPTTA, from a 45kda protein of M. tuberculosis (K. M. Dobos et al., Infect Immun 63 (1995) 2846-53). This latter peptide was shown to be)-glycosidically linked to a disaccharide composed of two hexose residues through a threonine residue.

Example 3 Identification of glycosylation residues Site directed mutagenesis was used to replace the threonine residues in the region of MPT83 identified above as the putative glycosylation site, with valine as shown in Table 2 below Table 2 Amino acid Recognition by ConA 1-20- FLAGCSSTKPVSQDTSPKPATSPAAPVITAAMADPAADLIGRG- PhoA +ve 1-20- FLAGCSSTKPVSQDTSPRPATSPAAPV W AAMADPAADLIGRG- PhoA-ve Immunoblotting was effected as described in Example 2. This showed that mutation of the threonine residues ablated ConA binding. There was also a noteable decrease in the apparent molecular weight of the mutated fusion protein. This result suggests that the peptide region of SEQ Id No. 1 is a signal motif for O-linked glycosylation in MPT83.

Example4 Glycosylation of MPT70 using the signal motif of MPT83 The corresponding regions of the MPT83 and MPT70 proteins is shown in Table 3.

Table 3 MPT83 S P A A P V T T A A M A D P A A 57 MPT70 S P P A A---------A G 31 In order to confirm that SEQ ID No. 1 indeed directs glycosylation of MPT83, a peptide comprising the amino acids indicated in bold type in Table 3 was introduced into MPT70 and fused to PhoA as illustrated in Figure 3.

Immunoblotting as described in Example 2 showed that the modified MPT70 fusion protein was recognised by ConA, thus demonstrating that the signal motif for 0-glycosylation in MPT83 can be used to glycosylate heterologous antigens in M. smegmatis.

During these experiments it was observed that ConA binding was reduced in both the truncated MPT83 fusion and in the chimeric MPT70/MPT83 fusion compared with the binding observed for the full length MPT83-PhoA fusion. This indicates the presence of other glycosylation sites within MPT83.

Example 5 Identification of an N-linked glycosylation site within MPT83 Further comparison of the amino acid sequence of MPT83 and MPT70 identified manner differences in amino acid sequence that might generate motifs for both O and N-linked glycosylation in MPT83.

A putative site for N-linked glycosylation was identifed within the carboxy terminus of MPT83 at asparagine residue 193. This

residue was mutated to glycine (the corresponding residue in MPT70) in both the native molecule and a recombinant MPT83 lacking the 0-glycosylation site at threonine residues 48 and 49 as shown in Figure 4. The recognition of the constructs by ConA is also shown in Figure 4.

Mutation of the asparagine residue to glycine at position 192 led to a reduction in the level of ConA binding compared to that of the native MPT-PhoA fusion. When the threonine residues within the motif for 0-glycosylation were mutated along with the asparagine residue at position 192, complete ablation of ConA binding was observed, suggesting that the asparagine residue is a site for N-glycosylation.

Thus a proposed model for glycosylation of the antigen for MPT83 is illustrated in Figure 5.