A lipopolysaccharide (LPS) -rich outer membrane surrounds gram- negative bacteria. The LPS molecules themselves are made up of three regions, a lipid A region, a core oligosaccharide, and an O side chain, which is highly diverse.

The phenotype of a bacterial strain, and in particular the lipopolysaccharide phase variation, can vary, and as a result, the properties, of the strain, for example, its applicability as a vaccine, can be affected.

For instance, it is known that virulent strains of F. tularensis segregate into a variety of colony types on culture in and on various growth media (EIGELSBACH, H. T. , et al (1951). Journal of Bacteriology 61,557-569). With regard to vaccine strains, it has been suggested that only some types of viable colony, those which have been designated as"blue", are immunogenic and induce a protective immune response when used for vaccination whereas other colony types, and particularly those described as"grey" do not induce a protective immune response (W. D. Tigertt (1962) Bacterial Reviews, 26,354-373).

However, in spite of the considerable period since these observations, the distinctions between the immunogenic and non- immunogenic phenotypes have not been resolved satisfactorily, and the existence of"blue"and"grey"types is largely anecdotal.

There is evidence that differences in the LPS are responsible for different phenotypes (COWLEY, S. C. , et al (1996). Molecular

microbiology 20,867-874), but the relationship of this to vaccine applications is not clear.

Furthermore, it has been reported that monoclonal antibodies directed against the lipopolysaccharide of F. tularensis might be useful in the diagnosis of disease (FULOP, M. J. , et al (1991). Journal of Clinical Microbiology 29, 1407-1412), but also that they may cross react with different phenotypes, including a Cap-strain, which is non-immunogenic.

The successful licensing and manufacture of F. tularensis LVS will require a number of quality control tests throughout the entire production process to ensure that batches of vaccine are safe, efficacious and have reproducible properties. The efficacy of a vaccine needs to be assessed as far as possible, prior to administration. Once administered, even monitoring of a patient's antibody response to the vaccine may not give a clear idea of the level of protection afforded, due to the complexity and variety of the immune response. The best way to ensure protection is to ensure that appropriate doses of immunogen are administered initially.

An ideal vaccine would comprise 100%"blue"phenotype. However, the proportion of blue and grey colony types can vary in different batches of vaccine.

A variety of simple tests might be used to validate the vaccines, which consider different properties (genetic, microbiological, biochemical and physical) of the product.

The FDA has indicated that it is essential that a reliable test for distinguishing between blue and grey colony types is available. A microscope test based on obliquely transmitted light is one possibility, but this test suffers from a degree of operator subjectivity. Of greater concern, the test requires the further growth of bacteria on agar. Furthermore, it has

been shown that lipopolysaccharide phase variation can occur spontaneously in culture (Cowley et al. supra. ) and therefore, the results of this test may not accurately reflect the proportions of blue and grey colonies in the original vaccine.

There is a need therefore, for a simple, reliable and reproducible assay which could be used for quality control (QC) of vaccines such as the LVS vaccines of F. tularensis at all stages in the production process.

According to the present invention there is provided a method for confirming the presence or amount of a particular immunogenic phenotype of a gram negative bacteria in a sample, said method comprising contacting a sample of said bacteria with a labelled binding agent which is specific for a lipopolysaccharide (LPS) of said particular phenotype of said bacteria, detecting the presence of a complex formed by said binding agent and LPS within the sample, and relating that to the presence or amount of said phenotype within the sample.

The method of the invention is particularly suitable for testing the quality of a vaccine strain, such as F. tularensis, where the applicants now believe that the immunogenicity and the levels of protection afforded by the vaccine, is affected by lipopolysaccharide phase variation.

In a particular embodiment of the invention, the binding agent is labelled with a visible label, and the presence of said complex is determined using flow cytometry. In this case, the any cells that are of the target phenotype, will bind to the labelled binding agent, and will provide a appropriate signal when passing through the flow cytometer. other cells will be distinguishable by the absence of label on them.

Particular labels are fluorescent labels such as fluorescein or the like. The fluorescent label is suitably one which can assist in detection in a conventional flow cytometer.

Alternatively, the sample is also contacted with a second labelled binding agent that specifically binds an LPS of a different phenotype, and wherein the presence of a complex between said second labelled binding agent and LPS in a sample is also detected. In this case, the second labelled binding agent carries a different label to the first labelled binding agent, so that this can be distinguished in the flow cytometer.

If desired however, further labelled antibodies may be added to the sample to detect the presence of a variety of phenotypes.

When looking at a range of variants, fluorophores should be selected such that their emission wavelength are suitably distinct, preferably with at least 50 nm difference in emission wavelength.

Such methods will provide a user with an objective, quantitative answer, which will be reproducible to accepted statistical standards. It is more accurate than the current microscope based method, and furthermore, is easier to perform, as the protocol could be followed by non-technical staff.

Furthermore, the method can be carried out in one reaction pot and doesn't require the growing up of colonies (which may add to inaccuracies). It is also much faster, as the analysis of a vaccine strain using the method of the invention may take only approximately 2 hours as compared to about 3 days using a conventional microscope method Suitably, the binding agent is an immunoglobulin, such as an antibody or a binding fragment thereof.

Antibodies or binding fragments thereof may be polyclonal or monoclonal, and these may be produced using conventional methods.

For instance, polyclonal antibodies may be generated by immunisation of an animal (such as a rabbit, rat, goat, horse, sheep etc) with the toxin or immunogenic subunits or fragments thereof, to raise antisera, from which antibodies may be purified.

Monoclonal antibodies may be obtained by fusing spleen cells from an immunised animal with hybridoma cells, and selecting cells which secrete suitable antibodies.

Antibody binding fragments include F (ab') 2, F (ab) 2, Fab or Fab' fragments, as well as single chain (sc) antibodies, FV, VH or VK fragments, but they may also comprise deletion mutants of an antibody sequence. Acronyms used here are well known in the art. They are suitably derived from polyclonal or monoclonal antibodies using conventional methods such as enzymatic digestion with enzymes such as papain or pepsin (to produce Fab and F (ab') 2 fragments respectively). Alternatively, they may be generated using conventional recombinant DNA technology.

Preferably, the labelled binding agent is a labelled monoclonal antibody.

As discussed above, the method is particularly applicable to the analysis of strains of gram-negative bacteria intended for use as a vaccine. Where the gram-negative bacteria is Francisella tularensis, this may therefore comprise F. tularensis live vaccine strain (LVS). Specifically, the method can be used to detect and/or quantitate the amount of"blue"phenotype, which has mutated to grey phenotype in a sample, to check its suitability as a vaccine.

In a particular embodiment, the or each labelled binding agent is specific for an 0 side chain of a lipopolysaccharide.

Specifically, the method of the invention will suitably be carried out by suspending a dry batch of bacterium, and in particular a dry live vaccine strain, in a suitable medium or solvent such as phosphate-buffered saline (PBS). After a suitably equilibration time, for instance, about 30 minutes, the sample may be diluted to the required concentration.

Suitably then, one or more labelled binding partners as described above, and in particular, fluorescently labelled antibodies, is added to the sample. One specific antibody in particular, an anti-LPS antibody, is added for each phenotype to be detected, but where there are more than one, each one is differently and distinguishably labelled. Suitably, the mixture is then incubated to allow antibody-antigen interaction to take place, for example at temperatures of from 5-45°C, such as about 37°C. Generally incubation will be carried out in the dark, for example for a period of from 5 minutes to 1 hour, and generally for about 15 minutes.

In a particular embodiment, the mixture is then added to a flow cytometer by dilution with carrier fluid. The flow cytometer may be a standard laboratory cytometer, provided with suitable excitation laser sources. Each cell passes sequentially in front of a focussed light beam, and the optical characteristics of the cell are determined. For example, the flow cytometer can be set to produce two dimensional charts of side vs forward scatter of the light beam, to sort the cells on the basis of size. Forward scatter vs fluorescence intensity will allow separation quantitation of labelled cells. This latter operating mode is particularly suitable in the context of the present invention.

Where there are more than one, say two labelled binding agents used, each labelled with a different fluorophore, the labelled bacteria can be analysed by 2 channel FACS analysis. The data from this analysis will provide a rapid readout of the proportion of the two respective phenotypes, for example, blue and grey bacteria, in the sample vaccine.

The test described above can be automated to judge vaccine strains acceptable or otherwise. For instance, the original 1955 Soviet vaccine control procedures stated that the 'immunogenic' (i. e. blue) type must constitute 20 to 30% of the total number of organisms in the vaccine'. It is thought that if the LVS vaccine lots can be shown to contain 80% blues or above, they will be considered acceptable.

The results of the test described above can be read and compared with this value using an appropriately programmed computer.

Predictions using this test with 80% blue and above varied by no more that 4%. If eighty percent is the vaccine cut off for acceptability, this test should be appropriate to indicate the percentage of blue phenotypes in a population.

Once this has been determined, the viability of the strain can be tested using conventional methods.

Thus the invention further provides a quality control method for determining the efficacy or immunogenicity or quality of a live vaccine, said method comprising subjecting said vaccine to a method as described above, and detecting the amount of immunogenic phenotype in the vaccine sample. In particular, the live vaccine sample is a vaccine strain of F. tularensis.

According to a further aspect of the invention, there is provided a kit for carrying out a method as described above, said kit comprising a labelled binding agent which is specific for a lipopolysaccharide of a particular phenotype of a gram-

negative bacteria, and in particular a fluorescently labelled antibody which is specific for an 0 side chain of a vaccine strain of gram-negative bacteria such as F. tularensis LVS, as outlined above.

Detection of LPS may be useful in other contexts. For instance, the presence of LPS in a vaccine such as a sub-unit vaccine may be unwanted for other reasons. For instance, the LPS of many types of gram-negative bacteria are toxic to animals. This applies to for example, the LPS of E. coli, which may be used to express recombinant sub-unit vaccines.

The labelled binding agents described above, which are specific for these unwanted LPS, may be used in detection methods as described above, to readily and rapidly detect LPS contaminants in vaccines.

Thus in a further aspect the invention provides a method for detecting LPS in a vaccine such as a sub-unit vaccine, said method comprising contacting a sample of said vaccine with a labelled specific binding agent for LPS, and detecting any complex formed between the complex and vaccine particles.

The invention will now be particularly described by way of example with reference to the accompanying diagrammatic drawings in which: Figure 1 shows a Western blot of F. tularensis LVS, blue, grey and cap-and F. novicida proteinase K digests probed with the anti-F. tularensis monoclonal antibody used in the test. The LVS control preparation was not treated with proteinase K. The characteristic ladder associated with binding of antibody to the 0-antigen of LPS is seen in the blue and LVS samples only.

Figure 2 shows a Western blot of F. tularensis LVS, blue and grey (harvested from vialt or BCGA plates*) and F. novicida,

proteinase K digests, probed with anti-F. novicida monoclonal antibody (FnMab : upper blot) or anti-F. tularensis monoclonal antibody (FtMab : lower blot).

Figure 3 shows the results of a FACS display from blue and grey phenotypes cultured on BCGA. Cells were stained with FITC- labelled anti-F. tularensis monoclonal antibody. Display (3a) shows cells sorted by size. The fluorescence with region R1 from a mixed preparation is shown in is shown by density plot (3b).

(3c) and (3d) show the staining pattern of individual blue and grey phenotypes, respectively.

Figure 4 is a graph showing the percentage of F. tularensis LVS blue phenotypes in NDBR lots of LVS vaccine estimated by the test.

Example 1 Antibody specificity to phenotypes A panel of monoclonal antibodies against F. tularensis and F. novicida (FtMab and FnMab) were tested for their ability to recognise strains of F. tularensis and to distinguish between blue and grey colony forms found in batches of F. tularensis LVS vaccine.

Separate F. tularensis LVS blue and grey cultures were tested and these are listed in Table 1. LVS source Phenotype NDBR lot 4 predominantly blue NDBR lot 4 predominantly grey LVS cap'mutant'grey'type LVS stock derived from predominantly blue NDBR lot 4 LVS NDBR Lot 2 Reported as 99% blue LVS NDBR Lot 4 Reported as 99% blue LVS NDBR Lot 11 Reported as 99% blue F. novicida U112, ATCC grey \ type 15482

Table 1-Source of cultures used for antibody testing The LVS blue cultures provided were'mostly', but not necessarily 100% blue. This could not be assessed by culture as grey colony types take longer to form visible colonies on BCG agar.

F. tularensis HN63 and Schu4 were also tested, as was LVS capsular-deficient strain (cap-) was as described by SANDSTROM, G. , et al. (1988). Infection and Immunity 56,1194-1202.

Cultures were taken from storage at-70°C, streaked on blood cysteine glucose agar (BCGA, with supplements) and incubated overnight. Viable count estimates were incubated at 37°C for 3-4 days for blue variants, and 6-8 days for greys Heat-killed cells or LPS preparations (proteinase-K treated heat killed cells) in PBS were run on SDS-PAGE at 5 ug protein per lane, followed by Western blot. The blots were probed with antibody and detection used the ECL plus kit (Amersham Pharmacia). Antibodies were used at 1/1000 and goat anti-mouse IgG3 conjugate at 1/40,000.

The anti-F. tularensis monoclonals were directed against the LPS 0 side chain present on the blue variant, an example of which is shown in figure 1. They also reacted with SCHU4 (weakly) and HN63 (data not shown) ; however, none of these antibodies reacted with LVS grey and cap-strains, or F. novicida.

The anti-F. novicida monoclonal reacted only with F. novicida (Figure 2). One of the monoclonal panel of antibodies that detected F. tularensis LPS was taken forward to use in the test to distinguish between blue and grey phenotypes of F. tularensis LVS (figure 1).

There results are quite surprising. It has been reported that grey phenotypes possess an abnormal LPS structure, which differentiates them from blue colony types. Cowley et al. supra. reported that the FtMab and FnMab reacted with F. tularensis LVS and F. novicida respectively without cross-reacting, indicating antigenically separate LPS forms. They found that a certain LVS opacity mutant showed a cross-reaction with both antibodies, prompting the initial assumption that F. novicida antibodies would therefore cross-react with the grey phenotype of LVS.

However, this Western blot analysis using the F. novicida antibody showed no cross-reactivity with any of the antigen preparations (including vaccine vial-derived LVS, purified blue and grey colony types and a capsule negative cap-strain of LVS), apart from F. novicida.

Example 2 FACS Analysis A Becton Dickinson FACScan flow cytometer was used to devise a rapid test.

The selected antibody was produced by a cultured cell line, purified by dialysis and conjugated to FITC. 45Xul of cells at 1 x106 dilution were mixed with FITC-labelled antibody, incubated for 15 minutes at 37°C in the dark, then diluted in pre-filtered

Isoton II fluid. This mixture was vortexed and analysed by FACS producing a count of labelled and non-labelled particles of the appropriate size.

BD CellQuest Pro (version 4.0. 1) was used for analysis. The typical presentation of a cell population is shown in figure 3.

The left hand figure (3a) is a density plot of the size profile of all cells measured by forward and side scatter. The right hand picture (3b) represents the fluorescence from an artificially mixed population of blue and grey phenotypes within the ringed region (R1 : operator set). This becomes the chosen area for analysis. The lower two displays show the fluorescence profile of the individual blue and grey cultures (c and d).

Division between the upper (blue) population and the lower (grey) population is set by the operator. The software will then express these populations as a percentage of the total in the R1 area.

Tests carried out on the frozen vials of separately provided grey and blue batches reflected the total amount of each (data not shown).

Example 3 Test of vaccine vials NDBR Vaccine lots 2,4 and 11 (stored at-20°C) were reconstituted in 2ml of sterile distilled water and left at room temperature for at least half an hour prior to dilution and testing by FACS as described in Example 2. All dilutions were made in PBS.

Tests were carried out on NDBR 101 lots 2,4 and 11 with cells at concentrations of 106 and 107 per test (in 45 gel). Figure 4 is a summary of the tests carried on out these vaccine lots.

Error bars are SEM. There was a small difference in estimates when the staining of 106 and 107 cells were compared.

Unexpectedly, the 107 aliquots appeared to give a more accurate

estimation of percentage blues. Estimates for all lots were above 96% for the 106 tests and 98% for 107.

The NDBR101 batch of LVS vaccine contained 99% blue and 1% grey colonies (WAAG, D. M. , et al. (1992). Journal of Clinical<BR> Microbiology 30, 2256-2264; WAAG, D. M. , et al. (1996). FEMS Immunology and Medical Microbiology 13,205-209). These figures are reflected in our results.