FIELD OF THE INVENTION The present invention relates to compounds and methods for treating and preventing bacterial and viral disease, and in particular, carbohydrate derivatives for inhibiting binding to cell receptors.

BACKGROUND The physiology, structure, and biochemical systems of infectious agents and their hosts are usually quite different. Current antimicrobial therapy relies on these differences.

Although the mechanisms of action of many antimicrobials are not well understood, the five major categories of action include inhibition of cell wall synthesis, inhibition of cell membrane function, inhibition of protein synthesis, inhibition of nucleic acid synthesis, and interference with intermediary metabolism. (See e. g., W. K. Joklik et al., [eds.], Zinsser Microbiology, 18th ed., Appleton-Century-Crofts, Norwalk. CT, [1984], p. 193).

Antibacterial Agents The best known antibacterial agent is penicillin. Like all (3-lactam drugs, it is a compound which selectively inhibits bacterial cell wall synthesis. Because of their relatively high concentration of peptidoglycan, gram-positive organisms tend to be much more susceptible to the effects of penicillin and other (3-lactams than gram-negative organisms. Importantly, because they affect cell wall synthesis, penicillin and the other (3-lactams are only effective against actively growing and dividing cultures.

Some organisms are naturally resistant to penicillin. Moreover, following years of use to treat various infections and diseases, penicillin resistance has become increasingly widespread in the microbial populations that were previously susceptible to the action of these drugs.

Tetracycline is the drug of choice for some bacteria, such as V. cholera.

However, resistance to this drug is also well-known. Indeed, levels of tetracycline resistance in East Africa, Bangladesh and parts of India have become high enough that V. cholera isolates from these areas are assumed to be tetracycline resistant until results of susceptibility testing are available. See J. B. Kaper et al.,"Cholera,"Clin. Micro. Rev.

8: 48 (1995).

The development of tolerance and resistance to antimicrobials represents a significant threat to the ability to treat disease. Many factors have contributed to this increased observance of resistant strains, including over-use and/or inappropriate administration of antimicrobials, the capability of many organisms to exchange genetic material which confers resistance (i. e., R plasmids), and the relatively rapid mutation rate observed with many bacteria, which allows for selection of resistant organisms.

Once an organism has developed resistance to a particular drug, it becomes important that an effective replacement drug be identified. If the organism develops resistance to this second drug, another replacement is needed. One example of the historical development of multiple drug resistance is gonorrhea. Prior to the 1930's, treatment for this disease usually involved mechanical means, such as irrigation and use of urethral sounds in males. In the late 1930's, sulfonamides were introduced and found to be effective in treating gonorrhea. After a few years, sulfonamide-resistant strains of N. gonorrhoeae were isolated. Fortunately, by this time, penicillin was available and found to be effective. However, by the 1970's, many isolates of N. gonorrhoeae were found to be penicillin-resistant. This required the use of alternative drugs such as spectinomycin. It can be expected that this trend will continue, with the development of strains that are resistant to sulfonamides, penicillin, spectinomycin, and other antimicrobials.

Thus, there remains a need to develop new strategies against bacteria. Ideally ; the approach should be effective even against multiple-drug resistant organisms.

Antiviral Agents An ever-increasing number of viruses are being identified as the source of human disease. The better known diseases caused by viruses include, chicken pox, measles, mumps, influenza, hepatitis, poliomyelitis, rabies, and now, of course, HIV. There are, however, many more virus-related diseases. Indeed, viral infections are estimated to be responsible for more than sixty-percent of human sickness occurring in developing countries.

In contrast to the somewhat successful story of antibiotics, efforts to treat viral infection have been largely ineffective. When individuals become infected, modern medicine can do little but ease the symptoms. Viral epidemics have only been avoided by treatment of uninfected individuals with vaccines.

The effort to treat viral infection has been hampered largely by the unique structural and functional characteristics of viruses. A virus is essentially nucleic acid surrounded by a lipid-protein envelope. A virus invades a host cell and uses the host cell's machinery to replicate itself. The latter characteristic makes it especially difficult to find drugs which kill the virus and leave the host unharmed.

With a more detailed understanding of viral function, more sophisticated and promising approaches to treatment have been suggested. One approach has come from the recognition that the viral envelope proteins may be involved in binding to the host cell and, ultimately, penetration of the virus into the host cell. This approach involves the use of competing proteins or parts of proteins to block the binding or"fusion"event. An example of this approach can be found in U. S. Patent Ser. No. 4,880,779 which describes inhibitory peptides to block retroviral fusion. Another approach has come from the recognition that some viruses use unique polymerases to replicate nucleic acid. This approach involves the use of competing, nucleotide derivatives to bind to the polymerase and stop replication. An example of this approach can be found in U. S. Patent Ser. No.

4,916,122 which describes synthetic deoxyuridine derivatives to block retroviral nucleic

acid replication.

In spite of these attempts, there is a severe need for new strategies for antiviral therapy. While treatment strategies are importantly, there is even a greater need for an approach to the prevention of viral disease.

SUMMARY OF THE INVENTION The present invention relates to compounds and methods for treating and preventing bacterial and viral disease, and in particular, carbohydrate derivatives for inhibiting binding to cell receptors. The present invention contemplates inhibitors to block the binding of microorganisms and/or their toxins to such receptors. In one embodiment, the present invention contemplates using oligosaccharide-derivatized dendrimers to 1) inhibit binding of microorganisms (e. g. viruses) to cell surface glycoconjugates, as well as 2) inhibit cells from binding carbohydrate residues present on the surface of the microorganism (e. g. bacteria).

The present invention contemplates that such inhibitors can be used for both treatment and prevention of disease (as well as preventing the spread of infection in a subject with nascent disease). However, the present invention is not limited to in vivo treatment. The present invention contemplates that the inhibitors of the present invention are useful in assays in vitro to assess the susceptibility of microorganisms to inhibition (and to better understand interaction with host cells). Specifically, glycosylated dendrimers can be used to study the carbohydrate requirements for adherence of a bacteria, virus, or bacterial toxin. Since the length of the outermost arms of a dendrimer can be extended by use of the appropriate chemical reactions prior to the addition of the oligosaccharides, the present invention contemplates using such in vitro reactions to obtain information about the optimum spacing between replicate oligosaccharide binding sites, allowing for better inhibitor design.

In addition, the present invention contemplates that the oligosaccharide derivatives can serve as carriers, i. e. to deliver drugs to microorganisms. In one embodiment, the present invention contemplates using (multivalent) oligosaccharide-derivatized dendrimers

(i. e. defined clusters) to deliver antibacterial or antiviral agents. In this manner, specificity is achieved (e. g. through the interaction of the oligosaccharide with the specific ligand on the microorganism) and systemic exposure to the antibacterial or antiviral agent (and the consequent toxicity) is avoided.

In one embodiment, the present invention contemplates methods for synthesizing compounds, comprising: a) providing: i) a source of oligosaccharide; ii) oligosaccharide treatment means; and iii) a polymer; b) treating said source with said treatment means under conditions such that oligosaccharide is released from said source to create free oligosaccharide; and c) mixing said free oligosaccharide with said polymer under conditions such that said free oligosaccharide is covalently attached to said polymer to create a carbohydrate derivative. In one embodiment, said source of oligosaccharide is selected from the group consisting of glycoproteins, glycosaminoglycans and glycolipids.

In another embodiment, glycolipids are used as a source and said glycolipids are selected from the group consisting of neutral glycosphingolipids, gangliosides and sulfatoglycosphingolipids. In another embodiment, gangliosides are used as the source and said gangliosides are selected from the group consisting of GM1 (including fucosylated GM1), GM2, GM3, GDla and GDlb. In another embodiment, neutral glycosphingolids are used as the source and said neutral glycosphingolipids are selected from the group consisting of glucosylceramide, lactosylceramide, globotriasylceramide, globotetraosylceramide and asialo GM1.

It is not intended that the present invention be limited by the treatment means. A variety of chemical and enzymatic treatment means are contemplated. In one embodiment, said treatment means is triethylamine.

It is not intended that the present invention be limited to the nature of the polymer.

In one embodiment, said polymer is a branched polymer. In a preferred embodiment, said branched polymer is a dendrimer and said carbohydrate derivative is multivalent.

The present invention also contemplates a method, comprising: a) providing: i) a source of oligosaccharide selected from the group consisting of glycoproteins, glycosaminoglycans and glycolipids; ii) oligosaccharide treatment means; and iii) a branched polymer; b) treating said source with said treatment means under conditions such

that oligosaccharide is released from said source to create free oligosaccharide; and c) mixing said free oligosaccharide with said polymer under conditions such that said free oligosaccharide is covalently attached to said polymer to create a multivalent carbohydrate derivative.

In still another embodiment, the present invention contemplates a method, comprising: a) providing: i) a glycolipid source of oligosaccharide, wherein said glycolipid is selected from the group consisting of neutral glycosphingolipids, gangliosides and sulfatoglycosphingolipids; ii) oligosaccharide treatment means; and iii) a dendrimer; b) treating said glycolipid source with said treatment means under conditions such that oligosaccharide is released from said source to create free oligosaccharide; and c) mixing said free oligosaccharide with said dendrimer under conditions such that said free oligosaccharide is covalently attached to said dendrimer to create a carbohydrate derivative.

The present invention also contemplates the resulting compounds. In one embodiment, the present invention contemplates a carbohydrate derivative comprising oligosaccharide covalently attached to a non-carbohydrate polymer. In one embodiment, said polymer is a branched polymer. In a preferred embodiment, said branched polymer is a dendrimer. Again, it is not intended that the present invention be limited to particular sources of oligosaccharides. In one embodiment, said oligosaccharide is derived from a glycolipid and said glycolipid is selected from the group consisting of neutral glycosphingolipids, gangliosides and sulfatoglycosphingolipids. In another embodiment, the oligosaccharide is derived from a ganglioside and said ganglioside is selected from the group consisting of GM1 (including fucosylated GM1), GM2 and GM3, etc.

The present invention specifically contemplates a carbohydrate derivative comprising glycolipid oligosaccharide covalently attached to a dendrimer, wherein said glycolipid is selected from the group consisting of neutral glycosphingolipids, gangliosides and sulfatoglycosphingolipids.

The present invention also contemplates methods of treatment (including both acute and preventative treatment). In one embodiment, the present invention contemplates a method of treatment, comprising: a) providing: i) a subject having symptoms of (viral

or bacterial) disease, and ii) a carbohydrate derivative comprising glycolipid oligosaccharide covalently attached to a dendrimer; b) administering said derivative to said subject under conditions such that said symptoms are reduced. Again, said glycolipid is selected from the group consisting of neutral glycosphingolipids, gangliosides and sulfatoglycosphingolipids, and said gangliosides are selected from the group consisting of GM1 (including fucosylated GM1), GM2 and GM3, etc. Preferably, said derivative is multivalent.

It is not intended that the present invention be limited to particular to particular diseases. In one embodiment, said symptoms are symptoms of cholera.

In yet another embodiment, the present invention contemplates a method of preventative treatment, comprising: a) providing: i) a subject at risk of (viral or bacterial) disease, and ii) a carbohydrate derivative (e. g. multivalent) comprising glycolipid oligosaccharide covalently attached to a dendrimer; b) administering said derivative to said subject. Again, said glycolipid is selected from the group consisting of neutral glycosphingolipids, gangliosides and sulfatoglycosphingolipids. While not limited to a particular disease, in one embodiment, said subject is at risk for bacterial disease and said bacterial disease is cholera.

DESCRIPTION OF THE DRAWINGS Figure 1 shows the chemical structure of the ganglioside GM1.

Figure 2 is a flow chart for the synthesis of the phenylisothiocyanoto derivative of oligo-GM1.

Figure 3 is a thin layer chromatograph showing oligo-GMl derivatives.

Figure 4 is a flow chart for the synthesis of tetra-and octa (propylene imine) dendrimers.

Figure 5 is a thin layer chromatograph showing adherence of horseradish peroxidase conjugated cholera toxin (HRP-CT) to GM1 and oligo-GMl-PITC derivatized dendrimers.

Figure 6 is a graph showing inhibition of adherence of 1211-labeled choleragenoid to GM1-coated plastic wells by oligo-GMl-containing ligands.

Figure 7 shows predicted lowest-energy conformation of the fully derivatized oligo-GMl-PITC and StarburstTM dendrimers.

Figure 8 shows a computer overlay of the crystal structure of choleragenoid with the predicted structure of the tetra (propylene imine) (oligo-GMl-PITC) 4 derivative.

Figure 9 shows two graphs depicting adherence of 25I-labeled choleragenoid to NCTC-2071 cells. Figure 9A shows the results when cells were grown in the presence of media containing different amounts of GM1 prior to harvest 18 hr later and incubation with 6 nM labeled choleragenoid for 1 hr at 16°C. Figure 9B shows the results when cells were grown in media containing 50nM GM1 for 18 hrs prior to a 1 hr incubation at 16°C with the indicated concentrations of labeled choleragenoid.

Figure 10 shows three graphs depicting adherence of'25I-labeled choleragenoid (Figure 10A), cholera toxin (Figure 10B), and the heat labile enterotoxin of E. coli (Figure 10C) to GM 1-treated NCTC-2071 cells in the absence or presence of (oligo-GMl- PITC) 7 octa (propylene imine).

Figure 11 is a graph showing the tryptophan fluorescence emission spectra for choleragenoid plus oligo-GMl-containing compounds.

Figure 12 shows two graphs depicting the tryptophan fluorescence emission spectra for cholera toxin (Figure 12A) and the heat labile enterotoxin of E. coli (Figure 12B) plus oligo-GMl-containing compounds.

DEFINITIONS To facilitate understanding of the invention, a number of terms are defined below.

The present invention contemplates oligosaccharides"made from"as well as oligosaccharides"based upon"a variety of sources of carbohydrate. Oligosaccharides "derived from"or"made from"such sources are natural oligosaccharides that are isolated by treatment of such sources (e. g. chemical or enzymatic treatment), while oligosaccharides"based upon"such sources are synthetic and may contain sugar substitutions or deletions, but contain at least three of the same sugars (in the same sequence and linkage) as the natural oligosaccharide. In one embodiment, the present invention contemplates isolating oligosaccharides from host cell glycoconjugates or

pathogen glycoconjugates, for the preparation of carbohydrate derivatives. The term "oligo-X"is meant to indicate the oligosaccharide moiety released ("free") and isolated from the source ("X"); thus, the term"oligo-GM1"is used to indicated the isolated oligosaccharide moiety of GM1 (a"glycolipid oligosaccharide"). In another embodiment, the present invention contemplates synthetic oligosaccharides based upon host cell glycoconjugates or pathogen glycoconjugates, for the preparation of carbohydrate derivatives.

"Carbohydrate derivatives"are molecules containing natural or synthetic oligosaccharides linked to non-carbohydrate (e. g. protein, protein polymers (such as poly- L-lysine), non-protein polymers, etc.). In one embodiment, the present invention contemplates linking oligosaccharides to dendrimers to generate carbohydrate derivatives.

It is not intended that the present invention be limited to derivatives wherein only one oligosaccharide is linked. Indeed, in a preferred embodiment, a plurality of oligosaccharides are linked to create a"multivalent"derivative.

"Carbohydrate conjugates"are molecule containing natural or synthetic oligosaccharides linked to carbohydrate (e. g. dextran).

"Dendrimers"are highly branched polymers that originate from a central core. A number of dendrimers are available commercially (e. g. from Aldrich).

The term"polymer"is intended to indicate all types of polymers (i. e. molecules with repeating units). In one embodiment, the present invention contemplates linking oligosaccharides to such polymers as dextran and polyethylene glycol.

The term"carrier"is intended to indicate that a molecule can carry or deliver a drug or other active ingredient (e. g. oxygen radical) to a target (e. g. host cell, pathogen, etc.). Typically, the carrier is delivered via the bloodstream to the target. In one embodiment, the present invention contemplates using the carbohydrate derivatives of the present invention as carriers.

"A subject with symptoms of (bacterial or viral) disease"is meant to indicate a human or animal (whether cow, horse, sheep, etc.) having detectable symptoms (i. e. detectable by observation or diagnostic testing of fluid and/or tissue samples) known to

those skilled in the art to be associated with disease. For example, HIV infection is associated with immune deficiencies that are readily detectable; on the other hand, HIV can also be associated with readily detectable retroviral nucleic acid in serum or plasma of the patient. Similarly, bacterial disease is typically associated with fever; on the other hand, specific diseases have diarrhea as a characteristic symptom. The present invention is not limited to any one symptom for any one disease.

Symptoms are"reduced"when there is a detectable quantitative reduction. For example, fever and/or respiration rate can be quantitatively detected and measured and thus a reduction in fever and/or respiration can be readily detected and quantitated.

Similarly, fluid loss is detectable and can be measured. It is not intended that the present invention be limited to precise levels or reductions of a particular magnitude. Most importantly, the present invention is not limited to"cures"or complete elimination of each and every symptom. It is sufficient that there is a reduction in one or more symptoms, regardless of the magnitude of the reduction. For example, in the case of cholera, a reduction in fluid loss can help to stabilize the patient, even though other symptoms of disease are present.

Importantly, the present invention contemplates both acute treatment and preventative treatment. In the case of preventative treatment, subjects"at risk for (bacterial or viral) disease"are treated. Subjects are"at risk"where disease is prevalent in the area (e. g. V cholera in East Africa, Bangladesh and parts of India) or detected in members of the immediate population (e. g. within a city for humans; within a farm or herd for animals). The present invention also contemplates that immune deficient patients and hospitalized patients (regardless of their immune state) are, by definition, at risk.

However, immune competent individuals are not at risk unless the above-discussed criteria are satisfied.

DESCRIPTION OF THE INVENTION The present invention relates to compounds and methods for treating and preventing bacterial and viral disease, and in particular, carbohydrate derivatives for inhibiting binding to cell receptors. While it is not intended that the present invention be limited to a precise mechanisms by which a benefit is achieved, it is contemplated that, in one embodiment, the carbohydrate derivatives of the present invention mimic the carbohydrate portion of the cell receptor and thereby provide an alternative target for the microorganism and/or a microbial toxin.

Carbohydrates are quite heterogeneous. Oligosaccharides can differ from one another in composition and sequence in a manner analogous to polypeptides.

Oligosaccharides with the identical composition and sequence can, moreover, differ in sugar linkages. Added to this level of diversity is the fact that a given carbohydrate chain can be found on either glycoproteins, glycosaminoglycans or glycolipids, each with their own spatial orientation in the cell membrane.

It is not intended that the present invention be limited to a particular source of carbohydrate. Derivatives made from carbohydrate chains of glycoproteins, glycosaminoglycans or glycolipids are contemplated. Within glycolipids, carbohydrate derivatives made from (or based upon) oligosaccharides of glycosphingolipids are, in particular, contemplated.

A. Glycosphingolipids Glycosphingolipids have three major structural features: a long-chain base, a fatty acid moiety, and a carbohydrate chain. Heterogeneity of the carbohydrate chain, as mentioned above for carbohydrates in general, can be found in glycosphingolipids with carbohydrate chains of different length (as small as one sugar to as large as twenty sugars), different sugar composition (glucose, galactose, glucosamine, galactosamine, fucose, and neuraminic acid are the most common), different sugar linkages (the glycosidic linkage can involve a number of different carbon atoms and can be found in two spatial orientations), and different sugar sequences.

Glycosphingolipids are usually broadly divided into two classes: those that contain

only neutral sugars (neutral glycosphingolipids) and those that contain sialic acid (acidic glycosphingolipids, or"gangliosides"). A third class, the sulfatoglycosphingolipids, which contain carbohydrate substituted with sulfate ester groups, is also recognized. The present invention contemplates oligosaccharides made from (or based upon) all three classes.

The sphingosine base and the fatty acid moiety (together called ceramide) comprise the hydrophobic region of the molecule (see Figure 1) whereas the carbohydrate chain makes up the hydrophilic region of the molecule. This amphiphilicity of glycosphingolipids makes them well-suited to a position in the cell membrane such that the ceramide portion is imbedded in the lipid bilayer and the carbohydrate portion extends outward from the membrane.

B. Binding To Glycosphingolipid Receptors A number of bacterial toxins, bacteria and viruses have been found to recognize and adhere to the oligosaccharide portion of glycosphingolipid receptors. The present invention contemplates inhibitors to block the binding of such organisms and/or their toxins to such receptors.

1. Bacterial Toxin Binding The present invention contemplates inhibiting bacterial toxin binding to cell receptors. A variety of toxins are known to bind carbohydrate. While not limited to particular toxins, the present invention contemplates, in particular, inhibiting cholera toxin, and the heat labile enterotoxins of E. coli and Campylobacter jejuni, all three of which recognize the oligosaccharide portion of ganglioside GM1 as a cell surface receptor.

Antigenic cross-reactivity has been observed between cholera toxin and both heat labile enterotoxins. The structures of cholera toxin and the heat labile enterotoxin of E. coli are quite similar. Both have five binding sites which are spaced at similar distances and are on one side of the protein. Cholera is a severe problem for third world countries and travelers'diarrhea is a problem for many people all over the world. At the moment there is no preventative for these toxin-induced diseases.

2. Binding Of Bacteria The present invention contemplates inhibiting binding of bacteria (or portions thereof) to cell receptors. A variety of bacteria are known to bind carbohydrate. For example, Pseudomonas aeruginosa pili mediate adherence to asialo-GMl [(See Lee et al., Molecular Microbiol. 11: 705 (1994)] and Helicobacter pylori can adhere to gangliotetraosylceramide, gangliotriaosylceramide, and phosphatidylethanolamine [ (See Lingwood et al., Infection and Immunity 61: 2474 (1993)]. Thus, the present invention specifically contemplates using asialo-GM1, as well as gangliotetraosylceramide and gangliotriaosylceramide, as sources of oligosaccharide for the preparation of carbohydrate derivatives.

In addition, Pneumocystis carinii (which causes pneumonia in patients with impaired immunity) adheres to cells that have mannose receptors. The bacteria has a mannose-rich surface component (glycoprotein A) that is recognized by mannose receptors present on macrophage within the lung (O'Riordan et al, Infection and Immunity 63: 779- 784,1995). In this instance, instead of using oligosaccharide-derivatized dendrimer to inhibit binding of bacteria to a cell surface glycoconjugate, the derivatized dendrimer (e. g. containing mannose or an oligosaccharide comprising mannose) is contemplated by the present invention to keep the cell from binding carbohydrate residues present on the surface of the bacteria.

3. Binding of Virus A variety of viruses are known to bind carbohydrate-containing receptors. A putative receptor for Human immunodeficiency virus type 1 (HIV-1) is gal-cer. Long et al. found that binding was critically dependent on the concentration of gal-cer in the target membrane suggesting that the binding was to glycolipid rich domains and that the gal-cer conformation might be important for gpl20 recognition. Long et al., J. Virol. 68: 5890 (1994). Thus, the present invention contemplates inhibiting HIV binding using a polyvalent galactosyl ligand (including but not limited to carbohydrate derivatives comprising dendrimers having galactose moieties or dendrimers having oligosaccharide

comprising galactose, such as oligosaccharides having terminal galactose residues).- Rotavirus is believed to adhere to carbohydrate receptors on epithelial cells in the small intestine. The binding is thought to be essential for infection and is probably carbohydrate mediated. Neutral lipids reported to function as receptors are gal (pal- 3) galNAc (ßl-4) glc (ßl-4) glc (pl-l) cer and pentaosylceramides with terminal galNAc residues. Srnka et al., Virol. 190: 794 (1992). Thus, the present invention specifically contemplates carbohydrate derivatives wherein in such oligosaccharides are employed.

C. Design Of Inhibitors To Receptor Binding A variety of inhibitors are contemplated by the present invention. As noted above, carbohydrate chains are present on a number of classes of molecules. In one embodiment, the present invention contemplates isolating the carbohydrate chain free of the rest of the molecule and attaching it to a dendrimer.

1. Isolation of the Carbohydrate Chain It is not intended that the present invention be limited to particular methods for isolating carbohydrate. A variety of methods are available. For example, several procedures have been developed for the cleavage and purification of the oligosaccharide portion of glycosphingolipids. See e. g., S. Hakomori, J. Lipid Res., 7: 789 (1966).

Typically, the initial reaction is the oxidative cleavage of the C-4-C-5 double bond of the sphingosine base. The free, intact oligosaccharide is then released by treatment with alkali.

A simpler procedure for the cleavage and isolation of the oligosaccharide portion consists of the selective oxidation (by e. g. 2,3-dichloro-5,6-dicyanobenzoquinone) of the allylic OH-3 of the sphingenine base. Then, the oligosaccharide can be cleaved from the 3-ketosphingolipid intermediate by a base-catalyzed p-elimination (e. g. by treatment with a treatment means such as triethylamine). See M. Miljkovic and C-L. Schengrund, "Oxidative Degradation of Glycosphingolipids Revisited: A Simple Preparation Of Oligosaccharides From Glycosphingolipids,"Carbohydrate Res. 155: 175 (1986).

2. Coupling Oligosaccharides to Macromolecules In one embodiment, the above-described oligosaccharides are coupled (covalently or non-covalently) to macromolecules. It is not intended that the present invention be limited to particular coupling chemistries or strategies. A variety of approaches are contemplated. In one embodiment, aryl-amine groups are introduced into the terminal reducing end of oligosaccharides by reacting them with 2- (4-aminophenyl)-ethylamine.

After subsequent conversion to the corresponding saccharide-phenylisothiocyanato derivatives, saccharides can be covalently linked to macromolecules. For example, saccharides can be linked to free lysylamine groups of proteins.

Some reasons for using the phenylisothiocyanate derivative instead of reductive amination to directly couple the oligosaccharide to the dendrimer include the fact that 1) the phenylisothiocyanate provides an additional 8.7A spacer which computer modeling indicates would provide a better"fit"between ligand and toxin, and 2) the reaction is more efficient. As others have noted, reductive amination reactions in aqueous medium are usually slow when molar amounts of aldehyde or amine are insufficient to drive the reaction. Synthesis of the phenylamine derivative by reductive amination is feasible because the molar amount of amine far exceeded the amount of oligo (such as oligo- GM1).

3. Dendrimers In one embodiment, the present invention contemplates attaching oligosaccharides (covalently or non-covalently) to dendrimers. It is not intended that the present invention be limited to particular dendrimers. In one embodiment, poly (propylene imine) dendrimers having four or eight primary amino groups and a Starburst (PAMAM) dendrimer having eight primary amino groups are used as core molecules, to which phenylisothiocyanate derivatized oligosaccharides are covalently attached to yield multivalent oligosaccharides. In a preferred embodiment, the oligosaccharides that are attached are isolated from glycolipids, and more particularly glycosphingolipids, and preferably glycosphingolipids involved in pathogen-binding or toxin-binding.

D. In Vivo Targeting As noted above, the present invention contemplates inhibiting of receptor binding both in vitro and in vivo. For in vivo inhibition, a variety of strategies are contemplated including but not limited to the use of dendrimers as carriers.

Goers et al. (U. S. Patent No. 4,867,973, hereby incorporated by reference) describe the use of antibody conjugated to antimicrobials. In contrast, the conjugates of the present invention do not require antibody. Instead, specificity is conferred by the carbohydrate interaction with the corresponding ligand.

It is not intended that the present invention be limited to the compound or drug conjugated to the glycosylated dendrimers of the present invention. A variety of drugs (such as those set forth by Goers, see above) that are toxic to microorganisms (e. g. bacteria, viruses, etc.) can be employed, as well as (in certain embodiments) those that are toxic to host cells (e. g. virally infected host cells and cancer cells).

E. Formulation And Delivery The carbohydrate derivatives of the present invention may be prepared either as liquid solutions or suspensions, or in solid forms. Oral formulations (e. g., for gastrointestinal diseases) usually include such normally employed additives such as binders, fillers, carriers, preservatives, stabilizing agents, emulsifiers, buffers and excipients as, for example, pharmaceutical grades of mannitol, lactose, starch, magnesium stearate, sodium saccharin, cellulose, magnesium carbonate, and the like. These compositions take the form of solutions, suspensions, tablets, pills, capsules, sustained release formulations, or powders, and typically contain 1%-95% of active ingredient, preferably 2%-70%. The compositions are also prepared as injectables, either as liquid solutions or suspensions; solid forms suitable for solution in, or suspension in, liquid prior to injection may also be prepared.

Delivery of the formulation containing the carbohydrate derivatives of the present invention may be achieved by a variety of routes, including but not limited to, intravenous injection. However, as noted above, for certain diseases, oral delivery is contemplated.

EXPERIMENTAL The following examples serve to illustrate certain preferred embodiments and aspects of the present invention and are not to be construed as limiting the scope thereof.

In the experimental disclosure which follows, the following abbreviations apply: eq (equivalents); u. (micron); M (Molar); uM (micromolar); mM (millimolar); N (Normal); mol (moles); mmol (millimoles); p. mol (micromoles); nmol (nanomoles); g (grams); mg (milligrams) ; ug (micrograms); ng (nanograms); L (liters); ml (milliliters); pl (microliters) ; cm (centimeters); mm (millimeters); pm (micrometers); nM (nanomolar) ; °C (degrees Centigrade); PBS (phosphate buffered saline); TLC (thin layer chromatography); HPTLC (high performance thin layer chromatography); HPLC (high pressure liquid chromatography); U (units); d (days).

In the experimental disclosure which follows, the following materials were obtained from commercial sources: V. cholera toxin B subunit horseradish peroxidase conjugate (CT-B-HRP), V cholera toxin B subunit (CT-B) and goat anti-choleragenoid antibody were obtained from List Biological Laboratories (Campbell, CA); generation 1 Starbursfm (PAMAM) dendrimer, acrylonitrile, 1,4-diaminobutane, borane-methyl sulfide complex (BH3 concentration of 10.0-10.2M), anhydrous methanol, anhydrous tetrahydrofuran, sodium cyanoborohydride, thiophosgene, and 2- (4-aminophenyl)- ethylamine from Aldrich (Milwaukee, WI); ganglioside standards from Matreya Inc.

(Pleasant Gap, PA); Supersignal CL-HRP Substrate System from Pierce (Rockford, IL); Bio-Gel P-2 [fine 45-90 mm (wet)] from Bio-Rad (Hercules, CA); centricon-3 and centricon'-10 filters from Amicon Corp. (Lexington, MA); PD-10 (Sephadex G-25M) columns, Sephadex G-25, and DEAE-Sephadex A-25 from Pharmacia LKB Biotechnology Inc. (Uppsala, Sweden); HPTLC (Silica Gel-60) plates from VWR (Bridgeport, NJ), Immunolon 1 Removawell strips from Dynatech Labs Inc. (Chantilly, VA); Alchemy' III (3D molecular modeling software) from Tripos Associates, Inc. (St. Louis, MO); a CarboPac PA1 (4x250mm) HPLC column and HPLC system from Dionex (Sunnyvale, CA), and X-OMAT AR film from Eastman Kodak Co. (Rochester, NY).

As noted above, the present invention contemplates isolating oligosaccharides from a variety of sources, including glycosphingolipids. In the case of the isolation of

gangliosides such as GM1, purification was carried out using bovine brain. The partial acid hydrolysis of disialo-, trisialo-, or mixed gangliosides was carried out. Briefly, 200 mg of gangliosides were dissolved by sonication in lml of O. 1N H2S04 and then incubated at 80°C for 45 min. After neutralizing the hydrolysate with ION NaOH, the sample was diluted to 5 ml with methanol/chloroform/water (2/1/0. 26, by vol) and centrifuged to precipitate insoluble material. GM1 was isolated from the supernatant by chromatography on a DEAE-Sephadex A-25 column (100 ml bed vol). The column was eluted with a step gradient of methanol/chloroform/water (60/30/8, by vol) to methanol/chloroform/l. OM sodium acetate (60/30/8, by vol). GM1 was recovered in the fraction eluted with methanol/chloroform/0. 2M sodium acetate (60/30/8, by vol). After the GM1 fraction was eluted, polysialylated lipids were recovered using methanol/chloroform/I. OM sodium acetate (60/30/8, by vol). Ganglioside containing fractions were rotoevaporated to remove organic solvents and dialyzed against water to remove salts prior to being dried under vacuum from the frozen state (lyophilized). Purity of each fraction was determined using HPTLC with chloroform/methanol/0.3% CaCl2 (60/35/8, by vol) or chloroform/ isopropanol/50mM KC1 (2/13.4/4.6, by vol). Sialic acid containing glycosphingo-lipids were visualized with resorcinol spray. GM1 was identified by its co-mobility with standard GM1 and its ability to function as a ligand for the binding subunit of cholera toxin. Recovered polysialylated gangliosides were rehydrolyzed and the procedure to isolate GM1 repeated.

To determine whether the GM1 used for the isolation of oligo-GM1 contained both N-acetyl-and N-glycolylneuraminic acid, the GM 1 was hydrolyzed and the released carbohydrate was analyzed by anion exchange chromatography using pulsed amperometric detection (Dionex, technical note 20,1989).

EXAMPLE 1 This example describes the synthesis of oligo-GMl derivatives. The oligosaccharide moiety of GM1 (oligo-GMl) was isolated chemically as described above (See Miljkovic and Schengrund, supra). Thereafter the phenylisothiocyanate derivative of oligo-GMl was synthesized. Figure 2 is a flow chart showing the synthesis scheme.

Compound 1 was converted to 2, the aminophenyl derivative, by reductive amination.

Compound 3, the phenylisothiocyanate derivative of oligoGMl was produced by reacting compound 2 with thiophosgene. The R in this particular case indicates the galpl- 3gaINAcpl-4 [sialic acid a2-3] galpl-4 portion of the oligosaccharide (although the reaction can be extended to other compounds such that R can comprise a variety of different sugars, linkages and sequences).

More specifically, 1 mM oligo-GMl was reductively aminated at 37°C for 90 hr in 122mM 2- (4-aminophenyl)-ethylamine in 200mM borate buffer, pH 8.0, containing 32mM sodium cyanoborohydride. The aminophenyl derivative of oligo-GMl was separated from unreacted oligo-GMl by chromatography on a Bio-Gel P2 column.

Twenty-five mM pyridine acetate buffer, pH 7.0, was used as the eluent and the aminophenyl derivative was eluted prior to oligo-GMl. Aliquots of each fraction were analyzed by HPTLC with chloroform/methanol/0.3% CaC12 (60/35/8, by vol) as the developing solvent and visualized using resorcinol. Fractions containing the aminophenyl derivative were combined and lyophilized.

Synthesis of the phenylisothiocyanate derivative was accomplished by incubating 9.6 moles of the aminophenyl derivative with 21.1 emotes of thiophosgene in 1 ml of 75% ethanol at room temperature for 1 hr. Volatile organic compounds were removed by evaporation under nitrogen, residual water by lyophilization. The remaining solid contained the phenylisothiocyanate derivative of oligo-GMl (oligo-GMl-PITC) plus any unreacted aminophenyl derivative (Figure 2). Purity of each derivative of oligo-GMl was determined by HPTLC with n-butanol/acetic acid/water (2/1/1, by vol) or methanol/n- butanol/water (2/1/1, by vol) used to develop the chromatograph and resorcinol to visualize sialic acid-containing compounds. Low resolution negative ion fast atom bombardment (-Fab) mass spectroscopy was used to determine the mass of the synthesized oligo-GMl-PITC.

A thin layer chromatograph of GM1 and the oligo-GMl derivatives is shown in Figure 3 (the origin is indicated by"O"). The compounds in each lane were as follows: a) GM1, B) oligo-GMl, C) the aminophenyl derivative of oligo-GMl, and D) the phenylisothiocyanate derivative of oligo-GMl.

The plate was developed in methanol/n-butanol/water (2/1/1, by vol) and the si-alic acid containing compounds visualized using resorcinol spray. Rf values were 0.84,0.76, 0.48, and 0.80, for GM1, oligo-GMl, oligo-GMl phenylamine, and oligo-GMl-PITC, respectively. Average percent yields were 31% for oligo-GMl, and 80% for the aminophenyl derivative. Purity of the derivatives as determined by densitometric scanning of bands on a thin layer chromatograph (Stratagene Eagle Eye scanner and NIH Image software) was essentially 100% for oligo-GMl and 90% for the oligo-GMl aminophenyl derivative. Densitometric scanning indicated that the phenylisothiocyanate derivative, which was not isolated from unreacted oligo-GMl aminophenyl derivative, was about 80% pure. It was not necessary to isolate the oligo-GMl-PITC derivative because it was the only product that would covalently bind to the amino groups of the dendrimer under the conditions used. The oligosaccharide contained N-acetyl and N-glycolyl derivatives of sialic acid since each was present in the GM1 used (94% and 6%, respectively). Low resolution-Fab mass spectroscopic analysis of the oligo-GMl-PITC, produced two major peaks with molecular masses of 1185 and 1201 (+ 0.5).

EXAMPLE 2 This example describes the synthesis of poly (propylene imine) dendrimers. First and second generation dendrimers, with 4 or 8 primary amino groups, respectively, were synthesized according to the method described by de Brabander-van den Berg and Meijer [(Angew. Chem. Int. Ed. Engl. 32: 1308 (1993)] except borane methylsulfide in anhydrous tetrahydrofuran (THF), instead of catalytic hydrogenation, was used to reduce nitriles on the dendrimer intermediates to the corresponding primary amines [(Brown et al., J. Org.

Chem. 47 : 3153 (1982)]. Figure 4 is a flow chart showing the synthesis scheme.

Synthesis of 1, an 0.5 generation poly (propylene imine) dendrimer was accomplished by reacting 1,4-diaminobutane with acrylonitrile. Borane mthyl sulfide was used to reduce compound 1 to produce the tetra (proplylene imine) dendrimer with four terminal primary amino groups, compound 2. Repetition of the addition and reduction reactions yields 3, the octa (propylene imine) dendrimer, with eight terminal primary amino groups.

Compound 3 is drawn in Figure 4 in its computer-modeled, lowest energy conformation.

The reduction was carried out as follows. Under a nitrogen atmosphere, 16.6 mmole dendrimer (CN) 4 was dissolved in 100ml anhydrous THF and heated to 50°C, prior to the gradual addition of 73.2 mmole of borane methylsulfide. After heating at 50°C for an hr, 1 OOmls of anhydrous methanol was slowly added, and excess borane was converted to methyl borate by bubbling anhydrous HCI through the mixture. The reaction mixture was then refluxed for one hr at 65°C to form the dendrimer generation 1-amine hydrochloride. The contents of the flask were rotoevaporated to dryness. Methyl borate was removed by resuspending the residue in 50 ml of methanol and drying it by rotoevaporation three times. Unreduced dendrimer was removed from the first generation product by dissolving the residue in 100ml of water and extracting it three times with 100 ml of chloroform. The extracted aqueous phase was dried, yielding the dendrimer- (NH2 HCl) 4 (generation 1.0). A second generation dendrimer was obtained by converting the dendrimer-(NH2 HC1) 4 to dendrimer- (CN) 8 and then reducing it to obtain dendrimer- (NH2 HCl) 8 (generation 2.0) using the procedures described above except that the molar ratio of borane methylsulfide to dendrimer-(CN) 8 was 8.5 to one (Figure 4).

The synthesis was facilitated by the solubilities of the different intermediates. The presence of terminal nitrile groups on the 0.5 and 1.5 generation poly (propylene imine) dendrimers, resulted in their being readily soluble in most organic solvents. In contrast, the terminal amino groups on the first and second generation poly (propylene imine) dendrimers made them soluble in water. The different solubilities permitted the use of solvent extraction as a purification tool.

IR was routinely used to monitor the presence of nitriles and primary amines during the stepwise synthesis of the tetra-and octa (propylene imine) dendrimers.

Generation 0.5 and 1.5, nitrile containing dendrimers, had a characteristic peak at-2260 cm-1 (data not shown). This peak was absent in generation 1.0 and 2.0 dendrimers which had four and eight primary amine groups, respectively, and absorbed in the frequency range indicative of primary amines, 3350-3500 cm-1. Analysis of generation 1.0 and 2.0 poly (propylene imine) dendrimers by HPTLC, using the solvents described above, showed that they remained at the origin. The compounds were strongly positive when visualized using ninhydrin spray. Generation 0.5 and 1.5, which lacked primary amines and moved

from the origin upon HPTLC, were ninhydrin negative. The mass of the first generation dendrimer as determined by low resolution-Fab mass spectroscopy was 317 (+ 0.5).

EXAMPLE 3 This example describes the synthesis of Oligo-GMl Dendrimers. Coupling of the oligo-GMl-PITC to each of the dendrimer cores was accomplished by combining oligo- GM1-PITC with the dendrimer in 0.2M borate buffer, pH 9.0. A 5: 1 molar ratio of oligosaccharide-PITC to dendrimer was used for dendrimers with four primary amino groups, and a 9: 1 molar ratio for dendrimers having eight primary amino groups. In each case, the mixture was stirred for 15 hr at 37°C.

Oligo-GMl dendrimers were isolated from lower molecular weight components by size exclusion centrifugation, using the appropriate centriconTM filter. Retentates were rinsed three times by addition of water followed by reconcentration prior to drying under vacuum from the frozen state. Recovery of oligo-GMl dendrimers was confirmed using HPTLC with either n-butanol/glacial acetic acid/H20 (2/1/1, by vol) or chloroform/methanol/0.3% calcium chloride (60/35/8, by vol) as the developing solvent.

Plates were visualized with both ninhydrin (to detect primary amines) and resorcinol (to detect sialic acid). Total sialic acid content of each oligo-GMl dendrimer was determined using the thiobarbituric acid procedure. Samples containing known concentrations of GM1 were treated in the same way in order to account for any loss of sialic acid due to incomplete hydrolysis or destruction of released sialic acid.

Centrifugation in centricons with appropriate molecular weight cutoff filters proved effective for separating oligo-GMl-PITC-dendrimers from underivatized dendrimers and oligo-GMl-PITC. Both the oligo-GMl-PITC and underivatized dendrimer core could be detected in the filtrate when analyzed by HPTLC (data not shown). Oligo-GMl-PITC- dendrimers, recovered in the retentate, remained at the origin when analyzed by HPTLC using the previously described mobile phases. Oligo-GMl-PITC-poly (propylene imine) dendrimers were resorcinol positive and ninhydrin negative.

Analysis of the sialic acid content of each of the modified dendrimers indicated that generation 1.0 poly (propylene imine) oligo-GM1-PITC-dendrimers had four sialic

acid residues, generation 2.0 poly (propylene imine) oligo-GMl-PITC-dendrimers had an average of seven sialosyl residues. Since each oligo-GMl-PITC moiety has only one sialosyl residue, the results indicate that under the conditions used, an average of seven of the eight primary amines present on the second generation poly (propylene imine) dendrimer were derivatized with oligo-GMl-PITC. The generation 1.0 poly (propylene imine) dendrimer was fully derivatized.

EXAMPLE 4 This example describes the synthesis of Oligo-GM 1 Dendrimers using the commercially available Starburst dendrimers. Coupling of the oligo-GMl-PITC was accomplished by combining oligo-GM1-PITC with the dendrimer in 0.2M borate buffer, pH 9.0. A 20: 1 molar ratio of oligosaccharide-PITC to dendrimer was used and the mixture was stirred for 30 hr at 37°C. Separation and analysis was carried out as described above for the synthesized dendrimers. The oligo-GMl-PITC-StarburstTM dendrimer was both resorcinol and ninhydrin positive, even though more rigorous conditions were used for its synthesis compared to those used for the linkage of oligo- GM1-PITC to the corresponding poly (propylene imine) dendrimer.

Analysis of the sialic acid content of the modified dendrimers indicated that StarburstTM oligo-GMl-PITC-dendrimers had an average of six sialosyl residues. Since each oligo-GMl-PITC moiety has only one sialosyl residue, the results indicate that under the conditions used, an average of six of the eight primary amines on the Starburst dendrimer were derivatized with oligo-GMl-PITC.

EXAMPLE 5 This example describes HPTLC overlay experiments to determine whether cholera toxin B subunit ("CT-B") would bind to the oligo-GMl dendrimers (made as described above) and whether that adherence involved the GM1 binding sites on the toxin. Briefly, 0.65 nmoles of GM1, asialo-GMl, and oligo-GMl-PITC dendrimers were chromatographed using HPTLC (methanol/glacial acetic acid/H2O, 15/7.5/0.25, by vol).

The plates were air dried, and dipped in an 0.075% solution of polyisobutyl methacrylate.

Nonspecific binding sites were blocked by incubating the plate in 0.1% BSA in phosphate buffered saline for 1 hr at room temperature. The plate was then incubated in 0.1% BSA in phosphate buffered saline containing CT-B-HRP conjugate (1: 50,000 dilution) for 1 hr at room temperature. Following incubation, the plate was rinsed seven times in phosphate buffered saline, pH 7.2. Adherence of the B subunit was detected using SuperSignalTM CL-HRP, an enhanced chemiluminescent substrate for HRP.

Chemiluminescent detection was used to monitor adherence of the toxin to A) asialo GM1, B) GM1, C) tetra (propylene imine) (oligo-GMl-PITC) 4, D) octa (propylene imine) (oligo-GMl-PITC) 7, and E) polyclonal anti cholera toxin binding subunit antibody (Figure 5). Lanes A'-E'show the same compounds as in A-E overlaid with HRP-CT that was preincubated with GM1. Plates were developed in methanol/glacial acetic acid/water (15: 7.5: 0.25, by vol.). The origin is indicated by the O.

From the results, it is clear that CT-B-HRP adhered to GM1, to first and second generation poly (propylene imine) oligo-GMl-PITC dendrimers, and to a polyclonal goat anti-choleragenoid antibody, adsorbed to the HPTLC plate after the plate was run (Figure 5). Oligo-GMl-PITC StarburstTM dendrimer was not tested under these conditions. CT- B-HRP did not adhere to asialo-GM1. When the HPTLC overlay experiment was repeated using CT-B-HRP preincubated with 6.5ßM GM1 (for 1 hr at 37°C), the toxin no longer bound to either GM1 or the poly (propylene imine) oligo-GMl-PITC-dendrimers, but did adhere to the anti-choleragenoid antibody.

EXAMPLE 6 In this example, well-binding assays were done to determine the effectiveness of the oligo-GMl dendrimers at inhibiting the adherence of cholera toxin B subunit to GM1- coated wells. Briefly, GM1 in methanol was added to plastic microtiter wells and allowed to dry. Potential nonspecific binding sites were blocked by incubating the wells with phosphate buffered saline containing 0.1% BSA for 1 hr at 37°C. After 1 hr, the buffer was removed and the wells used for the binding assay. Wells lacking GM1, but blocked in the same way, were used to determine nonspecific binding. Labelled cholera binding subunit (-6nM) was preincubated for 1 hr at 37°C in the presence or absence of inhibitor

in phosphate buffered saline (pH 7.2) containing 0.1% BSA and then added to GM1- coated plastic wells. After incubating for 1 hr at 37°C, the toxin was removed, the wells washed seven times with PBS, and bound 121 I-labeled toxin determined by counting in a gamma counter.

The results are shown in Figure 6. Each point is the average of quadruplicate samples. The closed circles indicate octa (propylene imine) (oligo-GMl-PITC) 7 ; the closed squares indicate tetra (propylene imine) (oligo-GMl-PITC) 4 ; the open circles indicate StarburstTM (oligo-GMl-PITC) 6 ; the closed triangles indicate GM1 ; and open triangles indicate oligo-GMl.

From the results it is clear that all oligo-GMl dendrimers were more effective at inhibiting the adherence of 1211-labeled cholera toxin B subunit to GM 1-coated wells than was GM1. The concentration of poly (propylene imine) oligo-GMl-PITC dendrimer needed to inhibit adherence of the binding subunit to GM1 by 50% (IC50) was 3nM for the octa (propylene imine) (oligo-GMl-PITC) 7 dendrimer and 7-8nM for both the tetra (propylene imine) (oligo-GMl-PITC) 4 and StarburstTM (oligo-GMl-PITC) 6 dendrimers compared to an average of 40nM for GM1. In contrast, the IC50 for oligo-GMl was 10pM. Comparable concentrations of underivatized or acetylated dendrimers did not inhibit adherence of the binding subunit to GMl-coated wells. Replicate experiments gave similar results.

EXAMPLE 7 This example describes molecular modeling to predict the lowest energy conformation and size of each of the fully derivatized oligo-GMl-PITC dendrimers. The software used was Alchemy III. Each of the oligo-GMl-PITC dendrimers was predicted to be large enough to span the diameter of the cholera toxin B subunit. Using the predicted structures, average molecular distances between the oligo-GMl moieties of the oligo-GMl-PITC dendrimers were compared to the geometry of the GM1 binding sites predicted by x-ray crystallographic structure determination of the toxin B subunit.

Figure 7 shows the predicited lowest-energy conformation of the fully derivatized oligo-GMl-PITC and Starburst dendrimers. The tetra (propylene imine) oligo-GMl-

PITC dendrimer is on the left, the octa (propylene imine) oligo-GMl-PITC dendrimer is in the center, and the StarburstTM oligo-GMl-PITC dendrimer is on the right. An oligo- GM1-PITC moiety can be seen at the end of each dendrimer arm. Structures shown are uniformly scaled to show their relative sizes.

The octa (propylene imine) oligo-GMl-PITC-dendrimer, which if completely derivatized was predicted to have its oligo-GMl-PITC moieties spaced approximately the same distance apart as the binding sites on the B subunit, was the best inhibitor. In contrast, the StarburstTM oligo-GMl-PITC-dendrimer, which, if fully derivatized, was predicted to have its oligo-GMl-PITC moieties too far apart for optimal interaction with the binding sites on the B subunit, was a less effective inhibitor.

Molecular modeling also predicted that the poly (propylene imine) dendrimer would be more flexible than the StarburstTM. The poly (propylene imine) dendrimer core is linear in nature, whereas the StarburstT dendrimer core contains amide bonds. The presence of the amide bond restricts the freedom of rotation at that site thereby decreasing its molecular flexibility. The three dimensional structure of the core molecules predicted more"intramolecular room"around each of the arms of the poly (propylene imine) dendrimers compared to those of the Starburst dendrimer. Therefore, although the oligo-GMl-PITC moieties on the tetra (propylene imine) dendrimer may not exactly fit the binding sites on the B subunit, it is possible that the molecular flexibility of the dendrimer arms would allow the oligosaccharides to move into appropriate positions.

Molecular modeling may also explain why the tetra (propylene imine) oligo-GMl- PITC-dendrimer was not as efficient a ligand as the octa (propylene imine) oligo-GMl- PITC-dendrimer. In order to effectively interact with the cholera toxin B subunit, the oligo-GMl-PITC moieties of the dendrimer need to be in appropriate alignment with the binding sites, which are all on one side of the binding subunit. The predicted lowest- energy, three-dimensional structure suggests that if fully derivatized, the octa (propylene imine) oligo-GMl-PITC-dendrimer would have a spherical-like conformation, with the oligosaccharide moieties distributed evenly (Figure 7). Because of this, some of its oligo- GM1 moieties would consistantly be held in position for optimal interaction with the cholera toxin B subunit. The tetra (propylene imine) oligo-GMl-PITC dendrimer, on the

other hand, is more planar. Its conformation predicts that effective binding with the toxin would most likely occur only when the toxin docked from the top or bottom of the derivatized dendrimer.

Figure 8 shows an overlay of the crystal structure of the choleragenoid with the Alchemy'III predicted structure of the tetra (propylene imine) (oligo-GM 1-PITC) 4 derivative. The relative sizes of the toxin (green), and the tetra (propylene imine) oligo- GM1-PITC dendrimer (multi-color) are shown. The structures indicate that the potential exists for the choleragenoid (GM 1-binding sites are located near the"points"of the toxin pentamer) to adhere to more than one of the oligo-GMl-PITC moieties at the end of each arm of the dendrimer. In addition, it appears that adherence of the choleragenoid to three oligo-GMl-PITC moieties would result in blockage of the central pore through which the A subunit putatively must pass in order to affect the adenylate cyclase activity of the cell.

Experimental evidence supports the molecular modeling prediction of an increased steric hinderence around the Starburst dendrimer core. First, synthesis of the StarburstTM oligo-GMl-PITC dendrimer required a longer coupling time and a higher concentration of oligo-GM1-PITC, compared to the conditions used for the poly (propylene imine) dendrimer cores ; and second, a quantitative assay for associated sialic acid residues indicated that the Starburst dendrimer has an average of six amino groups derivatized with oligo-GMl-PITC, while the octa (propylene imine) dendrimer has an average of 7, and all of the amino groups of the tetra (propylene imine) dendrimer were derivatized.

EXAMPLE 8 This example describes the ability of oligo-GMl-PITC derivatized dendrimers to inhibit adherence of choleragenoid, cholera holotoxin, and the heat labile enterotoxin of E. coli to GM1 on the surface of viable NCTC-2071 cells (chemically transformed murine fibroblasts) was investigated. Since each binding subunit of the toxins has a single tryptophan residue, the effect that adherence of the toxin to either GM1, oligo-GMl, or derivatized dendrimer had on their tryptophan fluorescence emission spectra was determined to ascertain whether each ligand induced a comparable change in the tryptophan microenvironment.

Chemically transformed, GMl-deficient, mouse fibroblast NCTC-2071 cells were cultured, asceptically, in defined NCTC-135 media (Sigma Chemical Co.) with L- glutamine and 0.22% sodium bicarbonate. Cultures were grown at 37°C in 95% air/5% C02 and 90% humidity. Confluent cells were harvested by first exposing them for one minute at 37°C to 0.05% trypsin/versene containing 0.1% glucose and then rapping the flask sharply to dislodge the cells. Trypsin activity was inhibited by addition of an approximately equal volume of NCTC-135 medium supplemented with 3% fetal bovine serum (FBS) and the cells pelleted by centrifugation at 200 X g for five minutes. The supernatant was discarded and the cells resuspended in fresh, unsupplemented, NCTC-135 media. Number of viable cells was determined by counting trypan blue negative cells in a hemocytometer. Aliquots containing-5 X 106 cells were then seeded into 75 cm2 flasks in a total volume of 10 ml of fresh NCTC-135 media.

Since NCTC-2071 cells contain little cell surface GM1, it was necessary to determine how much GM1 had to be added to the cells to provide binding sites for the toxins, and how much labeled toxin would be used in the assays. Therefore, NCTC-2071 cells were grown for 18 hrs prior to harvest in media supplemented with increasing amounts of GM1 (0 to 500nM). After harvesting cells by scraping them into PBS, they were recovered by low speed centrifugation and washed three times with PBS. After the third rinse cells were resuspended in a small volume of NCTC-135 media and the number of viable cells determined by counting trypan blue negative cells in a hemocytometer.

Aliquots containing 3 X 105 viable cells in 100 ul were then added to an equivalent volume of media containing 12nM labeled choleragenoid. After one hr at 16°C samples were transferred to 0.2 urn filter microcentrifuge tube inserts. Nonspecific binding sites on the filters were blocked prior to use by filtering 500 1ll of PBS containing 0.1 % BSA through them. Samples were centrifuged at 1,000 X g for five min. Cells retained by the filters were rinsed three times by repetitive addition of 200 pL1 of PBS followed by centrifugation. Bound label was determined by counting that associated with cells retained on the filter inserts in a gamma counter. Binding to added GM1 was obtained by subtracting the counts associated with cells that were grown in media alone from those associated with cells grown in media containing GM1.

Figure 9A shows the results where cells were grown in the presence of media containing different amounts of GM1 prior to harvest 18 hr later and incubation with 6nM 1211-labeled choleragenoid for 1 hr at 16°C. Figure 9B shows the results cells were grown in media containing 50nM GM1 for 18 hrs prior to a 1 hr incubation at 16°C with the indicated concentrations of labeled choleragenoid. Cell-associated labeled choleragenoid was determined by counting in a gamma counter. Values shown are the average counts obtained for quadruplicate samples and replicate experiments gave similar results. Error bars indicate the standard deviation in counts obtained for four samples.

The results indicate that incubation of the cells with as little as 50nM GM1 resulted in the adherence of a significant amount of labeled choleragenoid and that the amount of choleragenoid that adhered to the cells was linearly related to the amount of GM1 added to the media (Figure 9A). When cells that had been incubated with 50nM GM1 were subsequently incubated for 1 hr at 16°C with increasing amounts of labeled choleragenoid, binding was found to be saturable (Figure 9B). To avoid saturation of binding sites on GM 1-treated cells, subsequent experiments were carried out using cells grown for 18 hr in media containing 200nM GM1 and a final concentration of 2nM labeled protein to monitor binding.

The effectiveness of (oligo-GMl-PITC) 7 octa (propylene imine) at inhibiting the adherence of labeled choleragenoid or toxin to GM 1-treated cells was determined as follows. Two nM labeled protein was preincubated for 1 hr at 37°C with different concentrations of (oligo-GMl-PITC) 7 octa (propylene imine) prior to its addition to GM1- treated or control (grown in media alone) cells. The concentration of ligand used was based on the amount found previously to inhibit adherence of labeled choleragenoid or holotoxin to GMl-coated plastic wells by 50% (IC50s). For 6nM choleragenoid and heat labile enterotoxin those values were 3nM and 6nM, respectively for (oligo-GMl-PITC) 7 octa (propylene imine), while for 6nM cholera toxin it was 7nM. To minimize endocytosis of labeled choleragenoid or toxin, initial incubations of the cells with toxin plus dendrimer or toxin alone were done at 16°C. Subsequently, analogous experiments were done at 37°C. After 1 hr, samples were transferred to the filter microcentrifuge tube inserts and treated as described above. Specific binding was defined as the amount of label bound to

GMl-treated cells minus that bound to the same number of control cells. Values for each experimental point are the average of quadruplicate samples and each experiment was usually done twice. Experiments were also carried out in which the preincubation of toxin with dendrimer was omitted. In these experiments, inhibitor was added to the cells just prior to the addition of labeled toxin: the rest of the procedure was unchanged. In some of the assays done at 37°C, lOOug of bacitracin was added/ml of incubation media to reduce the possibility of receptor-mediated endocytosis during the incubation period.

Figure 10 shows the adherence of"'I-labeled choleragenoid (Figure 10A), cholera toxin (Figure l OB), and the heat labile enterotoxin of E. coli (Figure 1 OC) to GM 1-treated NCTC-2071 cells in the absence or presence of (oligo-GMl-PITC) 7 octa (propylene imine).

Average cell-associated counts are indicated by striped bars for experiments in which labeled choleragenoid was preincubated with (oligo-GMl-PITC) 7 octa (propylene imine) for 1 hr at 37°C prior to incubation with cells for 1 hr at 16°C, by open bars for experiments in which labeled choleragenoid or toxin was preincubated with derivatized dendrimer for 1 hr at 37°C prior to incubation with cells for 1 hr at 37°C, by solid bars for experiments in which dendrimer and choleragenoid or toxin were added directly to the cells which were then incubated for 1 hr at 37°C, and by dotted bars for experiments in which toxin and dendrimer were added directly to cells which were then incubated for 1 hr at 37°C in media containing 1 OOpg/ml of bacitracin. Cell-associated label was determined by counting in a gamma counter. The concentration of choleragenoid or toxin (2nM) and number of cells (3 X 105) grown in the presence of 200 nM GM1 was the same for each assay. Each value shown is the average of at least three separate measurements. Error bars indicate the standard deviation.

The results provide a quantitative assessment of the efficacy of (oligo-GM 1-PITC) 7 octa (propylene imine) as an inhibitor. A significant reduction in adherence of'25I-labeled choleragenoid, cholera holotoxin, or heat-labile enterotoxin of E. coli to GMl-treated cells was seen when (oligo-GMl-PITC) 7 octa (propylene imine) was present (Figure 1 OA- C). Inhibition of choleragenoid binding to GM 1-treated cells by (oligo-GMl-PITC) 7 octa (propylene imine) depended upon the concentration of derivatized dendrimer used.

While significant inhibition (r=0. 0002,2-tailed unpaired t test) of choleragenoid adherence

to GMl-treated cells was obtained when the choleragenoid and (oligo-GM1-PITC) 7 octa (propylene imine) were added directly to GMl-treated cells, it was less than that obtained when the choleragenoid was preincubated with the oligo-GM1-PITC derivatized dendrimer prior to addition to the cells (51% and 94%, respectively). Bacitracin reduced the amount of cell-associated cholera holotoxin (Figure 10B). Counts per min shown in the graphs were obtained by subtracting label associated with control cells incubated with choleragenoid or holotoxin alone. This was done because the addition of 4 or 16 nM (oligo-GMl-PITC) 7octa (propylene imine) plus labeled choleragenoid or holotoxin directly to control cells resulted in the adherence of as much label to the control cells as was found in association with GMl-treated cells. Interestingly, this binding was always significantly less than the amount of label associated with GM1 treated cells incubated with choleragenoid or toxin alone. At a concentration of derivatized dendrimer equal to or greater than-50 nM, binding to control cells went down to about 2-3 times that seen with control cells minus derivatized dendrimer (usually 2,000-3,000 compared to < 1000 cpm for control cells incubated with choleragenoid or holotoxin alone). This effect was not as pronounced when labeled choleragenoid was preincubated with the derivatized dendrimer prior to addition to control cells. The cause of the enhanced binding to control cells was not determined.

EXAMPLE 9 The ability of (oligo-GMl-PITC) 7 octa (propylene imine) to inhibit adherence of 6nM choleragenoid to GMl-treated and control cells was also monitored using immunofluorescence. Cells (3 X 105) were seeded and grown in each of eight separate chambers on a microscope slide. Cells in half of the chambers were treated with GM1 as above, the others with media alone. Choleragenoid (6nM) was pre-incubated at 37°C for 1 hr in the presence or absence of 30nM (oligo-GM1-PITC) 7 octa (propylene imine) or 30nM GM1, in PBS containing 0.1% BSA and then added to cells that had been rinsed three times with 500 ul PBS containing 0.1% BSA. After a 1 hr incubation at 5°C, cells were rinsed three times with PBS and then fixed at room temperature in 200 p1 of 4% paraformaldehyde. Fixative was removed after 10 min and cells rinsed three times to

remove unbound choleragenoid prior to incubation at room temperature for 1 hr with goat anti-choleragenoid IgG in PBS containing 0.1% BSA. Cells were washed to remove excess primary antibody and bound antibody fluorescently labeled by incubating the cells with rabbit anti-goat IgG Cy3 conjugate in PBS containing 0.1% BSA. Cells were then washed with PBS and immunofluorescence observed using a fluorescence microscope equipped with a rhodamine filter. Nonspecific fluorescence was accounted for by monitoring a cell sample not exposed to choleragenoid but fixed and stained in the same way.

The immunofluorescent analysis of choleragenoid binding to GM 1-treated and control cells (data not shown) indicated that more choleragenoid adhered to cells grown in media containing 200 nM GM1 than to cells grown in media alone. Experiments to determine the effectiveness of GM1 and (oligo-GMl-PITC) 7 octa (propylene imine) at inhibiting adherence of choleragenoid to GM 1-treated cells indicated that preincubation of choleragenoid with (oligo-GM1-PITC) 7 octa (propylene imine) markedly reduced the amount of choleragenoid associated with cells compared to that seen when the derivatized dendrimer was replaced with an equivalent concentration of GM1 (data not shown). Cells prepared in the same manner, but not exposed to choleragenoid, showed little fluorescence indicating that there was minimal nonspecific association of the antibodies used.

EXAMPLE 10 This example describes experiments to determine the effect of (oligo-GM1-PITC) 7 octa (propylene imine) on NCTC-2071 cell viability. Cconfluent cells were grown in media containing the dendrimer at a concentration of 500nM. This concentration was selected because it was ten times the highest concentration of derivatized dendrimer used in most studies. After 18 hrs, media was removed, the cells harvested into PBS by scraping, and the number of trypan blue negative and positive cells counted using a hemocytometer.

The results indicate that exposure of the cells to 500 nM (oligo-GMl-PITC) 7 octa (propylene imine) for 18 hr at 37°C had no sigificant effect on cell viability as determined by trypan blue exclusion. Cell counts obtained for quadruplicate samples

indicated that the average percent of trypan-blue negative (viable) cells was 77% (+2.6%) for cells grown in media containing (oligo-GM1-PITC) 70cta (propylene imine) compared to 78% (+4.9%) for cells grown in media alone.

EXAMPLE 11 This example describes tyrptophan fluorescence analyses. The effect of potential ligands on the intrinsic tryptophan fluorescence of choleragenoid and the holotoxins was monitored using an Aminco-Bowmang series 2 luminescence spectrometer and associated software from SLM-Aminco (Rochester, NY). Choleragenoid or toxin (0.5 uM) in PBS, pH 7.2, was incubated at 37°C for 1 hr and then allowed to come to room temperature prior to monitoring its tryptophan fluorescence emission spectra. Samples were excited at 282 nm and the resulting fluorescence monitored over the wavelength range of 290- 430 nm. Samples containing choleragenoid or toxin plus the ligand to be studied in a 1: 5 molar ratio (toxin: ligand), were incubated at 37°C for 1 hr prior to recording their emission spectra. A five-fold concentration of ligand was used because five molecules of oligo-GM1 were needed to occupy each of the binding sites in one molecule of toxin and to be consistent the same ratio was used in each experiment. However, the observations that 1) the mass of a labeled choleragenoid-derivatized dendrimer complex appeared to be less than 100,000 using a molecular weight cut-off filter, and 2) the IC50 for the dendrimer was about half the concentration of choleragenoid used (see above), indicated that one dendrimer molecule was adhered to by one pentameric binding subunit.

Figure 11 shows the tryptophan fluorescence emission spectra for choleragenoid plus oligo-GM1-containing compounds. Choleragenoid was incubated at 37°C for 1 hr in the absence (solid circles), or presence of a five-fold molar excess of GM 1 (open cirlces), oligo-GMl (solid squares), (oligo-GMl-PITC) , octa (propylene imine) (open squares), or (oligo-GMl-PITC) 4 tetra (propylene imine) (open triangles).

Choleragenoid, excited by ultraviolet light with a wavelength of 282 nm, had a tryptophan emission spectra with a maximum at approximately 350 nm. Similar emission spectra were observed when the choleragenoid was incubated with a five-fold molar excess of either of three gangliosides, GDla, GTlb, or asialo-GMl, that fail to function

as ligands for the binding subunit. In contrast, incubation of choleragenoid with a five- fold molar excess of GM1, oligo-GMl, (oligo-GMl-PITC) 7 octa (propylene imine), or (oligo-GM1-PITC) 4 tetra (propylene imine) resulted in almost identical-12 nm blue-shifts in the wavelength at which maximum fluorescence emission was observed (Figure 11).

Similar changes in fluorescent emission maxima were observed when either cholera holotoxin (Figure 12A) or the heat labile enterotoxin of E. coli (Figure 12B) was incubated with the same oligo-GMl-containing compounds at 37°C for 1 hr in the absence (solid circles) or presence of a five-fold molar excess of GM1 (solid triangles), (oligo- GM1-PITC) 7 octa (propylene imine) (solid diamonds), (oligo-GMl-PITC) 4 tetra (propylene imine) (open diamonds), oligo-GMl (solid circles), or asialo-GMl (open cirlces). No blue-shift was observed when the holotoxins were incubated with asialo-GMl. The reduction in maximum fluorescent intensity seen with the derivatized dendrimers may be due to quenching by the PITC residues.

From the above, it should be clear that derivatized dendrimers (such as the oligo- GMl-PITC-derivatized dendrimers) are readily synthesized. Moreover, by presenting oligosaccharide in a clustered arrangement, are as effective or better ligands than the natural source (e. g. GM1) for binding pathogens and/or their toxins (e. g. cholera toxin).

The observation that an oligo-GM1-PITC-derivatized dendrimer was a better inhibitor of the adherence of the binding subunit than was native GM1 is important because it may provide a model for developing compounds that can inhibit the adherence of pathogens or toxins that recognize glycosphingolipids as a ligand. The lipid per se can not be used because it has been shown that exogenous glycosphingolipids can become functional components of a cell's plasma membrane. As a result, a previously nonsusceptible cell can be made susceptible to the toxic agent.

By varying the attached carbohydrate, the mode of attachment, or the dendrimer core,"multivalent"oligosaccharides that function as efficient ligands for specific pathogens can be designed and synthesized. This approach may be useful for preventing or treating illnesses in which the binding of a pathogenic agent to a clustered carbohydrate ligand is an essential step in the etiology of the disease.

A possible reason for the efficacy of the oligosaccharide derivatized dendrimers is that the dendrimer core promotes the radial distribution of the added oligosaccharide moieties, effectively clustering them into an"artificial"micelle. The use of dendrimers for the synthesis of multivalent oligosaccharides has the advantage of providing more defined products than are obtained using protein, peptide, or polyacrylamide cores.