WO2004052394A1 - Broad-spectrum lps based vaccines of unencapsulated strains of haemophilus influenzae and other pathogenic species of gram-negative bacteria - Google Patents

Abstract

Vaccines prepared using purified LPS of unencapsulated (non-typeable) Haemophilus influenzae detoxified by a strategy which leaves structurally intact the native Lipid A moiety, presereing broadly cross-reactive epitopes expressed by an antigenic supramolecular structure composed of at least 50 LPS monomers. The detoxification strategy preferentially involves the formation of the equimolar complex between by the Lipid A moiety of LPS and the cyclic synthetic peptide with the sequence NH2-Lys-Thr-Lys-Cys-Lys-Phe-Leu-Leu-Leu-Cys-COOH, CysA-Cys10 disulfide. The same strategy of detoxification resulting in the preservation of the supramolecular, micelle-like, structure of native NTHi LPS is extended to LPS antigens purified from other pathogenic species of bacteria, in order to generate the homologous endotoxoid vaccines.

Description

BROAD-SPECTRUM LPS-BASED VACCINES OF UNENCAPSULATED STRAINS OF Haemophil us infl uenzae AND OTHER PATHOGENIC SPECIES OF GRAM-NEGATIVE BACTERIA

The obligate parasite Haemophilus influenzae is a bacterial species which belongs to the genus Haemophilus, which is part of the normal flora of the respiratory tract of humans and many animal species. Within the species, there are strains containing a carbohydrate capsule (encapsulated strains) and strains that do not express the carbohydrate capsule (unencapsulated strains) . Unlike the .encapsulated H. influenzae type-species, the non-typeable H. influenzae lacks the carbohydrate capsule responsible for the known serological classification (type a, b, c, d etc.) obtained by using the carbohydrate capsule-specific polyclonal and monoclonal antibodies. Hence, the unencapsulated strains of H. influenzae are also known as non-typeable H. influenzae (NTHi) . This species is often responsible for a serious pathology in infants and young children, known as chronic otitis media that can develop systemic meningitis. NTHi strains colonize about 85 % of humans and is one of the most common Otitis Media pathogen, accounting for about 30-35 % of all cases of this disease, the other significant pathogens being Streptococcus pneυmoniae, capsulated Haemophil us infl uenzae type b and Moraxella catharralis . It has been calculated that by three years of age, 50 to 70 % of all children will have experienced at least one episode of Acute Otitis Media (AOM) . AOM is associated with mastoiditis, brain abscesses and acute meningitis. The most important toxic antigen of NTHi strains is Endotoxin or Lipopolysaccharide (LPS) , which is responsible for the damages to the tissues of the internal ear, once the NTHi strain has established a local infection.

An attempt to classify various unencapsulated strains of H. influenzae (NTHi strains) from the serological point of view led, in the past decade, to the identification of five sub-types on the basis of the anti-Lipopolysaccharide (LPS) monoclonal antibody reactivity (Campagnari et al . , Infect. Immun. 55 : 882- 887, 1987) . The reason for the identification of multiple sub-types through the use of monoclonal antibodies, seems to reside at least in part, in the phenotypic variation of LPS which is expressed on the surface of the bacterial cell as the most important toxic antigen of NTHi. This phenomenon is associated to the phase-variation of the species Haemophilus infl uenzae (Weiser , Trends Microbiol., 8 : 433-435, 2000) . Indeed, structural variations within the molecular structure of the oligosaccharide moiety of LPS, like the phosphorylcholine (ChoP) and the

disaccharide sequence Gal α-l,4-Gal, have been reported in the literature and associated to genotypic variations (Weiser et al., Cell, 59 : 657-665, 1989 ; High et al . , Mol. Microbiol., 9 : 1275-1282, 1993). For instance, the phase-variation associated to ChoP is the result of changes in the number of 5'-CAAT-3' repeats within the lie 1A allele which encodes a choline kinase. Furthermore, post-translational effects on the carbohydrate moiety of LPS due to the rapid variation of the bacterium exposed to the selective pressure within the host, are responsible for further heterogeneity within the molecular structure of LPS. A specific case of this phenomenon has been reported in the case of non-hexose groups like Phosphocholine and Phosphoethanolamine (PEtN) which, in different strains of NTHi bacteria, may be linked to different monosaccharide residues or be completely absent ( Risberg et al . , Eur. J. Biochem., 243 : 701-707, 1997 ; Risberg et al . Eur . J.Biochem. , 265 : 1067-1074, 1999). At this purpose, it has been reported that the exchange of lie ID alleles resulting in a kinase sequence differing in 29 out of 226 amino acids between two strains is sufficient to switch the position of ChoP on the oligosaccharide moiety of NTHi LPS (Weiser and Pan, Mol.Microbiol. 30, 767-775, 1998) .

Thus, the different phenotypes present within the NTHi species derive from the expression of various antigenic determinants (epitopes) . It is therefore inferred, that the possibility to develop a candidate LPS-based vaccine with very few or, ideally, just one LPS antigen is crucially dependent from the identification of the molecular requisites that NTHi LPS must express in vivo, in a mammalian host, for the recognition by the host immune system of candidate shared epitope(s) among LPS of various NTHi strains.

The five NTHi LPS sub-types above referenced, were identified by the corresponding five monoclonal antibodies (Mabs) which could recognise different epitopes within the molecular structure of the various NTHi LPS. Very recently, a panel of three murine Mabs has been reported, recognising altogether about 90 % of the 155 NTHi strains clinically isolated, although none of them exceeded singularly the 29 % of specificity (Ueyama et al . Clin. Diagn. Lab. Immunol. 6 : 96-100, 1999) .

These findings show that, so far, there have been significant problems in the identification of a common, broadly shared determinant (epitope) , within the structure of NTHi LPS and suggest that the approach of using monoclonal antibodies (Mabs) , while most appropriate for investigation of the antigenic variability within the molecular structure of the carbohydrate moiety of LPS (intra-strain variability) , can be hardly useful for identification of a common epitope (s) present among the variety of molecular entities (epitopes) of LPS belonging to various strains of NTHi bacteria (inter-strain variability) recognised by the host immune system which reacts with LPS exclusively through serum polyclonal antibodies (PAbs) .

To clarify this important issue, it is fundamental to consider the physico-chemical structure of NTHi LPS and the relation of it with the expression of its imiuunochemical features. NTHi LPS is a R (rough) -type LPS, also known as LOS (lipooligosaccharide) in that its structure contains a lipid A moiety (responsible for the toxicity of the molecule) and a short carbohydrate motif Heptose-Heptose-Heptose-KDO which serves as backbone supporting additional monosaccharide and functional residues, depending from the phenotypic heterogeneity of the bacterial strain (Helander et al . , Eur. J. Biochem., 177 : 483-492, 1988 ; Gibson et al . , J. Bacteriol., 175 : 2702-2712, 1993). In addition, NTHi LPS shows, as any other species of LPS, a variety of physical forms in different environmental conditions, which range from monomeric units (one single molecule encompassing lipid A covalently bound to the carbohydrate moiety) to multimeric units (association of several monomeric units through the Lipid A moiety) as well as aggregate super-structures that may form micelle-like systems (Seydel et al., Eur. J.Biochem., 186 : 325-332, 1989 ; Brogden and Phillips, 1 : 261-278, 1988 ; Rustici et al . , Science, 259 : 361-365, 1993 ; Brandenburg et al . , 218 : 555- 563, 1993) . When in supramolecular structure LPS expresses, through its Lipid A moiety, the binding site for surface receptors present on macrophages. Binding of Lipid A to these specialised cells of the immune system triggers the cascade of events responsible for the biological activity of LPS. The binding site has been mapped and selectively saturated on the basis of synthetic peptide structures (Rustici et al., Science, 259 : 261-278, 1993) that have opened the possibility to safely immunise a mammalian host by a stable complex antigen peptide-LPS which retains the immunogenic characteristics without the toxicity of LPS (Velucchi et al . , J. Endotox.. Res., 4 : 261-272, 1997) .

The present status of the art, however, does not provide information on the role that the physical form of LPS (monomeric, multimeric or micelles) , particularly NTHi LPS, plays in the optimal induction of antibodies related with their optimal specificity, upon LPS presentation to the immune system of mammalians.

Otitis media is a chronic infection of the middle ear due to several species of pathogenic bacteria. NTHi strains are the most frequent bacterial isolates from the middle ear effusions. NTHi LPS is actually found in the middle ear effusions and is responsible for the release of the pro-inflammatory cytochines alpha-tumour necrosis factor (TNF alpha) and interleukine 1 beta (IL-1 beta) . A correlation has been found between the concentration of LPS and that of TNF-alpha and IL-1 beta, in the middle ear effusions. As a consequence of the persistence of LPS in the tissues of the middle ear, even when the bacteria are no longer alive due to antibiotic treatment, the ciliated epithelial cells are damnaged and chronic otitis media is established. More details of the clinical process of otitis media as well as of the importance of NTHi LPS in the establishment of this pathology, can be found in the following bibliography :

Li DJ et al . , Ann. Otol. Rhinol. Laryngol . , 99 : 33-34 (1990) ; Gu XX et al . , Infect. Immun., 63 : 4115-4120 (1995) ; Morrison DC et al . , J.Endotox. res., 1 : 71-83 (1994) ; Willet DN et al.,Annals Othol . Rhinol. Laryngol., 107 :28-33 (1998).

The approach of using monoclonal antibodies (Mabs) for selecting a shared antigenic determinant (epitope) among the variety of NTHi LPS belonging to pathogenic NTHi strains (therefore an epitope useful to predict that it may be significantly representative of the antigenic repertoire recognised by the polyclonal antibodies (PAbs) specifically induced in a mammalian host by various strains of NTHi LPS) is limited by at least two factors : 1- The low probability to select, in the screening process, the useful MAb which recognises the searched shared epitope among a variety of other antigenic determinants

2- The physical form of LPS may not be significantly relevant in the interaction with the monoclonal antibody (MAb) . In fact, the Mab would react only with a structurally simple determinant present in the carbohydrate sequence of LPS, independently from whether LPS is showing a supramolecular (e.g. : multimeric and micelle-like) structure or a monomeric structure. This is due to the fact that the combining site of a monoclonal antibody, as that of any other antibody, cannot accomodate more than six to eight monosaccaride residues (Kabat, J. Immunol., 97: 1-11, 1966) and due to the clonal nature of the MAb, all the antibodies of a given clone are directed against the same target. Accordingly, a rigorous preparation of monoclonal antibody should not exhibit any precipitating activity for the corresponding antigen, as it does not have the capability to simultaneously bind to two different epitopes and form a high- molecular weight network that can precipitate from aqueous solutions. In contrast, a polyclonal population of antibodies (PAbs) contains many clones of antibodies for different specific antigenic determinants (epitopes) . To simplify, one may think that all the antibodies showing the same specificity for a given epitope of the antigen (even if their affinity for that epitope is not strictly the same) may be represented as a clone within the polyclonal population present in the antigen-specific antiserum.

It is therefore inferred, from these considerations, that in order to identify a population of antibodies with broad specificity for NTHi strains expressing heterogeneous LPSs, not a monoclonal but rather a polyclonal population of antibody should be used, in order to detect not just the specificity for the simple carbohydrate structure of the monomeric NTHi LPS but, and most importantly, the specificity of the entire population of polyclonal antibodies for any epitope potentially expressed through the supramolecular structure of NTHi LPS (multimeric, aggregate and eventually micellar) which contains several copies of the simple (monomeric) unit of LPS. The work of Gu et al., Infect. Immun., 64 : 4047-4053 (1996) and that of Wu and Gu (Infect. Immun., 67 : 5508-5513 (1999) teach how some preparations of chemically-detoxified NTHi LPS (O-deacylated LPS) may express a certain degree of immunogenicity by induction of polyclonal antibodies in animal models, only after coupling with a T-lymphocyte dependent carrier protein like Tetanus Tocxoid and OMP (outer membrane proteins) . However, the authors' conclusions in the former work were that multivalent conjugates derived from few types of NTHi LOSs would be necessary to cover most NTHi strains, while the authors' conclusions of the latter work were that NTHi-dLOS conjugates were specific for only 40 % of the NTHi strains analysed. In both referenced works, the molecular characteristics of NTHi LPS were completely ignored with respect to their role played in the specificity of the immune response induced. Thus, on the basis of the present knowledge, one can conclude that the current literature is silent on the rationale for the design of a LPS-based vaccine expressing optimal features of cross-reactivity for all the pathogenic strains of NTHi. The invention provides an antigenic material for preparing a vaccine which contains LPS of non-typeable Haemophil us expressing broadly cross-reactive epitopes through an antigenic supramolecular structure composed of several units of (LPS monomers) _n , where n is an integer with a minimum value 50 monomeric units ; each LPS monomer being composed of the following minimal structure:

R"" Heptose (l-»2) Heptose (l->3) Heptose (l-»5) [2-keto-3-deoxy- octulosonic acid]

R' R" R'" n

2

I 6 J

R4'-GlcN 1 -» 6 GlcN-Rl R2', 3' R2,3 R5", 6" where :

Rl and R4' are independently either hydrogen or phosphate or phosphoethanolamine residues at position 1 and 4' of the two GlcN residues of the Lipid A moiety;

R2,2' are amide-linked fatty acid residues and R3,3' are ester-linked fatty acid residues at the respective positions of the two GlcN residues of the Lipid A moiety;

R5", 6" are two additional fatty acid residues respectively linked, via ester-bond, to each of the two fatty acid residues in position R2',3' ;

R' , R" and R"' are independently either mono-hexose or di-hexose residues at the positions 2, 6, 4 of the three Heptose residues, respectively, which mono or di- hexose may be partly phosphorylated by phosphocholine residues;

R"" are independently either hydrogen or phosphate or phosphoethanolamine residues at position 4 of the 2- keto-3-deoxy-octulosonic acid.

Preferred examples of Rl, R4 are : phosphate groups of R2,2' are : -0-CO-(3)- hydroxytetradecanoic acid of R3,3' are : -NH-CO-(3)- hydroxytetradecanoic acid of R5'',6''are : -O-CO-tetradecanoic acid Preferred examples of R',R'',R''' are : glucose or galactose or lactose residues phosphorylated by one residue of phosphocholine .

Preferred examples of R' ' ' ' are : phosphoethanolamine residues.

The invention also discloses vaccines which are made by optimal detoxification the antigenic NTHi LPS and methods of inducing immunity to infections by unencapsulated (non-typeable) Haemophil us strains.

Accordingly, it is a primary object of the invention to provide means for the identification and selection of the minimal and optimal features necessary to the molecular structure of NTHi LPS, whether alone or in conjugated form, for inducing in a mammalian host a polyclonal antibody immune response which is broadly specific for the pathogenic strains of NTHi (specific for at least 70 % of the pathogenic NTHi strains) . It is also the object of this invention to provide a method for the preparation of a broadly cross-reactive, non-toxic, LPS-Peptide complex (Endotoxoid) -based vaccine for prevention of otitis media and meningitis due to pathogenic, unencapsulated, NTHi strains. Finally, it is also the object of the invention to transfer the knowledge of the discovered optimal immunogenic features of NTHi LPS to LPS of other species of pathogenic Gram-negative bacteria for the preparation of safe and effective vaccines.

These and other objects of the invention will become apparent from the appended specification.

Three different physical forms of NTHi LPS have been prepared for properly investigating the minimal and optimal features of the molecular structure of NTHi LPS which retains complete antigenic properties :

1- monomeric

2- multimeric in supramolecular structure 3- multimeric in aggregate supramolecular structure forming micelle-like systems.

The source of LPS was a pathogenic strain of non- typeable Haemophil us infl uenzae, isolated from the middle ear effusion fluid of a child suffering of recurrent otitis media (Strain NTHi BY47) . The bacterial strain was cultivated basically in the conditions reported in the literature (Gu et al . , Infect. Immun. , 64 : 4047-4053, 1996). Purification of NTHi BY47 LPS was performed by the classic phenol-water method followed by pelletting of LPS by ultracentrifugation at 10⁵x g x 8 hours. Absence of contamination by proteins was ascertained by amino acid analysis which yielded a protein content lower than 0.5 % (w/w). Absence of contaminating nucleic acids was ascertained by measuring the OD of the LPS solution at the wavelength of 260 nm, in comparison to a reference solution of nucleic acids at the

concentration of 50 μg/ l which gives OD = 1.000. Contamination by nucleic acids was estimated to be less than 0.5 % (w/w). The three physically different forms of LPS were prepared from a stock solution of highly purified NTHi BY47 LPS, at the concentration of 1 g/ml in PBS.

Monomeric NTHi LPS is prepared by dissolving NTHi LPS BY47 in the Ca⁺² and Mg ²⁺ chelating reagent 0.1 M sodium citrate pH = 6.5, at the concentration of 1 mg/ml. The Calcium ions present in the aqueous buffers used during the purification of LPS as well as in the organic fluids of mammalians, serves as ""molecular glue" that allows several monomeric units of LPS to stick together through non-covalent chemical bonds involving bivalent Calcium ions and the negatively charged Phosphate groups of the Lipid A moiety of LPS. Through Calcium ions, LPS preparations may form multimeric aggregates and super-aggregates in micellelike systems that contribute to the supramolecular characteristics of the molecule in a given environment. By chelating the Calcium ions in opportune conditions, this reagent dissociates (in non- denaturating conditions, that is without chemical manipulation of the immunodeterminants or epitopes) LPS multimers down to monomers, that can be then kept as such in this buffer. The NTHi LPS solution treated with Sodium Citrate was then sieved by molecular sieving on a Sephadex G-100 column (Pharmacia, Uppsala, Sweden) and the peak of LPS eluted at the Vi (total volume) of the column was recovered and analysed with respect to the chemical composition relative to Glucosamine content, Phosphate residues, primary Amino goups and total hexoses as well as with respect to biological activity of Lipid A using the classic LAL test (Limulus Amoebocyte Lysate) . The Vi of a Sephadex G-100 column elutes molecules with a molecular mass in the range 2,000 - 4,000 Daltons. When investigated by MALDI-TOF technique, the most abundant entity of LPS collected at the Vi gave a molecular mass corresponding to 3,350 Daltons, in the range value of the theoretical mass calculated for the monomeric unit of the deep-rough NTHi LPS structure, containing Lipid A and the motif Hep-Hep-Hep-KDO supporting few phosphorylated monosaccharide residues (Helander et al . , Eur. J.Biochem., 177 : 483-492, 1988 ; Risberg et al . , Eur. J. Biochem., 265 : 1067-1074, 1999; e.i. structure of the LPS of Haemophilus influenzae mutant strain RM 118-26) . When this preparation of monomeric NTHi LPS

(100 μg/ml) was exposed to the direct observation through electron microscopy using the quick-freeze, deep-etch, rotary-replication transmission technique at 52,000 magnifications, neither aggregates nor micellelike systems could be observed in comparison to the native, purified, solution of NTHi LPS (Fig. 1A) .

Multimeric NTHi having a supramolecular structure is prepared by dispersing anhydrous NTHi LPS BY 47 (50 mg) in absolute ethanol and treating the dispersion for 2 hours at 25 °C with the nucleophilic reagent Sodium Hydroxide-Ethyl Alchool (95 %) in order to hydrolize the ester-linked residues (O-acyl residues) of the Lipid A moiety. This treatment (Seid and Sadoff, J. Biol. Chem., 256: 7305-7310, 1981) only hydrolyses four out of the six fatty acid residues present in the structure of Lipid A, but does not touch the phosphate esters of the residues present in the NTHi LPS structure. The procedure yields NTHi LPS which is still in multimeric, supramolecular form, although unable to form micelle-like systems, as ascertained by electron microscopy performed in the conditions and with the technique above reported. The treated NTHi LPS was then sieved on Sepharose 6B, where it was eluted as a simmetric peak (Kd = 0.2) . The eluted fractions were pooled and assayed for chemical identity as well as for LAL activity, as above reported. When investigated by MALDI-TOF technique, the most abundant entity present in the poly-dispersion system eluted as a symmetric peak with Kd = 0.2, gave a molecular mass of 10⁵'² Daltons. Given the average molecular mass of the monomeric unit above reported (3,350 Daltons) these values correspond to an average number of 50 associated monomers present in the most abundant molecular entity of the O-deacyl NTHi LPS.

Investigation of this NTHi LPS preparation (100 μg/ml) by electron microscopy, at 52,000 magnifications vis-avis with the native, purified, preparation of NTHi LPS BY47, did not give evidence of supramolecular aggregate structures forming micelle-like systems (Fig. IB) . Multimeric NTHi having a supramolecular aggregate structure expressing a micelle-like system, is prepared as follows. Native, purified, NTHi LPS BY47 is solubilized by ultrasounds in buffered saline (0.15 M Sodium chloride-0.01 M sodium phosphate, pH = 7.2 ) at the concentration of 100 ug/ml, keeping the temperature at 37 °C. In these conditions NTHi LPS BY47 appears at the electron microscope (52,000 magnifications) as a supramolecular aggregate structure forming a micellelike system (Fig.lC). This physical form is explainable on the basis of the LPS monomeric structure which contains both hydrophobic (all six fatty acid residues of Lipid A) and hydrophilic (the oligosaccharide chain) moieties, mimicking the structure of a detergent or surfactant. Under appropriate conditions (aqueous systems containing high concentrations of Ca⁺² and Mg⁺² ions, like physiological buffers as well as serum, plasma and blood) LPS monomers have the property to stick together assuming the configuration of multimeric aggregates which, as in a micelle-like system, are in a state of minimal energy appropriate to stabilise the molecule in a hydrophilic environment. This configuration requires that the hydrophilic moiety (the oligosaccharide chain of LPS) be exposed to form hydrogen bonds with water, while the hydrophobic moiety (the fatty acid residues of Lipid A) be internalised and confined within an area containing a minimal amount of water. This configuration, defined as tail-to-tail association, is the typical architecture presented by the lipids in the cell membranes and well explains the electron microscope-detected LPS structure as LPS is a glycolipid molecule constituent of the outer membrane of Gram-negative bacteria. It therefore appears that this configuration is also a characteristic of native, purified, LPS in aqueous environment, as shown in Fig.lC. This preparation of NTHi LPS, when processed by molecular sieving on Sepharose 6B was completely excluded from the gel matrix and recovered at the Vi of the column. However, in a Sepharose 2B column, the preparation showed a symmetric peak at Kd = 0.5. The eluted fractions were pooled and analysed by chemical composition and by LAL activity as above reported. By MALDI-TOF technique, this preparation yielded a molecular mass of 10⁵'⁹ for the most abundant entity. This value corresponds to an average number of 250 monomeric units for the most abundant molecular entity of NTHi LPS heterogeneous molecular mass system, with a minimal value of 50 monomeric units.

Very recently, new insights on the fundamentals of the formation of self-assembling systems and micellar systems have been achieved through the use of Block

Copolymers (van Hest et al . , Science, 268 : 1592-1595,

1995 ; Zimmerman et al . , Science, 271 : 1095-1098,

1996 ; Harada and Kataoka, Science, 283 : 65-67, 1999) . Those studies proved that, in order to achieve an aggregate status and multimolecular micellization, pairwise recognition through hydrogen bonds of oppositely charged polymer strands had to occur on the basis of length. In other words, only pair of block copolymers of equivalent length could match and form poly-ionic complex micelles. These findings might offer a theoretical explanation of how the observed multimeric association in LPS preparations does occur in aqueous milieu (Fig. 1C) , since each LPS monomer can be considered as an entity composed of substantially equal-length monosaccharide residues supporting charged phosphate residues and hydrophobic fatty acid residues. The Lipid A binding site was determined in the three preparations of NTHi LPS as follows: LPS expresses its biological activity by interacting with proteins and receptors of the immunosystem through its Lipid A moiety. The molecular mapping concerning the characteristics required by the binding site of Lipid A in the recognition of peptide structures was achieved a few years ago (Rustic! et al . , Science, 259 : 361- 365, 1993) . Specific synthetic peptides with restricted conformation (cyclic) were designed in that study, which optimally saturated the Lipid A binding site and inhibited the biological activity of LPS on the basis of selective competition (antagonism) with LPS-binding receptor proteins (Porro, Trends Microbiol., 2 : 65-66, 1994) . The selectivity of binding of Synthetic Anti- Endotoxin Peptides (SAEP) was assessed with LPS of different species, all showing a supramolecular structure in aqueous solutions.

When the supramolecular structure of LPS was lost by the use of the Ca⁺²- chelating reagent Sodium Citrate, LPS showed a significantly reduced toxicity in LAL assay and SAEP could not bind Lipid A anymore

(Velucchi et al . in : Vaccines '94, Norrby, Brown,

Chanock, Ginsberg Eds., Cold Spring Harbor Laboratory

Press pp. 141-146, 1994), a finding suggesting that the Lipid A binding site can be expressed only by multiple association of LPS monomers. On these basis, a further assessment of the retention or loss of the supramolecular structure for LPS, would be the capability to selectively bind SAEP through its Lipid A moiety. Table 1 shows that only the two preparations of NTHi LPS conserving a multiple association of monomers resulting in a supramolecular structure were capable to selectively bind SAEP and Polymyxin B, while the monomeric preparation of NTHi LPS did not show any binding activity. The finding would therefore have direct implications on the biological (toxicological and immunological) activity of LPS, suggesting that the degree of supramolecular structure expressed "in vivo" by NTHi LPS could result in the corresponding expression of biological activity. Accordingly, we have prepared three formulations of vaccines with the three different physical forms of NTHi LPS, for directly investigating "in vivo" their immunogenic potential in terms of antibody immune response induced and specificity of the antibodies elicited.

Immunological studies were carried out in Swiss Webster mice to determine the immunological properties of the three preparations of NTHi LPS. Previous immunological studies have clarified the crucial role played by an integral Lipid A structure within the whole structure of LPS. In particular, Lipid A is responsible for the T-cell activity of LPS as well as of its immunological characteristic to behave as a T-cell dependent antigen (Tough et al., J. Exp. Med., 185 : 2089-2094, 1997 ; Velucchi et al . , J. Endotox, Res., 4 : 261-272, 1997). According to these studies, only LPS possessing the entire Lipid A structure can work as a true immunogen. Therefore, the two physical forms of NTHi LPS showing either an incomplete (O-deacylated) Lipid A moiety or a monomeric LPS loosing the Lipid A-driven superstructure, would not have the capability to work as immunogens per se. For this reason, both physical forms of NTHi LPS BY47 have been covalently linked to the T-cell dependent carrier protein BSA, in the conditions formerly reported (Velucchi et al . , J. Endotox. Res., 4 : 261-272, 1997).

In contrast, NTHi LPS prepared as multimeric aggregate superstructure in a micelle-like system can be an immunogen per se as it possess the entire, biologically active, structure of Lipid A (Velucchi et al . , J. Endotox. Res., 4 : 261-272, 1997). However, in order to achieve homogeneity of the immunization protocol, this form of NTHi LPS was also prepared as a glycoconjugate, using BSA as carrier protein and the same conditions of reaction. The physico-chemical characteristics and the dose of immunisation for all the immunogens prepared are reported in Table 2.

Groups of five SW mice each, were injected with three doses, three weeks apart, of each immunogen. Bleedings were performed just before and two weeks after each injection was given. Serological analysis was performed by ELISA assay, measuring the level of IgG and IgM isotype antibodies to NTHi LPS BY47 . The results of animal immunizations are reported in Table 3.

The specificity of the polyclonal antibodies induced in SW mice by the three different physical forms of NTHi LPS was determined using murine antisera. The concept of Specificity used in this Application for the polyclonal antibody populations present in the murine antisera is that to quantitate the amount of antibodies specifically binding to an antigen expressing multiple epitopes, according to the classic work of Berzofsky and Schechter (Molec. Immunol., 18 : 751-763, 1981). Since the concept of specificity of antibodies is complementary to the concept of cross-reactivity of antigens, in this work the Specificity of the polyclonal antibody populations induced by the three different physical forms of NTHi LPS has been determined by calculating the ratio between the Minimal Inhibitory Concentration of the reference antigen (MIC Ag ref, that is native, purified NTHi LPS) and that of the antigen under investigation (MIC Ag inv, that is purifird NTHi LPS in different physical forms) . Each MIC was determined in inhibition-ELISA assay by determining the inhibitory activity of the competitor antigens (purified NTHi LPS in the three different physical forms) on the binding curve obtained by the native, purified, NTHi LPS BY47 reacting with its homologous murine antiserum. The experimental basis of determining the Specificity of a polyclonal antiserum by the MIC values of the antigens in binding competition assay was previously reported by Porro et al. (Molec. Immunol., 22 : 907-919, 1985). The inhibition-ELISA assay, using the native highly purified NTHi LPS BY47 as coating agent of plastic plates, was performed basically in the conditions reported for Neisseria meningi tidis LPS (Velucchi et.al., J. Endotox. Res., 4 : 261-272, 1997). The MIC values expressing the results of the relative inhibitory power of the various NTHi LPS preparations, that is their antigenic cross-reactivity which express the complementary specificity of the murine PAbs, are reported in Table 4.

The parallel specificity of the murine monoclonal antibody MAHI-3 for the three different physical forms of NTHi LPS was determined and reported in Table 5. The specificity of MAHI-3 for the "inner-core" sequence Hep-Hep-Hep-KDO of NTHi LPS has been reported by Borrelli et al . (Infect. Immun., 63 : 2665-2673, 1995). The epitope target of this antibody is very important as it determines the specificity of an antibody for an epitope highly conserved within the structure of NTHi LPS. Accordingly, the specificity of MAHI-3 for the three different physical forms of NTHi LPS BY47 is reported in Table 5. As one can see, specificity of MAHI-3 for the monomeric structure of NTHi LPS BY47 is lost, while the specificity for LPS in supramolecular aggregate, micelle-like, structure is not significantly different from that determined for the multimeric LPS in supramolecular structure. Comparison of the specificity of the murine polyclonal antibodies (antiserum) induced by NTHi LPS BY47 with respect to that of the murine monoclonal antibody MAHI- 3 was assessed using binding competition of the two sets of antibody populations for NTHi LPS BY47. In fact, a polyclonal population of antibodies may be imagined as a family of monoclonal antibodies specific for the various, different, epitopes present in an antigenic structure. If a given monoclonal antibody is significantly present within the polyclonal population considered, that monoclonal antibody can be revealed by binding competition with the antiserum containing the same specific antibody, assuming that the two equally specific antibodies have comparable affinity for the same epitope recognised. In contrast, if either the specificity or the affinity of the two antibodies for a given target epitope were different, then one should not observe complete, quantitative, displacement of the former antibody by binding competition ; rather, the parallel binding of the two antibodies should occur. In the case of the present Application, the MAHI-3 monoclonal antibody was used in binding competition for NTHi LPS BY47, against the murine antiserum induced by the NTHi LPS BY47. Given the specificity of MAHI-3 for the "inner-core" sequence Hep-Hep-Hep-KDO, if a significant amount of the antibodies contained in the antiserum were directed against the same epitope target with comparable affinity to MAHI-3 monoclonal antibody, then pre-incubating the MAHI-3 monoclonal antibody with NTHi LPS BY47 would result in a significant inhibition of the binding activity of the antiserum to NTHi LPS BY47. In contrast, if the antiserum would not contain a significant amount of an antibody population specific for the same target sequence recognized by the Mab, together with antibodies specific for additional targets, then one would observe a parallel, additive, binding of the MAHI-3 antibody and the antiserum. Fig.2 shows that the latter case is the one which occurs when ELISA assay is performed with both (Mab and PAbs) antibody preparations, suggesting that the populations of antibodies present within the polyclonality of the antiserum (induced by NTHi LPS BY47) is specific for epitopes that include, but are not limited, to the sequence Hep-Hep-Hep-KDO. Most importantly, the supramolecular structure of NTHi LPS is crucial to express these additional epitopes that may contribute to broaden the specificity of the immune response induced. The results point out that monomeric NTHi LPS, although containing the basic, antigenic, "core" sequence Hep-Hep-Hep-KDO, is not per se sufficient to express the whole antigenic repertoire of native NTHi LPS, which is achieved through the formation of a superstructure featuring the minimal characteristics above reported. In fact, type 2 specificity (partial specificity or determinant sharing) and not type 1 specificity (or complete specificity) is detected when the monomeric form of NTHi LPS is used as inhibitor of the immuno-complex formed by native NTHi LPS and the PAbs (murine antiserum) induced against it (Table 4 and Fig.3). Comparable results were obtained when LPS purified from Neisseria meningi tidis and Salmonella typhi were used in the same analytical model with their homologous murine antisera (Fig. 4 and 5) .

Comparison of the models with respect to the epitope composition within the supramolecular structure of NTHi LPS and the optimal multimeric aggregation required for inducing a broad spectrum specificity to NTHi LPS is summarised in Table 4, on the basis of the MIC value. Complete Specificity (or type 1 Specificity) of the PAbs present in the murine antiserum is only obtained with the two NTHi LPS preparations expressing a supramolecular structure through an optimal multimeric association of 50 to 250 and more monomeric units of NTHi LPS. Each of these monomeric LPS units encompasses the Lipid A structure typical of NTHi LPS (Helander et al . , Eur. J. Biochem., 177 : 483-492, 1988) which is covalently linked to the backbone sequence Hep-Hep-Hep-KDO supporting the phase variable- associated monosaccharide residues which may be phosphorylated by phosphoethanolamine and phosphocholine residues as reported in some examples of the current literature (Risberg et al . , Eur. J. Biochem., 265 : 1067-1074, 1999). However, in order to investigate on the minimum number of LPS monomeric units able to express the supramolecular structure responsible for the expression of a complete antigenic repertoire, which then corresponds to the possibility to induce type 1 Specificity (or complete specificity) in a mammalian host, a further degradation of the molecular size of NTHi LPS BY47 was performed, achieved through a longer specific hydrolytic reaction (18 hours, in the conditions above reported) of the O-acyl residues of the Lipid A moiety. This preparation of NTHi LPS BY47 showed on Sepharose 6B a symmetric peak- with a Kd value equal to 0.5, corresponding to an average molecular mass of 10⁴"⁷, therefore composed of an average number of 15 monomeric units. The antigen was used as inhibitor of the specific reaction performed between native, purified, LPS BY47 and its specific murine antiserum. The results, reported in Table 6, demonstrate that either the MIC50 value or the MIC100 value of LPS in supramolecular structure formed by an average number of 15 monomeric units are significantly higher with respect to those of native, micelle-like, NTHi LPS. The significantly higher MIC value of the 15 mers NTHi LPS preparation reflects a lower amount of antigen bound by the antibody and therefore a lower specificity of the antiserum for that LPS structure .

The direct demonstration of the results shown in Table 6, was achieved by estimating the quantitative amount of affinity-purified IgG polyclonal antibodies binding to the two physically different structures of NTHi LPS. Experimentally, the two forms of NTHi LPS were incubated at comparable concentrations, 30 min. at 37 °C, with a convenient amount of the murine anti NTHi LPS BY47 polyclonal antibodies. Each solution was then sieved on Sepharose 6B and the peaks containing protein antibody and Lipid A were quantitated on the basis of the w/w ratio amino acid (antibody) : glucosamine (Lipid A) . Table 7 shows that a comparable amount of protein antibody binds an amount of native, micellelike, NTHi LPS BY47 which is about 1 log (9.5 times) higher of the amount relative to the 15 mers LPS preparation. The result directly proves that the specificity of the anti NTHi LPS polyclonal antibodies for a native, micelle-like, form of NTHi LPS is significantly higher than that relative to a system of LPS composed of few monomeric units. Once again, this finding stresses the point that an optimal supramolecular structure of NTHi LPS is crucial for inducing an antibody immune response of broad specificity and high affinity.

Table 1 shows that the Lipid A-binding site of NTHi LPS in monomeric form is lost but it remains conserved when LPS retains its supramolecular structure. Here, using the 15-mers LPS preparation, one can show that even the association of few units of monomeric LPS, like in the case of the 15-mers NTHi LPS preparation, is not per se sufficient to properly express the Lipid A binding site. Determination of the optimal expression of the Lipid A binding site has been performed using the criterion of binding Selectivity by the synthetic cyclic peptide SAEP2, as previously reported (Rustici et al . , Science, 259 : 361-365, 1993). Experimentally, equal amounts (0.5 mg) of the 15-mers NTHi LPS preparation and the native, micelle-like, NTHi LPS were mixed together and added of the required amount of SAEP2 sufficient for saturating just the 50% of the Lipid A present in the total amount of NTHi LPS (total of 0.17 mg SAEP / mg Lipid A) . In this way, binding competition of SAEP2 for the two physically different forms of NTHi LPS takes place. The molecular mixture was then sieved on Sepharose 6B and the peak eluted at Kd = 0 (native LPS) was separated from that eluted at Kd = 0.5 (15-mers LPS). The ratio Lipid A / SAEP2 determined in each peak as glucosamine content (Lipid A) and total amino acid content (SAEP2) has been used as parameter to define optimal binding activity and therefore optimal expression of the Lipid A binding site. Table 8 shows the significant difference relatively to the values of the ratio Lipid A / SAEP2 for the two different structures of NTHi LPS. The results outline the finding that optimal expression of the Lipid A binding site in NTHi LPS is also related to the optimal expression of the supramolecular structure of LPS. Therefore, two parameters are useful in clarifying the importance of the supramolecular structure of NTHi LPS : 1- Optimal expression of the cross-reactive epitopes for obtaining the required broad-spectrum Specificity of the polyclonal antibodies

(antiserum) and ; 2- Optimal expression of the Lipid A binding site, detectable as selective saturation by

SAEP2, which results in the Lipid A-mediated immunological activity of LPS.

A murine antiserum induced by a non-toxic form of native NTHi LPS ( the Endotoxoid below described into the details, starting from the native and highly purified NTHi LPS BY47), was then used for probing its specificity against NTHi bacterial strains isolated from hospitalised patients from various European geographical areas. Assessment of the degree of specificity achieved by a polyclonal population of antibodies (antiserum) induced against a randomly- selected LPS from wild type NTHi bacterial strains isolated from different geographic areas, is highly significant for proving the concept of broad antigenic cross-reactivity. Accordingly, more than one hundred and fifty wild type NTHi strains isolated from immunocompetent and immunodeficient hospitalised patients in European clinical centres were assayed by whole-cell ELISA with the murine antiserum induced by the NTHi LPS BY47. The Specificity of the antiserum for the various wild type bacterial strains is reported in Fig.6 and 7. The results show that the murine antiserum to NTHi LPS BY47 contains antibodies able to bind LPS expressed on the bacterial cell of more than 90 % of the NTHi strains clinical isolates. The degree of specificity was detected within a range of approximately 0.5 log, a result expected on the basis of the chemical heterogeneity of the R residues linked to the conserved Hep-Hep-Hep-KDO backbone structure to whom such a significant part of the antibodies are directed (see Fig. 2) .

These results, predictable on the basis of the serum Specificity as detected by the MIC values of different physical forms of NTHi LPS (Table 4), prove that the mammalian immune system recognises an optimal physical form of NTHi LPS, which appears to play a crucial role for inducing an antibody immune response of broad Specificity.

The bacterium non-typeable Haemophilus infl uenzae is a human-restricted pathogen and therefore there are no animal models significantly predictive of the degree of protection achievable by anti-NTHi LPS antibody titres in humans on the basis of the anti-NTHi LPS antibody titres induced in animals. However, some animal models have been proposed (prevention of experimentally induced NTHi species-specific bacteremia in mice / rats and otitis media in chinchillas) that migh be useful for providing some experimental observations on the biological functionality of antigen-specific antibodies. On these premises, the biological functionality of antigen-specific antibodies has been addressed to the bactericidal complement- dependent activity and opsonophagocytic activity (Zagurski et al . Infect. Immun. 68 : 2525-2534, 2000 ; Sun et al . , Vaccine 18:1264-1272, 2000). Accordingly, in order to prove the biological functionality of the polyclonal antibodies induced by the two immunogenic physical forms of NTHi LPS of the present Application, the ability of the induced anti-NTHi LPS antibodies to fix and activate the complement cascade of different mammalian species was assessed, since antigen-specific antibody-mediated activation of the complement system is the most efficient pathway by which antibodies express bactericidal activity. The antigen-specific, complement-dependent PIH (Passive Immuno Hemolysis) method was used to test this characteristic of the anti-NTHi LPS antibodies, in the conditions basically reported by Velucchi et al . (J. Endotox. Res., 4 : 261- 272, 1997) . Analytical comparison of the PIH titers relative to the antisera induced by the two physical forms of NTHi LPS is shown in Table 9.

Preparation of the Vaccine in supramolecular, micelle- like, structure using the Endotoxoid concept

The NTHi LPS proposed as a vaccine must be properly detoxified using "ad hoc" techniques of detoxification, since the vaccine must be safe and immunogenic, therefore retaining the appropriate molecular conformation while avoiding the induction of LPS- specific side-effects in a mammalian host. Different strategies may be considered for the purpose. Among these : 1 - The use of synthetic anti-endotoxin peptides binding with high affinity and selectivity to the Lipid A moiety of LPS (Rustici et al . , Science, 259 : 361- 365, 1993) for detoxification of either free or protein-conjugated LPS-micelles (Velucchi et al., J. Endotox. Res., 4 : 261-272, 1997).

2 - Entrapment in mono-lamellar and multi-lamellar liposomes to reduce the Lipid A-related toxic effects (Bennet-Guerrero et al . , Infect. Immun., 68 : 6202- 6208, 2000) .

3 - O-deacylation of LPS by nucleophilic bases in organic solvents followed by covalent conjugation to a protein carrier in order to recover immunogenicity

(Seid and Sadoff, J. Biol. Chem., 256 : 7305-7310, 1981) .

The preferred embodiments related to a method allowing the most selective detoxification of the NTHi LPS antigen for fully retaining the conformational features required for the optimal expression of the antigenicity and immunogenicity proper of the native NTHi LPS, is here below reported :

LPS is purified from NTHI cells by any conventional method which may include cold phenol extractions followed by fractionated precipitation in 20-80% ethanol and ultracentrifugation at 10⁵ x g x 4-6 hours. The pellet of NTHi LPS is then collected, finely suspended in distilled water and dialysed against 0.1 M CaCl₂ solution for 8 hours in order to form the calcium salt of NTHi LPS. It is then re-precipitated by 80 % ethanol, re-suspended in sterile water and freeze-dried at the concentration of 1.0 to 5.0 mg/ml. This bulk product is then re-suspended in saline and sterile-

filtered by 0.22 μm membrane. A desired amount of such NTHi LPS material (1-5 mg/ml, pH = 6.2), which contains about 50 % of its molecular structure in the form of Lipid A structure (bearing the binding site for mammalian receptors and therefore being responsible for inducing the well known LPS-related toxic effects) is then saturated by an appropriate synthetic conformationally-restricted, anti-endotoxin peptide (SAEP) with the sequence :

NH₂ - Lys-Thr-Lys-Cys-Lys-Phe -Leu-Leu-Leu-Cys-COOH S S by adding at 37 °C, a sterile aqueous solution of SAEP (5-10 mg/ml, pH = 7.6), in a way that the final ratio SAEP / Lipid A (LPS) = 1 (mol/mol) . In these conditions, being the concentration of LPS well above the CMC (Critical Micellar Concentration) , the non- toxic, equimolar, complex SAEP-LPS (named Endotoxoid) precipitates. The precipitate is collected and washed several times with sterile water. It can then be re- suspended in sterile water and freeze-dried, then stored at - 20 °C.

The physical-chemical, immunopharmacological and immunological characteristics of the NTHi Endotoxoid vaccine are reported in Table 10.

The same procedure can be used when a protein conjugate of NTHi LPS is considered, in the place of native NTHi LPS. The procedure for generating such a semi-synthetic glycoconjugate may be the one formerly reported by Velucchi et al., J. Endotox. Res., 4: 261-272 (1997) for LPS immunotype L8 derived from pathogenic strains of Neisseria meningitidis Group A or B or C. Binding of SAEP to Lipid A of NTHi LPS which results in the detoxified SAEP-LPS complex, is performed according to the method above here reported.

The dose of vaccine is very much dependent from the technology used to detoxify at acceptable levels the NTHi LPS preparation. In the case of an Endotoxoid, it is advisable that a range dose between 0.5-5μg NTHi LPS/dose is sufficient to confer protective immunogenicity to the host, based on the pre-clinical experiments here reported. However, those experts in the art of clinical dosing may ascertain the most appropriate dose for humans.

The use of mineral adjuvants such as Al(OH)₃ and AIPO4 (0.1-1.0 mg/ml), widely used in human vaccines, can be also advisable especially in the vaccine formulations.

The invention also includes the preparation Endotoxoid- based vaccines for other types of Haemophilus LPS. Specifically, LPS purified from Haemophilus pleuropneumoniae, once properly detoxified as an Endotoxoid which allows the retention of the optimal physico-chemical characteristics found in the case of NTHi LPS, can be used as vaccine in veterinary medicine to prevent this infection in swine, particularly prone to get this pulmonary pathology (Udeze et al., Am. J. Vet. Res. , 48 : 768-773, 1986 ; Paradis et al . , Infect. Immun., 62 : 3311-3319, 1997).

Furthermore, the present findings on the crucial role played by the supramolecular structure of LPS in the expression of the optimal immunogenic characteristics of the antigen, can be extended to LPS antigens of other significant Gram-negative pathogens, namely the capsulated strains of Haemophilus influenzae, Neisseria meningi tidis , Salmonella typhi , Shigella flexneri _r Vibrio cholerae etc . As an example, Fig. 8 shows the protective activity of S. typhimuri um Endotoxoid at a dose as little as 0.5 ug/mouse, in a CD1 mice population which has been challenged by the homologous pathogenic microrganism, a bacterial strain isolated from humans getting infected by eating contaminated pork meat and well characterized from the genetic point of view (Pontello et al., Epidemiol. Infect. 120 : 209- 214, 1998) .

Tab. 1 Expression of the lipid A binding site within the NTHi LPS structure in the three different preparations

LPS preparation Moles of SAEP2/mg LPS Moles of PmxB/mg LPS

MICELLE-LIKE SYSTEM 0.17+0.02 0.18+0.03

MULTIMERIC 0.16+0.04 0.18+0.04

MONOMERIC

None None

Tab. 2 Physico-chemical characteristics and dose/mouse of the three NTHi LPS-based imπ-unogens

Immunogen LPS/Protein Ratio Dose LPS

⁽w/w⁾ (μg/mouse)

MICELLE-LIKE SYSTEM 5.0

MICELLE-LIKE SYSTEM-BSA 1.35 5.0 conjugate

MULTIMERIC-BSA coniugate 1.48 5.0

MONOMERIC-BSA conjugate 1.55 5.0

Tab. 3 End-point ELISA titers specific for NTHi LPS, of mice immunized by three s.c. injections, three weeks apart, of the iπππunogens listed in Table 2.

ELISA titer

Iirrnunogen IgG IgM

MICELLE-LIKE SYSTEM 51,200+15,000 14,000+4,000 MICELLE-LIKE SYSTΞM-BSA 40,000+10,000 10,000+2,000 conjugate MULTIMERIC-BSA conjugate 25,000+ 6,000 4, 000±1,000

MONOMERIC-BSA conjugate 12,000+4,000 2,000+ 400

P < 0.01 by t-test of the onomeric-BSA conjugate against each of the other three imitiunogens Tab. 4 Inhibitory activity of the three LPS's physical forms with respect to the complex formed between native NTHi LPS and murine PAbs.

NTHi LPS preparation MIC₅₀ (μg/ml) MIC₁₀₀ (μg/ml)

MICELLE-LIKE SYSTEM 10 50

MULTIMERIC 25 200

MONOMERIC 150 >1000

Tab . 5 Specificity of MAHI3 for the three different forms of LPS .

NTHi LPS preparation _MICso (μg/ml) MIC₁₀₀ (μg/ml)

MICELLE-LIKE SYSTEM 15 100

MULTIMERIC 30 200

MONOMERIC >1000 >1000

Tab . 6 Specifici ty of murine PAbs induced to native LPS for either the homologous native LPS or the 15- mers LPS population

NTHi LPS preparation MIC₅₀ (μg/ml) MIC₁₀₀ (μg/ml)

MICELLE-LIKE SYSTEM Ϊ0 50

15-MERS 50 400

Tab . 7 Quantitative amount of affinity-purified murine IgG polyclonal antibody specific for NTHi LPS binding to the native LPS and to the 15-mers LPS preparation

NTHi LPS preparation mg LPS / mg of protein Ab

MICELLE-LIKE SYSTEM 34 / 24= 1 . 42

15-MERS 4 / 27= 0 . 15 Tab. 8 Selectivity of binding of SAEP2 for the 15-mers NTHi LPS preparation in direct binding competition with native NTHi LPS

NTHi LPS preparation μmoles _SAEP2*/mg LP_S Selectivity (%)

MICELLE-LIKE SYSTEM 0.13+0.01 76.5+5

15-MERS 0.04+0.002 23.5±5

* total amount (100 %) of SAEP2 in the reaction mixture

= 0.17 μmoles

Tab. 9 Passive Haemolytic Inhibition (PHI) titers induced by the two forms of NTHi LPS LPS preparation PHI titer (serum dil^{" x})

MICELLE-LIKE SYSTEM 500 MULTIMERIC 200

Tab. 10 Physical-chemical, immunopharmacological and immunological characteristics of NTHi Endotoxoid

BY47

a) Physical-chemical

Molar Ratio SAEP : NTHi LPS in the Endotoxoid 1.0 Weioat Ratio SAEP : NTHi LPS in the Endotoxoid 0.5

Stability of the Endotoxoid at high ionic strength (0.5 M)

As determined by the release of SAEP 100 %

Stability of the Endotoxoid in the pH range (2-11)

As determined by the release of SAEP 100

Stability of the Endotoxoid to serine proteases (Trypsin) as determined by molecular analysis

(AA for SAEP and GlcNH; for Lipid A, in the defined molar ratio) of the recovered (by precipitation) and purified complex antigen, following incubation in the time-range: Trypsin

2 hs at 37 °C 100 %

18 hs at 37 ° 95 %

Stability of the Endotoxoid after storage in sterile PBS At 37 °C for up to 6 months, as determined by the release of SAEP 100 %

b) Immunopharmacologiαal

NTHi LPS NTHiEndotoxoid

Biological activity estimated ϋy LAL 'assay Expressed as concentration (pg/ l) and 25 25,000

(^'Log reduction) - (3.0)

Minimum dose of antigen (ng) releasing Significant serum upon i.v. injection in CDl mice 5 1,000

(Log reduction) - (2.3)

Minimum dose of O-deacylated NTHi LPS Releasing comparably significant TNF and IL-6 ^Λin vivo" upon i.v. injection in CDl mice 5 1,000

(Log reduction) - (2.3)

Minimum dose of antigen (ng) releasing Significant serum TNF and IL-6 "in vivo"

Upon s.c. injection in CDl mice 100 > 10,000 (Log reduction) - (> 2.0)

c) Immunological

Mean of ELISA-detected titres of IgG and IgM isotypes specific for NTHi LPS induced in SW mice by 5 ug of NTHi Endotoxoid and NTHi LPS, respectively, injected by s.c. route. Passive protection test* then followed by using the antisera.

NTHi Endotoxoid

IgG Ratio Wx/Wo IgM Ratio Wx/Wo

Week 0 1, 000 500 _

(CFTJ/ml = 2,750)

Week 2 15, 000 15 1,50 3.0

Week 5 23, 000 23 1,000 2.0

(CFU/ml = 2) p < 0.01

NTHi LPS

Week 0 1, 500 - 800

(CFU/ml = 2,200)

Week 2 20,000 13 1,600 2.0

Week 5 28,000 19 1,200 1.5

(CFU/ml = 4) p < 0.01

* Passive protection test in -day-old infant rats (10 / group) immunized i.p. by 0.1 ml (dil. 1/10) of the anti-NTHi LPS mouse antiserum induced against the NTHi Endotoxoid (week 5) and challenged i.p. 24 hours later with 0.1 ml containing about 40-50 CFU of the homologous NTHi strain BY47. Geometric mean of the detected CFU/ml of blood, in parenthesis, 24 hours post-challenge.

Claims

I Claim

1. An antigen comprising LPS of unencapsulated

Haemophil us influenzae in a detoxified form which allows LPS to retain the supramolecular micelle-like structure through the native Lipid A moiety, such an antigen comprising a Lipid A-mediated association of multiple units responding to the formula (LPS monomers) _n , where _n is an integer with a minimum value of 50 ; each LPS monomer being composed of the following structure :

R""

Heptose (1—2) Heptose (l-»3) Heptose (1-45 ) [2-keto-3-deoxy- octulosonic acid] R' R" R"' n 2

6

R4'-GlcN 1 → 6 GlcN-Rl R2',3' R2,3

R5",6' where :

Rl and R4' are independently either hydrogen or phosphate or phosphoethanolamine residues at position 1 and 4' of the two GlcN residues of the Lipid A moiety; R2,2' are amide-linked fatty acid residues and R3,3' are ester-linked fatty acid residues at the respective D2

positions of the two GlcN residues of the Lipid A moiety;

R5", 6" are two additional fatty acid residues respectively linked, via ester-bond, to each of the two fatty acid residues in position R2',3';

R' , R" and R"' are independently either mono-hexose or di-hexose residues at the position 2,6, 4 of the three Heptose residues, respectively, which mono or di-hexose optionally are partly phosphorylated by phosphocholine residues;

R"" are independently either hydrogen or phosphate or phosphoethanolamine residues at position 4 of the 2- keto-3-deoxy-octulosonic acid.

2. A vaccine which comprises the antigen of claim 1 detoxified through the binding to the Lipid A moiety of

NTHi LPS, alone or in conjugate form, of the cyclic synthetic peptide with formula :

NH₂-Lys-Thr-Lys-Cys-Lys-Phe-Leu-Leu-Leu-Cys-COOH _{s S}

Such a binding resulting in the non-toxic, equimolar, complex Peptide-NTHi LPS .

3. A vaccine for inducing immunity in a host against unencapsulated strains of Haemophilus influenzae according to Claim 2.

4. A vaccine according to Claim 2 which is combined with a vaccine against Streptococcus pneumoniae and/or Moraxella catharralis for the prophylaxis of otitis media and bacterial meningitis.

5. A vaccine according to Claim 2 which is combined with a vaccine against the capsulated strains of Haemophil us infl uenzae and/or Neisseria meningi tidis for the prophylaxis of bacterial meningitis.

6. A vaccine comprising LPS purified from Neisseria meningi tidis in a detoxified form which allows LPS to retain the supramolecular micelle-like structure through the native Lipid A moiety.

7. A vaccine according to claim 6 wherein the Lipid A moiety of Neisseria meningi tidis LPS is bound, alone or in conjugated form, to the cyclic synthetic peptide having the sequence:

NH₂-Lys-Th-Lys-Cys-Lys-Phe-Leu-Leu-Leu-Cys-COOH _{s s}

such a binding resulting in the non-toxic, equimolar, complex Peptide-Neisseria meningi tidis LPS.

8. A vaccine comprising LPS purified from Salmonella typhi in a detoxified form which allows LPS to retain the supramolecular micelle-like structure through the native Lipid A moiety.

9. A vaccine according to claim 8 wherein the Lipid A moiety of Salmonella typhi LPS is bound, alone or in conjugated form, to the cyclic synthetic peptide having the sequence: NH₂-Lys-Thr-Lys-Cys-Lys-Phe-Leu-Leu-Leu-Cys-COOH

_{S s}

such a binding resulting in the non-toxic, equimolar, complex Peptide-Salmonella Typhi LPS.

10. A vaccine comprising LPS purified from Shigella fl exneri in a detoxified form which allows LPS to retain the supramolecular micelle-like structure through the native Lipid A moiety.

11. A vaccine according to claim 10 wherein the Lipid A moiety of Shigella fl exneri LPS is bound, alone or in conjugated form, to the cyclic synthetic peptide having the sequence:

NH₂-Lys-Thr-Lys-Cys-Lys-Phe-Leu-Leu-Leu-Cys-COOH _{s s}

such a binding resulting in the non-toxic, equimolar, complex Peptide-SΛigella flexneri LPS.

12. A vaccine comprising LPS purified from Vibrio chol erae in a detoxified form which allows LPS to retain the supramolecular micelle-like structure through the native Lipid A moiety.

13. A vaccine according to claim 12, wherein the Lipid

A moiety of Vibrio cholerae LPS is bound, alone or in conjugated form, to the cyclic synthetic peptide having the sequence:

NH₂-Lys-Thr-Lys-Cys-Lys-Phe-Leu-Leu- eu-Cys-COOH _{s s} such a binding resulting in the non-toxic, equimolar, complex Peptide- Vibrio cholerae LPS.