This invention relates to HCMV proteins, and particularly to those that may be useful, directly or indirectly, in diagnostic procedures, prophylaxis and therapy.

Derived from DNA sequence analysis of HCMV we present for the first time evidence for a viral encoded MHC class I antigen. Major histocompatibility complexes, MHC- loci, have been identified so far only in vertebrates including mammals, birds, amphibians and probably reptiles and fish. The MHC encoded proteins, the MHC-antigens , are known to play a vital role in the self versus non-self recognition in the immune response. They are divided into class I and class II molecules, each consisting of a set of cell-surface glycoproteins . The MHC class I antigens are expressed on almost all nucleated somatic cells where they are noncovalently associated with 2~ ^m: **- ^c:rc *.=lob lin ( β2 ^m ) (Klein, J. Natural History of the Major Histocompatibility Complex, Wiley - Interscience Publication (1987) ) .

We have identified an open reading frame within the Hindlll O-fragment of HCMV strain ADI69 encoding a potential glycoprotein which ( i ) fulfils most

characteristic features of a typical MHC class I antigen and (ii) shows significant homology in the deduced amino acid sequence. The biological implications of an HCMV encoded class I antigen are of course significant for the development of a vaccine against HCMV. Recent publications have shown, (i) that HCMV exists in_ vivo (in urine) as β2 ^m ~ coated particles (McKeating, et al, J. Gen. Virol. 68, 7δ5~ 792 (1987)) . (ϋ) that cultured HCMV strain ADI69 binds β m and competes for ₂ binding sites on its host cells (human fibroblasts) (Grundy, et al J. Gen. Virol. 68, 793"803 (1987)) and (iii) that the addition of β2 to the culture medium increases the amount of infectious extracellular HCMV (Grundy et al, supra). Based on immunoprecipitation experiments with β specific antibodies, the same group also reported that 2 ^m might bind to two viral envelope proteins of molecular masses 36,000 and 65,000 daltons respectively (Grundy, et al, J. Gen. Virol. 68, 777-784 (1987)). Derived from these and further data the authors proposed that the interaction of the two non-class I viral envelope proteins with HLA class I antigens via β2m displacement might be a pathway for HCMV receptor-mediated infection. Here, we suggest evidence, that the m- receptor on the viral surface is an MHC class I like glycoprotein which is not acquired from the host cell during budding but is encoded by the virus itself.

Summary of the invention

According to one aspect of the present invention there is provided a polypeptide which comprises at least part of the a ino acid sequence encoded by the nucleotide sequence 1*^3 to 12-47 as depicted in Fig. 2 hereof, or an allele or variant thereof, and having at least one epitope characteristic of HCMV. The invention also includes sub- genomic DNA sequences encoding such a polypeptide, recombinant cloning and expression vectors containing such DNA, recombinant microorganisms and cell cultures capable of producing such a polypeptide, and vaccines against HCMV incorporating such a polypeptide or a recombinant viral vector capable of expressing the polypeptide in a mammal immunized therewith. The invention further includes monoclonal and polyclonal antibody binding specifically to such polypeptide. The polypeptide can be used in diagnostic kits and procedures to detect HCMV specific antibodies in a clinical sample. The antibodies hereof can be used in diagnostic kits and procedures to detect HCMV antigens in a clinical sample, and they can also be used therapeutically or prophylactically for administration by way of passive immunisation against HCMV infection.

Brief description of the drawings

Figure 1 shows a physical map of HCMV strain ADI69. The linear double stranded DNA genome of HCMV consists of a

long unique (U- _Q ) and a short unique (Ug) region both of which are flanked by repeat sequences (boxed) which allow the virus to exist in four equimolar isomers (Oram, et al, J.Gen. Virol. 59, 111-129 (1982)). The scale shows the genome size in kilobases (kb). Below the scale the Hindlll map of the generally accepted prototype orientation is shown. The Hindlll O-fragment is indicated by shadowing and the approximate position and of the predicted HCMV-H301 gene is indicated by an arrow.

Figure 2 shows the nucleotide sequence of the HCMV- H301 gene in 5'~3' direction. The predicted coding sequence starting at position 1 3 (start codon, ATG) to position 1247 (stop codon, TGA) has been translated into the corresponding amino acid sequence which is written above the nucleotide sequence in the one letter code. The potential TATA-box (position 100) at -42 upstream of the start codon and the potential polyadenylation site AATAAA (position 1722) at +473 downstream of the stop codon are indicated by boxing. A second polyadenylation site (position 132) of an upstream open reading frame, ending at position 3 (stop codon, TAA) , is also indicated by boxing. Another short open reading frame downstream of the HCMV- H301 gene is present from position 1246 (start codon, ATG) to position 1 40 (stop codon, TGA). The potential leader sequence (LD) and the transmembrane region (TM) of the deduced amino acid sequence are also boxed in.

Class I antigens are located within the MHC-locus, named HLA complex in human, H-2 complex in mouse, RT1 complex in rat, RLA complex in rabbit and many more (Klein et al supra) . The structure of class I genes has been studied in detail (Klein et al , supra, and Hood, et al , Ann.Rev. Immunol. , 29-568 (1983)) . They are between 3,000 and 12,000 bp long, depending upon the variable length of their introns . Throughout all species studied so far, class I genes show a very conserved gene structure. The coding sequence of each gene is about 1,100 bp long and is divided into 8 exons, separated by f introns. Exon 1 encodes the leader sequence (LD) , exons 2, 3 and 4 encode the αl , a.2 and α3 domain respectively, exon 5 encodes the transmembrane region (TM) and finally exons 6, 7 and 8 encode the cytoplasmic domain (CY) and the 3' untranslated region.

In comparison to this, the structure of the HCMV- H301 gene is quite different. There are no indications of suitable GT/AG splicing signals for the typical upstream and downstream intron boundaries in class I genes (Hood et al , supra) The predicted coding sequence of 1 , 104 bp (position 142 - 1246, Fig. 2) , however, coincides perfectly with the average length of the coding sequence of known class I genes. We therefore assume, that the HCMV-H301 gene is not spliced. The predicted gene shows all characteristic elements of a complete gene. It has a

potential promotor (TATA-box) at -42 upstream of the start codon and a potential polyadenylation site (AATAAA) at +473 downstream of the stop codon. Furthermore, it is closely flanked by other open reading frames at both ends (only partly shown in Fig. 2) which is a typical feature of herpesvirus gene arrangement (Baer, et al Nature (London) 310, 207-211 (1984)) .

MHC class I glycoproteins are cell surface molecules and accordingly their α chain can be divided into an extracellular region, a hydrophobic anchor or transmembrane region and an intracellular or cytoplasmic region. Like other membrane proteins class I precursors also have a short, hydrophobic leader sequence at the NH2~terminus which becomes cleaved off before the molecules are expressed on the cell surface.

The deduced HCMV-H301 amino acid sequence was aligned to some α chains of class I antigens from mouse (H- 2) , rat (RT1), human (HLA) and rabbit (RLA) (Klein et al, supra, Klein, et al , Immunol. Today 7, 41-44 (1986) , Claverie et al , Proteins 1, 60-65 (1986)). In Fig. 2 homologous amino acids are marked by ( : ) for 50% - 100% ho ology and (.) for 25 - 50 homology. Four highly conserved cysteines (C) in the a.2 and α3 domains are highlighted by asterisks (*). All potential N-linked ^" glycosylation sites (N_-X-T/S) and the HCMV-H301 transmembrane region (TM) are underlined.

The HCMV-H301 leader sequence (LD) , consisting of 18 amino acids, is 3 amino acids shorter than the average LD sequence of some HLA and H-2 antigens and shows only little homology (average 15%)- This is not surprising because it does not form part of the mature protein and the uninterrupted sequence of 1 mostly hydrophobic amino acids appears to be a good candidate for a functional LD sequence.

The bulk of the class I α chain consists of the extracellular region which is divided into three domains , αl , α2 and α3, respectively. Each domain consists of about 90 amino acids. The main structural features of the extracellular region are four highly conserved cysteines (C), two in the α2 domain and two in the α3 domain. The αl domain normally contains no cysteine. It is presumed, that the two cysteines in both the α2 and α3 domain, form a covalent disulfide bridge by looping out the intervening amino acid sequence similar to Ig molecules. This creates a loop of 62 amino acids in the α2 domain and a loop of 55 amino acids in the c_3 domain.

These characteristic, structural elements, common to all class I molecules, are also present in the HCMV-H301 α chain. It also can be divided into the corresponding αl , α2 and α3 domain. The αl domain also contains no cysteine and shows an average homology of 30% in the amino acid sequence to various αl domains compared. The α2 domain

shares the two highly conserved cysteines and shows an average homology of 26% in the amino acid sequence. However, the corresponding loop between the two cysteines consists of 72 amino acids rather than 62 amino acids found in all the other α2 domains compared. The average homology of 27% in the amino acid sequence of the HCMV-H301 α3 domain seems to be clustered in mainly 3 distinct regions (overlined in Fig. 2) . As the exact binding site of 2m to the α3 domain of class I α chains is not known and since HCMV has been shown also to bind β2m (Grundy, et al, J.Gen.Virol. 68, 777"784 (1987)). These three regions make likely candidates for a β2ia binding site. The HCMV- H301 α chain shows a potential TM sequence (boxed in Fig. 2), consisting of 23 uncharged, mainly hydrophobic amino acids.

The cytoplasmic terminal region of HCMV-H301 contains no potential phosphorylation site and shows no significant homology in the amino acid sequence. Assuming that the predicted HCMV-H301 gene is transcribed, translated and expressed on the surface of the infected cell, the absence of any phosphorylation site in the HCMV- H301 cytoplasmic region could have a signal significance in the viral budding process in order to distinguish between membrane patches expressing host or viral encoded proteins, and in a similar way, the 13 potential glycosylation sites of the HCMV-H301 α chain could also present a signal

significance on the surface of an infected cell by increasing the negative charge on the cell surface.

Uses of HCMV 'HLA' Gene H301 and its Expression Polypeptide Both the HCMV 'HLA' gene H301 and its expression polypeptide having most characteristic features of an MHC class I antigen have a variety of commercial uses. The expression polypeptide may be produced either in a recombinant expression system using the HCMV 'HLA' gene H301 , or by any other preparative process which utilises the gene sequence information as provided herein and with reference to Fig. 1.

The main commercial uses can be summarised as follows : i) use of the HCMV 'HLA' gene H301 and of its expression polypeptide in vaccination systems; ii) use of the HCMV 'HLA' gene H301 and of its expression polypeptide (in crude preparations, as a purified product or as a recombinant) to raise antibodies (monoclonal, polyclonal and engineered) suitable for use in diagnostic tests and as therapeutic and prophylactic agents; and iii) use of the expression polypeptide as reagents in diagnostic kits.

These uses are discussed in more detail below: i) Vaccines

Hitherto experimental vaccines have been based on attenuated, non-pathogenic forms of HCMV. However, these vaccines can have undesirable side effects. The present invention provides an alternative approach to the production of a vaccine against HCMV by providing: a) a sequence of HCMV 'HLA ¹ gene H301 which encodes b) an expression polypeptide having most characteristic features of an MHC class I antigen and acting as a receptor for 2 . The provision of both the gene sequence and the expression polypeptide sequence provides for two alternative vaccination routes .

For a recombinant virus vaccine the identified HCMV 'HLA' gene H301 sequence is isolated and introduced into a suitable mammalian virus vector by conventional genetic engineering techniques (see for example the detailed description below) and transferring the plasmid into the host e.g. the human for vaccination. Suitable virus vectors are the poxviruses such as vaccinia virus.

Of course the receptor protein itself may be utilised as a vaccine according to techniques well known in the art. These techniques comprise compounding the protein with a suitable adjuvant or excipient of the kind conventionally employed in vaccine formulations . This form of vaccination might be more appropriate than the

recombinant vaccine for example in immunosuppressed individuals . ii) Antibody Preparation

The vaccination techniques described above are simply techniques used to stimulate a given hosts immune response and thereby prime the immune system to a foreign antigen. These same techniques may be applied to experimental animals, from which antibodies produced against the antigen (in this case the HCMV 'HLA' gene expression polypeptide) may be harvested. Thus the host animal is immunised with either a recombinant virus vector carrying the HCMV 'HLA' gene H301 or with the expression polypeptide itself. Antibodies specific to the expression polypeptide are then extracted from the host animals antiserum, using standard techniques well known in the art. Monoclonal antibodies may also be prepared from the cells of the immunised animals using standard techniques well known in the art.

Antibodies are prepared to a high degree of specificity by contacting them with the expression polypeptide immobilised on a suitable support such as an affinity column gel and then separating bound antibody from the expression polypeptide by eluting the affinity column with a reagent which destroys the polypeptide-antibody binding.

iii) Use of Antibodies for Diagnosis- Therapy and

Prophylaxis and Use of Antigens for Diagnosis

Both the antibodies produced as described above and the expression polypeptide itself, can be used in methods for, and incorporated as parts of kits for, detecting infection with HCMV.

Where antibody is used as a detection reagent, HCMV is detected in a biological sample (usually sera) taken from a patient. Alternatively the expression polypeptide is used as a reagent to detect antibodies to HCMV the biological sample. The presence of serum antibodies to HCMV indicates past or present infection with HCMV.

The methods for detecting infection using both the expression polypeptide and antibodies thereto comprise a variety of conventional assay procedures, based for example on ELISA, RIA or immunofluorescence. Typically the expression polypeptide could be immobilised on a support (such as an ELISA plate or a dip-stick) and then contacted with the clinical sample. After washing, the support is contacted with labelled anti-human i munoglobulin which binds to any HCMV antibody which has been found by immobilised HCMV protein. The label may be an enzyme and binding of the labelled anti-human immunoglobulin is detected by applying a suitably chromogenic substrate. Alternative labelling/detection systems are also well known in the art.

Expression of HCMV 'HLA' gene H301 and production of antibody to this protein 1. Expression of the HLA gene in bacteria

Inspection of the nucleotide sequence and the deduced restriction endonuclease sites of the HLA gene showed that a DNA fragment that included the whole open reading frame minus the first 39 amino acids could be excised with restriction enzyme Hindi. This fragment was cloned into the polylinker of pSP65 (Kreid et al (1984) Nucleic Acids Res. 12, 7057-7070) so that the fragment could be excised with restriction enzymes BamHI and Pstl. The BamHI-Pstl fragment was cloned into plasmid pEX-3 that had been cut with BamHI and Pstl so that the HLA gene was in frame, as a fusion protein downstream, of the body of β- galactosidase. The resulting plasmid was called CXI. The pEX-3 plasmid and E_. coli strain POPS are designed so that high levels of β-galactosidase fused to other protein coding sequences may be synthesized in an inducible manner (Stanley, et al, (1984) . EMBO J. 3_ l429~34) . When plasmid CXI was introduced into E_. coli POPS and expression of the β-galactosidase/HLA fusion was induced, a major protein band of approximately M _r 150K was apparent. This contrasts with the M _r 110K β-galactosidase produced from plasmid pEX- 3- This 150K protein was purified from polyacrylamide gels, emulsified in Freund's adjuvant and used to immunise two rabbits (designated 4l and 42) by intramuscular

injection. Serum taken from these animals after repeated immunisation was tested against cells infected with recombinant vaccinia viruses expressing the HLA gene and against HCMV-infected cells and virus (see below).

2. Expression of the HLA gene in vaccinia virus

The nucleotide sequence of the HLA gene showed that an additional ATG codon was present upstream of the ATG used to initiate translation of the HLA gene and before the first convenient upstream restriction enzyme site. Consequently, before this gene could be adequately expressed in vaccinia virus it was necessary to remove this additional ATG. This was done as follows. First, the HLA gene was isolated from a plasmid pAT 153/HCMV Hindlll 0 which contained the HCMV genomic Hindlll 0 fragment (Oram et al, (1982) J. Gen. Virol., 59, 111-129). pAT153/HCMV Hindlll 0 was cut with Hindlll, the 8683bp HCMV Hindlll 0 fragment isolated and cut with Stul into fragments of 6728 and 1955 bp. The 6728 bp Hindlll-Stul fragment was further digested with BamHI-EcoRI and a 1370 bp BamHI-EcoRI fragment isolated and cloned into plasmids pUClδ and pUC19 (Yanisch-Perron et al, (1985), Gene 33, 103-119) that had been cut with BamHI and EcoRI, to form plasmids pSB-l8/02 and pSB-19/02, respectively. Plasmid pSB-19/02 was used for the removal of the additional ATG upstream of the HLA gene by Bal31 exonuclease trimming. BamHI cut pSB-19/02

was digested with Bal31 for 18, 22, 26 or 30 minutes, the ragged ends of DNA repaired with Klenow DNA polymerase and then digested with EcoRI. Isolated DNA fragments were cloned into Smal and EcoRI out pUClδ. Thirty six clones were isolated and sequenced and one identified that had lost the first two nucleotides of the upstream ATG codon. This clone was called pSBH301-26.

The modified HLA gene was transferred from pSBH301- 26 into vaccinia virus by first cloning it into plasmid vectors pGS62 (Smith et al (1987). Virology, 160, 336-345) and pRK19- (pRK19 is constructed in a similar way to pGS62 (Smith et al, supra) but instead of the 7» promoter it has the vaccinia late 4b gene promoter (Rose et al , (1985). J. Virol. , 56, 830-836). A 1261 bp BamHI-EcoRI fragment was purified from pSBH301-26 and inserted downstream of vaccinia promoters p7.5K (Smith et al , supra) or p4b (Rose et al , supra) in plasmids pGS62 and pRK19, respectively. Resultant plasmids pl02 and pl03 were used to generate TK ^" recombinant vaccinia viruses V102 and V103 using previously established methods (Mackett, et al (1984). J. Virol. 4_9_, 857-64) .

Expression of the HLA gene by recombinant vaccinia viruses V102 and V103 was demonstrated by infecting CV-1 cells with these viruses, radiolabelling the cells with 35s-methionine and analysing extracts of the infected cells by polyacrylamide gel electrophoresis and autoradiography.

Cells infected with recombinant vaccinia virus V102 or V103 but not HCMV gB-VAC (Cranage, et al, (1986). EMBO J. 5_, 3057-63) or WT vaccinia virus or uninfected cells synthesise a diffuse protein of M _r 71K. This represents the product of the HLA gene and Is considerably larger than the primary amino acid chain of 39-5 kD predicts. The difference is most likely attributable to the 13 potential N-linked glycosylation sites. The diffuse nature of the band is consistent with the presence of glycosylation.

Next we tested whether antibody raised against the HLA protein expressed in bacteria would recognise the protein made by recombinant vaccinia virus V103- Antibodies from rabbits 4l and 42 were able to immunoprecipitate a M _r 71 protein from V103 infected cells but not from uninfected cells or cells infected with another TK~ vaccinia recombinant (H307). Similarly, these antibodies were able to detect the same polypeptide in V103 infected cells by Western blotting.

It will be apparent that the nucletide sequence of Fig. 2 could be varied, within the constraints of the genetic code, to encode the same polypeptide. Moreover, the polypeptide itself may be varied from that shown in Fig. 2, for example by way of amino acid addition, deletion, substitution, insertion or inversion, while retaining the essential antigenicity characteristic of HCMV. Such variants may arise as natural alleles , or may

be produced synthetically by methods well known in the art. Antibodies of the present invention include not only entire immunoglobulins, but also fragments thereof having the characteristic complementarity determining regions (see eg. WO 86/01533). The monoclonal antibodies hereof may be those raised by immunization and produced from hybridomas , or alternatively they can be expressed from recombinant host cells transformed with DNA encoding the immunoglobulin. The antibody may, furthermore, be a hybrid having different binding characteristics in its variable regions, and/or a hybrid comprising the variable or complementarity determining regions produced by immunizing a non-human animal with the polypeptide hereof and constant regions obtained from a different antibody such as a human immunoglobulin (see eg. GB 2188638A and EP 173494A) .