Background

1. Field of the Invention

The present application relates to the field of growth factor-glycosaminoglycan interactions.

2. Background of the Invention

Glycosaminoglycan Structure

Glycosaminoglycans (GAG) are naturally-occurring carbohydrate-based molecules implicated in regulation of a number of cellular processes, including blood coagulation, angiogenesis, tumor growth, nerve cell development, smooth muscle cell proliferation, and gene expression, most likely by interaction with effector molecules. GAG's are linear, non- branched chains of repeating two-sugar (disaccharide) units which may be up to 150 units in length, and are well known and described in the art. See, for example, Jackson, et al., (1991) Physiological Reviews 71 : 481-539 and Kjellen, et al., (1991) Ann. Rev. Biochem. 60: 443- 475. GAG's are often, but not always, found covalently bound to protein cores in structures called proteoglycans. Proteoglycan structures are abundant on cell surfaces and are associated with the extracellular matrix around cells.

Glycosaminoglycans (also referred to herein and the art as "glycans") can be divided into four main classes on the basis of the repeating disaccharide unit in the backbone. Typically, one sugar is a uronic acid, and the other is either an N-acetylglucosamine or an N- acetylgalactosamine. The classes are exemplified by the following four GAGs: (1) heparan sulfate (HS) (D-glucuronic acid/N-acetyl- or N-sulfo-D-glucosamine); (2) chondroitin/dermatan sulfate (D-glucuronic acid or L-iduronic acid/N-acetyl-D- galactosamine); (3) keratan sulfate (D-galactose/N-acetyl-D-glucosamine), and (4) hyaluronic acid. All GAGs, with the exception of hyaluronic acid, contain sulfate groups variously esterified to the ring hydroxyl groups of the sugars. These negatively charged groups are believed to figure prominently in the biological properties attributed to glycosaminoglycans. The naturally-occurring forms of GAGs, particularly heparin, heparan sulfate. chondroitin sulfate and dermatan sulfate. in fact are complex hetero-oligosaccharides composed of mixtures of differentially sulfated sugar residues.

One of the most thoroughly studied glycosaminoglycans is the widely used anticoagulant heparin. Heparin is a highly sulfated form of heparan sulfate, which is found in most cells. As a commercial product, heparin is a hetero-ohgosaccharide composition of about 20-60 monomeric units, having an overall extended length of about 100-300 A, having no protein associated with it, and its anticoagulant properties can be ascribed exclusively to the specific sulfation patterns found on the carbohydrate chains. So-called "low molecular weight" hepann typically is a hetero-ohgodisacchaπde composition of about 25-30 monomeric units, having an overall extended length of about 4θA. Heparin is known to have a variety of potentially useful biological activities beyond its ability to inhibit blood coagulation including, for example, the ability to block complement activation, smooth muscle cell proliferation and tumor growth. However, the toxicity of heparin at the levels required to manifest these activities in vivo has limited its clinical use. Heparan sulfate, the predominant GAG on cell surfaces, contains fewer sulfate groups than heparin and has been shown to contain regions of high sulfation interspersed among regions of low or no sulfation.

Other polysulfated compounds described in the art and asserted to have clinically useful activities analogous to those attributed to hepann include fractions or fragments of the naturally-occurnng GAGs, pentosan polysulfate (PPS), dextran sulfate, chondroitin sulfate, and keratan sulfate; and suramin, a polysulfonated naphthylurea whose structural motif likely mimics that of a GAG sequence. As for heparin, the toxicity of these compounds at the levels required for therapeutic utility has limited their clinical use. A representative listing of publications describing these compounds and their asserted biological activities includes: US Patent No. 5.158,940 issued October 27, 1992; US Patent No. 4.826,827. issued May 2. 1989; international patent publication Nos. WO 90/15816 (public December 27, 1990), WO 91/13624 (public September 19, 1991 ), and WO 93/07864 (public April 29, 1993); Wellstein, et al., (1991) J. Natl. Cancer Inst. 83: 716-720; and Jentsch, et al., (1987) J. Gen. Virol. 68: 2183-2192.

GAG Binding Proteins and GAG Binding Specificity

Many important regulatory proteins bind tightly to heparin. including chemokines. growth factors (including cytokines). enzymes and proteins involved in lipid metabolism This binding property was, for a long time, thought to arise only from non-specific ionic interactions involving positively charged regions on the proteins with the negatively charged sulfates of heparin. However, recent results with two proteins. Antithrombin III (AT III)

and basic fibroblast growth factor (bFGF). demonstrate that the interactions between heparin and AT HI or bFGF can show specificity. The specific interaction involves complex binding sites on the protein molecule and infrequently occurring sequences in the heparin GAG chain. See. for example, EPO patent publication 0 509 517 A2, published October 21, 1992; Turnbull. et al., (1992) J. Biol. Chem. 267: 10337-10341 ; Gallagher, et al., ( 1992) Glvcobiologv 2: 523-528; Habuchi. et al., (1992) J. Biochem. 285: 805-813; Yayon, et al., ( 1991) Cell 64: 841-848; and Rapraeger. et al.. (1991) Science 252: 1705-1708.

That specific protein binding sequences might exist in the carbohydrate chain of heparin was first suggested by the observation that some preparations were more effective than others in inhibiting coagulation. Careful studies in 1987 revealed that there is a defined five sugar sequence (pentasaccharide) with a characteristic sulfation pattern that represents the specific binding site for AT III. a protease inhibitor that blocks the action of thrombin and other enzymes which initiate blood coagulation. The Kd for the binding between AT III and this specific GAG recognition site is about 10 nM (10 ^~ 8 M), which qualifies it as a high affinity interaction. Although weaker and less specific binding of these proteins to other regions of heparin can occur, virtually all of the anticoagulant activity of heparin is attributable to this five sugar sequence. This pentasaccharide, generally known as the AT UI binding site, now has been synthesized chemically and shown to possess the appropriate activities of the naturally occurring sequence. Binding of AT III to this site is thought to provide the basis for heparin's anticoagulant activity by positioning and "presenting" the enzyme inhibitor to the proteases thrombin and Factor Xa.

A second example of a somewhat specific binding site has been reported for fibroblast growth factor. This GAG sequence, isolated from fibroblast heparan sulfate. was found to represent the tightest binding fraction present. It is not clear, however, whether other molecules such as other heparin binding growth factors can bind to this sequence, nor is it clear that the affinity of this binding is as high as the binding between bFGF and heparin. The interaction between the isolated GAG sequence and bFGF might, at present, best be described then as selective, rather than absolutely specific.

Growth Factors and Cytokines

It is well recognized that the endogenous heteroligodisaccharides heparan sulfate and heparin bind with appreciable affinity to a wide spectrum of the mitogenic proteins termed cytokines and growth factors, although the strength of these interactions varies considerably among the different factors. Among the growth factors and cytokines described as

heparin/HS -binding proteins are: TGF-β, endothelial cell growth factor, IL3 and GM-CSF, interferon-γ, hepatocyte growth factor, fibroblast growth factor (FGF) -1 (acidic FGF), FGF- 2 (basic FGF), FGF-3 (int-2), FGF-4 (Hst-1, K-FGF), FGF-5, FGF-6 (Hst-2) and FGF-7 (keratinocyte GF). For example, heparin will release TGF-β from inactive complexes with α2-macroglobulin and will potentiate TGF-β action. The stability in solution of acidic and basic FGF (aFGF and bFGF) is enhanced in the presence of HS/heparin, and the polysaccharides potentiate the mitogenic activity of the FGFs, especially of aFGF. These effects are presumed to be due to the formation of complexes between FGF and heparin which prolong the biological lifetime of the proteins by protecting them from proteolysis and thermal denaturation. In tissues, aFGF and bFGF can be detected in the extracellular matrix and basement membranes, where they are bound to HS. It has been proposed that the action of heparinases or proteases that degrade heparan sulfate proteoglycans will release FGFs from the basement membranes enabling them to act on nearby target cells. In addition to effects on FGF stability and tissue localization, a central role has now been described for HS in controlling the interaction of bFGF with cell signalling receptors.

Neuregulins

A recently described family of growth factors, the neuregulins (reviewed by Mudge, ( 1993) Curr. Biol. 3:361 : Peles and Yarden, (1993) Bioessavs 15:815). are synthesized by neurons (Marchionni, et al., (1993) Nature 362:313) and by mesenchymal cells from several parenchymal organs (Meyer and Birchmeier, (1994) PNAS 91 : 1064). The neuregulins and related factors that bind pl85 ^erD -B2 have been purified, cloned and expressed (Benveniste, et al., PNAS, 82:3930, 1985; Kimura, et al., (1990) Nature 348:257: Davis and Stroobant, ( 1990) J. Cell Biol. 110: 1353; Wen. et al., (1992) Ceil 69:559; Yarden and Ullrich, (1988)

Ann. Rev. Biochem. 57:443: Dobashi. et al., (1991) Proc. Natl. Acad. Sci. 88:8582: Lupu. et al., (1992) Proc. Natl. Acad. Sci. 89:2287: Wen, et al., (1994) Mol. Cell. Biol. 14: 1909).

Recombinant neuregulins have been shown to be mitogenic for peripheral glia (Marchionni, et al., (1993) Nature 362:313) and have been shown to influence the formation of the neuromuscular junction (Falls, et al., (1993) C l 72:801; Jo. et al., (1995) Nature 373: 158;

Chu, et al., Cell 14: 329, 1995).

The neuregulin gene consists of at least thirteen exons. The neuregulin transcripts are alternatively spliced and these encode many distinct peptide growth factors, which are referred to as the neuregulins (Marchionni, et al., Nature 362:313, 1993). DNA sequence comparisons revealed that neu differentiation factor (NDF) (Wen. et al., (1992) Q \ 69:559) and heregulins (Holmes, et al., (1992) Science 256: 1205), which were purified as ligands of

the pl85 ^erD -B (also known as neu or HER2) receptor tyrosine kinase, also are splicing variants of the neuregulin gene. The acetyichohne receptor inducing activity (ARIA) also is a product of the neuregulin gene (Falls, et al., (1993) Cell 72:801). A common structural feature of the neuregulins is the presence of a single epidermal growth factor-like (EGF) domain.

The sites of neuregulin gene expression have been characterized by use of nucleic acid probes to analyze RNA samples by a variety of methods, such as Northern blotting, RNase protection, or in situ hybridization. Transcripts have been detected in the nervous system and in a variety of other tissues (Holmes, et al., (1992) Science 256:1205; Wen, et al., (1992) Ceil 69:559; Meyer and Birchmeier, (1994) PNAS 21:1064). Sites of gene expression have been localized in the brain and spinal chord and in other tissues. (Orr- Urteger, et al., (1993) PNAS 90:1867: Falls, et al., (1993) Cell 72:801 ; Marchionni, et al., (1993) Nature 362:313; Meyer and Birchmeier, (1994) PNAS 91 :1064: Chen, et al., (1994) J. Comp. Neurol. 349:389; Corfas, et al., (1995) Neuron 14: 103). Specifically in the retinal neurepithelium, expression of neuregulin transcripts has been detected at embryonic day 18 in rat (Meyer and Birchmeier, (1994) PNAS 91 : 10641.

Although a large number of neuregulins may be produced by alternative splicing, they can be broadly sorted into the putative membrane-bound and the soluble isoforms. The former contains a putative trans-membrane domain and may be presented at the cell surface. Membrane-anchored peptide growth factors may mediate cell-cell interactions through cell- adhesion or juxtacrine mechanisms (reviewed by Massague and Pandiella, (1993) Ann. Rev. Biochem. 62:515). Alternatively, the putative membrane-bound isoforms may be cleaved from the cell surface and function as soluble proteins (Wen, et al., (1992) Cell 69:559: Falls, et al., (1993) Cell 72:801 ). The soluble neuregulin isoforms contain sequences corresponding to the extracellular domains of the putative membrane-bound isoforms, but terminate before the transmembrane domain. These neuregulin isoforms may be secreted, and hence could affect cells at a distance; or they may be present in the cytoplasm, but could be released upon cellular injury. In the latter case, neuregulins may function as injury factors, as has been postulated for the ciliary neurotrophic factor (Stockli, et al, (1989) Nature 342:920). Any one of these modes of action of the neuregulins may occur.

Cellular targets of peptide growth factors are those which bear receptors for the factor(s) and those that are shown to respond in a bioassay either in vitro or in vivo. Based on studies demonstrating phosphorylation on tyrosine residues or cross-linking experiments, neuregulins are candidate ligands for the receptor tyrosine kinases pl85 ^er °B2 (or HER-2 in

human), pl85 ^{erbB 3} (HER-3 in human), pl85 ^erbB4 (or HER-4 in human) or related members of the EGF receptor (EGFR) gene family. Collectively, these receptors can be referred to as erbB receptors. Though the precise ligand-receptor relationship of each neuregulin protein with each member of the EGFR family is yet to be clarified, several lines of evidence suggest that binding of ligands is mediated by either erbB3 and erbB4 , but signaling occurs through either erbB2. erbB4 and heterodimers of the various subunits (e.g., Carraway and Cantley, (1994) Cell 78:5). These receptors are known to be present on

Schwann cells and muscle cells (Jo, et al., (1995) Nature 373: 158 . and other neuregulin targets, such as cell lines derived from various tumor tissues, such as breast and gastric epithelia. Sites of expression of the HER-4 gene have been localized by in situ hybridization to several regions of the brain, including: hippocampus, dentate gyms, neo cortex, medial habenula, reticular nucleus of the thalamus, and the amygdala (Lai and

Lemke. ( 1991 ) Neuron 6:691 ).

Neuregulins have been shown to have a variety of biological activities depending on the cell type being studied. Several neuregulins, including native bovine GGFI. II and III and recombinant human GGF2 (rhGGF2) are mitogenic for Schwann cells (Marchionni, et al., (1993) Nature 362:313), as is heregulin B l (Levi, et al, (1995) J Neurosci. 15: 1329). Further activities of rhGGF2 on Schwann cells include the stimulation of migration and the induction of neurotrophic factors, such as nerve growth factor (Mahanthappa, (1994) Soc. Neurosci. Abst #691.7). On human muscle culture, rhGGF2 has a potent trophic effect on myotubes (Sklar, et al., U.S. Pat. Applic. # 08/059, 022). The differentiation response to rhGGF2 also includes induction of acetyichohne receptors in cultured myotubes (Jo, et al.. (1995) Nature 373: 158). This activity is associated with other forms of neuregulin. including ARIA (Falls et al.. ( 1993) CeU 72:801 ) and heregulin B l (Chu. et al., (1995) Neuron 14:329). as well as with rhGGF2. Further, ARIA has been shown to induce synthesis of voltage-gated sodium channels in chick skeletal muscle (Corfas and Fischbach, (1993) J. Neurosci. 13:2118). Glial growth factor (GGF), and more specifically rhGGF2, can restrict neural crest stem cells to differentiate into glial cells in vitro (Shah, et al, (1994) Cell 77:349). In summary, there are examples of neuregulin proteins affecting proliferation, survival and differentiation of target cells.

Neuregulins provide signals for growth and differentiation of cells by binding and activating several members of the EGFR subfamily of receptor tyrosine kinases: erbB2. erbB3, and erbB4 (Padhy, et al., (1982) CeU 23:865-872: Coussens, et al. (1985) Science

230: 1 132-1 139: Kraus. et al., ( 1993) Proc Natl Acad Sci USA 90:2900-2904: Plowman, et al., (1993) Nature 366:473-475).

Information relating to neuregulins (also referred to as heregulins and erbB2 ligands) provided in U.S. Patent No. 5.367,060, issued November 22, 1994 and U.S. Patent No. 5,530,109, issued June 25, 1996, is hereby incorporated by reference.

Proteoglycans are mediators of growth factor function.

Cell surface proteoglycans and the extracellular matrix (ECM) are important components in the regulation of growth, motility, and differentiation (for review, see Adams, et al., (1993) Development 1 17: 1 183-1 198). Of particular relevance to growth factor signaling are the cell surface and ECM forms of heparan sulfate proteoglycans (HeSPGs). HeSPGs are known to bind fibroblast growth factors, interieukin 2. hepatocyte growth factor, platelet-derived growth factor B, and many others (Adams, et al. (1993) Development 1 17: 1 183-1 198). Methods for the purification of NDF (Yarden, et al., (1991) Biochem 30:3543-3550: Peles, et al., (1992) CeU 69:205-216), HRG-α (Holmes, et al., (1992) Science 256: 1205-1210), and ARIA (Falls, et al., (1993) CeU 72:801-815) all made use of heparin-affmity chromatography.

Cancer

Activation or altered expression of erbB receptors seems to be an important step in the development of certain cancers. A specific amino acid replacement in the transmembrane region of pl85 ^erD -B2 produced a constitutively active receptor tyrosine kinase leading to oncogenic transformation in EtNU-treated rats. Studies of transgenic mice expressing this oncogenic form of neu (neuT) have shown that systemic administration of anti-erbB2 antibodies prevented the development of breast tumors for up to 90 weeks of age. Although this specific mutation has not been observed in human cancers, amplified expression of the several different receptors of this family has been associated with malignancy. Among these cancers are: for erbB2 (HER-2, in human) — adenocarcinoma of breast, stomach, and ovary; for EGFR—malignant gliomas and squamous cell carcinomas, and; for erbB3 and erbB4 ~ breast carcinomas. Hence, both animal studies and molecular diagnosis of cancer patients have implicated deregulated or ectopic expression of erbB receptors in oncogenic transformation of a anti-HER2 (erbB2) monoclonal antibodies in patients with metastatic breast cancer are most gratifying. Complete or partial remission was observed in 1 1.6% of treated patients following 10 weekly doses of rhuMab HER2. These data build a strong case for pharmaceutical intervention directed at this signalling pathway.

Continuous stimulation of the erbB receptor signalling pathway could also be mediated by excessive production of neuregulins. For example, mitogenic activity isolated from bilateral acoustic neuromas and Schwannomas showed chromatographic properties similar to native GGF purified from bovine pituitary. More recently the neoplastic phenotype of a mutagenized, transformed Syrian hamster fibroblast cell line has been attributed to a hyperactive autocrine loop mediated by neuregulins. Hence neuregulins are expressed at a place and a time which could trigger continuous proliferation of cells. It thus would be highly desirable to develop therapeutic agents for the treatment of cancers which involves neuregulins.

Summarv of the Invention

It now has been discovered that the activity of neuregulins is modulated by the interaction of neuregulins with specific, determinable oligodisaccharide structures

("glycans") pendant from proteoglycans immobilized on a cell or extracellular surface. The specific binding between a neuregulin and the oligodisaccharides is determined by the structure and sequence of saccharides, typically disaccharides, and usually including the sulfation pattern defined within the oligodisaccharide unit, all of which together define a binding site having relatively high affinity and specificity for a given neuregulin

("Glyceptor"). These oligodisaccharide surface-immobilized binding sites differ from true cell surface receptors in that they typically lack the transmembrane signaling function associated with ligand-receptor binding and, typically, bind to a site on a neuregulin different from that recognized by the receptor binding site. This new observation presents an opportunity for modulation and control of the physiological function of neuregulins.

In one aspect, the method of the invention comprises the step of substantially preventing or otherwise interfering with the interaction of a neuregulin of interest with its Glyceptor sequence. By interfering in some way with the interaction between a neuregulin and its Glyceptor sequence, one effectively interferes with the ability of a neuregulin to interact with its receptor or other protein. The step of substantially preventing or otherwise interfering with the interaction of a neuregulin with its Glyceptor sequence may be achieved by administering to an animal a molecule that acts as a Glyceptor sequence analog ("Glyceptor sequence antagonist"), and which competes with the Glyceptor sequence for neuregulin binding. The Glyceptor sequence antagonist may act by preventing the protein from interacting with its Glyceptor sequence, and or by competitively displacing a protein from its Glyceptor sequence seat. Useful Glyceptor binding sequence antagonists

contemplated by the invention include soluble forms of the Glyceptor binding sequence, or any other synthetic or natural-sourced sequences that constitute or functionally mimic the structure of the Glyceptor sequence, and which have a specific, predetermined composition.

It has not been previously demonstrated or predicted that the interaction between the vast majority of neuregulins and surface-immobilized GAG chains can show any degree of specificity. The unanticipated discovery of such specificity now enables the development of a kind of inhibitory molecule, not previously envisioned, that can specifically antagonize the action of a given neuregulin. For example, one can now more specifically antagonize the action of neuregulins while not significantly affecting the action of other heparin-binding growth factors. Moreover, the discovery now enables the development of analogs of specific gylcosaminoglycan sequences that can act as agonists or have other utilities in vivo including, for example, as imaging or other tissue-targeting agents.

Provided herein is an enabling description of the fundamental discovery and resulting concept which permits the identification of such therapeutically useful compounds. Also provided are an enabling description of a process for identifying and isolating the full range of specific GAG binding sequences and an enabling description of a process for utilizing such sequences to screen for therapeutically useful compounds, as well as a description of the characteristics defining useful natural source-derived or synthetically produced analogs, including those that act as antagonists to the protein-glycan interaction.

Thus, it is an object of this invention to provide means for modulating a biological effect induced by a neuregulin-receptor or neuregulin-protein interaction by modulating, including preventing or otherwise interfering with, the interaction between a neuregulin and the glycan sequence that binds it. Another object is to teach a method for identifying and isolating analogs of a glycan sequence having specificity for a given neuregulin, and to teach the use of these analogs as agonists or antagonists. Another object is to provide means for testing various types of cancer by controlling undesired cell growth and proliferation, in vivo. Yet another object of the invention is to provide means for therapeutic and prophylactic manipulation of neuregulins and related biological molecule function, including providing novel compositions, and providing a process for discovering useful novel compositions. Such compositions have utility for altering pathologic responses by inhibiting or enhancing the action of one or more members of the neuregulin family of growth factors.

These and other objects and features of the invention will be apparent from the description, drawings and claims which follow.

The foregoing and other objects of the invention, the various features thereof, as well as the invention itself, may be more fully understood from the following description, when read together with the accompanying drawings.

Brief Description of the Drawings

Figure 1 shows the effects of distinct glycosaminoglycans on rhGGF2-induced phosphorylation of pl85 (putative neuregulin receptor) as detected by anti-phosphotyrosine immunoblotting.

Figure 2A shows the effects of varying doses of heparin on rhGGF2-induced DNA synthesis in cultured Schwann cells.

Figure 2B shows the effects of varying doses of heparan sulfate on rhGGF2-induced DNA synthesis in cultured Schwann cells.

Figure 2C shows the effects of various glycosaminoglycans on rhGGF2-induced DNA synthesis in cultured Schwann cells.

Figure 3 shows the effects of 4-methyllumbelliferyl-β-D-xyloside (an inhibitor of proteoglycan biosynthesis) on rhGGF2-induced DNA synthesis in cultured Schwann cells.

Figure 4A is a schematic diagram of how neuregulins signal cells by interacting with "Glyceptors".

Figure 4B is a schematic diagram of how "Glyceptors" sequence antagonists interfere with neuregulin signaling by preventing normal neuregulin binding to heparan sulfate proteoglycans.

Figure 4C is a schematic diagram of how ligand antagonists can interfere with neuregulin signaling by preventing normal neuregulin binding to heparan sulfate proteoglycans by occupying neuregulin binding sites on heparan sulfate proteoglycans.

Figure 5A shows the retardation of ¹²⁵ I-heparin mobility by varying concentrations of rhGGF2 (affinity co-electrophoresis gel shown in inset).

Figure 5B shows the retardation of ¹²⁵ I-heparin mobility by varying concentrations of rhGGF2 (affinity co-electrophoresis gel shown in inset).

, Figure 6 shows the effects of increasing rhGGF2 concentrations on the retention of rhGGF2/complexes in the filter binding assay.

Figure 7 shows the effects of three different polyanion compounds on rhGGF2/heparin complex formation in the filter binding assay.

Figure 8A shows the effects of three polyanion compounds on rhGGF2-induced DNA synthesis.

Figure 8B shows that the polyanion effects on rhGGF2-induced DNA synthesis are reversible as evidenced by the washout ("w.o.") conditions.

Figure 9 illustrates the chemical reaction of a primary amine, an aldehyde or Ketone, a carboxylic acid and an isonitrile to form an acylaminoacid.

Figure 10 illustrates potential functional groups for the four sets of reactants; acids, aldehydes or ketones, amines and isonitriles.

Figure 11 represents the 96-well format used for the construction of a pilot combinatorial library with various aldehydes and amines as reactants.

Figure 12 represents a generic formula for acylamino acid amides.

Figure 13 illustrates the effect of identified combinatorial compounds in the schwann cell proliferation assay.

Figure 14 illustrates the structures of the four identified combinatorial compounds tested in the schwann cell proliferation assay.

Detailed Description of the Invention

The invention provides a method for modulating the interaction between neuregulins and their Glyceptor sequences. As used herein, the term Glyceptor sequence refers to an oligodisaccharide sequence, including sulfated disaccharides, contained within a given glycosaminoglycan immobilized on a cell or extracellular matrix surface and which binds, with specificity, to a glycan-binding protein. Thus, as used herein, Glyceptor sequence and "neuregulin-specific glycosaminoglycan sequence" are used interchangeably and are understood to be synonyms. The molecular surface structure on the neuregulin that interacts specifically with a given Glyceptor sequence is referred to herein as a "glycan-binding site."

Typically, the Glyceptor sequence-effector protein interaction alone has no transmembrane signal transducing or other direct effect. Without being limited to any given theory, interaction of a neuregulin with its Glyceptor sequence serves to enable, or otherwise enhance, the ability of the neuregulin to interact with its receptor or other protein and/or to facilitate signalling. The ligand-receptor or other protein-protein interaction may occur on the same cell surface to which the Glyceptor sequence is immobilized, may occur on an adjacent cell, or may occur on an extracellular matrix surface.

For example, exogenous heparin and HS inhibit both early and late Schwann cell responses to rhGGF2. The rapid phosphorylation of erbB2 family members is a consistent, early event in cells responding to neuregulins (Peles, et al., ( 1993) BioEssays 15:815-824). In the case of cultured primary rat Schwann cells, a protein of M _r = 185 kD (pl85), a size consistent with erbB family members, is phosphorylated within 2 minutes of exposure to rhGGF2 (Marchionni. et al., (1993) Nature 362:312-318). If HeSPGs play a critical role in neuregulin binding prior to receptor tyrosine kinase activation, then exogenously applied heparin or heparin-like molecules should serve as competitive, soluble neuregulin receptors and thereby inhibit pl 85 phosphorylation. Schwann cells were exposed to various GAGs, or to 15 ng/ml of rhGGF2 in the presence of these GAGs. As can be seen in Figure 1. heparin inhibits p l 85 phosphorylation in a dose-dependent manner and completely blocks phosphorylation at 1.0 mg/ml. Though significantly less potent than heparin, HS treatment exerts some inhibition at 1.0 mg/ml. Neither keratan sulfate. dermatan sulfate, nor chondroitin sulfate show any inhibition of p 185 phosphorylation at the doses tested.

While pl85 phosphorylation in Schwann cells is detected within 2-3 minutes of exposure to rhGGF2, detectable DNA synthesis in response to the factor is assayed over a 48 hour period of exposure. Heparin and HS also inhibit this mitogenic response (Figures 2A

and B). As in the case of inhibiting pl85 phosphorylation. heparin is a more potent inhibitor of rhGGF2-induced DNA synthesis than HS. At the highest dose of rhGGF2 tested (32 ng/ml) heparin inhibits DNA synthesis by 90% at a concentration of 1.0 mg/ml; at a concentration of 10 mg/ml HS inhibits DNA synthesis by 50-60 %. In marked contrast, keratan sulfate, dermatan sulfate, and chondroitin sulfate show only modest inhibitory activity at a concentration of 10 mg/ml (Figure 2C). Thus, heparin and the closely related HS are the only GAGs tested that block Schwann cell responsiveness to rhGGF2. This result is suggestive of a direct interaction between rhGGF2 and HeSPGs that can be inhibited by exogenous heparin-like molecules.

Further, a known inhibitor of proteoglycan biosynthesis inhibits Schwann cell responsiveness to rhGGF2. Rather than competing with cell surface HeSPGs in the binding of rhGGF2, the effect of exogenous heparin and heparin-related molecules might be inteφreted as solely a result of sequestering rhGGF2 away from the cell surface through interactions akin to heparin-affinity chromatography. In order to perturb directly the expression of Schwann cell HeSPGs, the cells were cultured in the presence of β-xyloside; cell permeable β-xylosides affect proteoglycan biosynthesis by inhibiting the attachment of GAG chains to proteoglycan core proteins (Robinson, et al., (1975) Biochem J 148:25-34; Galligani, et al., (1975) J Biol Chem 250:5400-5406). In the case of Schwann cells, culture in the presence of 0.5-1.0 mM β-xyloside for 1 week results in a 75% to 80% reduction in GAG attachment to cell associated proteoglycans, and a 10- to 12-fold increase in free GAG chains in the culture medium (Carey, et al., (1987) J Cell Biol 1987:1013-1021). As can be seen in Figure 3, culture of Schwann cells in medium containing 0.5-1.0 mM β-xyloside for 48 hours followed by culture for an additional 48 hours in medium containing β-xyloside plus rhGGF2 results in a 60% to 90% decrease in maximum DNA synthesis relative to control cells grown in medium containing rhGGF2 plus equivalent dilutions of the β- xyloside solvent, dimethylsufoxide. Thus the degree to which Schwann cell responsiveness to rhGGF2 decreased in the presence of β-xyloside was proportional to reported decreases in cell-attached GAGs induced by this inhibitor of proteoglycan biosynthesis (Carey, et al., (1987) J Cell Biol 1987:1013-1021).

Thus the activity of neuregulins is modulated by the interaction of these proteins with specific, determinable oligodisaccharide structures ("glycans") pendant from proteoglycans immobilized on a cell or extracellular surface. The specific binding between glycan-binding proteins and the oligodisaccharides is determined by the structure and sequence of saccharides, typically disaccharides, and usually including the sulfation pattern defined within the oligodisaccharide unit, all of which together define a binding site having

relatively high affinity and specificity for a given glycan-binding protein. These oligosaccharide surface-immobilized binding sites differ from true cell surface receptors in that they typically lack the transmembrane signaling function associated with ligand- receptor binding and, typically, bind to a site on the protein different from that recognized by the receptor binding site. The above observations present an opportunity for modulation and control of the physiological function of neuregulins.

In one aspect, the invention provides a method for modulating the biological effect induced by a neuregulin by interfering with, or otherwise preventing interaction of, a given growth factor with its Glyceptor sequence which may be on the same cell as the growth factor receptor, a neighboring cell, or the extracellular matrix. Thus, the method of the invention comprises the step of substantially preventing or otherwise interfering with the interaction of a neuregulin with its Glyceptor sequence. By interfering in some way with the interaction between the neuregulin and its Glyceptor sequence, one effectively interferes with the ability of the protein to interact with its receptor or other protein. The step of substantially preventing or otherwise interfering with the interaction of the neuregulin with its Glyceptor sequence may be achieved by administering to an animal a molecule that acts as a Glyceptor sequence analog ("Glyceptor sequence antagonist"), and which competes with the Glyceptor sequence for neuregulin binding. The Glyceptor sequence antagonist may act by preventing the neuregulin from interacting with its Glyceptor sequence, and/or competitively displace a neuregulin from its Glyceptor sequence. Useful Glyceptor sequence antagonists contemplated by the invention include soluble forms of the Glyceptor binding sequence, or any other synthetic or natural-sourced sequences that constitute or functionally mimic the structure of the Glyceptor sequence, and which have a specific, predetermined composition.

The method is anticipated to be particularly useful in inhibiting undesired cell proliferation, such as can occur in a hyperproliferative disease, including cancers. As neuregulins have been shown to stimulate migration of cells in culture, the method can be useful in inhibiting metastasis of tumors.

Alternatively, a molecule may be administered that is a glycan-binding protein analog ("ligand antagonist" or "decoy ligand") which competes with the neuregulin for Glyceptor sequence binding and. when bound, prevents or substantially inhibits the protein from binding to the Glyceptor sequence, and or competitively displaces protein from its Glyceptor sequence. Such neuregulin antagonists include antibodies recognizing one or more epitopes on the Glyceptor sequence and capable of blocking or otherwise interfering

with the ligand binding site on the Glyceptor sequence. In one embodiment, useful neuregulin antagonists include modified, soluble forms of a neuregulin that can still bind the Glyceptor sequence with specificity but which, as modified, can not interact with the other protein or receptor necessary for effecting the biological activity in vivo. Ligand antagonists, including the modified glycan-binding effector protein, have an additional utility as in vivo targeting agents. For example, an imaging or cytotoxic agent can be complexed with the modified glycan-binding protein using standard means, such as by covalent attachment, and be targeted to the site of action of the ligand thereby. Methods for creating target-specific complexes are well-known and are well described in, for example, the cancer therapeutic art. Still other useful ligand antagonists include synthetic organics defining a molecule capable of mimicking the glycan binding site on the ligand.

In still another aspect, the invention contemplates a chimeric synthetic molecule comprising at least two Glyceptor sequences covalently linked and having a conformation sufficient to allow concurrent binding of each said protein-specific glycosaminoglycan sequence to a specific glycan-binding protein. Preferably, each protein-specific glycosaminoglycan sequence binds to a different protein-specific glycosaminoglycan binding site. The two binding sites may be tethered by means of a linker capable of acting as a spacer as well as a crosslinking means. Preferably, the linker also allows free rotation of the two sites independent of one another. The chimeric molecule is anticipated to have particular utility as an agonist functioning, for example, to evoke receptor dimerization and/or to help present a glycan-binding effector protein to a receptor, by binding both a soluble effector protein and a surface bound protein.

In one embodiment, the Glyceptor sequence analogs of the invention useful as antagonists and agonists have a binding affinity for neuregulins defined by a dissociation constant in the range of lO'^M to 10" ^M, preferably having a dissociation constant of less than 5 x 10 ^' ^M, more preferably less than 10"^M, or even 10'^M. As will be appreciated by those skilled in the art, the higher the binding affinity of the analog, the lower the concentration needed to induce a therapeutic effect in vivo, and the less likely the molecule is therefore, to induce a toxic response.

In another embodiment, the overall length of the Glyceptor sequence analog preferably does not exceed about 40A. and preferably is less than 40A, on the order of about 15-20A. Where the isolated Glyceptor sequence or an analog to be used comprises an oligodisaccharide sequence, the molecule preferably has fewer than 20 monomer units, preferably fewer than 16 monomer units, most preferably between 4-15 units, inclusive.

Smaller oligosaccharide sequences may reduce specificity and larger sequences may enhance toxicity. Preferred oligosaccharide antagonists also have an overall length of less than 40 A. In all cases, the oligodisaccharide analogs contemplated have a specific, predetermined composition, which serves to distinguish the compositions of the invention from the endogenous soluble heterogenous oligosaccharide mixtures that may be found in the body.

Non-oligodisaccharide molecules useful as Glyceptor sequence analogs include antibodies or other peptides capable of interacting specifically with the Glyceptor sequence binding site on neuregulins. Still another useful class of molecules includes synthetic organic molecules whose chemical structure functionally mimics that of a Glyceptor sequence in binding specifically with a glycan-binding protein. These synthetic constructs may or may not include carbohydrate and amino acid sequences. For example, suramin. a polysulfonated naphthylurea, interferes with Glyceptor sequence-effector molecule interactions, presumably by competing with the Glyceptor sequence for the ligand binding site. Here, the naphthylurea likely provides a scaffolding or backbone structure with an appropriate distribution of sulfonates disposed about the heterocyclic rings to functionally mimic the sulfated oligodisaccharide sequence that defines the protein-specific glycosaminoglycan sequence. Thus, other synthetic organics can be generated having unique backbone structures, and on which are disposed constituents of appropriate charge and size. However derived, the Glyceptor sequence analog preferably has an appropriate distribution of functional substituents capable of interacting specifically with the glycan binding site. For synthetic organics, the appropriate substituents may be provided by pendant carboxylates, phosphates, sulfonates, hydroxylates, amino groups, alkyl and aromatic moieties.

In another embodiment, the synthetic organic Glyceptor sequence antagonist is derived from the class of molecules whose structure is based on features of the glycan binding site on the neuregulins with which the glycan analog interacts, the characteristics of the class being defined by the generic structure as described in detail herein below. As will be appreciated by those having ordinary skill in the art of chemical synthesis, a combinatorial library can be constructed containing multiple candidate sequences created based on the generic structure, and the candidates tested in the screening assay presented herein to identify useful analogs, including antagonists, having appropriate affinity for a ligand. Similarly, a combinatorial library "kit" can be constructed containing isolated, captured candidate molecules defined by the generic structure, a neuregulin. and a means for

screening the candidates for their ability to bind said glycan binding protein with an affinity above a preselected threshold level.

Thus, in still another aspect, the invention provides a method for identifying specific oligodisaccharide sequences and functional analogs thereof which interact specifically with neuregulins. As described herein, selected oligodisaccharide sequences having a defined pattern of charged groups and a desired binding specificity and affinity for a given ligand may be identified and used to create serum-soluble Glyceptor sequence analogs, useful per se, or as screening reagents or templates for the rational design of polypeptide or organic- based Glyceptor sequence analogs. In still another aspect, the invention provides a high flux screening assay for identifying candidate analog molecules.

The invention essentially consists of compounds that are selected for their ability to mimic a specific Glyceptor sequence-neuregulin interaction, and methods for their selection and use. In one embodiment the compounds specifically inhibit the interaction of a neuregulin with its cognate binding sequence in a glycosaminoglycan (GAG) chain, and have particular utility for inhibiting that interaction in vivo. Figure 4 illustrates one mechanism of the compounds provided by the invention, as it pertains to neuregulin- Glyceptor sequence interactions, and where the Glyceptor sequence is a cell surface-bound sequence. The invention also is anticipated to be useful for neuregulin-Glyceptor sequence interactions where the Glyceptor sequence is bound to an extracellular matrix surface.

Figure 4A shows cells 210 having both Glyceptor sequences 218 and transmembrane receptors 216 on their surfaces. Neuregulin molecules 214 are aided in binding to receptors 216 by Glyceptor sequences 218. analogous to a Glyceptor sequence "hand" guiding a neuregulin "key" into a receptor "lock" to activate cell proliferation through a transmembrane signal. In Figure 4B, Glyceptor sequence antagonists 225 are reacting in a solution with neuregulin 214 and competitively stripping off neuregulin bound to Glyceptor sequences 218 to modulate (here diminish) the activity of the neuregulin 214 on the cells 210. Figure 4C shows ligand antagonists 230 similarly modulating neuregulin activity. In the case of neuregulins, use of the antagonists of the invention can interfere with, for example, undesired cell growth.

Useful Glyceptor sequence analog compounds of the invention may include antibodies or other related molecules capable of interacting and interfering with neuregulin- Glyceptor sequence interaction by binding the "Glyceptor." Antibodies can be made by standard means well known and described in the art (see, for example, Immunology. Roitt,

et al.. eds. Haφer and Row, New York, 1989) using an isolated oligodisaccharide Glyceptor sequence of interest as the antigen.

Other useful Glyceptor sequence analogs include soluble forms of GAG oligodisaccharides or may be derived from the endogenous cognate GAG binding sequences. Alternatively, the analog may be a mimetic compound of the cognate GAG sequence, including a non-carbohydrate mimetic compound. Briefly, the Glyceptor sequence analog compositions obtained through this enablement may be sulfated glycosaminoglycan oligodisaccharides of a predetermined, defined sequence, including that of a Glyceptor sequence found in nature, or they may be synthetic mimetics thereof. A process is provided herein for the selection of such suitable compounds.

A means for designing useful synthetic Glyceptor sequence analogs is to compare sequences of a number of glycan-binding growth factors and compare regions of homology and non-homology. This information, together with an investigation of the known three- dimensional structure of several such proteins permits one to identify a physical "map" of the glycan binding site on the effector protein. (See. for example, Baldwin, et al., ( 1991 ) PNAS 88: 502, and Clore, et al. (1991 ) J. Mol. Biol. 22: 61 1.) Characteristically, the glycan binding site is an extended band that stretches across the surface of the protein rather than, for example, defining a pocket, as may occur in enzyme-substrate interactions. Moreover, the glycan binding site typically is defined by a particular distribution of positively charged residues that interact favorably with the anionic charge on the GAG, and also may include other residues that can prevent or limit interaction with particular GAGs, either by steric or ionic interference.

The synthetic organic analog molecules useful in the invention are synthetic molecules that mimic the action of naturally-occurring GAG binding sequences, whether the synthetic molecule is naturally derived, synthetically produced, substantially oligodisaccharide in nature, or substantially free of carbohydrate. Such compounds may contain sulfate esters or negatively charged groups at precise locations in their structures that interact with the basic side chains that characterize the glycan binding site. In this case the analog specifically mimics the binding structure of the natural-sourced sequence. Alternatively, the analog may comprise functional groups that interact with other, different side chains in the glycan binding site, sufficient to allow specific binding interaction of the analog with the glycan binding site, but by means of different contacts. In either case the analog can be said to functionally mimic the protein binding structure of the native

Glyceptor sequence. Of course, a molecule that is a specific structural mimetic also will be a functional mimetic.

In one embodiment, candidate compounds obtained from nature may be screened as described herein. Alternatively, candidate compounds can be formulated utilizing an approach that includes consideration of size and charge distribution of the Glyceptor sequence and glycan binding site with which it interacts. Interaction may be achieved by contacts analogous to those made by the endogenous glycan sequence, or by different contacts that produce a functionally equivalent specific interaction at the site. As will be appreciated by those having ordinary skill in the art, analogs having higher or lower binding affinities than that of the endogenous sequence can be obtained by this method.

Preferably, and as described herein, a combinatorial "library" containing a group of designed candidates is created, each molecule having a different composition, and the group screened for molecules that bind the ligand of interest above a threshold affinity. A threshold affinity of a candidate Glyceptor sequence analog for a preselected glycan-binding protein readily may be determined by means of a standard competition assay, and/or by gel shift assay, as described and exemplified in Examples 3, 7 and 8, below. Preferably, candidates will exhibit binding affinities represented by low dissociation constants, e.g., having Kd values in the range of 10-7M to 10- 1 M, preferably less than 10-8 ₀ r even less than 10-9M.

Screening Protocol

The screening procedure for identifying lead compounds selective for binding to and inhibiting neuregulins is as follows:

1. Combinatorial synthesis reaction products— are. the contents of individual wells including solvent, starting materials, intermediates and final products. This represents the initial endpoint of the synthetic chemistry and the beginning of the biochemical screening steps.

2. Primary binding hits—are combinatorial synthesis reaction products that block a specified level of the binding of heparin to rhGGF2 when screened in a 96 well format.

3. Confirmed binding hits— τe primary binding hits that have reproducible blocking of heparin binding to rhGGF2 (to a specified level) when analyzed in individual assays. Complete binding inhibition curves are performed to determine the approximate IC50 values for all confirmed binding hits.

4. Selected binding candidates— are confirmed binding hits whose IC50 values are comparatively low and whose final products are present in sufficient yield in the synthesis reaction products to warrant resynthesis, product purification and extensive analytical chemistry and binding studies. Some of the selected binding candidates will advance to the biological screening assays as pure compounds.

5. Preliminary bioactive candidates— are selected binding cαndidαtes-t at are non-toxic and block responses of cultured Schwann cells to rhGGF2 in in vitro assays.

6. Confirmed bioactive candidates— are preliminary bioactive candidates that are tested on human tumor cell lines and found to be non-toxic and active in growth arrest assays.

7. Lead compounds— are confirmed bioactive candidates that are least 10-fold selective for binding to and inhibiting rhGGF2 compared to other heparin-binding growth factors.

A) Preliminary Screening

The solution phase binding assay described above is formatted into a high throughput screening system to select compounds that antagonize the binding of heparin to rhGGF2. Briefly, a 20nM solution of rhGGF2 is incubated for 60 minutes at room temperature with the test compound(s) and a trace amount of high affinity GAG fragment, which is tyraminated at the reducing end and iodinated. The screening assays can be carried out in 96 well Hybridot Manifold (BRL) that allows for rapid filtration through a nitrocellulose sheet. The sheet contains 96 radioactive "spots", which are subsequently counted in a Wallac microbeta microtiter plate scintillation counter. This approach has proven to be very efficient, and can be used routinely to screen up to 12 microtiter plates (960 wells)/week. To reduce the potential for artifacts, any binding effects of the starting components can be assayed. In all cases, the individual components of the isonitrile chemistry, the acid, amine and aldehyde, cannot be due to unreacted starting materials.

Libraries are analyzed first in a single point survey. Primary binding hits fall into two groups: 1) wells that blocked > 50% of binding and 2) wells that blocked 20-50% of binding. All of the primary binding hits are rescreened and approximate IC50 values are calculated on the confirmed binding hits. These compounds also are tested for color or chemical quenching, and any compound that interferes is eliminated. In order to maximize the possibility of finding an adequate number of selected binding candidates from the biochemical screening of these libraries, a "cut-off of 40 uM can be set.

In order to ascertain quickly if the library construction rationale is correct, the first few confirmed binding hits can be analyzed more extensively. Early confirmed binding hits are resynthesized and purified; and chemical structure is characterized by HRES-MS, to show a single. calculated molecular ion peak, as well as characteristic fragmentations, demonstrating that these confirmed binding hits are indeed the predicted acylaminoacid amides.

Once the prioritized confirmed binding hits are identified (IC50<10 uM, etc.), the compounds can be synthesized independently on a larger scale. For each synthesis, first the crude reaction is again checked by HPLC analysis, and is reassessed in the binding assay. If these analyses provide the expected data (i.e., retention time and binding inhibition similar to the initial screens), then the scaled up product is subjected to preparative-HPLC. All major peaks are collected and assayed. Fractions that show activity in the binding assay are then taken to dryness, and the solid material is analyzed by high-resolution mass

spectroscopy (HRES-MS), NMR or chemical analysis. Binding analyses of the pure compounds to rhGGF2 can be repeated several times to obtain IC50 values. Data from the analysis of each of these selected binding candidates can then be compared in order to prioritize those compounds that will advance to the next stage. Next, selected binding candidates are screened in cell culture assays to validate their potential as antagonists of neuregulin-erbB signaling.

Cultured primary rat Schwann cells provide the first bioassay system for testing the bioactivity of rhGGF2 antagonists. The purification, cloning and expression of glial growth factors, including rhGGF2, is based on the proliferative response of Schwann cells to neuregulins (Marchionni, et al., (1993) Nature 362: 312-318), and this response provides a highly reliable in vitro assay system to monitor the activity of rhGGF2.

Compounds (selected binding candidates) can be tested first for any overt toxicity on cultured cell lines (e.g., NIH-3T3) or primary rat Schwann cells. Non-toxic compounds can be tested for the inhibition of both early (neuregulin receptor phosphorylation) and late (DNA synthesis) Schwann cell responses to rhGGF2. and subsequently for reversibility of action. Using a high throughput assay for DNA synthesis, the first bioassay for selected binding candidates will be inhibition of rhGGF2-stimulated Schwann cell DNA synthesis. IC50 values can be determined and compared to the binding data.

Once inhibition of rhGGF2-stimulated DNA synthesis has been established for a specific compound, then pl 85 neuregulin receptor phosphorylation can be assayed as follows: Schwann cells are prepared as described above for the DNA synthesis assay and plated in 24-well plates at a concentration of 250.000 cells/O.5ml in DME/5. Test samples (rhGGF2 and various concentrations of selected binding candidates are premixed for 60 min. to equilibrate binding) are added to the cells for 2.5-3 minutes at 37°C. The media is aspirated and 50 ul of SDS reducing sample buffer (preheated to 65°C), is added to each well. After scraping and triturating the wells, the contents are transferred to microcentrifuge tubes, boiled for 10 minutes, and subjected to polyacrylamide gel electrophoresis. Proteins are electrotransferred to nitrocellulose membranes. The membranes are pre-soaked in transfer buffer containing 200 uM sodium orthovanadate (Sigma) and the transfer is performed using the same buffer. Membranes are probed with the recombinant anti- phosphotyrosine antibody. RC20H (Transduction Laboratories), and immunoreactive bands visualized using ECL chemiluminescence reagents (Amersham) per manufacturer ^' s instructions. Compounds that block both the receptor phosphorylation and the DNA

synthesis activities of rhGGF2 will be analyzed further for reversibility of action in washout experiments, as described in Sudhalter. et al. (Sudhalter, et al., (1996) Glia 17:28-38).

The results from these initial bioassays will identify compounds with bioactivity in the well-characterized Schwann cell system. At the conclusion of this stage of the screen, potential antagonists of neuregulin-erbB signaling should satisfy several key criteria, including: 1) reversibly block both the early and the late responses to rhGGF2; and 2) display a rank order of potency in vitro that matches the biochemical analyses. Compounds that pass successfully through this first stage of cell-based screening are designated as primary bioactive candidates. In order to determine if primary bioactive candidates have activity on cells that are more relevant to cancer cells, a second tier has been added to the cell-based screening stage, which involves the use of human tumor cell lines.

A number of human tumor cells lines are known to respond to neuregulins. Some lines also express one or more neuregulin isoforms. For example, the breast adenocarcinoma lines MDA-MB231 and MDA-MB-453 were utilized in the erbB2 receptor phosphorylation bioassays that were used to monitor the purification of NDF and heregulin (Wen, et al.,

(1992) CeU 69: 559-572; Holmes, et al., (1992) Science 256:1205-1210). Glial tumor lines

U87MG (astrocytoma) and U 13876 (glioblastoma) express neuregulin and erbB receptor transcripts detected by northern blotting. These lines can be obtained from the American

Type Culture Collection and will use them to further evaluate primary bioactive candidates.

Since erbB receptor activation and proliferation may be stimulated via paracrine or autocrine neuregulin signaling in these lines, the lines provide opportunities for two types of bioassays. First, in a manner similar to Schwann cells, we will determine if test compounds can block the responses to exogenous rhGGF2. Second, it can be determined if the compounds can arrest the growth of cell lines grown in low-to-moderate concentrations of fetal calf serum (FCS) without added rhGGF2. The concentration of FCS to be used can be determined empirically for each line.

Thus these experiments with human tumor lines should help to demonstrate the applicability of the primary screening system beyond neural cell growth and further validate the overall approach.

The instant application also provides novel process for selecting useful analogs of the glycan-binding protein Glyceptor sequence interaction from a collection of randomly obtained or rationally designed candidate compounds. In one aspect, compounds which are

selected by the process described herein will have the useful property of specifically displacing neuregulins, from their functionally active locations.

For example, use of neuregulin antagonists of the invention can interfere with undesired cell growth. And thus, can be used to prevent tumor growth.

Using this type of analysis, a generic structure useful in creating candidate Glyceptor sequence analogs can be generated, particularly useful as part of a combinatorial library of candidates. A preferred generic structure is presented in Figure 12. The generic structure defines an oligomeric structure composed of at least two monomeric units and one or more functional groups pendant therefrom. The position and composition of the pendant functional groups in the generic structure are designed to provide a "library" of useful substituents which can interact with the side chains defining a glycan binding site, whether by making the same contacts as that of the naturally occurring Glyceptor sequence, or by making different contacts.

The Glyceptor sequence antagonist or neuregulin antagonist for use as a therapeutic agent prepared as described herein may be provided to an individual by any suitable means, preferably directly or systemically, e.g., parenterally, preferably in combination with a pharmaceutically acceptable carrier. As used herein, "a physiologically acceptable carrier" includes any and all solvents, dispersion media, antibacterial and antifungal agents that are non-toxic to human, and the like. Particularly contemplated are pharmaceutically acceptable salts, which may be base salts, alkali metal salts, and alkaline metal salts. The use of such media and agents as pharmaceutically active substances are well known in the art.

Where the therapeutic agent is to be provided directly (e.g., locally, as by injection, to a desired tissue site), or parenterally, such as by intravenous, subcutaneous, intramuscular, intraorbial, ophthalmic, intraventricular, intracranial, intracapsular, intraspinal, intracisternal, intraperitoneal. buccal. rectal, vaginal, intranasal or by aerosol administration, for example, the therapeutic agent preferably comprises part of an aqueous solution. The solution is physiologically acceptable so that in addition to delivery of the desired therapeutic agent to the patient, the solution does not otherwise adversely affect the patient ^' s electrolyte and volume balance. The aqueous medium for the therapeutic agent thus may comprise normal physiologic saline (0.9% NaCl, 0.15M), pH 7-7.4.

Useful solutions for oral or parenteral administration may be prepared by any of the methods well known in the pharmaceutical art, described. for example, in Remington's

Pharmaceutical Sciences, (Gennaro, A., ed., Mack Pub., 1990). Formulations may include, for example, useful excipients to control the release of the therapeutic agent in vivo. The therapeutic agents also provided herein may be administered alone or in combination with other molecules known to have a beneficial effect in, for example, modulating the inflammatory response, and or which may enhance targeting of the agent to a desired tissue or cell surface. Other useful cofactors may include symptom-alleviating cofactors. including , antiseptics, antibiotics, antiviral and antifungal agents and analgesics and anesthetics.

The compounds provided herein can be formulated into pharmaceutical compositions by admixture with pharmaceutically acceptable nontoxic excipients and carriers. As noted above, such compositions may be prepared for parenteral administration, particularly in the form of liquid solutions or suspensions; for direct administration, in the form of powders, nasal drops or aerosols.

Compounds of the invention can be employed, either alone or in combination with one or more other therapeutic agents as discussed above, as a pharmaceutical composition in mixture with conventional excipient, i.e., pharmaceutically acceptable organic or inorganic carrier substances suitable for parenteral, enteral or intranasal application which do not deleteriously react with the active compounds and are not deleterious to the recipient thereof. Suitable pharmaceutically acceptable carriers include but are not limited to water, salt solutions, alcohol, vegetable oils, polyethylene glycols, gelatin, lactose, amylose, magnesium stearate, talc, silicic acid, viscous paraffin, perfume oil, fatty acid monolgycerides and diglycerides, petroethral fatty acid esters, hydroxymethyl-cellulose, polyvinylpyrrolidone, etc. The pharmaceutical preparations can be sterilized and if desired mixed with auxiliary agents, e.g., lubricants, preservatives, stabilizers, wetting agents, emulsifiers. salts for influencing osmotic pressure, buffers, colorings, flavorings and/or aromatic substances and the like which do not deleteriously react with the active compounds.

The compositions can be formulated for parenteral or oral administration to humans or other mammals in therapeutically effective amounts, e.g., amounts which provide appropriate concentrations of the agent to target tissue for a time sufficient to inhibit the desired ligand - Glyceptor sequence interaction of interest, as described above.

As will be appreciated by those skilled in the art, the concentration of the compound described in a therapeutic composition will vary depending upon a number of factors, including the dosage of the drug to be administered, the chemical characteristics (e.g.,

hydrophobicity) of the compounds employed, and the route of administration. The preferred dosage of drug to be administered also is likely to depend on such variables as the type and extent of tissue loss or defect, the overall health status of the particular patient, the relative biological efficacy of the compound selected, the formulation of the compound, the presence and types of excipients in the formulation, and route of administration. In general terms, the compounds of this invention may be provided in an aqueous physiological buffer solution containing about 0.001 to 01 % w/v compound for parenteral administration. Typical dose ranges are from about 10 ng/kg to about 1 g/kg of body weight per day; a preferred dose range is from about 0.1 ug/kg to 100 mg/kg of body weight. It will be appreciated by those having ordinary skill in the art that analogs having higher binding affinities typically will require lower total administration concentrations than those having comparatively lower binding affinities.

ABBREVIATIONS

Throughout the present specification the following abbreviations and terms are used:

Glyceptor sequence - protein-specific glycosaminoglycan sequence; HS - heparan sulfate; HSPG - HS proteoglycan; GF - growth factor; dp - degree of polymerization (e.g. for a disaccharide, dp=2, etc.); GlcA - glucuronic acid; IdoA - iduronic acid; IdoA (2s) - iduronic acid 2-sulfate; GlcNAc - N-acetyl glucosamine; GlcNSOβ - N-sulfated glucosamine; GlcNSOβ (6s) - N-sulfated glucosamine 6-sulfate; GlcA (2s) - glucuronic acid 2-sulfate; PBS, phosphate-buffered saline.

Growth factors known to bind heparin

- Basic fibroblast growth factor (bFGF)

- Acidic fibroblast growth factor (aFGF)

- Heparin binding epithelial growth factor (HB-EGF)

- Recombinant human Glial Growth Factor 2 (rhGGF2)

Provided below are descriptions of experiments disclosing how to identify useful oligodisaccharide sequences having specificity for neuregulins (Examples 1, 2, 5 and 6), and how to test the effectiveness of identified Glyceptor sequences and their analogs for their effectiveness in blocking neuregulin-Glyceptor binding (Example 3) and on cultured cells to evaluate activity in vitro (Examples 4 and 7).

Example 1 Heparin and HS bind rhGGF2 with high affinity - demonstration by affinity co-electrophoresis.

To measure the binding affinity of peptide growth factors for various GAGs, horizontal agarose gel electrophoresis of [ ¹²⁵ I]-GAG chains was performed as described by Witt and Lander (Witt, et al., (1994) Curr Biol 4:394-400).

Heparin from porcine intestinal mucosa (Sigma) was derivatized with tyramine and radiolabeled with ¹²⁵ I (San Antonio, et al., ( 1993) Biochem 32:4746-4755). This material was desalted on a PD-10 column and subjected to gel filtration on a Sephadex G-50 column. Samples of approximate molecular weight 4000 were used in this study. HS lόmer (HSI 6) was purified from porcine intestinal mucosa. HSI 6 was labeled with * ²⁵ I and subjected to gel filtration on a Sephadex G-50 column (Lee, et al., (1991) Proc Natl Acad Sci USA 88:2768-2772).

Varying concentrations of rhGGF2 were incoφorated into individual lanes in a 1% agarose gel, and [ ¹²⁵ I]-heparin or [ ¹²⁵ I]-HS (-3000 cpm per sample) were electrophoresed through the gel in a buffer system of 0.1 M sodium acetate, 50 mM MOPSO, 0.5% CHAPS, pH 7.0. Immediately following electrophoresis, gels were dried with forced warm air and the distribution of radioactivity visualized using a phosphorimager (Fuji BAS 1000). Electrophoresis of the negatively charged oligosaccharides through the protein-containing wells results in an alteration of mobility of the GAG chains that reflects the equilibrium binding of the protein to the GAG. Retardation coefficients (R) can be calculated for each concentration of protein. The apparent dissociation constants are equated to the protein concentration at which the GAG is half maximally retarded using the formula R = R./(l+K _d /[protein]) (Lee, et al., (1991) Proc Natl Acad Sci USA 88:2768-2772).

This technique allowed the determination of apparent dissociation constants by observing the retardation in electrophoretic mobility of [ ¹²⁵ I]-heparin (Figure 5 A, inset) and [ ^I25 I]-HS (Figure 5B, inset) by a range of rhGGF2 concentrations embedded within agarose gels. The dissociation constants were calculated for heparin (Figure 5A) and HS (Figure 5B) by determining the concentration of rhGGF2 that produced a half-maximal mobility shift of the labeled GAGs (Lee, et al. (1991 ) Proc Natl Acad Sci USA 88:2768-2772). The values observed were K _d = 9.7 nM for heparin, and Kd = 22 nM for HS.

Example 2 Heparin binds rhGGF2 with high affinity - demonstration by solution phase binding and trapping of heparin/rhGGFl complexes on nitrocellulose.

Another method for quantifying the binding interactions between proteins and GAGs is to allow such binding to occur in solution, and then trap the resultant protein-GAG complexes on nitrocellulose filters by suction filtration. Filter binding was used as an assay for compounds that block rhGGF2 binding to a radiolabeled heparin fragment. Equilibrium binding was achieved in PBS with the rhGGF2 present at varying concentrations, and the concentration of the [ ^I25 I]-heparin fragment maintained below the YL_. Once equilibrium was reached (1.5 h), rhGGF2-GAG complexes were captured by suction filtration onto a 0.45 μM nitrocellulose filter. The filter was then dried at 42°C for 30 minutes and evaluated using a matrix array scintillation detector (Wallac 1205 Betaplate). Only protein-GAG complexes are retained on the filter, and thus the amount of heparin retained on the filter was plotted. As can be seen in Figure 6. [ ^{l 25} I]-heparin retention on the filter is dependent on the concentration of added rhGGF2. If the starting concentration of GAG chain is below the K _d for rhGGF2 binding, then the apparent K will be the rhGGF2 concentration at which half the input GAG chains are retained on the filter. In this case, that concentration was 5.5 nM, a value comparable with the K determined in Example 1.

Example 3 Heparin-rhGGF2 binding is inhibited by synthetic antagonists, but not by other highly sulfated GAGs.

The filter binding assay is also a useful tool for comparing quantitatively the ability of various compounds to competitively inhibit the binding of rhGGF2 to labeled heparin:

Filter binding was used as an assay for compounds that block rhGGF2 binding to a radiolabeled heparin fragment. Equilibrium binding was achieved in PBS with the rhGGF2 concentration maintained at 40 nM (just above the Kd) for each reaction, and the concentration of the [ ^{1 5} I]-heparin fragment maintained below the K _< j. Test compounds were added to the assay solution at 10 μM and were serially diluted 2-fold for a total of 9 concentrations per compound. Once equilibrium was reached (1.5 h), rhGGF2-GAG complexes were captured by suction filtration onto a 0.45 μM nitrocellulose filter and analyzed as described in Example 2; the amount of heparin retained on the filter in the absence of competitor was plotted as 100%. Disruption of protein [ ^I25 I]-heparin complexes by competitor results in decreased radioactivity retained on the filter; the point at which 50% of the counts are retained was calculated to be the IC ₅₀ for the drug.

Figure 7 shows the dose-dependent inhibition of rhGGF2 binding to heparin by suramin and two related polyanion compounds, GL12 and NF066. The competition curves were used to calculate the IC ₅₀ of each of these compounds (Table 1).

Growth Factor Tested rhGGF2 bFGF GL12 0.9 mg/ml 3.8 mg/ml

Suramin 2.2 mg/ml 21.6 mg/ml

NF066 2.0 mg/ml 25.0 mg/ml

Table 1. Comparison of IC ₅₀ values of polyanion compounds on rhGGF2 and bFGF binding to heparin. K _d of binding to heparin is 5.5 nM for rhGGF2, and 5.0 nM for bFGF. IC ₅₀ values were determined using the filter binding assay. IC ₅₀ values were determined by calculating the dose necessary to achieve half-maximal inhibition in Figure 5B.

Note that the rank order of inhibitory activity of these compounds, GL12 » suramin > NF066. is the same for both rhGGF2 and for bFGF. The filter binding assay also demonstrates that the same GAGs do not compete with rhGGF2-heparin binding at the doses tested. This lack of competition by a variety of highly sulfated GAGs demonstrates the specificity of the interaction between rhGGF2 and heparin-type sulfated polymers (Table 2).

Glycosaminoglycan ICsn to displace rhGGF2 from heparin Heparin .06 mg/ml Chondroitin sulfate 30.0 mg/ml

Dermatan sulfate > 12.5 mg/ml

Keratan sulfate > 25.0 mg ml

Table 2. Comparison of IC ₅₀ values of GAGs on rhGGF2 binding to heparin. K of binding to heparin is 5.5 nM for rhGGF2. IC ₅₀ values were determined using the filter binding assay, and were determined by calculating the dose necessary to achieve half- maximal inhibition as in Table 1.

Example 4 Synthetic antagonists of heparin-GGF binding inhibit Schwann cell responsiveness to rhGGF2.

The data concerning suramin and its analogs (Example 3) present an additional pharmacological means of directly perturbing rhGGF2 interactions with Schwann cell surface HeSPGs:

Sciatic nerve Schwann cells from 3 day old rats were purified by the methods of Brockes (Brockes (1987) Meth Enzymol 147:217-225). Cells were plated on tissue culture plastic precoated with poly-D-lysine (PDL, Collaborative Research) in low glucose Dulbecco's Modified Eagel's Medium (DMEM, Fisher/Mediatech) supplemented with 10% heat-inactivated fetal bovine serum (Hyclone). This medium is referred to as DMEM/10. After 24 hours the medium was replaced with fresh DMEM/10 containing 10 mM cytosine arabinoside (Sigma). Two to 3 days later, the medium was removed and replaced with DMEM/10 supplemented with partially purified bovine GGF (the carboxymethylcellulose fraction of bovine pituitary extract, GGF-CM (Goodearl, et al., (1993) J Biol Chem 268:18095-18102)) and 5 mM forskolin (Calbiochem). When the cells reached confluence, they were trypsmized and treated with anti-Thy 1 (Tl lD7e, Serotec), and rabbit complement (Gibco) to remove contaminating fibroblasts. Cells were then plated in DMEM/10 supplemented with GGF-CM and forskolin. Upon further expansion, cell stocks were used for assay puφoses detailed below or were frozen for future use.

The DNA synthesis assay was performed according to the method of Brockes (Brockes, (1987) Meth Enzvmol 147:217-225) with slight modifications. Schwann cells were prepared for assay by growing to confluence on PDL coated plastic in the presence of GGF-CM and 5 mM forskolin. The cells were then withdrawn from growth factor and forskolin for 3 days, trypsinized, and plated in 96-well plates at a concentration of 10,000 cells/100 ml in DMEM supplemented with 5% serum (DMEM 5). The next day test samples were added to the wells and 24 hours later either [ ¹²⁵ I]-UdR. or [methyl- ³ H]thymidine was added. The cells were harvested approximately 18 hours later using a To tec Harvestor and the samples counted using a Wallac 1450 MicroBeta liquid scintillation counter.

Figure 8A demonstrates that not only do GL12, suramin, and NF066 inhibit rhGGF2-induced DNA synthesis in a dose-dependent manner, but the rank order of their potency matches that of their potency in inhibiting rhGGF2 binding to heparin in the cell- free assay (Figure 7 and Table 1 , above). When cultured in the presence of 60 ng/ml

rhGGF2. half maximal inhibition of Schwann cell DNA synthesis by GL 12 is - 5 mM, suramin is -20 mM, and NF066 is -35 mM.

The effects of suramin and its analogs are reversible, and not due to non-specific cytotoxicity. This was demonstrated by "washout" experiments where Schwann cells were maintained in normal medium. 100 mM suramin, or 100 mM GL12 for 24 hours. After washing with fresh control medium, cultures were transferred to one of four experimental media with labeled nudeoside (control medium, 60 ng/ml rhGGF2. 100 mM test compound, or rhGGF2 plus test compound). Total DNA synthesis was assayed 48 hours later as in the previous experiments. As can be seen in Figure 8B, cells maintained in test compounds for the first 24 hours and then transferred to control medium (washout conditions) displayed levels of DNA synthesis identical to those of cells maintained in control medium or test compound through the entire culture period. No sign of overt cytotoxicity was observed by light microscopy in any of the above conditions. Cells transferred from control medium to medium containing rhGGF2 showed expected increases in DNA synthesis. Schwann cells transferred from test compound medium to compound-free medium containing rhGGF2 displayed DNA synthesis levels comparable to rhGGF2-treated cells that had not been pretreated with test compounds. Thus the inhibition of rhGGF2-induced DNA synthesis by suramin and its analogs appears to occur through the direct inhibition of rhGGF2 binding to cell surface HeSPGs.

Example 5 Combinatorial chemistry library construction.

Combinatorial library construction exploited the significant advantage of isonitriles in that they can be used to carry out numerous single step, high yield solution phase reactions (Gokel. G.. et al.. in Isonitrile Chemistry. (Ed. Ugi, I.), Academic Press, New York, 40). A combinatorial isonitrile (Ugi) chemistry technology was developed which produced organic compounds in a 96-well format. By not relying on solid phase chemistry, it was feasible to synthesize many compounds in parallel with one (or a small number) of compounds per well. This avoided artifacts which can arise from assaying complex mixtures of weakly active compounds (non specific binders) and simplifies structural identification of "hits". We investigated the reaction of a primary amine, an aldehyde or ketone, a carboxylic acid and an isonitrile to form an acylaminoacid amide as shown in Figure 9.

This type of compound was synthesized by mixing equimolar amounts of the four reactants in a polar solvent. For the libraries described below R2 is not H and thus each reaction produces two isomeric compounds. For simple alkyl substituted starting

materials, yields in excess of 90% have been reported (Lee. et al , (1991) PNAS USA 88_.2768-2772) ι _n addition, there are thousands of commercially available amines, aldehydes and carboxylic acids which enable sets of compounds to be synthesized with virtually any type of functionality

Validation of Starting Materials Prior to using the isonitrile chemistry, the starting matenals were determined to have had the solubility and reactivity properties that are consistent with high yield product formation Typically, each potential reactant was substituted for the corresponding amine, aldehyde or carboxylic acid from the ' model reaction" descnbed above Results from reverse phase HPLC analyses showed that this chemistry is extremely robust in that excellent yields of product are obtained from aromatic as well as aliphatic amines, ketones as well as aldehydes and a range of sterically hindered carboxylic acids. From this type of experiment, a set of 20 amines, 16 aldehydes/ketones and 24 carboxylic acids were qualified which consistently produced high yields. In addition, a series of isonitriles were evaluated and n-butyl, cyclohexyl, benzyl and methyl acetate all gave excellent results The functional groups represented in these four sets of reactants are shown in Figure 10 (the arrow on each structure indicates the point of attachment to the ammo acid scaffold).

The sets of reactants were chosen to have minimal overlap in structure Thus the amme substituents are different and complementary to the aldehyde/ketone, or carboxylic acid substituents The result was that the current set of reactants contained 58 different pharmacophores. The functionality was chosen to- 1 ) mimic the side chains of amino acids, 2) to contain a variety of heterocyclic ring systems as is found in many existing protein-binding drugs, and 3) to be rich hydrogen bond forming functional groups such as phenols, alcohols, esters, ethers, etc The later feature is based on the assumption that hydrogen bond forming functionality is appropnate for a drug which will bind a carbohydrate binding pocket on the growth factor One additional desired feature for the library is anionic functional groups Since carboxylic acids are inherently incompatible with this chemistry, methyl and ethyl esters which can subsequently be hydrolyzed to the corresponding carboxylic acids (see below) were chosen to be introduced into the library

Pilot combinatorial hbrarv construction In order to confirm that the selected reactants could be used to create libraries, a series of experiments was carried out in which compounds were synthesized in a 96 well format, where all wells contain 3- furoic acid and N-butyl isonitrile and the rows and columns contain a specified aldehyde

and amine. as indicated in Figure 1 1 The experimental details were as follows A 2M stock solution of each reactant was prepared in the appropriate solvent An equal amount (50 ml) of the four appropriate reactants was added to 80 wells of a 96 well polypropylene plate in which the well volume was 1 ml The reactions in each column contained a specified amme and the indicated aldehyde was added to each row. A single carboxylic acid. 3-furoιc acid and n-butyl isonitrile was added to all 80 wells Columns 1 1 and 12 are reserved for controls which are added prior to screening. After 20 hours, the reactions were diluted by addition of 200 ml of DMSO and 600 ml of methanol, a sequence which was empirically determined to maximize product recovery in those wells in which the product precipitated Assuming a 50% yield, product concentration in these plates is 50 mM The products are stable and do not precipitate when stored at -20° C An aliquot was then taken from each well and analyzed by HPLC. In each chromatogram the assignment of product was confirmed by an algorithm which enables the retention time of the product to be predicted within 2 minutes from the incremental contributions of the reactant side chains The yields for each reaction are shown in the corresponding wells. A yield of >50% was observed in 70/80 (88%) of the reactions and >60% m 63/80 (79%) of the reactions. In several wells (G3. H3) the yield was low due to selective precipitation of the product. In other wells (E3, E6) the principle side- product was the Schiff base formed from the am e and aldehyde in the first step of the reaction sequence. Thus, the principal impurities are starting mateπals or Schiff base. Eleven additional plates were synthesized by this protocol in which the different amines, aldehydes/ketones and carboxylic acid were evaluated For these plates samples were taken from ten wells per plate for analysis by HPLC Of the 1 10 reactions analyzed, 96 (87%) showed product yield of >50%.

A protocol also was developed to introduce carboxylates by hydrolyzing the ester-containing compounds in a 96 well plate format Ester containing wells were diluted 5-fold with methanol to a product concentration of approximately 10 mM A 50 ul aliquot of the compound solution was combined with 10 ml of IM K2HPO4 (pH=13.3) to increase the pH to >12 The hydrolysis proceeded overnight at room temperature and the solution was neutralized by addition of 10 ml of 1 M KH2PO4 (pH=2.2) These conditions were empirically determined to provide quantitative hydrolysis without precipitation of phosphate or the ester

Example 6 Pilot Library Synthesis/Screening

Since all of the components and operations of our strategy for combinatorial synthesis appeared to be in working order, a combined synthesis/screening technology was implemented in a moderate scale (several thousand compounds) compound screen. The combinatorial synthesis technology was applied to the construction of two libraries of compounds with chemical functionality biased towards GAG-like structures. In the first library, each compound contains a carboxylic acid moiety which was introduced in the Ugi reaction as an ester which was subsequently hydrolyzed to the corresponding acid. The second library has been constructed with reactants containing sulfonate groups to mimic the sulfonate groups found in GAGs and the sulfonate groups found in suramin, GL-12, and NF-066. The detailed composition of the two libraries is shown in Table 1 below:

Table 1

Carboxylate Library

Plate # Ri R2 R3 R4 # Compounds

8104,8105,8113 24 20 -CK 480

8106.8107,8114 24 20 480

8110-8111 20 16 320 8112 8 2 32 128

Sulfonate Library

Plate # Rl R2 R3 R4 # Compounds

indicates connection to root structure

Following the synthesis of a set of 2544 carboxylates and 1280 sulfonate derivatives we employed the rhGGF2 binding assay adapted to a 96 well format to identify an initial set of hits. Briefly, a 20 nM solution of rhGGF2 was incubated for 60 minutes at room temperature with the test compound(s) and a trace amount of a high affinity GAG fragment which has been tyraminated at the reducing end and iodinated. Assays for bFGF and VEGF were run under similar conditions except that the protein concentrations are set at 7.5 nM and 15 nM, respectively. The screening assays were carried out in 96 well Hybridot Manifold (BRL) which allowed rapid filtration through a nitrocellulose sheet which then contains 96 radioactive "spots". Radioactivity on the filter was quantitated on a up to 10 sheets of nitrocellulose per run. This approach has proven to be very efficient and has been routinely used to screen up to 12 microtiter plates (960 wells)/week. Since each plate contained 80 reaction products, 16 wells were available for positive and negative controls which included purified samples of the relevant amines, aldehydes, carboxylic acids and isonitrile reactants. Other controls included wells with no compound added and wells that contained excess "cold" heparin.

The two libraries were analyzed first in a single point survey. Positive hits fell into two groups: 1 ) wells that block >50% of binding and 2) those that block 20-50% of binding. Wells from the second group are rescreened to determine if the blocking was reproducible, and if so, they were selected along with those wells in the first group for complete binding inhibition curves to determine compound IC50 values. Compounds with IC50 < 30 uM were selected for further analysis. These selected compounds were tested for color or chemical quenching and some were eliminated. A few of were purified by HPLC and were re-analyzed. The current status of these screens are given in Table 2. In Table 3, the IC50 data is reported for the eight most advanced compounds in the screen.

These efforts have resulted in the identification of 17 preliminary lead compounds. 2 carboxylates and 15 sulfonates, which are at various stages of analysis. The composite hit rate of 0.45% is inteφreted to mean that the general design of our approach is working as planned and it bodes well for additional screening efforts.

Table 2. rhGGF2 LIBRARY SCREENING

Library Initial survev ³ First rescreenb Further screening ⁰

2544 carboxylates 133 @ 20-50% (5.2%) 8 selected 41 @ >50% (1.6%) 10 selected

Total hits 174 18 selected 2 selected for

(0.7% of 2544) product purification--* (0.07% of 2544)

1280 sulfonates 90 @ 20-50% (7.0%) 7 selected 29 @ >50% (2.3%) 20 selected

Total hits 119 27 selected 15 selected for

(2.1% of 1280) product purification

(1.17% of 1280) compound concentration between 30-40 μM; those compounds that are 20-50% inhibitors are rescreened as single points: those compounds that are >50% inhibitors are rescreened with a complete binding inhibition curve to determine compound IC50 single point rescreens that repeat as 20-50% inhibitors are selected tor complete binding inhibition curves to determine compound IC50; compounds with IC50 < 30μM are selected for further analysis

IC50 is repeated 2-3 times to determine reproducibility; compounds are tested for color or chemical quenching of signal; compounds are eliminated for non reproducibility and signal quenching puπfied compounds will be analyzed further biochemically including activity confirmation (IC50 values; confirmation of structure (mass spectroscopy), and; specificity binding assays against other growth factors

Table 3. IC50 ^a Summary of 8 Selected Compounds

Library Compound rhGGF2 rhGGF2 bFGF bFGF unpurified purified unpurified purified

Carboxylate 8108 B3 15 3 50 3.5

8200hE3 12 20 15 >100

Sulfonate 8301F10 40 63 >100

8303B8 15 14 1 1 18

8303E9 25 27 34 >100

8304B9 35 38 38 >100

8401F10 40 85

8404H3 15 45 a Values are in uM

Example 7 Effect of identified combinatorial compounds in the schwann cell proliferation assay

Three compounds that were identified in the binding assay were resynthesized on a larger scale, then purified by HPLC and tested for inhibition of Schwann cell proliferation. The samples were provided in 100% methanol and were diluted into media at least 100-fold at the highest concentration used in the bioassay. These compounds were analyzed for biological activity in the Schwann cell proliferation assay using the method described by Sudhalter, et al.. (Sudhalter, et al., (1996) GHa 17:28-38). Figure 13 shows the results of this assay.

The compounds and controls were diluted into culture media and mixed with rhGGF2 (60 ng/ml final concentration) to achieve the final concentrations ranging from 0.195uM to 200uM. After incubating at room temperature for 90 minutes the mixtures were used to replace the media on the cells, and then the assay was performed according to the standard protocol. Stocks of individual compounds were in 200 mM methanol (except ROY20.2 and ROY20.3, which were at 20 mM).

Compound ROY20.1 was 100% methanol, the solvent control. When it was present at up to 0.1 % in the assay it did not interfere with the mitogenic response of Schwann cells to rhGGF2. Compound ROY22.2 was suramin, the positive control, which shows an IC50 of roughly 50 uM in this assay. Compound ROY22.3 appeared to be the most potent in the batch, but also was quite toxic to the cells. Compounds ROY22.4-6 represent the HPLC- purified compounds that were positive in the binding assay. Inhibition of rhGGF2- stimulated Schwann cell proliferation was seen in the 100 uM range.