This invention relates to a method of identifying a compound useful in modulating Notch signalling. In particular, it relates to a method of identifying an antibody or a phospholipid mimic useful in modulating Notch signalling. It also relates to novel agents, including novel antibodies or phospholipid mimics, useful in modulating Notch signalling for use in medicine, and the treatment of cancer in particular.

The Notch pathway is a universally conserved, core signal transduction system in metazoan organisms playing key roles in cell fate determination, cell proliferation and apoptosis during development with a crucial impact on most tissues and organs. In adults Notch has key roles in tissue homeostasis by regulating stem cell maintenance and function, immune system activation and angiogenesis. Dysregulation of the pathway results in a number of inherited and acquired disorders, including various cancers, and it is therefore a key target for therapeutic intervention. Despite the importance of the pathway, little is known about the structural basis of Notch receptor/ligand interactions and the molecular mechanisms that transduce this recognition event into signal activation. Notch ligands fall into two structurally related families: Jagged/Serrate and Delta-like. Notch signalling requires cell surface expression of a heterodimeric trans-membrane Notch receptor. Ligand binding by one of the Jagged/Serrate or Delta families, is followed by ligand-endocytosis, coupled with transendocytotic removal of the bound Notch extracellular domain, this exposes the remaining membrane-tethered Notch moiety to further proteolytic cleavages. The final intramembrane cleavage, which may require an endocytic step, releases the intracellular domain of Notch. The intracellular domain of Notch then translocates to the nucleus, binds to a transcription factor of the CBF 1 , Suppressor of Hairless, Lag- 1 (CSL) family and relieves repression of genes of the HES and Hey families. These transcriptional repressors in turn inhibit expression of genes that drive cells to adopt a differentiated state. Interactions with the Notch receptor can activate or inhibit Notch signalling, dependent upon whether ligands are presented to Notch on adjacent cells (in trans), or on the same cell (in cis). Notch ligand activity is also sensitive to the modification of O-fucosylated Notch by Fringe . This can potentiate or prevent signalling by different ligands by mechanisms that are poorly characterised but are important in controlling embryonic patterning and boundary formation between adjacent developmental compartments.

It is an aim of the present invention to provide modulators of Notch signalling and methods of identifying modulators of Notch signalling.

According to a first aspect the invention provides a method for identifying a compound useful in modulating the Notch signalling pathway, the method comprising the step of determining whether the compound binds to the N-terminal C2 domain of a Notch ligand.

The method may further comprise the step of determining whether binding of the compound to the C2 domain of a Notch ligand modulates Notch signalling. The invention may also provide a method for identifying a compound useful in modulating the Notch signalling pathway, the method comprising the step of determining whether the compound binds to Notch and reduces binding of a C2 domain to Notch. The method may further comprise the step of determining whether binding of the compound to Notch modulates Notch signalling.

The compound useful in modulating the Notch signalling pathway may be a phospholipid, a glycolipid,a sphingolipid, a small molecule or an antibody.

The compound may increase Notch signalling. The compound may decrease Notch signalling.

The Notch ligand may be a protein from the Jagged, human Delta-like- 1 or Drosophila Serrate protein family. The Notch ligand may, for example be mammalian or human Jagged- 1 (JAG1), Jagged-2 (JAG2), Delta-like 1 (DLL 1), Delta-like 2 (DLL2), or Delta-like 3 (DLL3). Table 1- Suggested Notch ligands from mammals with their equivalents from two model organisms C. elegans and D. melanogaster

The Notch ligand may be a protein from the Jagged, Delta-like or serrate protein family. The protein ligand may be one of the ligands in table 2.

In the method of the invention the N-terminal C2 domain of the Notch ligand may have a sequence having: at least 70, 75, 80, 85, 90, 95, 98, 99% or more sequence identity to SEQ ID NO. 1 ; at least 70, 75, 80, 85, 90, 95, 98, 99% or more sequence identity to SEQ ID NO. 2; at least 70, 75, 80, 85, 90, 95, 98, 99% or more sequence identity to SEQ ID NO. 3 ; at least 70, 75, 80, 85, 90, 95, 98, 99% or more sequence identity to SEQ ID NO . 4; or at least 70, 75 , 80, 85, 90, 95, 98, 99% or more sequence identity to SEQ ID NO. 5.

Amino acid identity may be calculated using any suitable algorithm. For example the UWGCG Package provides the BESTFIT program which can be used to calculate homology (for example used on its default settings) (Devereux et al ( 1984) Nucleic Acids Research 12, 387-395). The PILEUP and BLAST algorithms can be used to calculate homology or line up sequences (such as identifying equivalent or corresponding sequences (typically on their default settings), for example as described in Altschul S. F. ( 1993) J Mol Evol 36:290-300; Altschul, S, F et al ( 1990) J Mol Biol 215 :403- 10. Software for performing BLAST analyses is publicly available through the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/). This algorithm involves first identifying high scoring sequence pair (HSPs) by identifying short words of length W in the query sequence that either match or satisfy some positive- valued threshold score T when aligned with a word of the same length in a database sequence . T is referred to as the neighbourhood word score threshold (Altschul et al, supra). These initial neighbourhood word hits act as seeds for initiating searches to find HSPs containing them. The word hits are extended in both directions along each sequence for as far as the cumulative alignment score can be increased. Extensions for the word hits in each direction are halted when: the cumulative alignment score falls off by the quantity X from its maximum achieved value; the cumulative score goes to zero or below, due to the accumulation of one or more negative-scoring residue alignments; or the end of either sequence is reached. The BLAST algorithm parameters W, T and X determine the sensitivity and speed of the alignment. The BLAST program uses as defaults a word length (W) of 1 1 , the BLOSUM62 scoring matrix (see Henikoff and Henikoff ( 1992) Proc. Natl. Acad. ScL USA 89: 10915 - 10919) alignments (B) of 50, expectation (E) of 10, M=5, N=4, and a comparison of both strands.

The BLAST algorithm performs a statistical analysis of the similarity between two sequences; see e.g., Karlin and Altschul ( 1993) Proc. Natl. Acad. Sci. USA 90: 5873 - 5787. One measure of similarity provided by the BLAST algorithm is the smallest sum probability (P(N)), which provides an indication of the probability by which a match between two polynucleotide or amino acid sequences would occur by chance . For example, a sequence is considered similar to another sequence if the smallest sum probability in comparison of the first sequence to the second sequence is less than about 1 , preferably less than about 0. 1 , more preferably less than about 0.01 , and most preferably less than about 0.001.

The variant sequences typically differ by at least 1 , 2, 3, 5, 10, 20, 30, 50 or more mutations (which may be substitutions, deletions or insertions of amino acids). For example, from 1 to 50, 2 to 40, 3 to 30 or 5 to 20 amino acid substitutions, deletions or insertions may be made. The substitutions are preferably conservative substitutions, for example according to the table below. Amino acids in the same block in the second column and preferably in the same line in the third column may be substituted for each other.

Aliphatic Non-Polar G A P

I L V

Polar - uncharged C S T M

N Q Polar - charged D E

K R

Aromatic H F W Y

The polypeptides of the invention may also be provided as a fusion protein comprising a polypetide of the invention genetically or chemically fused to another peptide. The purpose of the other peptide may be to aid detection, expression, separation or purification of the protein. Alternatively the protein may be fused to a peptide such as an Fc peptide to increase the circulating half life of the protein. Examples of other fusion partners include beta-galactosidase, glutathione-S-transferase, or luciferase.

The polypeptides of the invention may be chemically modified, e.g. post- translationally modified. For example, they may be glycosylated, pegylated, phosphorylated or comprise modified amino acid residues. They may be modified by the addition of histidine residues to assist their purification or by the addition of a signal sequence to promote insertion into the cell membrane . Such modified polypeptides fall within the scope of the term "polypeptide" used herein.

Polypeptides of the invention may be in a substantially isolated form. It will be understood that the polypeptide may be mixed with carriers or diluents which will not interfere with the intended purpose of the polypeptide and the polypeptide will still be regarded as substantially isolated. A polypeptide for use in the invention may also be in a substantially purified form, in which case it will generally comprise the polypeptide in a preparation in which more than 50%, e .g. more than 60%, 70%, 80%, 90%, 95% or 99%, by weight of the polypeptide in the preparation is a polypeptide of the invention. Polypeptides of the invention may be made synthetically or be recombinantly produced. For example, a recombinant polypeptide may be produced by transfecting mammalian, fungal, bacterial or insect cells in culture with an expression vector comprising a nucleotide sequence encoding the polypeptide operably linked to suitable control sequences, culturing the cells, extracting and purifying the polypeptide of the invention produced by the cells. The amino acid sequence of polypeptides for use in the invention may be modified to include non-naturally occurring amino acids or to increase the stability of the compound. When the polypeptides are produced by synthetic means, such amino acids may be introduced during production. The polypeptides may also be modified following either synthetic or recombinant production. Polypeptides of the invention may also be produced using D-amino acids A number of side chain modifications are known in the art and may be made to the side chains of the polypeptides of the invention providing the activity of the polypeptide is retained. In the method of the invention the N-terminal C2 domain of the Notch ligand may have a sequence having: at least 70, 75, 80, 85, 90, 95, 98, 99% or more sequence identity to the mature peptide amino acids S32 to K198 of SEQ ID NO. 1 ;

at least 70, 75, 80, 85, 90, 95, 98, 99% or more sequence identity to the mature peptide amino acids S22 to V 190 of SEQ ID NO. 2;

at least 70, 75, 80, 85, 90, 95, 98, 99% or more sequence identity to the mature peptide amino acids A27 to R189 of SEQ ID NO. 3 ;

at least 70, 75, 80, 85, 90, 95, 98, 99% or more sequence identity to the mature peptide amino acids S27 to R186 of SEQ ID NO. 4;

at least 70, 75, 80, 85, 90, 95, 98, 99% or more sequence identity to the mature peptide amino acids M27 to K209 of SEQ ID NO. 5.

With reference to SEQ ID NOs 1 to 5, Figures 10 and 1 1 include the full sequence of the C2 domains of the above ligands. These are shown including the signal sequence for each protein. The signal sequence of each protein is shown in italics and is cleaved off to provide the full-length protein.

Table 2-details the percentage sequence identity in the C2 domain (without the signal sequence) of Notch ligands JAG1, DLL1, DLL3, DLL4 and JAG 2 (calculated using ClustalW2).

JAG1 DLL 1 DLL3 DLL4 JAG2

JAG1 100% 3 1 % 15% 3 1 % 50%

DLL 1 100% 28% 46% 34%

DLL3 100% 29% 17%

DLL4 100% 3 1 %

Loop one and loop three are highlighted in figure 10 for each of JAGl , JAG2, DLL l , DLL3 and DLL4. These loops have been shown to be important in binding to phospholipids. The compound identified by the method of the present invention may bind to loop 1 and/or loop 3.

The compound may be a phospholipid or phospholipid mimic that binds to loop 1 of JAGl , JAG2, DLL l , DLL3 and/or DLL4. The compound may be a phospholipid or phospholipid mimic that binds to loop 3 of JAGl , JAG2, DLL l , DLL3 and/or DLL4.

The compound may be a glycolipid or glycolipid mimic that binds to loop 1 of JAG l , JAG2, DLL l , DLL3 and/or DLL4. The compound may be a glycolipid or glycolipid mimic that binds to loop 3 of JAGl , JAG2, DLL l , DLL3 and/or DLL4. The compound may be a sphingolipid or sphingolipid mimic that binds to loop 1 of JAGl , JAG2, DLL l , DLL3 and/or DLL4. The compound may be a sphingolipid or sphingolipid mimic that binds to loop 3 of JAGl , JAG2, DLL l , DLL3 and/or DLL4.

The compound may be an antibody that binds to loop 1 of JAGl , JAG2, DLL l , DLL3 and/or DLL4. The compound may be an antibody that binds to loop 3 of JAG l , JAG2, DLL l , DLL3 and/or DLL4.

Whether a molecule binds to the C2 domain may be determined using any suitable technique, for example ELISA or a biophysical technique, for example isothermal titration calorimetry, surface plasmon resonance, thermofluor melting profiles, nuclear magnetic resonance or chemical shift mapping.

Whether a molecule binds to the C2 domain and also increases or decreases Notch signalling may be measured by any suitable technique, for example competitive inhibition/potentiation of cellular activation or signalling assays.

The compound identified by the method of the invention may be an antibody that binds to the N-terminal C2 domain. The compound identified by the method of the invention may be a phospholipid mimic. A phospholipid mimic may be a small molecule compound with a similar structure to a natural phospholipid that binds to the C2 domain and modulates Notch activity. The phospholipid mimic may prevent or reduce binding of natural phospholipids to the C2 domain.

The compound identified by the method of the invention may be a sphingolipid mimic. A sphingolipid mimic may be a small molecule compound with a similar structure to a natural sphingolipid that binds to the C2 domain and modulates Notch activity. The sphingolipid mimic may prevent or reduce binding of natural sphingolipids to the C2 domain.

The compound identified by the method of the invention may be a glycolipid mimic. A glycolipid mimic may be a small molecule compound with a similar structure to a natural glycolipid that binds to the C2 domain and modulates Notch activity. The glycolipid mimic may prevent or reduce binding of natural glycolipids to the C2 domain.

The compound identified by the method of the invention may not directly block the binding of the Notch ligand to Notch but may alter the activity of the Notch ligand by binding to the C2 domain.

The compound identified by the method of the invention may bind to Notch and block or reduce the binding of the C2 domain to Notch.

The compound identified by the method of the invention may bind to the C2 domain and block or reduce binding of the C2 domain to Notch.

The compound of the present invention may competitively inhibit the binding of a C2 domain to Notch.

Any compound identified by the method of the present invention may be formulated into a pharmaceutical composition with pharmaceutically acceptable diluents, excipients or carriers. In another aspect the present invention provides a compound identified by a method of the present invention.

In another aspect the present invention provides an agent, such as a small molecule, an isolated antibody a phospholipid mimic, a glycolipid mimic or a sphingolipid mimic, that specifically binds to the N-terminal C2 domain of a Notch ligand.

In another aspect the present invention provides an agent, such as a small molecule, an isolated antibody a phospholipid mimic, a glycolipid mimic or a sphingolipid mimic, that specifically binds to Notch and blocks or reduces binding of Notch to the C2 domain of a Notch ligand.

Binding of the agent, antibody or phospholipid mimic, to the C2 domain of a Notch ligand may increase Notch signalling activity in a cell. Binding of the antibody to the C2 domain of a Notch ligand may decrease Notch signalling activity in a cell. The C2 domain may be the C2 domain of a Jagged protein, in particular human Jagged- 1.

The agent, isolated antibody or phospholipid mimic of the present invention may be formulated in a pharmaceutical composition with one or more pharmaceutically acceptable diluents, excipients or carriers.

In another aspect, the present invention provides a method of modulating Notch signalling in a cell, comprising contacting the cell with an effective amount of an agent, an isolated antibody or a phospholipid mimic according to the present invention.

Modulating Notch signalling may be increasing Notch signalling in a cell or decreasing Notch signalling in a cell. According to a further aspect the invention provides an agent, a small molecule an isolated antibody a phospholipid mimic, glycolipid mimic or a phospholipid mimic of the present invention for use in medicine.

The present invention provides an antibody or an antibody fragment comprising a polypeptide sequence having at least 70%, at least 80%, at least 90%, at least 98% or having 100% sequence identity to SEQ ID no 18 and/or SEQ ID NO 21 for use in medicine .

The present invention provides an antibody or antibody fragment encoded by a nucleic acid comprising a sequence having at least 70%, at least 80%, at least 90%, at least 98% or having 100% sequence identity to SEQ ID no 16 and/or SEQ ID NO 20 for use in medicine.

According to a further aspect the invention provides an agent, a small molecule an isolated antibody a phospholipid mimic, glycolipid mimic or a phospholipid mimic of the present invention for use in stem cell derivation.

According to a further aspect the invention provides an agent, a small molecule an isolated antibody a phospholipid mimic, glycolipid mimic or a phospholipid mimic of the present invention for use in the treatment or prevention of immune disorders.

According to a further aspect the invention provides an agent, a small molecule an isolated antibody a phospholipid mimic, glycolipid mimic or a phospholipid mimic of the present invention for use in the treatment or prevention of cancer. The agent, isolated antibody or phospholipid mimic may be for use in the treatment of a condition or disorder caused by dysfunction of Notch signalling.

The agent, antibody or phospholipid mimic may be for use in the treatment of cancer. Cancer refers to the physiological condition in mammals in which a population of cells are characterized by unregulated cell growth. Examples of cancer include, but are not limited to, carcinoma, lymphoma, blastoma, sarcoma, and leukemia. More particular examples of such cancers include squamous cell cancer, small-cell lung cancer, non-small cell lung cancer, adenocarcinoma of the lung, squamous carcinoma of the lung, cancer of the peritoneum, hepatocellular cancer, gastrointestinal cancer, pancreatic cancer, glioblastoma, cervical cancer, ovarian cancer, liver cancer, bladder cancer, hepatoma, breast cancer, colon cancer, colorectal cancer, endometrial or uterine carcinoma, salivary gland carcinoma, kidney cancer, liver cancer, prostate cancer, vulval cancer, thyroid cancer, hepatic carcinoma and various types of head and neck cancers. According to a further aspect the invention provides the use of an agent, an isolated antibody or a phospholipid mimic of the present invention in the preparation of a medicament for the treatment of a condition or disorder caused by dysfunction of Notch signalling. The condition or disorder may be a cancer.

According to a yet further aspect, the invention provides a method of treating a condition or disorder caused by dysfunction of Notch signalling comprising administering to a subject in need thereof an effective amount of an agent, an isolated antibody or a phospholipid mimic of the present invention. The condition or disorder may be cancer.

Reference herein to "pharmaceutically acceptable carriers, diluents or excipients" includes without limitation any adjuvant, carrier, excipient, glidant, sweetening agent, diluent, preservative, dye/colorant, flavor enhancer, surfactant, wetting agent, dispersing agent, suspending agent, stabilizer, isotonic agent, solvent, or emulsifier that has been approved, for example, by the United States Food and Drug Administration or other governmental agency as being acceptable for use in humans or domestic animals.

The compounds of the present invention may be administered in the form of pharmaceutically acceptable salts. In such cases, pharmaceutical compositions in accordance with this invention may comprise a salt of such a compound, preferably a physiologically acceptable salt, which are known in the art. A pharmaceutically acceptable salt" includes both acid and base addition salts.

Pharmaceutical formulations will typically include one or more carriers acceptable for the mode of administration of the preparation, be it by injection, inhalation, topical administration, oral, lavage, or other modes suitable for the selected treatment. Suitable carriers are those known in the art for use in such modes of administration.

Suitable pharmaceutical compositions may be formulated by means known in the art and their mode of administration and dose determined by the skilled practitioner. For parenteral administration, a compound may be dissolved in sterile water or saline or a pharmaceutically acceptable vehicle used for administration of non-water soluble compounds such as those used for vitamin K. For enteral administration, the compound may be administered in a tablet, capsule or dissolved in liquid form. The table or capsule may be enteric coated, or in a formulation for sustained release. Many suitable formulations are known, including, polymeric or protein microparticles encapsulating a compound to be released, ointments, gels, hydrogels, or solutions which can be used topically or locally to administer a compound. A sustained release patch or implant may be employed to provide release over a prolonged period of time. Many techniques known to skilled practitioners are described in Remington: the Science & Practice of Pharmacy by Alfonso Gennaro, 20 ^th ed., Williams & Wilkins, (2000). Formulations for parenteral administration may, for example, contain excipients, polyalkylene glycols such as polyethylene glycol, oils of vegetable origin, or hydrogenated naphthalenes. Biocompatible, biodegradable lactide polymer, lactide/glycolide copolymer, or polyoxyethylene-polyoxypropylene copolymers may be used to control the release of the compounds. Other potentially useful parenteral delivery systems for modulatory compounds include ethylene-vinyl acetate copolymer particles, osmotic pumps, implantable infusion systems, and liposomes. Formulations for inhalation may contain excipients, for example, lactose, or may be aqueous solutions containing, for example, polyoxyethylene-9-lauryl ether, glycocholate and deoxycholate, or may be oily solutions for administration in the form of nasal drops, or as a gel.

The compounds or pharmaceutical compositions according to the present invention may be administered by oral or non-oral, e.g., intramuscular, intraperitoneal, intravenous, intracisternal injection or infusion, subcutaneous injection, transdermal or transmucosal routes. In some embodiments, compounds or pharmaceutical compositions in accordance with this invention or for use in this invention may be administered by means of a medical device or appliance such as an implant, graft, prosthesis, stent, etc. Implants may be devised which are intended to contain and release such compounds or compositions. An example would be an implant made of a polymeric material adapted to release the compound over a period of time . The compounds may be administered alone or as a mixture with a pharmaceutically acceptable carrier e.g., as solid formulations such as tablets, capsules, granules, powders, etc.; liquid formulations such as syrups, injections, etc.; injections, drops, suppositories, pessaries. In some embodiments, compounds or pharmaceutical compositions in accordance with this invention or for use in this invention may be administered by inhalation spray, nasal, vaginal, rectal, sublingual, or topical routes and may be formulated, alone or together, in suitable dosage unit formulations containing conventional non-toxic pharmaceutically acceptable carriers, adjuvants and vehicles appropriate for each route of administration. The compounds of the invention may be used to treat animals, including mice, rats, horses, cattle, sheep, dogs, cats, and monkeys.

Compounds of the invention may be intended for administration in combination with other chemotherapy and/or radiotherapy treatments. Treatment with an agent, isolated antibody or phospholipid mimic of the invention may occur prior to, concurrently with, or subsequent to administration of radiation therapy and/or another chemotherapy agent.

The term "antibody" is used to mean an immunoglobulin molecule that recognizes and specifically binds to a target, such as the N-terminal C2 domain of a Notch ligand, through at least one antigen recognition site within the variable region of the immunoglobulin molecule . In certain embodiments, antibodies of the present invention include antagonist antibodies that specifically bind to the N-terminal C2 domain of a Notch ligand and interfere with, for example, ligand binding or receptor binding. In certain embodiments, antibodies of the invention may promote, for example, ligand binding or receptor binding.

As used herein, the term "antibody" encompasses intact polyclonal antibodies, intact monoclonal antibodies, antibody fragments (such as Fab, Fab', F(ab')2, and Fv fragments), single chain Fv (scFv) mutants, multispecific antibodies such as bispecific antibodies generated from at least two intact antibodies, chimeric antibodies, hybrid antibodies, humanized antibodies, human antibodies, fusion proteins comprising an antigen determination portion of an antibody, and any other modified immunoglobulin molecule comprising an antigen recognition site so long as the antibodies exhibit the desired biological activity. An antibody can be of any the five major classes of immunoglobulins: IgA, IgD, IgE, IgG, and IgM, or subclasses (isotypes) thereof (e.g. IgGl , IgG2, IgG3, IgG4, IgA l and IgA2), based on the identity of their heavy-chain constant domains referred to as alpha, delta, epsilon, gamma, and mu, respectively. The different classes of immunoglobulins have different and well known subunit structures and three-dimensional configurations. Antibodies can be naked or conjugated to other molecules such as toxins, radioisotopes, etc.

As used herein, the term "antibody fragment" refers to a portion of an intact antibody and refers to the antigenic determining variable regions of an intact antibody. Examples of antibody fragments include, but are not limited to Fab, Fab', F(ab')2, and Fv fragments, linear antibodies, single chain antibodies, and multispecific antibodies formed from antibody fragments. An "Fv antibody" refers to the minimal antibody fragment that contains a complete antigen-recognition and -binding site either as two-chains, in which one heavy and one light chain variable domain form a non-covalent dimer, or as a single-chain (scFv), in which one heavy and one light chain variable domain are covalently linked by a flexible peptide linker so that the two chains associate in a similar dimeric structure. In this configuration the complementary determining regions (CDRs) of each variable domain interact to define the antigen-binding specificity of the Fv dimer. Alternatively a single variable domain (or half of an Fv) can be used to recognize and bind antigen, although generally with lower affinity. A "monoclonal antibody" as used herein refers to homogenous antibody population involved in the highly specific recognition and binding of a single antigenic determinant, or epitope . This is in contrast to polyclonal antibodies that typically include different antibodies directed against different antigenic determinants. The term "monoclonal antibody" encompasses both intact and full-length monoclonal antibodies as well as antibody fragments (such as Fab, Fab', F(ab')2, Fv), single chain (scFv) mutants, fusion proteins comprising an antibody portion, and any other modified immunoglobulin molecule comprising an antigen recognition site . Furthermore, "monoclonal antibody" refers to such antibodies made in any number of manners including but not limited to by hybridoma, phage selection, recombinant expression, and transgenic animals.

The term "humanized antibody" refers to forms of non-human (e.g. rodent) antibodies that are specific immunoglobulin chains, chimeric immunoglobulins, or fragments thereof that contain minimal non-human sequences. Typically, humanized antibodies are human immunoglobulins in which residues from the complementary determining regions (CDRs) within the antigen determination region (or hypervariable region) of the variable region of an antibody chain or chains are replaced by residues from the CDR of a non-human species (e.g. mouse, rat, rabbit, hamster) that have the desired specificity, affinity, and capability. In some instances, residues from the variable chain framework region (FR) of a human immunoglobulin are replaced with the corresponding residues in an antibody from a non-human species that has the desired specificity, affinity, and capability. The humanized antibody can be further modified by the substitution of additional residue either in the variable framework region and/or within the replaced non-human residues to refine and optimize antibody specificity, affinity, and/or capability. In general, the humanized antibody will comprise substantially all of at least one, and typically two or three or four, variable domains containing all or substantially all of the CDR regions that correspond to the non- human immunoglobulin whereas all or substantially all of the FR regions are those of a human immunoglobulin consensus sequence. The humanized antibody can also comprise at least a portion of an immunoglobulin constant region or domain (Fc), typically that of a human immunoglobulin. Examples of methods used to generate humanized antibodies are described in U. S. Pat. No. 5,225,539.

The term "human antibody" as used herein means an antibody produced by a human or an antibody having an amino acid sequence corresponding to an antibody produced by a human made using any technique known in the art. This definition of a human antibody includes intact or full-length antibodies, fragments thereof, and/or antibodies comprising at least one human heavy and/or light chain polypeptide such as, for example, an antibody comprising murine light chain and human heavy chain polypeptides.

"Hybrid antibodies" are immunoglobulin molecules in which pairs of heavy and light chains from antibodies with different antigenic determinant regions are assembled together so that two different epitopes or two different antigens can be recognized and bound by the resulting tetramer.

The term "chimeric antibodies" refers to antibodies wherein the amino acid sequence of the immunoglobulin molecule is derived from two or more species. Typically, the variable region of both light and heavy chains corresponds to the variable region of antibodies derived from one species of mammals (e.g. mouse, rat, rabbit, etc) with the desired specificity, affinity, and capability while the constant regions are homologous to the sequences in antibodies derived from another (usually human) to avoid eliciting an immune response in that species.

Embodiments of the present invention will now be described, purely by way of example, with reference to the accompanying drawings, in which:

Figure 1 shows the structure of Jagged 1N-EGF3 reveals a C2 domain N- terminal to the DSL motif, (a) The structure of J- 1N-EGF3 in a ribbon representation with the N-terminus towards the left of the figure and the C- terminus towards the right. Disulphide bonds are shown as sticks within semi- transparent spheres and the two glycans as stick models on the upper side, (b) Overlay of the N-terminal C2 domain of J- 1N-EGF3 on the Munc l 3- 1 C2b domain, (c) Alagille syndrome-associated missense mutations that affect the C2 domain (shown as darker spheres) are mapped onto the structure of J- 1N-EGF3 (cartoon and solid surface) and are located within the core of the domain (inset panel shows section at the level indicated by dotted line) , (c) Structure-based sequence alignment of the Munc l 3- 1 C2b domain and the N-terminal domains of Notch ligands. Most sequences associated with β-strands contain some regions of absolute(light grey) or high(dark grey) sequence homology, while loop regions are more variable . Residues coordinating the Ca2+ ion within the Munc l 3 domain are indicated, and residues mutated in this study in J- l (*);

Figure 2 shows that Notch ligands contain a functional Ca ²⁺ dependent phospholipid domain at their N-termini, (a) J- 1 _N - _E G _F3 incubated with trypsin in the presence and absence of Ca ²⁺ demonstrates a Ca ²⁺ -dependent proteolytic protection, consistent with the presence of Ca ²⁺ binding sites in the C2 domain (UD-undigested). (b) N-EGF3 constructs of three diverse Notch ligands (J- l , Dll- 1 and Drosophila Serrate) all bind liposomes made from a PC/PE/PS mixture whilst constructs lacking the N-terminal C2 domain do not (c) All three ligands are Ca ²⁺ -dependent in liposome binding, but they differ in their affinity for liposomes of a constant composition (comparison of J- l and Dll- 1 shown);

Figure 3 shows mutation of residues within the β5-β5 loop of the Jagged- 1 C2 domain leaves Notch binding intact, but disturbs phospholipid binding and Notch activation (a) Notch- 1 binding of WT J- 1 _N - _E G _F3 and mutants assessed by ELISA. F207A affects the Notch binding face of the DSL domain, D 140A/D 144A are putative Ca ²⁺ binding residues within the β5-β6 loop, (b) Liposome binding by WT J- 1 _N - _E G _F3 and mutants assessed by ELISA using fluorescently labeled liposomes (PC/PE/PS). (c) The ability of WT J- 1 _N - _E G _F3 F207A control and D→A mutants to activate receptor in Notch 1 -transfected cells was assessed using a split luciferase reporter system (d) Structure of WT J- 1 _N - _E G _F3 at pH7.5 in the presence of 7.5mM Ca ²⁺ reveals a single calcium coordinated between the β 5-β6 and β ΐ - β2 loops. J- l is shown as a cartoon representation (cyan - apo structure, teal - Ca ²⁺ -bound form) with key residues highlighted as stick representations and the bound Ca ²⁺ (pink) and a water molecule (teal) involved in coordinating it are shown as spheres;

Figure 4 shows Notch- 1 and Jagged- 1 extracellular domain architecture. The domains of Notch- 1 and Jagged- 1 that interact with each other for signalling are boxed;

Figure 5 shows the structure of the Jagged- 1 construct consisting of the N- terminal MNNL, DSL and EGF l -3 domains (J 1NE3) reveals that the MNNL domain is a C2-phospholipid binding domain, (a) cartoon of domain organization of the Jagged- 1 construct expressed for new structural studies J 1NE3 (b) Semi-transparent surface and cartoon representation of the X-ray structure of J J 1NE3 with MNNL domain coloured red and Notch- 1 binding face as previously defined coloured purple (c) location of disulphide bonds within the MNNL domain (d) cartoon representations of the MNNL domain overlaid on the

C2 domain of phospholipase C with the calcium ions bound to phospholipase C shown as red spheres (PDB XXX);

Figure 6 shows the MNNL/C2 domain of Jagged- 1 binds phospholipids in a calcium dependent fashion. Upper Panel: 2ug protein coated on Protein A sepharose beads and blocked with gelatine. 0.5uM phospholipid added as liposomes (PC:PS:PE) and mixed thoroughly. Washed with reaction buffer and liposomes solubilised with 1 % Triton X- 100. Bound lipid was quantified using fluorescent probe (Fluorescene-PE) in liposomes. Lower Panel: lug protein coated on Maxisorp plates for lhr @ 37C and blocked with gelatine. 25uM phospholipid added as liposomes (PC:PS:PE) and incubated. Washed with reaction buffer and liposomes solubilised with 1 % Triton X- 100. Bound lipid was quantified using fluorescent probe (Fluorescene-PE) in liposomes;

Figure 7 shows addition of calcium but not magnesium confers resistance to trypsin digestion on the J 1NE3 construct;

Figure 8 shows Mutation of residues predicted to be involved in phospholipid binding destroys Notch- 1/Jagged- l signalling at the same level as a mutation within the Jagged- 1 binding site for Notch- 1. Purified hJag-Fc ligands were immobilised onto cell culture plates at the indicated concentrations. After aspirating unbound ligand, Notch FL LCI reporter cells were seeded onto the plates (@ 1.25e4 cells/well in 0.5 μg/ml DOX containing medium) . Imaging was performed the next day. Photon flux data were normalized to the wells pre- incubated with TBS buffer alone . We have previously demonstrated that the mutation F207A prevents Notch- 1 binding by Jagged- 1 and so directly disrupts signalling. Mutation of the D 140/D 144 cluster to alanine is seen to disrupt signalling to the same level as the F207A mutation within the binding site strongly implicating these residues as putative calcium ligands required for phospholipid recognition. Data from a single ligand concentration are presented in the LH panel and from a wider range of mutants and at a range of ligand concentrations in the RH panel;

Figure 9 shows conservation in the N-terminal domain of Notch-ligands. The sequence alignment reveals blocks of higher sequence conservation amongst human Notch-ligands (Jagged- 1 , -2, Delta Like- 1 , -3 and -4) that correspond to secondary structural elements in the Jagged- 1 N-terminal domain structure implying that the C2 fold is conserved amongst these ligands. Mapping conservation onto the surface of the structure using Con Surf (http://consurf.tau.ac.il; maroon - highly conserved, cyan - highly variable) demonstrates that the surface of the C2 domain is not highly conserved suggesting that the ligands may differ in their phospholipid specificity;

Figure 10 shows aligned versions of predicted mature sequences of Human Notch-ligand predicted (Jag-2, Dll- 1 , Dll-3, Dll-4) or observed (Jag- 1) C2 domain mature sequences aligned and loops likely to be involved in modified lipids or other small molecule binding annotated. Residues in bold font are those in loop l and loop3 of the Jagged- 1 C2 structure and are predicted to be the sites of interaction with modified-lipids or other small molecule modulators of signalling - these are the same region as indicated bold in Figure 1 1 ; and

Figure 11 shows independent Sequences from European Bioinformatics Databases - Residues shown in italics predicted to be signal sequences and to be cleaved in the mature proteins. Residues in bold font are those in loop l and loop3 of the Jagged- 1 C2 structure and are predicted to be the sites of interaction with modified-lipids or other small molecule modulators of signalling - these are the same region as indicated in bold in Figure 10,

Figure 12 shows the cDNA and protein sequences of JDSL3 1 ,

Figure 13 shows the crystal Structure of Jagged- 1 NE3 shows C2 domain includes a single Calcium ion, shown as a sphere. Cell Rep _^ 2013 Nov 27;5(4) : 861 -7. doi: 10. 1016/j .celrep.2013.10.029. Epub 2013 Nov 14.

Structural analysis uncovers lipid-binding properties of Notch ligands.

Chillakuri CR ¹ , Sheppard D, Ilagan MX, Holt LR, Abbott F, Liang S, Kopan

R, Handford PA, Lea SM,

Figure 14 shows a summary of Screening of C2 Domain Monoclonal Antibodies,

Figure 15 shows the results of an ELISA assay to test whether C2 domain antibodies block Notch binding, C2 antibodies do not block Notch/Jagged interaction in an ELISA assay. In contrast 84b, an antibody with an epitope in J l DSL-EGF3 did block the interaction. N 1 _{NE I 4} (200 ng) was immobilised on the surface of a 96-well MaxiSorp plate . Interaction with h Jagged 1 _NE3 (50 nM) was determined in the absence of antibody and in the presence of a 1 : 1 dilution of hybridoma supernatant,

Figure 16 shows C2 antibodies do not block interaction of N l n-o with Jagged 1 expressing cells. The hybridoma supernatants were screened by flow cytometry for their ability to block the interaction between cell expressed full length human J l (HEK293 transfectants) and biotinylated N l n_ i ₃ soluble recombinant protein bound to avidin-coated fluorescent beads. No binding was observed with a control protein from human fibrillin 1 (solid grey) . The shift to the right indicates binding in the absence (grey line) or presence (black line) of hybridoma supernatant. Addition of mAbs 10c (left) and 3 1 (centre) had no effect on binding. In contrast J 1 - 183D (right), a characterised antibody with an epitope in the DSL domain completely inhibited binding.

Figure 17 shows calcium dependence of C2 domain mAb 3 1 , mAb 3 1 exhibits calcium-dependence in an ELISA assay. Experiment 1 : A 96-well MaxiSorp plate was coated with NE3 at a concentration of 5 μg/ml in Tris buffered saline and incubated at 4°C overnight. The wells were washed with TBS, 0.05% Tween-20 containing either 5 mM CaCl ₂ (dark grey)or 5 mM EDTA (light grey). The plate was blocked for I hr at room temperature with the above buffers containing 2% BSA. It was then incubated with a 1 : 100 dilution of each hybridoma supernatant in buffer containing 5 mM CaCl ₂ or 5 mM EDTA. After 3 washes the plate was incubated with 1 : 1000 dilution anti-mouse HRP conjugate . The plate was washed with buffers containing CaCl ₂ or EDTA and developed with 2,2'-Azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) diammonium salt (Sigma-Aldrich). Error bars indicate Mean ±SD,

Figure 18 shows calcium dependence of C2 Domain mAb 3 1 , Experiment 2: This was carried out similarly with a 1 :200 dilution of the hybridoma supernatants in buffers containing either 5 mM CaCl ₂ (dark grey) or 50 mM EDTA (light grey),

Figure 19 shows testing C2 domain antibodies in Notch signalling assay, Blocking of Notch activation signal by mAb3 1 which recognises C2 domain of hJ l . Using a split luciferase assay, mAb 3 1 but not I OC, was found to block Notch activation , similar to a control J l mAb raised to the DSL domain, known to block Notch binding and activation. The effect of mAb 3 1 is mediated through the binding of mAb to the C2 domain of J l ( present in J l construct NE3, but not DE3). Error bars Mean ± SD (n=5), Figure 20 shows testing of C2 domain antibodies in Notch signalling assay Expt 2 - performed as described for Figure 19,

Figure 21 shows a liposome binding assay to show J- l NE3 binding to liposomes is reduced by mAb3 1 ,

Data shown as mean±SD, n>3

NE3 binding is dependent on Ca2+, as expected.

J- l 65D is an anti DSL control antibody, Figure 22 shows a co-culture assay to show Notch activation by full length J l is reduced by mAb3 1 ,

Dark shading -full length J- l ( with C2 domain)

Light shading - J- l ( without C2 domain)

Mean of 2 replicates shown.

The Notch signalling system has a crucial impact on development and homeostasis of most tissues and organs. Disregulation of the pathway results in a number of inherited and acquired disorders, including various cancers, and it is a key target for therapeutic intervention. Despite the importance of the pathway, little is known about the structural basis of Notch receptor/ligand interactions and the molecular mechanisms that transduce this recognition event into signal activation. The Notch receptor exists as a heterodimeric trans-membrane protein which upon binding to ligands from Jagged/Serrate or Delta-like families results in trans-endocytic removal of metalloprotease-cleaved Notch extracellular domain. The final intra-membrane cleavage by gamma secretase releases the intracellular domain of Notch (NICD), which translocates to the nucleus and binds to a transcription factor of the CBF l , Suppressor of Hairless, Lag- 1 (CSL) family. This complex, together with the co- activator MAML, relieves repression and activates genes of the repressor Hes and Hey repressor families. Interactions with the Notch receptor can activate or inhibit Notch signalling, dependent upon whether ligands are presented to Notch on adjacent cells (trans), or on the same cell (cis) . O-glycosylation of Notch also plays an important role in regulating Notch signaling in a ligand-specific manner through the action of Fringe glycosyltransferases. Direct contact between the epidermal growth factor (EGF)-like domain EGF 12 of Notch and the Delta/Serrate/Lag-2 (DSL) domain, approximately 150 residues from the N-terminus of the ligands, confers specificity to the interaction. The role of the very N-terminal portion of the ligands is unknown, although its importance is implied by the fact that all Notch ligands through evolution position the DSL domain adjacent to it and deletion of this region in a C. elegans ligand LAG-2 abolished its function.

The structure of a Notch ligand construct consisting of the full N-terminal extension, the Notch-binding DSL domain and the three adjacent EGF domains from human Jagged- 1 (J- 1N-EGF3) (Figure la) has been solved. This structure demonstrates that the N-terminal portion of Jagged- 1 folds as a C2-phospholipid-recognition domain with strong structural homology to the Munc l 3 C2B domain (RMSD 2.7A vs. PDB 3kwt, 10% sequence identity) (Figure lb). The C2 domain packs closely on top of the Notch-binding DSL domain without altering the structure of the DSL-EGF 1 -3 portion of the construct, when compared to our earlier structure for this region in isolation, and therefore extends the near linear domain organization. Many missense mutations affecting this domain of Jagged- 1 cause Alagille Syndrome, a developmental disorder usually associated with J- l haploinsufficiency, and mapping these onto our new structure suggest that they destabilize the hydrophobic core and prevent correct folding of the C2 domain (Figure lc) . This is supported by our attempts to express J- 1N-EGF3 constructs containing these Alagille-associated substitutions which largely resulted in no or little protein secretion, and emphasize the importance of the native C2 domain fold for cell surface targeting of ligand. Sequence alignments with other Notch-ligands (Figure I d) suggest that the C2 domain is present in all mammalian Notch ligands with conserved residues mapping to those playing key structural roles, whilst residues exposed on the surface of the domain are more variable . Even evolutionarily distant Notch ligands such as Drosophila Serrate and the C. elegans LAG-2 retain N-terminal regions of similar lengths and their sequences are compatible with a C2 fold. These data suggest an important role for the ligand C2 domain in the Notch signalling pathway. Since the strongest structural homology of the J- l C2 domain is to calcium-dependent phospholipid-binding domains (e.g. the Munc l 3- 1 C2b domain), studies were undertaken to investigate whether or not J- 1N-EGF3 bound Ca2+. Our limited proteolysis assay showed that Ca2+, but not Mg2+ or EDTA, protects the J- 1N-EGF3 fragment from tryptic digestion, indicating the presence of previously unrecognized calcium-binding sites (Figure 2a). This is a C2 domain-specific effect, since a construct lacking the C2 domain (J- 1 AC2) did not show Ca2+-dependent protection (data not shown), and is consistent with published data that demonstrate that C2 domain loops involved in Ca2+ coordination adopt a more structured conformation in the presence of Ca2+ .

It was investigated whether the Jagged C2 domain is functionally active as a phospholipid-binding domain. Both solution and plate-based assays were used to study protein binding to fluorescently labeled liposomes consisting of mixtures of PC/PS/PE. These assays demonstrated that the J- 1N-EGF3, but not the J- 1 AC2 construct, bound liposomes in a Ca2+-dependent manner (Figures 2b) supporting the structure-based hypothesis that the C2 domain at the N-terminus of Jagged- 1 is a functional phospholipid-recognition domain. Since sequence comparisons implied conservation of the C2 domain in other Notch ligands we repeated these assays with human Delta-like 1 (35 % identity in C2 domain) and Drosophila Serrate (40% identity), and demonstrated that these Notch-ligands also bind liposomes in a Ca2+- and C2 domain-dependent fashion (Figure 2b). However, as expected from sequence diversity in the ligand-binding loops, comparative analysis of human J- l and DLL- 1 constructs showed different levels of liposome binding (Figure 2c).

To further dissect the functional consequences of phospholipid recognition, a series of site-directed J- 1N-EGF3 mutants were designed targeting aspartate residues within the loops homologous to those involved in Ca2+ and phospholipid binding in Munc l 3- 1 C2b (Figure I d). Structural and sequence comparisons suggested that aspartate residues in the β5-β6 loop (D 140 and D 144) were most likely involved in Ca2+ coordination (located in structurally equivalent positions to residues D757 and D759 in the Munc l 3- 1 C2b domain and conserved through human Notch ligand proteins) although involvement of aspartates in the β ΐ - β2 loop which are less conserved through human Notch ligand proteins could not be ruled out. Single and double mutants of J- 1N-EGF3 with alanine substitutions in place of aspartates in the β5 - β6 loop were cloned and expressed. Loss of calcium-dependent protection against trypsin proteolysis was observed in D 140A and in a D 140A/D 144A double mutant. Phospholipid binding was also reduced, consistent with a role for these residues in Ca2+ coordination and hence in phospholipid binding. It was investigated whether J- l C2 domain mutants affected Notch activation. An ELISA-based Notch-binding assay demonstrated that the aspartate mutants were competent for Notch binding, in contrast to a previously identified ablative DSL domain mutant (F207A) control (Figure 3a). By contrast F207A was WT-like in its ability to bind liposomes, whilst aspartate mutants showed reduced liposome binding with the double mutant being most profoundly affected (Figure 3b). Mutant proteins were next analysed using a quantitative split luciferase Notch activation reporter system (Figure 3c) where Notch- 1 was expressed on the cell-surface and Fc fusions of Jagged- 1 were presented on the well surface (to mimic Notch/ligand interactions in trans). This assay was chosen because it is a receptor-proximal reporter whose output is directly proportional to the amount of NICD released, making it well-suited for quantifying the extent of a deficit caused by ligand mutations. Wild-type J- 1N-EGF3 was able to promote Notch activation in this assay, whilst the F207A mutant, which cannot bind Notch, was strongly impaired. The double mutant D 140A/D 144A ablated Notch-dependent signalling to level similar to that of F207A, and either D 140A or D 144A alone significantly reduced Notch activation even though all of these mutants were competent for Notch- 1 binding.

To explore the molecular basis of the effects of these mutations another J- 1N-EGF3 structure was determined from crystals grown at a physiological pH in the presence of 7.5mM Ca2+. As for other Ca2+-dependent C2 domains, this new structure demonstrated that calcium bound in a cleft between the β 1 -β2 and β5-β6 loops at the top of the C2 domain (Figure 3d), ordering them despite the increased length of these loops in Jagged- 1 compared to other C2 domains (Figure S4). The structure demonstrates that a single Ca2+ ion is bound under these conditions, coordinated by the side chains of Asp72 (OD 1) and Asp l 40 (OD 1 and OD2), the main chain carbonyl of Serl 41 and an ordered water molecule (Figure 3d inset). Bidentate coordination of the bound Ca2+ by Asp 140 explains why substitution of this side chain by Ala significantly reduces phospholipid binding but our structure does not explain why mutation of Asp 144 also has this effect. It is noteworthy that, even in the presence of Ca2+, the β5-β6 loop remains highly mobile (as indicated by mean main chain atomic temperature factors of 67A2 compared to 4 lA2 for the whole structure) and we cannot therefore exclude a further rearrangement of this loop leading Asp 144 to be directly or indirectly (via Ca2+ coordination) involved in phospholipid binding. Collectively, our molecular and cellular data strongly imply that, in addition to the Notch- 1/Jagged- l trans interaction, signalling may also require Ca2+ dependent phospholipid binding by Jagged- 1. Our experimental data demonstrating that this activity is conserved in other Notch-ligands (Figure 2), together with the observed sequence conservation (Figure 1), further suggests that all Notch signalling will have a lipid-binding element. It is of interest that three disease-causing mutations associated with extrahepatic biliary atresia map to the loop structures of the C2 domain, rather than the hydrophobic core, emphasizing the importance of this region for function.

Notch-dependent signalling is a complex phenomenon with a series of extra- and intra- cellular events required for signal generation under tightly controlled circumstances. Additional complexity was recently added to these pathways when a Notch-ligand, Jagged- 1 , was demonstrated to interact with and be sequestered from productive Notch interactions by another cell-surface protein CD46. The structural and functional data now add another component to this complex pathway suggesting that C2 domain-mediated lipid binding is a modulator of the signalling process.

Resolution of the structure of the C2 domain

Notch/Ligand interactions are driven primarily by direct contact between epidermal growth factor (EGF) like domains in the centre of the Notch extracellular domain (EGF 1 1 - 13) and the Delta/Serrate/Lag (DSL) domain approximately 150 residues from the N-terminus of the ligands. Although all Notch-ligands through evolution position the DSL domain in the same relationship to the N-terminus, the role of this portion of the ligands is unknown although loss of function associated with deletion of this region in a C. elegans ligand implies a critical role in signalling. A Notch-ligand construct consisting of the full N-terminal extension, the Notch-binding DSL domain and the three EGF domains C-terminal to the DSL from human Jagged- 1 (J- 1 _N - _E G _F3 ) has been expressed and the structure solved, as summarised in Figure 1. The structure demonstrates that the N-terminal portion of Jagged- 1 folds as a C2-phospholipid recognition domain with strong structural homology to the Munc l 3 C2B domain (RMSD 2.722A vs. PDB 3kwt, 10% sequence identity) and that this domain packs closely on top of the Notch-binding DSL domain without altering the structure of the DSL-EGFi_ ₃ portion of the construct when compared to our earlier structure for this region in isolation. Many mutations in Jagged- 1 are associated with human disease, particularly with Alagille Syndrome, and mapping these mutations onto the new structure suggested that these largely act by preventing correct folding of the C2- domain. This was supported by attempts to express J- 1 _N - _E G _F3 constructs containing the disease-associated mutations which largely resulted in no or little protein expression suggested mis-folding of the C2-domain led to intracellular retention implying that Alagille Syndrome is a disease caused by low/no expression on the cell-surface of Jagged- 1 rather than by more subtle alteration of Jagged- 1 function due to local structural/functional alterations. Conservation of the C2 domain

Sequence alignments with other Notch-ligands (Figure ID) suggest that the C2 domain is conserved throughout all mammalian Notch-ligands with the conserved residues mapping to those playing key structural roles whilst residues exposed on the surface of the domain are more variable between different ligands. Even in more evolutionarily distant Notch-ligands such as Drosophila Serrate and the C. elegans LAG proteins the size of the N-terminal regions is conserved and the sequences are compatible with a C2 fold as assessed using threading algorithms suggesting conservation of fold. The N-terminal domain of Jagged-1 binds phospholipids in a calcium-dependant manner

Conservation of the C2 domain in evolutionary distant Notch-ligands suggests a functional role of the domain in Notch signalling. Since the strongest structural homologies were with C2 phospholipid domains that were Ca ²⁺ dependent in their recognition (e.g. the Munc l 3 - 1 C2b domain) and, particularly due to the previously established role for Ca ²⁺ binding in Notch signalling (Notch interactions with ligands require Ca ²⁺ binding in the Notch EGF domains). Jagged- 1 was found to bind Ca ²⁺ . Inspection of the structure of the J- l C2 domain revealed that, as for other C2 domains in the absence of Ca2+, in the Ca ²⁺ free structure the loops likely to be involved in Ca ²⁺ and hence phospholipid binding, were disordered (Figure 2). In other C2 domains these loops have been demonstrated to adopt more structured conformations in the presence of Ca ²⁺ . Limited proteolysis of J- 1 _N - _E G _F3 and J- 1 _D S _L - _E G _F3 constructs in the presence or absence of Ca ²⁺ /EGTA/Mg ²⁺ was used to investigate whether the J- l C2 domain similarly refolded. These studies demonstrate (Figure 2b) that the J- l C2 domain became more proteolytically resistant in the presence of Ca ²⁺ suggesting that, as for other C2 domains, loops involved in Ca2+ binding adopt more structured conformations in the presence of Ca2+. The Jagged C2 domain is functionally active as a phospholipid binding domain. Solution and plate based assays were used to study protein binding to fluorescently labelled liposomes consisting of mixtures of PC/PE/PS . These assays demonstrated that the J- 1N-EGF3 but not the J- 1DSL-EGF3 construct bound liposomes in a Ca ²⁺ dependent manner supported our structure-based hypothesis that the C2 domain at the N-terminus of Jagged- 1 is a functional phospholipid recognition domain (Figure 2c) . To further dissect the molecular basis of phospholipid recognition and to probe its functional consequences the J- 1N-EGF3 structure was analysed (Figure 3) . Structural and sequence comparisons suggested that aspartate residues in the β5 -β6 loop (D 140 and D 144) were likely involved in Ca ²⁺ coordination (located in structurally equivalent positions to residues D757/D759 in the Munc l 3 - 1 C2b domain and conserved through human Notch-ligand proteins) although involvement of aspartates in the β 1 -β2 loop (D63 , D69 and D70; positioned similarly to residues D705/D71 1 in Munc l 3 - 1 C2b domain, but not conserved through human Notch-ligand proteins) could not be ruled out. J- 1N-EGF3 constructs were cloned and expressed containing these aspartate residues mutated to alanines as both single and pairs of mutations . Limited proteolysis revealed that mutation of the D 140/D 144 pair led to a construct that showed no difference in proteolytic sensitivity in both the presence of absence of Ca2+ suggesting that these residues are involved in coordination of the ion and hence in phospholipid binding . All constructs generated were assayed for Notch binding using an ELISA-based assay with the Jagged- 1 DSL domain (F207A) as a control for ablation of Notch binding by Jagged. These assays demonstrated that, with the exception of J- 1N-EGF3 -F207A, all the WT and mutant constructs were competent for Notch binding . Next these mutations were analysed using a split luciferase Notch-signalling reporter system where Notch- 1 was expressed on the cell-surface and Jagged- 1 constructs expressed as Fc fusions presented as an oriented array on the well surface to mimic Notch/ligand interactions in trans . Wild-type J- 1N-EGF3 was able to stimulate Notch-dependent signalling in this assay whilst, as predicted, J- 1N-EGF3 -F207A which cannot bind Notch, does not. Whilst mutation of D69A/D72A did not alter signalling, surprisingly the paired D 140A/D 144A mutations ablated Notch-dependent signalling to a similar level to F207A and either D63A, D 140A or D 144A alone significantly reduced the ability to stimulate Notch-dependent signals even though all of these mutants are competent for Notch- 1 binding . Taken together with the limited proteolysis data and the structural comparisons the data strongly implies that in addition to requiring direct interactions between Notch- 1 and Jagged- 1 in trans, signalling is critically dependent on phospholipid binding by Jagged- 1 . The level of sequence conservation furthermore implies that the C2 domain is conserved at the N-termini of all Notch-ligands further suggesting that all Notch signalling will be dependent on phospholipid binding . Materials and Methods

Protein production :

Jagged and Notch proteins were produced in HEK 293T cells using a transient transfection system (Aricescu AR, Lu W, & Jones EY (2006) Acta Crystallogr D Biol Crystallogr 62(Pt 10) : 1243 - 1250) . Proteins were purified using Ni-chelating sepharose followed by gel-filtration. Protein used for crystallization was produced in 293 S GnTI -/- cells .

Notch binding assay:

Human Notch- 1 _n _ _EG FI4 was coated on the MaxiSorp plates . WT and mutant Fc tagged constructs of J- 1N-EGF3 were added at Ι ΟΟηΜ concentration and anti human Fc antibody-HRP conjugate was used to detect binding .

Limited proteolysis assay:

WT and mutant constructs of Histidine tagged J- 1N-EGF3 were treated with trypsin in the presence of l OmM Ca ²⁺ , Mg ²⁺ or EDTA . Samples were collected at different time intervals and analysed on SDS-PAGE.

Phospholipid binding assay:

Liposomes were prepared using PC: PS : PE-Fluorescein (80 : 15 : 5) by ultrasonication. The bead based assay was performed as described earlier but using Protein A- sepharose beads . In the plate based assay equimolar concentrations of proteins were coated on a MaxiSorp plate . After blocking with 1 % gelatin Ι ΟΟμΜ phospholipids were added in the presence of 2mM free Ca ² / Mg ²⁺ or 0.5mM EGTA. Bound liposomes were solubilized with 0.3 % Triton X- 100 and fluorescence was measured using a NanoDrop 3300. Crystallisation and Structure Solution:

Crystals for the 2.5 A apo structure were grown in 0.3 μΐ sitting drops using vapour diffusion at 3.2mg/ml with 0.97M Sodium Citrate (pH5.0), 19.4% Polyethylene glycol 6000 mother liquor at 25% after initial screening using commercially available reagents from Molecular Dimensions. Diffraction data was collected at ESRF on beamline ID29 with a Pilatus 6M-F detector. Phasing was solved through molecular replacement using the program Phaser and previously published structure of Jagged W-EGF 123 (Cordle J, et al. (2008) Nat Struct Mol Biol 15(8): 849-857 and McCoy AJ, et al. (2007) Phaser crystallographic software. J Appl Crystallogr 40 (Pt 4):658-674). The C2 domain was modelled using a combination of autobuild, buccaneer and manual modelling (Adams PD, et al. (2010) PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr D Biol Crystallogr 66(Pt 2):213-221 and Cowtan K (2006). The Buccaneer software for automated model building. 1. Tracing protein chains. Acta Crystallogr D Biol Crystallogr 62(Pt 9) : 1002- 101 1).

Both calcium bound forms were crystallised at 5.0mg/ml using commercial reagents from Molecular Dimensions using the sitting drop vapour diffusion method with a drop size of 0.2μ1 and 25% mother liquor. The 2.38A form was crystallised with 0.2M sodium acetate, 0. 1M Bis-Tris propane (pH 7.5) and 20% Polyethylene glycol 3350. The 2.84A form was crystallised using 0.2M sodium iodide, 0. 1M Bis-Tris propane (pH 6.5) and 20% Polyethylene glycol 3350. In addition l OmM CaCl ₂ was added to the initial protein solution. Data were collected at the Diamond Synchrotron facility on beamline 103 with a Pilates 6M-F detector. Phases were determined by molecular replacement using the apo structure and the program Phaser. Refinement for all structures was performed using autobuster and COOT (Emsley P, Lohkamp B, Scott WG, & Cowtan K (2010) Acta Crystallogr D Biol Crystallogr 66(Pt 4):486-501 and Smart OS, et al. (2012) Acta Crystallogr D Biol Crystallogr 68(Pt 4) :368-380). Notch activation assays

Purified J-1N-EGF3 Fc-fusion proteins (10 μg/ml in TBS) were immobilized onto culture plates overnight at 4°C. The plates were washed once with TBS and then, the Notch split luciferase reporter cells (HelaTetON cells stably expressing Notch 1-NLuc and CLuc-RBPji ) were seeded onto the immobilized ligand in the presence of 0.5 μg/ml Doxycycline to induce reporter expression. After 24 hours, the cells were assayed for bioluminescence as previously described (Ilagan MX, Lim S, Fulbright M, Piwnica-Worms D, & Kopan R (2011) Sci Signal 4(181):rs7).

Table 3- Crystallographic Data Collection and Refinement Statistics

Values denoted in brackets refer to inner shell *Refinement of the structure was performed at 2.5A, at 2.8A Completeness (%) = 89% and Rmerge = 30.0%

Sequencing

The VH and VL sequences of JDSL3 1 were successfully determined by V-region PCR and are shown in figure 12.

Hybridoma sequencing

The variable heavy and variable light chains were amplified using degenerate forward primers that bound either in the signal peptide or framework region 1 and a reverse primer that bound in the antibody constant region. The amplified genes were sequenced following a standard approach.

DNA sequencing and analysis

DNA was sequenced by conventional Sanger sequencing and data analysed using DNASTAR Lasergene software. Signal peptide and variable domain sequences were identified by comparison with known sequences in the IMGT database .

Variable heavy domain sequencing

The heavy chain gene for JDSL3 1 was successfully isolated and sequenced by V- region PCR. Alignment of the DNA sequence of the start of the CHI domain with mouse germline sequences confirms that the antibody has a mouse IgGl heavy chain. A BLAST search of the DNA sequence of the identified VH domain shows that the identified sequence is novel. Analysis of the protein sequence of the identified VH domain is consistent with that expected for a functional heavy chain. It was not possible to identify the full signal peptide sequence for the heavy chain.

Variable light domain sequencing

The light chain gene for JDSL3 1 was successfully isolated and sequenced by V-region PCR. See figure 3 in the appendix section for the sequencing chromatogram. Alignment of the DNA sequence of the start of the CL domain with mouse germline sequences confirms that the antibody has a mouse kappa chain. A BLAST search of the DNA sequence of the identified VL domain shows that the identified sequence is novel. Analysis of the protein sequence of the identified VL domain is consistent with that expected for a functional light chain. The sequence of the signal peptide could not be identified.

CLAIMS

1. A method for identifying a compound useful in modulating the Notch signalling pathway, the method comprising the step of determining whether the compound binds to the N-terminal C2 domain of a Notch ligand.

2. The method according to claim 1 further comprising the step of determining whether binding of the compound to the C2 domain modulates notch signalling. 3. The method according to claim 1 or claim 2 wherein the compound useful in modulating the Notch signalling pathway is a phospholipid, a glycolipid, a sphingolipid, a small molecule or an antibody.

4. The method according to any one of the preceding claims wherein the compound increases Notch signalling.

5. The method according to any one of the preceding claims wherein the compound decreases Notch signalling. 6. The method according to any one of the preceding claims wherein the Notch ligand is a protein selected from the Jagged, human Delta-like- 1 or the Drosophila Serrate protein family.

7. The method according to any one of the preceding claims wherein the Notch ligand is mammalian or human Jagged- 1 (JAG1), Jagged-2 (JAG2), Delta-like 1

(DLL 1), Delta-like 2 (DLL2), Delta-like 3 (DLL3) or Delta-like 4.

8. The method according to any one of the preceding claims wherein the Notch ligand binds to a mammalian or human protein comprising a C2 domain having at least 70%, 75%, 80%, 85%, 90%, 95%, 98%, 99% sequence identity to SEQ ID NO . 1 , SEQ ID NO. 2, SEQ ID NO. 3, SEQ ID NO. 4 or SEQ ID NO. 5.