WO1993008304A1

WO1993008304A1 - Inhibition of cytokine activity

Info

Publication number: WO1993008304A1
Application number: PCT/US1992/008847
Authority: WO
Inventors: Charles A. Dinarello; Reuven Porat; Burton D. Clark; Sheldon M. Wolff
Original assignee: New England Medical Center Hospitals, Inc.
Priority date: 1991-10-17
Filing date: 1992-10-15
Publication date: 1993-04-29

Abstract

In general, the invention features a method for inhibiting infection by an infectious microorganism in a patient. The method includes administering to the patient a cytokine antagonist. In a related aspect the invention features a method for screening candidate compounds to identify compounds capable of acting as cytokine antagonists. The method includes: a) measuring the growth rate of an infectious microorganism in a growth medium including the cytokine both in the presence and absence of the candidate compound, the growth rate of the microorganism being capable of being stimulated by the cytokine in the absence of the candidate compound and; b) determining whether the candidate compound decreases the growth rate of the microorganism in the presence of the cytokine.

Description

INHIBITION OF CYTOKINE ACTIVITY Background of the Invention This invention relates to treatment of infection by virulent bacteria and to inhibition of interleukin-1 action.

Gram-negative bacilli are one of the most frequent causes of death from infection. The development and outcome of such infection is influenced by a number of factors, one of the most important of which is resistance to killing by normal human serum. The importance of resistance to serum killing is illustrated by the fact that while the vast majority of isolates from the gastrointestinal tract are sensitive to killing (virulent) , more than 85% of the isolates from systemic clinical infections are resistant to such killing (avirulent) .

Numerous explanations have been put forth to explain the resistance of clinical isolates to serum killing. Cell wall components including lipopolysaccharide, capsular antigens and outer membrane proteins have been suggested as elements responsible for resistance. For example, Porat et al. (Infection and Immunity. 55:320, 1987) have suggested that the lipopolysaccharides of virulent E. coli isolates have a higher proportion of highly O-antigen-substituted subunits than the lipopolysaccharides of avirulent strains.

In extreme cases, patients with bacterial infection develop bacterial sepsis leading to shock.

Many of the symptoms of septic shock are caused by host cytokines, e.g., interleukin-1 and tumor necrosis factor, induced by bacterial endotoxin. Okusawa et al. (J^". Clin . Invest . 81:1162, 1988) report that interleukin-1 can induce symptoms like those of septic shock when injected into rabbits.

Summary of the Invention In general, the invention features a method for inhibiting infection by an infectious microorganism in a patient. The method includes administering to the patient a cytokine antagonist. In a preferred embodiment, the cytokine is interleukin-1. In an even more preferred embodiment the antagonist is IL-lra. In another preferred embodiment the infectious microorganism is virulent E. coli . In a related aspect the invention features a method for screening candidate compounds to identify compounds capable of acting as cytokine antagonists. The method includes: a) measuring the growth rate of a infectious microorganism in a growth medium including the cytokine both in the presence and absence of a the candidate compound, the growth rate of the microorganism being capable of being stimulated by the cytokine in the absence of the candidate compound and; b) determining whether the candidate compound decreases the growth rate of the microorganism in the presence of the cytokine. In preferred embodiments the microorganism is a virulent strain of E. coli; and the cytokine is interleukin-1. In a related aspect, the invention features a cytokine antagonist which includes all or a portion of a cell surface protein derived from an infectious microorganism.

In another related aspect, the invention features a cytokine antagonist which includes all or a portion of a cell surface carbohydrate derived from an infectious microorganism.

In another related aspect, the invention includes a cytokine antagonist which includes all or a portion of a cell surface lipopolysaccharide derived from an infectious microorganism.

In yet another related aspect, the invention includes a cytokine antagonist which includes all or a portion of a molecule secreted by an infectious microorganism. By "antagonist" is meant a molecule capable of interferring with the stimulatory effect of a cytokine on cell growth.

In a preferred embodiment the cytokine antagonist is derived from E. coli . In a more preferred embodiment the cytokine is interleukin-1.

In a preferred embodiment, the cell surface carbohydrate is present on the surface of a virulent, but not an avirulent strain of E. coli . Other features and advantages of the invention will be apparent from the following description of the preferred embodiments thereof, and from the claims.

Detailed Description The drawings are first briefly described. Fig. 1 is a set of graphs illustrating the effect of incubation time (min) in media containing 100 ng/ml (triangles) , 10 ng/ml (squares) , 1 ng/ml (circles) or 0 ng/ml (X) IL-lβ on the number of colony forming units. Panel A illustrates the results for a virulent strain and panel B illustrates the results for an avirulent strain. Fig. 2 is a set of graphs illustrating the effect of incubation time (min) under various conditions on the number of bacteria in culture. Panel A illustrates the effect of IL-lα (circles) and IL-13 (squares) on growth of a .virulent strain. Panel B illustrates the effect of IL-ljS on growth of a virulent strain (squares) and an avirulent strain (circles) . Panel C illustrates the effect of IL-1_/3 (squares) and IL-lra (circles) on growth of a virulent strain. In each of panels A, B and C, a control having no added IL-l or IL-lra is included (triangles) .

Fig. 3 is a bar graph illustrating the effect of incubation time on bacterial number for growth in the presence of various ratios of IL-lra to IL-l as well as in the presence of IL-l only or heat-inactivated IL-l only.

Fig. 4 is a graph illustrating the effect of bacterial number on binding of [¹²⁵I]-labeled IL-13 (cpm) for a virulent (filled circles) and an avirulent stain (open circles) .

Fig. 5 is a graph illustrating the effect of [¹²⁵I]-labeled IL-l concentration (pg) on binding of [¹²⁵I]-labeled IL-l (cpm) for a virulent (filled circles) and an avirulent (open circles) strain.

Fig. 6 is a graph illustrating the effect of unlabeled IL-l (open circles) and IL-lra (filled circles) concentration (pg) on binding of [¹²⁵I]-labeled IL-l to a virulent E. coli strain. E, coli Are Responsive to Interleukin-1

Experiments described below demonstrate that virulent, but not avirulent, strains of E. coli are responsive to interleukin-1 (IL-l) when grown in normal human serum. Specifically, IL-l increased the growth rate of a virulent strain while the growth rate of an avirulent strain was unaffected. The IL-l receptor antagonist (IL-lra) , a naturally occurring polypeptide that binds IL-l receptor without triggering IL-l activity, significantly decreased the growth-enhancing effect of IL-l on virulent E. coli. A separate set of experiments described below demonstrate that IL-l specifically bound virulent, but not avirulent E. coli .

These results suggest that compounds which compete with IL-l for binding to E. coli (e.g., IL-lra) can be used to inhibit growth of virulent E. coli under conditions where IL-l is present, for example, in bacterial infections of human patients.

Further, the results indicate that virulent strains of E. coli can serve as a source of molecules that will bind IL-l. Such molecules can act to block IL- 1 activity by blocking binding of IL-l to one or more of its receptors present on responsive eukaryotic cells. These blocking molecules are useful for inhibiting unwanted alloreactivity or blocking IL-l mediated inflammatory effects.

IL-l Increases the Growth Rate of Virulent E. coli in Brain Heart Infusion

Six virulent and four avirulent strains of E. coli (isolated from blood, feces or urine samples) were characterized by their ability to survive when exposed to human serum. Briefly, bacteria in log growth phase were diluted in phosphate buffered saline (PBS) containing 5 x 10~ M MgCl₂. 100 μl of 10~⁶ dilutions of bacteria were plated to determine the number of colony forming units (CFU) . In parallel, 50 μl of 10^"6 dilutions of bacteria were added to microtiter wells containing 50 μl PBS, and 100 μl of human serum or, as a control, 100 μl heat- inactivated (56°C, 30 minutes) human serum. After 2 hours of incubation aliquots were plated to determine the number of CFU. Bacterial killing of less than 5% in human serum defined a virulent strain and killing of greater than 95% in human serum killing defined an avirulent strain.

The effect of IL-l on the growth of virulent and avirulent strains was determined as described below.

Bacteria were grown in brain heart infusion (BHI, Difco, Detroit, MI) broth for 16-18 hrs at 37°C, washed and resuspended in BHI to a final concentration of 10⁶ bacteria/ml. Virulent and avirulent bacteria (3 x 10⁴) were incubated in rotating Eppendorf tubes at 37°C in BHI broth in the presence of various concentrations of IL-13 (Glaxo Institute for Molecular Biology, Geneva, Switzerland) . Otherwise identical cultures with heat inactivated (90°C, 30 min) IL-l instead of IL-ljS were used as a control. The cultures were diluted and plated on agar for bacterial counts at 30 minute intervals from time 0 to 4 hours of incubation. The number of CFU was determined 18 hours after plating. For statistical analysis, the log of bacterial counts between 60-180 minutes of incubation (log phase growth before reaching stationary phase) were plotted against time, and a least squares fit was used to determine the slopes. The differences in growth rates of bacteria were then assessed by comparing these slopes using analysis of variance followed by Fisher's least significant difference test.

Referring to Fig. 1, panel A, growth of a particular virulent strain was enhanced in the presence of 10 ng/ml IL-ljS (p < 0.001) or 100 ng/ml IL-1/3 (p < 0.0001) compared to the control. The effect of 1 ng/ml IL-l? on growth was not statistically significant. In contrast, the growth rate of an avirulent strain was unaffected by the addition of IL-1,9 to 100 ng/ml (Fig. 1, panel B) . IL-l Increases the Growth Rate of Virulent E. coli in Tissue Culture Medium

The effect of IL-l on bacterial growth in a defined tissue culture medium, RPMI-1640 (Gibco, Grand Island, NY) , was determined as described below. Bacteria were grown in RPMI for 16-18 hours at 37°C, washed and resuspended in RPMI to a final concentration of 10⁵ bacteria/ml. The concentration of duplicate samples was measured by optical density (405 nm) for serial dilutions of the bacterial suspensions. In some experiments, duplicate aliquots from these serial dilutions were also plated on agar and bacterial colonies were counted following a 24 hour incubation. Optical density values were correlated with the number of CFU for virulent and avirulent strains, in different culture media and under different conditions (IL-lα, IL-ljS, IL- lra, combination of cytokines, heat-inactivated IL-l, and no cytokines) . Separate standard curves for each strain and each growth condition served thereafter as a reference for determination of bacterial numbers by correlating diluted samples of the corresponding strain/condition/media using the linear part of the curve.

For growth studies, 1.5 ml Eppendorf tubes were prepared with 0.8 ml RPMI, 0.1 ml bacteria (10⁴) and 0.1 ml IL-l or RPMI to a final volume of 1 ml. The tubes were then slowly rotated, 12 turns per minutes, at 37°C. Tubes were removed from the incubator at 15 min intervals and sodium azide (0.1% final concentration) was added to stop growth. At the end of the experiment, samples from each tube were diluted for optical density readings. The differences in growth rates of bacteria were assessed by comparing the rate of bacterial growth between 5 xlO⁴ and 5 xlO⁶ bacteria. Statistical analysis was performed as described above. Referring to Fig. 2, panel A, IL-13 (100 ng/ml) and IL-lα (100 ng/ml) increased (p < 0.0001 and p < 0.05, respectively) the growth of virulent bacteria compared to the control (heat-inactivated IL-l) . Referring to Fig. 2, panel B, 100 ng/ml IL-ljS increased growth of a virulent, but not an avirulent strain (p < 0.0001) compared to the control.

The IL-l receptor antagonist (IL-lra) is a naturally occurring polypeptide that binds to IL-l receptors without manifesting agonist activities (Eisenberg et al., Nature 343:341, 1990). In a growth experiment performed as described above, IL-lra blocked 75% to 85% of the IL-1-induced increase in growth rate (Fig. 2, panel C) . In this experiment IL-lra (Synergen, Boulder, CO) was present at 10 μg/ul and IL-lj8 was present at 100 ng/ml. IRAP from Upjohn, Inc gave similar results. In the absence of IL-l, IL-lra alone did not affect bacterial growth.

Fig. 3 presents the results of growth assays performed on a virulent stain grown in the presence of various ratios of IL-lra to IL-ljS. In these experiments IL-lra (10-10,000 ng/ml) was added to virulent bacteria (3 x 10⁴) for 5 minutes at room temperature, followed by IL-ljS (10 ng/ml). Cultures were incubated at 37°C. At 30 minutes intervals aliquots were removed and dilated for determination of optical density. Statistical analysis, performed as described above, was used to determine the the growth rates during log phase growth for various IL-lra to IL-l ratios. Growth in the presence of heat-inactivated IL-l alone or IL-l alone was measured for purposes of comparison.

At a molar ratio of IL-lra to IL-13 of 1, there was a consistent, but not statistically significant reduction, in IL-ljS-induced growth, at ratios of 10:1 and 100:1, progressively greater inhibition of IL-ljø-induced growth was observed (p<0.0005 and p<0.00005, respectively) . Increasing the IL-lra concentration to 1,000-fold molar excess over IL-ljS did not reverse the effect of IL-l.

IL-lfl Binds to Virulent E. Coli Iodinated IL-1/3 was used to study the specific binding of IL-l to E. coli. Briefly, bacteria grown and washed in RPMI were resuspended at 10⁴-10⁷ CFU/ l in binding buffer (RPMI containing 20 mM HEPES and 1% bovine serum albumin) . For binding experiments, 0.1 ml bacteria, 0.1 ml [¹²⁵I]-labeled IL-ljS (prepared as described by Savage et al., Cytokine 1:23, 1989), 0.1 ml 2% sodium azide in RPMI, and 0.1 ml binding buffer or unlabelled cytokine were added to each tube. The tubes were slowly rotated at 4°C for 18-20 h, and then centrifuged (10,000 x g) for 10 min at 4°C, followed by immediate removal of the supernatants. Radioactivity in the pellet was counted.

Referring to Fig. 4, virulent bacteria (filled circles) bound [¹²⁵IJ-labeled IL-l whereas avirulent (open circles) bacteria did not bind a significant amount of t¹²⁵I]-labeled IL-l.

Referring to Fig. 5, at 10⁵ bacteria, increasing concentrations of [¹²⁵I]-labeled IL-l resulted in increased binding of [¹²⁵I]-labeled IL-l to the virulent, but not the avirulent, strain.

Referring to Fig. 6, the binding of [¹²⁵I]-labeled IL-l was specifically blocked by competition with unlabeled IL-ljS or IL-lra. Fifty percent inhibition of the binding was achieved by 150 pg of unlabeled IL-ljS (filled circles) , whereas 25 times more IL-lra was required for 50% inhibition (open circles) .

TNF (1,000 ng/ml) or IL-2 (100 U/ml) did not inhibit the binding of IL-l, although observed a small, but statistically significant, enhancement of E. coli growth was observed in the presence of human IL-2. Ouantitation of IL-l Binding Sites

Virulent bacteria, in the presence of 0.1% sodium azide, were incubated for 18 hr at 4°C with various amounts of radioactively labeled IL-l in both the presence and absence of excess unlabeled IL-l. Scatchard analysis of the binding data indicated that there are 20,000 - 40,000 IL-l binding sites per bacterium. Virulent E. coli as a Source of IL-l Recognition Factors Because virulent stains of E. coli respond to IL- 1, they must possess a structure that recognizes and binds IL-l. As noted above, virulent and avirulent E. coli stains differ in the nature of their cell wall carbohydrate moieties, thus it is likely that the IL-l recognition structure is a carbohydrate derivative. The fact that 20,000-40,000 IL-l binding sites are present on each cell (see above) suggests that the recognition structure may not be a protein since on mammalian cells, where the IL-l binding receptors are known to be polypeptides, there are far fewer binding sites (200-400 on human lymphocytes and 4,000 - 8,000 on murine cell lines) . Despite this, it is possible that the IL-l binding site is primarily a protein. Described below are a series of experiments which permit the identification and isolation of IL-l binding structures present on virulent bacterial strains. Protein Receptors

To ascertain if a protein plays a significant role in IL-l binding, virulent E. coli are exposed to trypsin (1-10 units ml) or glutoraldehyde (1%) at 37°C for 60 min. Following this treatment an IL-l binding assay

(described above) is performed. A greater than 30% loss of binding indicates that a protein plays a significant role in binding.

The membrane protein o pA may play a role in the virulence of certain E. coli strains. This possibility can explored by extracting ompA and testing the isolated protein for IL-l binding. To extract ompA, virulent bacteria are shaken vigorously in 0.9 mM CHAPS at 4°C for up to 60 min, cell debris are removed by centrifugation and the supernatant is tested for the presence of factors capable of binding IL-l as follows. Duplicate serial dilutions are applied to wells of a multichamber manifold containing a sheet of nitrocellulose. After incubation for 60 minutes, the nitrocellulose is removed and cut into strips separating the duplicates which are then placed in sealed bags containing either [¹²⁵I]-labeled IL- ljS or [¹²⁵I]-labeled IL-ljS plus a 100 fold-excess of unlabeled IL-ljS. After overnight incubation at 4°C, the strips are removed, washed and autoradiographed. If immobilized ompA binds IL-l specifically, this experiment is repeated using excess IL-lra. The results of these assays are compared to the results obtained using ompA isolated from avirulent E. coli . If such analysis indicates that ompA binds IL-l, fragments of ompA, generated by proteolytic digestion or recombinant DNA techniques, can be tested for their ability to bind IL-l.

Southern hybridization can be used to search for E. coli proteins related to the IL-l receptor. Briefly, total E. coli DNA isolated from virulent and avirulent strains is digested with a variety of restriction endonucleases and the resulting fragments are separated by gel electrophoresis. The fragments are then transferred to a nylon membrane and probed with labeled cDNA encoding human IL-l receptor type I or IL-I receptor type II using standard technqiues (Current Protocols in Molecular Biology, Ausubel et al., eds., illey Interscience, New York, 1991) .

Antibodies directed against IL-l receptor type I or IL-l receptor type II can also be used to look for IL-l binding sites. Briefly, E. coli are washed and suspended in binding assay buffer (Savage et al., Cytokine 1:23, 1989) and incubated at 4°C for 2 hours in the presence of anti-ILl receptor antibodies (10 jug/ml) followed by the addition of [¹²⁵I]-labeled IL-ljS. Cells are then incubated overnight at 4°C and specific [¹²⁵I]- labeled IL-ljS binding is determined using the assay described above.

It may be possible to identify IL-l binding structures by cross-linking IL-l to virulent E. coli . For this procedure E. coli are grown overnight in BHI broth, washed thoroughly at 4°C in PBS and incubated with saturating concentrations of [¹²⁵I]-labeled IL-ljS as described above. Unlabeled IL-lj3 or IL-lra, in 100 and 1,000-fold molar excess respectively, are added to control suspensions. After incubation at 4°C for two hours, the cells are centrifuged, washed and the cross- linking reagent BS₃ is added (2.7 mM) for two hours as described by Granowitz et al. (J^*. Biol . Chem. 266:14147, 1991) . The cells are then lysed in CHAPS lysis buffer and immunoprecipitated with anti-human IL-lj3 as described by Granowitz et al. (supra) . Carbohydrate Receptors

Enzymes that remove specific carbohydrate moieties can be used to identify carbohydrate structures involved in IL-l binding. This approach is similar to that used to identify carbohydrate moieties on the IL-l receptor of mammalian cells (Mancilla et al. , Cytokine 1:95, 1989; and Speziale et al., Lymphokine Res. 8:1, 1989).

A virulent strain of E. coli is first serotyped in order to ascertain the composition of the O-antigen polysaccharide side chains. This information is used to select the appropriate enzymes which cleave relevant carbohydrates from this strain. For example, if the serotype reveals multiple mannose chains, Endo-b-N- acetylglucosaminidase (Endoglycosidase H) at 40-400 U/mg of LPS will be used to remove mannose. The treated bacteria are then subjected to an IL-l binding assay (described above) . Removal of carbohydrate residues that are part of hte IL-l receptor on virulent strains wil decrease IL-l binding. Carbohydrates whose structures are based on the the removed residues can be immobilized on nitrocellulose and tested for their ability to bind IL-l.

All or part of lipopolysaccharide (LPS) may be the IL-l receptor on virulent bacteria. To determine if IL-l binds LPS of virulent strains, bacteria are grown in casamino acid broth supplemented with yeast extract and glucose. LPS is extracted by the hot phenol-water method (Westphal et al., Z . Naturforsch 76:148, 1956), dialyzed, lyophilized and then treated with 0.1N NaOH at 37°C for 18 h. The pH is adjusted to 7.0-7.2 with HC1, and the LPS preparations are dialyzed against distilled water for 3 days and lyophilized. Purified LPS is subjected to SDS-PAGE for quantitative analysis of the length of the O-polysaccharide side chains as described by Porat et al. (Infection and Immunity, 55:320, 1987). Dilutions of various LPS preparations are then immobilized in parallel on nitrocellulose and used in an IL-l binding assay as described above. Similar preparations of LPS removed from virulent and avirulent strains of E. coli are subjected to SDS- PAGE (7.5%) and then transferred to nitrocellulose by electroelution. As described above, the nitrocellulose is cut into strips and placed in bags containing either ¹²⁵I-IL-ljS or ¹²⁵I-IL-lj8 plus a 100 fold-excess unlabeled IL-ljS. This experiment will establish whether IL-l binds specifically to the LPS chains stripped from virulent E. coli and the relative molecular weight of such binding structures. If IL-l binds LPS from virulent, but not avirulent, bacterial strains, portions of LPS can be tested for IL-l binding as described above.

The outer membrane of Gram-negative bacilli consists of LPS and different proteins. LPS includes: lipid A, core oligosaccharides (3-deoxy-D-manno- octulosonic acid (KDO) and heptoses) and polysaccharide O-antigen side chains (repeating sub-units of oligosaccharides) . The length and distribution of these O-antigen polysaccharide side chains of LPS have been implicated in conferring bacterial sensitivity or resistance to the bactericidal activity of human serum (Porat et al. , Prog. Clin. Biol . Res. 272:103, 1988; and Porat et al. , Infect. Immunit. 55:320, 1987), a characteristic which distinguishes the response of E. coli to IL-l-induced growth enhancement. It has been shown that virulent organisms contain more of the longer carbohydrate side chains in the O-antigen region (Goldman et al., J^". Bacteriol . 159:877, 1984; and Grossman et al. , biol . Immunol . 227:859, 1990) . In contrast, bacteria which have a greater proportion of short carbohydrate side chains in the LPS are usually sensitive to the killing effect of human serum. In fact, mutant clones of Gram-negative bacilli (R595 for Salmonella minnesota and J5 and E. coli) , lacking their O-antigen polysaccharide region are extremely sensitive to serum. If IL-l binds O-antigen, the following analysis will permit determination of the nature of the O-antigen bound by IL-l. Briefly, LPS is disaggregated with 0.1M Tris buffer (pH 8.5) containing 0.6% sodium deoxycholate, and then applied to a Sephadex G-200 column (2.5 x 55 cm) , equilibrated with the same buffer, and eluted at 37°C at an approximate rate of 12 ml/h (Porat et al.. Infect. Immunol. 55:320, 1987; and Vukajlovich et al., J. Immunol . 130:2804, 1983). Fractions are analyzed for hexose content by the anthrone method (Scott et al., Anal. Chem. 25:1656, 1953) or by the cysteine-sulfuric acid method (Wright et al. , Anal , biochem. 49:307, 1972); KDO content is determined by the thiobarbituric acid method (Osborn et al., Proc. Natl . Acad. Sci . USA 50:499, 1963) . It is possible to separate two classes of O- polysaccharide side chains of LPS according to their molecular weights by calculating the ratios of hexose:KDO.

The different carbohydrate moieties are screened for the ability to bind IL-l. Following chromatography and determination of hexose:KDO ratios, the LPS fractions are dialyzed against Tris buffer for 5 days and then against distilled water to remove the DOC. Each fraction is applied to nitrocellulose and binding studies using ¹²⁵I-IL-l/3 are carried out. O-polysaccharide fractions are also treated with acetic-acid (1%) at 100°C, the lipid A portion precipitated and washed with Tris buffer and the supernatants dialyzed against PBS (pH 7.4) for 3 days. The resulting lipid A-free carbohydrates are separated by SDS-PAGE, transferred to nitrocellulose and used for binding studies as described above.

A liposome model can be used to assess the minimal amounts of O-polysaccharide side chains needed for the binding to IL-l and the critical ratios between long and short carbohydrate chains required binding. Multilamellar liposomes are first prepared (Porat et al.. Prog. Clin . Biol . Res . 272:103, 1988; and Porat et al.. Infect . Immunol . 55:320, 1987). Three lipids: dimyristyl phos^'phatidyl choline (25 mM in chloroform) , cholesterol (75 mM in chloroform) and dicetyl phosphate (3 mM in 1:1 {vol/vol} methanol-chloroform) will be mixed in a 10 ml pear-shaped flask in a molar ratio of 2:1.5:0.22, respectively. Chloroform and methanol are evaporated creating a thin film of dried lipids. LPS preparations (in gelatin veronal buffered saline, pH 7.4, at 200 μl) are added to the lipids and the mixture agitated vigorously for 5 minutes. The resulting multilamellar liposomes with imbeded LPS in the external layer are washed and resuspended in gelatin veronal buffered saline. Liposomes prepared with LPS preparations, consisting of different long:short carbohydrate side chains ratios, are incubated with [¹²⁵I]-labeled IL-l for 16 h at 4°C with slow rotation. Controls include labeled IL-l with the buffer without liposomes, and liposomes with heat-inactivated [¹²⁵I]-labeled IL-l. The liposomes are collected by centrifugation, washed and the bound cpm are measured. Bound IL-l will be calculated by subtracting non-specific binding (cpm in liposomes, not incorporated with LPS, but incubated with [¹²⁵I]-labeled IL-l) from total binding (cpm of washed LPS-incorporated liposomes with [¹²⁵I]-labeled IL-l).

The liposome model can also be used for determination of the inhibition of binding by IL-lra. In addition, the effect of treatment of the different 0- polysaccharide fractions with N-glycanase and O-glycanase (see above) on binding to IL-l can be assessed using this model.

Lectins can inhibit IL-l binding to mammalian cells. Accordingly ConA, PHA, wheat germ agglutinin (WGA) and peanut agglutinin (PNA) can be incubated with virulent strains of E. coli at various concentrations (0.1-100 μ-g/ml) at 4°C for 30 minutes, and then [¹²⁵I]- labeled IL-lJ is added and IL-l binding is assessed as described above. If bacterial binding to [¹²⁵I]-labeled IL-ljS is blocked by one or more of these lectins, lectins may affect the growth of IL-1-stimulated bacteria. This can be tested by pretreating virulent E. coli with lectins (at concentrations found to block IL-l binding) at 4°C for 30 minutes. IL-l, at 10 ng/ml, is added and growth curves are generated and analyzed as described above. If IL-l-induced bacterial growth enhancement is affected by the lectins, the effect of specific sugars (e.g., N-acetyl glucosamine/N-acetyl galactosamine and D- mannose) IL-1-stimulated growth can be tested. Those that block IL-1-stimulated growth can be used as IL-l antagonists.

IL-l binds to uromodulin when immobilized on plastic surfaces, and uromodulin inhibits IL-l-initiated biological activities. It is known that IL-l binding to uromodulin is inhibited by oligosaccharides derived from /3-2-D-glucose and D-fucose, suggesting that N- glycosylation of uromodulin is responsible for its ability to recognize and bind IL-l. Since a similar N- glycosylation maybe involved in IL-l binding to mammalian cells, it seems likely that IL-l binding to E. coli is similarly due to a N-glycosylated carbohydrate moiety. This may be tested by adding increasing concentrations of the oligosaccharides derived from jS-2-D-glucose (1-0- methyl jS-D-glucose 125mM, αjS-2-deoxy glucose 62.5mM, αjS- D-glucosamine 62.5mM, αjS-6-deoxy glucose 62.5mM) or D- fucose (αj8-6-deoxy D-galactose 31mM) to virulent E. coli at 4°C and then performing binding assays. These oligosaccharides can be tested for their ability to affect the growth of E. coli alone and in cultures of E. coli stimulated by IL-l.

Claims

1. Use of a cytokine antagonist in the preparation of a medicament for inhibiting infection by an infectious microorganism in a patient.

2. The use of claim 1 wherein said cytokine is interleukin-1.

3. The use of claim 1 wherein said infectious microorganism is virulent E. coli.

4. The use of claim 2 wherein said antagonist is IL-lra.

5. A method for screening candidate compounds to identify compounds capable of acting as cytokine antagonists comprising: a) measuring the growth rate of a infectious microorganism in a growth medium including said cytokine both in the presence and absence of a said candidate compound, the growth rate of said microorganism being capable of being stimulated by said cytokine in the absence of said candidate compound and; b) determining whether said candidate compound decreases the growth rate of said microorganism in the presence of said cytokine.

6. The method of claim 5 wherein said microorganism is a virulent strain of E. coli.

7. The method of claim 6 wherein said cytokine is interleukin-1.

8. A cytokine antagonist comprising all or a portion of a cell surface protein derived from an infectious microorganism.

9. A cytokine antagonist comprising all or a portion of a cell surface carbohydrate derived from an infectious microorganism.

10. A cytokine antagonist comprising all or a portion of a cell surface lipopolysaccharide derived from an infectious microorganism.

11. A cytokine antagonist comprising all or a portion of a molecule secreted by an infectious microorganism.

12. The cytokine antagonist of any of claims 8, 9, 10, or 11 wherein said infectious microorganism is E. coli .

13. The cytokine antagonist of claim 11 wherein said cytokine is interleukin-1.

14. The cytokine antagonist of claim 12 wherein said cell surface carbohydrate is present on the surface of a virulent, but not an avirulent strain of E. coli .