WO2001012655A1 - Peptides consisting of fragments of gli-1 and sufuh and their use - Google Patents

Abstract

The present invention relates to the field of mammalian signalling pathways, and more precisely to the physical interaction between two components of the Sonic hedgehog (Shh)-Patched (Ptch) signalling pathway, namely GLI-1 and SUFUH. The invention provides peptides consisting of fragments of GLI-1 and SUFUH, respectively which are able to specifically bind to SUFUH and GLI-1, respectively. The invention also provides monoclonal antibodies and antibody fragments specifically binding to these peptides, as well as pharmaceutical compositions containing the peptides, antibodies and/or antibody fragments, said pharmaceutical compositions being useful for treating cancer and diseases influencing cell differentiation and tissue development.

Description

New peptides, corresponding antibodies, and compositions thereof.

The present patent application relates to the field of mammalian signalling pathways, and more precisely to the physical interaction between two components of the Sonic hedgehog (Shh)-Patched (Ptch) signalling pathway, namely GLI-1 and SUFUH.

The Sonic hedgehog (Shh) - Patched (Ptch) signalling pathway is of key importance for both normal development and carcinogenesis, as shown by the presence of muta- tions in genes encoding components of this pathway in human malformation and can- cer-prediposing syndromes (holoprosencephaly [Roessler et al., Hum. Mol. Genet. 6:1847-1853 (1997)], nervoid basal cell carcinoma syndrome [Hahn et al., Cell 85:841- 851 (1996); Johnson et al., Science 272:1668-1671 (1996)] and Greig cephaloplysyn- dactyly [Vortkamp et al., Nature 352:539-540 (1991)].

Intracellular transduction of the Hedgehog signal in Drosophila involves derepression of the seven-transmembrane-domain protein Smoothened and release of a large complex containing Costal-2 (Cos-2), Fused (Fu), Suppressor-of-Fused (Su(fu)) and the transcriptional effector Cubitus interruptus (Ci) from microtubules at which they are anchored, as well as blockage of the proteolytic conversion of full-length Ci to its ami- no-terminal repressor form [Ruiz i Altaba, Cell 90:193-193 (1997); Monnier et al., Curr. Biol. 8:583-586 (1998); Methot et al., Cell 96:819-831 (1999)]. In Drosophila only the repressor form of Ci is detected in the nucleus indicating that a specific mechanism^) may prevent nuclear entry of full-length Ci. A separate activation step has been implicated in the conversion of full-length Ci to a transcriptional activator [Methot et al., Cell 96:819-831 (1999); Ohlmeyer et al., Nature 396:749-753 (1998)].

In mammals, functions corresponding to that of Ci have been assigned to the zinc- finger-containing and DNA-binding proteins Gli 1-3, which are expressed in an over- lapping pattern adjacent to cells secreting Shh or the homologous Indian hedgehog (Ihh) and Desert hedgehog (Dhh). GLI-3 mutations detected in human disorders result in expression of truncated GLI-3 proteins that mimic natural Ci processing with respect to their altered subcellular localization and transactivation properties in HeLa cells (the full-length protein is cytoplasmic) [Shin et al., Proc.Natl.Acad.Sci. USA 96:2880-2884 (1999)]. In the case of GLI-1, initial studies using the D259MG glioma cell line, which contains an amplified GLI-1 locus, showed that this protein is nuclear in localization

[Kinzler et al., Mol.Cell.Biol. 10:634-642 (1990)]; nuclear GLI-1 localization was also seen after GLI-1 cDNA was transfected into COS cells [Dahmane et al., Nature

389:876-881 (1997)]. In contrast, GLI-1 in human basal cell carcinomas (BCCs) was again cytoplasmic [Dahmane et al., Nature 389:876-881 (1997)]. Most recently it has been shown that, upon overexpression, all three GLI proteins may show either cytoplasmic or nuclear localization, depending on the cellular context [Ruiz i Altaba, Development 126:3205-3216 (1999)]. Analysis of the proteolytic processing of vertebrae Gli proteins using mouse embryo extracts has shown the appearance of shorter variants of endogenous Gli-3, but not of Gli-1 [Dai et al., J. Biol. Chem. 274:8143-8152

(1999)], whereas after overexpression in frog embryos shorter variants of all three Gli proteins were observed [Ruiz i Altaba, Development 126:3205-3216 (1999)]. However, as overexpressed full-length Gli proteins can be detected in the nucleus, proteolytic processing does not appear to be necessary for nuclear import. Taken together, the- se data strongly indicate that, in mammalian cells, there may be mechanisms that regulate the subcellular localization of both full-length and processed Gli proteins. Such mechanisms could involve the regulation of interactions of the Gli proteins with anchoring proteins, or the modification of the Gli proteins themselves, including proteolytic processing.

The above mentioned component of this system, the Suppressor of Fused, Su(fu), in Drosophila [Pham et al., Genetics 140:587-598 (1995)], has sofar not been been identified in man. In insects, Su(Fu) has been shown to form a complex together with Cos- tal-2 (Cos2) and the transcription factor Cubitus interruptus (Ci) [Monnier et al., Cur- rent Biol. 8:583-586 (1998)] which all are involved in the transduction of the Hedgehog signal.

The hedgehog signalling pathway is influenced by numerous autoregulatory and other control mechanisms as for example transcriptional regulation of Patched and Gli by Gli itself [Alexandre et al., Genes Dev. 10:2003-2013 (1996)], and proteolytic generation of Ci and Gli isoforms with opposite transactivation properties [Jiang et al., Nature 391:493-496 (1998); Ohlmeyer et al, Nature 396:749-753 (1998); Methot et al., Cell 96:819-831 (1999)]. Therefore, it is reasonable to assume that Su(fu) activity would also be controlled at various levels.

Accordingly, it would be highly desirable to find the control mechanisms behind the Sonic Hedgehog (Shh) - Patched (Ptch) signalling pathway, and to be able to use them as a basis for a pharmaceutical preparation for treating cancer and diseases influencing cell differentiation and tissue development.

Summary of the Invention

The present invention relates to the field of mammalian signalling pathways, and more precisely to the physical interaction between two components of the Sonic hedgehog (Shh)-Patched (Ptch) signalling pathway, namely GLI-1 and SUFUH. The invention provides peptides consisting of fragments of GLI-1 and SUFUH, respectivelwhich are able to specifically bind to SUFUH and GLI-1, respectively. The invention also provides monoclonal antibodies and antibody fragments specifically binding to these pepti- des, as well as pharmaceutical compositions containing the peptides, antibodies and/or antibody fragments, said pharmaceutical compositions being useful for treating cancer and diseases influencing cell differentiation and tissue development.

Detailed description of the invention As already mentioned, the present invention relates to the physical interaction between

GLI-1 and SUFUH. Accordingly, the present invention provides polypeptides derived from GLI-1 and SUFUH, respectively, which polypeptides show affinity to SUFUH and GLI-1, respectively.

Nucleic acid sequences, peptides and proteins are referred to in the description, claims and sequence listing by the following abbreviations:

SEQ.ID.NO.l discloses the amino acid sequence of SUFUH. This sequence has also been disclosed in WO 99/32517.

SEQ.ID.NO.2 and SEQ.ID.NO.3 disclose sequences corresponding to amino acid residues 380 - 484 and 387 - 484, respectively, of SUFUH. These fragments comprise a GLI-1 binding site.

SEQ.ID.NO.4 discloses a sequence corresponding to amino acids 387 - 405 of SUFUH. Peptides comprising this sequence have been shown to physically interact with GLI-1.

SEQ.ID.NO.5 presents the sequence of GLI- 1. It is disclosed in Kinzler et al., Nature 332:371-4.

SEQ.ID.NO.6 discloses a sequence corresponding to amino acid residues 1 - 407 of GLI-1. This fragment comprises sites physically interacting with SUFUH.

SEQ.ID.NO.7 discloses a sequence corresponding to amino acids 381 - 407 of GLI-1. This fragment comprises sites controlling shuttling of GLI-1 and is blocked when SUFUH interacts with this protein. SEQ.ID.NO.8 presents a sequence corresponding to amino acids 62 - 82 of GLI-1.

This fragment comprises sites binding to SUFUH.

SEQ.ID.NO.9 shows the amino acid sequence of SUFU from Drosophila melanogas- ter.

SEQ.ID.NO.10 presents the amino acid sequence of SUFUHΔ, a SUFUH variant that has been truncated in the carboxy terminal after amino acid 385.

SEQ.ID.NO.11 shows the amino acid sequence of haematoglutinin-tagged GLI-1.

SEQ.ID.NO.l 2 shows the amino acid sequence of myc-tagged SUFUH.

SEQ.ID.NO.13 shows the amino acid sequence of a haematoglutinin-tagged fragment comprising amino acids 1-407 of GLI-1

SEQ.ID.NO.14 shows the amino acid sequence of myc-tagged SUFUHΔ.

SEQ.ID.NO.15 shows the amino acid sequence of an amino-terminal fragment of GLI- 1 comprising amino acids 1 - 525.

SEQ.ID.NO.16 shows the amino acid sequence of an amino-terminal fragment of GLI- 1 comprising amino acids 1 - 525 in which leucines 495 and 501 have been altered to alanines.

SEQ.ID.NO.17 shows the amino acid sequence of SU(FU)-XL.

SEQ.ID.NO.18 shows the amino acid sequence of SU(FU)-LK.

SEQ.ID.NO.19 shows the amino acid sequence of myc-tagged SU(FU)-Tt. SEQ.ID.NO.20 shows the amino acid sequence of myc-tagged SU(FU)-XL.

SEQ.ID.NO.21 shows the amino acid sequence of myc-tagged SU(FU)-LK.

SEQ.ID.NO.22 shows the amino acid sequence of myc-tagged SU(FU)-Tt.

SEQ.ID.NO.23 shows the DNA sequence of SUFUH.

SEQ.ID.NO.24 shows the DNA sequence of SUFUH-XL.

SEQ.ID.NO.25 shows the DNA sequence of SUFUH-LK.

SEQ.ID.NO.26 shows the DNA sequence of SUFUH-Tt.

SEQ.ID.NO.27 shows the DNA sequence of GLI-1.

SEQ.ID.NOs.28-31 show the DNA sequence of four oligonucleotide primers useful in the chromosomal localization of SUFUH.

The polypeptides of the invention have one or more of the following properties. In a first embodiment, the polypeptides comprise fragments and/or derivatives of SUFUH that interacts with GLI-1. In a second embodiment, the polypeptides comprises fragments and/or derivatives of GLI-1 that binds to SUFUH. By contacting GLI-1 and SU- FUH with the polypeptides of the invention in vivo, it should be possible to affect the Sonic hedgehog (Shh) - Patched (Ptch) signalling pathway which could be useful in treatment of cancer.

The nucleic acid sequences of the present invention are preferably DNA, though they may be RNA. Nucleic acid sequences of the invention will typically be in isolated or substantially isolated form. For example up to 80, up to 90, up to 95 or up to 100% of the nucleic acid material ir a preparation of a nucleic acid of the invention will typically be nucleic acid according to the invention. Alterations, isolations or syntheses of the nucleic acid sequences of the invention may be performed by any conventional method, for example by the methods of Sambrook et al (Molecular Cloning: A Laboratory Manual; 1989).

A substitution, deletion or insertion may suitably involve one or more amino acids, typically from one to five, one to ten or one to twenty amino acids, for example, a sub- stitution, deletion or insertion of one, two, three, four, five, eight, ten, fifteen, or twenty amino acids. Typically, a polypeptide of the invention has at least 70% at least 80%, at least 90%, or at least 95% sequence identity to the sequence of any one of SEQ. ID. Nos. 2 to 4, or SEQ.ID.Nos 6 to 8.

In general, the physicochemical nature of the sequence of SEQ. ID. Nos. 2 to 4 or SEQ.ID.Nos 6 to 8 should be preserved in a polypeptide of the invention. Such sequences will generally be similar in charge, hydrophobicity and size to that of SEQ. ID. Nos. 2 to 4 or SEQ.ID.Nos 6 to 8. Examples of substitutions that do not greatly affect the physicochemical nature of amino acid sequences are those in which an amino acid from one of the following groups is substituted by a different amino acid from the same group:

H, R and K

I, L, V and M

A, G, S and T

D, E, Q and N. Where polypeptides of the invention are synthesised chemically, D-amino acids (which do not occur in nature) may be incorporated into the amino acid sequence at sites where they do not affect the polypeptides biological properties. This reduces the polypeptides' susceptibility to proteolysis by the recipient's proteases.

The nucleic acid sequences encoding the polypeptides of the invention may be extended at one or both ends by any non-wild-type sequence.

Thus, the polypeptides of the invention may be extended at either or both of the C- and N- termini by an amino acid sequence of any length. For example, an extension may comprise up to 5, up to 10, up to 20, up to 50, or up to 100 or 200 or more amino acids.

A polypeptide of the invention may be subjected to one or more chemical modifications, such as glycosylation, sulphation, COOH-amidation or acylation. In particular, polypeptides that are acetylated at the N-terminus are preferred, as are polypeptides having C-terminal amide groups. Preferred polypeptides may have one or more of these modifications. For example, particularly preferred peptides may have a C-terminal amide group and N-terminal acetylation.

A polypeptide of the invention may form part of a larger polypeptide comprising multiple copies of the sequence of one or more of SEQ. ID. Nos. 2 to 4 or SEQ.ID.Nos. 6 to 8 or a sequences related to them in any of the ways defined herein.

Polypeptides of the invention typically comprise at least 15 amino acids, for example 15 to 20, 20 to 50, 50 to 100 or 100 to 200 or 200 to 300 or 300 to 400 amino acids.

Polypeptides according to the invention may be purified or substantially purified. Such a polypeptide in substantially purified form will generally form part of a preparation in which more than 90%, for example up to 95%, up to 98% or up to 99% of the peptide material in the preparation is that of a polypeptide or polypeptides according to the in- vention.

A nucleic acid sequence according to the invention may be included within a vector, suitably a replicable vector, for instance a replicable expression vector.

A replicable expression vector comprises an origin of replication so that the vector can be replicated in a host cell such as a bacterial host cell. A suitable vector will also typically comprise the following elements, usually in a 5' to 3' arrangement: a promoter for directing expression of the nucleic acid sequence and optionally a regulator of the promoter, a translational start codon and a nucleic acid sequence according to the invention encoding a polypeptide having the ability to either bind to GLI-1 or SUFUH. A non-replicable vector lacks a suitable origin at replication whilst a non-expression vector lacks an effective promoter.

The vector may also comprise one or more non-coding sequences 3' to the sequence encoding the polypeptide of the invention. These may be from H. sapiens (the organism from which the sequences of the invention are derived) or the host organism which is to be transformed with the vector or from another organism.

In an expression vector, the nucleic acid sequence of the invention is operably linked to a promoter capable of expressing the sequence. "Operably linked" refers to a juxtaposition wherein the promoter and the nucleic acid sequence encoding the polypeptide of the invention are in a relationship permitting the coding sequence to be expressed under the control of the promoter.

The vector may be of any type. The vector may be in linear or circular form. For example, the vector may be a plasmid vector. Those of skill in the art will be able to prepare suitable vectors comprising nucleic acid sequences encoding polypeptides of the invention starting with widely available vectors which will be modified by genetic engineering techniques such as those described by Sambrook et al (Molecular Cloning: A Laboratory Manual; 1989).

In an expression vector, any promoter capable of directing expression of a sequence of the invention in a host cell may be operably linked to the nucleic acid sequence of the invention.

Such vectors may be used to transfect or transform a host cell. Depending on the type of vector, they may be used as cloning vectors to amplify DNA sequences according to' the invention or to express this DNA in a host cell.

A further embodiment of the invention provides host cells harbouring vectors of the invention, i.e. cells transformed or transfected with vectors for the replication and/or expression of nucleic acid sequences according to the invention. The cells will be cho- sen to be compatible with the vector and may for example be bacterial cells. Transformed or transfected bacterial cells, for example E. coli cells, will be particularly useful for amplifying nucleic acid sequences of the invention as well as for expressing them as polypeptides.

The cells may be transformed or transfected by any suitable method, such as the methods described by Sambrook et al (Molecular cloning: A Laboratory Manual; 1989). For example, vectors comprising nucleic acid sequences according to the invention may be packaged into infectious viral particles, such as retroviral particles. The constructs may also be introduced, for example, by electroporation, calcium phosphate precipitation, biolistic methods or by contacting naked nucleic acid vectors with the cells in solution.

In the said nucleic acid vectors with which the host cells are transformed or transfected, the nucleic may be DNA or RNA, preferably DNA. The vectors with which the host cells are transformed or transfected may be of any suitable type. The vectors may be able to effect integration of nucleic acid sequences of the invention into the host cell genome or they may remain free in the cytoplasm. For example, the vector used for transformation may be an expression vector as defined herein.

The present invention also provides a process of producing polypeptides according to the invention. Such a process will typically comprise transforming or transfecting host cells with vectors comprising nucleic acid sequences according to the invention and expressing the nucleic acid sequence in these cells. In this case, the nucleic acid sequence will be operably linked to a promoter capable of directing its expression in the host cell. Desirably, such a promoter will be a "strong" promoter capable of achieving high levels of expression in the host cell. It may be desirable to overexpress the poly- paptide according to the invention in the host cell. Suitable host cells for this purpose include yeast cells and bacterial cells, for example E. coli. However, other expression systems can also be used, for example baculovirus systems in which the vector is a baculovirus having in its genome nucleic acid encoding a polypeptide of the invention and expression occurs when the baculovirus is allowed to infect insect cells.

The thus produced polypeptide of the invention may he recovered by any suitable method known in the art. Optionally, the thus recovered polypeptide may be purified by any suitable method, for example a method according to Sambrook et al (Molecular Cloning: A Laboratory Manual).

The polypeptides of the invention may also be synthesised chemically using standard techniques of peptide synthesis. For shorter polypeptides, chemical synthesis may be preferable to recombinant expression. In particular, peptides of up to 20 or up to 40 amino acid residues in length may desirably be synthesised chemically.

The present invention also provides antibodies to the polypeptides of the invention. These antibodies may be monoclonal or polyclonal. For the purposes of this invention, the term "antibody", includes fragments of whole antibodies which retain their binding activity for a target antigen. Such fragments include Fv, F(ab') and F(ab').sub.2 fragments, as well as single chain antibodies.

The antibodies may be produced by any method known in the art, such as the methods of Sambrook et al (Molecular Cloning: A Laboratory Manual; 1989). For example, they may be prepared by conventional hybridoma techniques or, in the case of modified antibodies or fragments, by recombinant DNA technology, for example by the ex- pression in a suitable host vector of a DNA construct encoding the modified antibody or fragment operably linked to a promoter. Suitable host cells include bacterial (for example E. coli), yeast, insect and mammalian cells. Polyclonal antibodies may also be prepared by conventional means which comprise inoculating a host animal, for example a rat or a rabbit, with a peptide of the invention and recovering immune serum.

The present invention also provides pharmaceutical compositions comprising polypeptides or antibodies of the invention. In a first embodiment, the compositions comprise antibodies, fragments and/or derivatives of SUFUH that interacts with GLI-1. In a second embodiment, the compositions comprises antibodies, fragments and/or derivati- ves of GLI-1 that binds to SUFUH. By administrating pharmaceutical compositions of the present invention to a patient in need thereof, it should be possible to affect the Sonic hedgehog (Shh) - Patched (Ptch) signalling pathway which could be useful in treatment of cancer.

The compositions of the invention may be administered to mammals including humans by any route appropriate. Suitable routes include topical application in the mouth, oral delivery by means of tablets or capsule and parenteral delivery, including subcutaneous, intramuscular, intravenous and intradermal delivery. Preferred routes of administration are injection, typically subcutaneous or intramuscular injection. The compositions of the invention may be administered to the subject alone or in a li- posome or associated with other delivery molecules. The effective dosage depends on many factors, such as whether a delivery molecule is used, the route of delivery and the size of the mammal being treated. Typical doses are from 0.1 to 100 mg of the po- lypeptide of the invention per day, for example 0.1 to 1 mg, and 1 to 5 mg, 5 to 10 mg and 10 to 100 mg per dose. Dosage schedules will vary according to, for example, the route of administration, the species of the recipient and the condition of the recipient.

While it is possible for polypeptides of the invention to be administered alone it is pre- ferable to present them as pharmaceutical formulations. The formulations of the present invention comprise at least one active ingredient, a polypeptide or an antibody of the invention, together with one or more acceptable carriers thereof and optionally other therapeutic ingredients. The carrier or carriers must be "acceptable" in the sense of being compatible with the other ingredients of the formulation and not deleterious to the recipients thereof, for example, liposomes.

Formulations suitable for parenteral administration include aqueous and non-aqueous sterile injection solutions which may contain anti-oxidants, buffers, bacteriostatis, bactericidal antibiotics and solutes which render the formulation isotonic with the blood of the intended recipient; and aqueous and non-aqueous sterile suspensions which may include suspending agents and thickening agents, and liposomes or other microparticulate systems which are designed to target the compound to blood components or one or more organs.

Of the possible formulations, sterile pyrogen-free aqueous and non-aqueous solutions are preferred. Also preferred are formulations in which the polypeptides of the invention are contained in liposomes. Injection solutions and suspensions may be prepared extemporaneously from sterile powders, granules and tablets of the kind previously described. It should be understood that in addition to the ingredients particularly mentioned above the formulations of this invention may include other agents conventional in the art having regard to the type of formulation in question.

We have described a human homologue of Drosophila Su(fu) and presented strong evidence that the function of this protein is highly conserved. Our results are consistent with, and provide a mechanistic explanation for, the genetic data obtained from studies of Drosophila that indicate an inhibitory role for Su(fu) in the Hedgehog signalling pathway [Ingham et al., EMBO J. 17:3505-3511 (1998)]. We have shown, for the first time, to our knowledge, that SUFUH inhibits GLI-mediated transcription and the cellular response to SHH, and our results implicate the nucleus as a new intracellular site of action for SUFUH.

Moreover, we have shown that active nuclear export is a central mechanism for exclu- sion of full-length GLI- 1 from the nucleus, offering a plausible explanation for the observation in Drosophila that Ci remains cytoplasmic in cells devoid of either Su(fu) or Cos2 [Monnier et al., Curr. Biol. 8:583-586 (1998); Ingham et al., EMBO J. 17:3505- 3511 (1998)]. Interestingly, it has been proposed [Methot et al., Cell 96: 819-831 (1999); Ohlmeyer et al., Nature 396:749-753 (1998)] that a separate, uncharacterized activation step is required to convert full-length Ci to a transcriptional activator and it is tempting to speculate that this step involves regulation of nuclear-cytoplasmic shuttling. Further support for the biological relevance of such a regulated event comes from studies showing that GLI proteins show different subcellular distributions in different cell types [Shin et al., Proc. Natl. Acad. Sci. USA 96:2880-2884 (1999); Ruiz i Altaba et al., Development 126:3205-3216 (1999)], and that for all analysed artificially truncated or proteolytically processed Ci or GLI 1-3 proteins, subcellular distribution is consistent with inclusion or exclusion of predicted NES motifs [Shin et al., Proc. Natl. Acad. Sci. USA 96:2880-2884 (1999); Ruiz i Altaba et al., Development 126:3205- 3216 (1999); Dai et al., J. Biol. Chem. 274:8143-8152 (1999); Aza-Blanc et al, Cell 89:1043-1053 (1997)] (Fig. 6b). It should be noted that in the two cases in which loca- lization has been determined in vivo without vas overexpression (that is, for full-length

Ci in the fly [Aza-Blanc et al., Cell 89:1043-1053 (1997)] and for GLI-1 in human BCCs [Dahmane et al., Nature 389:876-881], the localization has been primarily cytoplasmic. A likely scenario, taking into account the role of GLI-1 as a transforming oncoprotein and a cell-fate determinant that responds to graded levels of SHH signalling, involves a finely tuned balance of activated nuclear entry and nuclear export.

The retention of GLI-1 in the cytoplasm by SUFUH when nuclear export is compromised, and the similar SUFUH-mediated retention in the cytoplasm of an otherwise constitutively nuclear truncated GLI-1 variant that lacks the NES, indicates that SUFUH could block nuclear enty of GLI-1, possibly by masking a nuclearlocalization signal, and thereby inhibit transcriptional activation of target genes (Fig. 8). Consistent with this idea, a truncated SUFUH variant unable to repress GLI 1 -induced transcriptional activation is also unable to modify the subcellular localization of GLI-1. Whether or not binding of SUFUH to GLI- 1 on DNA, or elsewhere in the nuclear compartment, actually acts to repress of block activation of transcription, alone or in combination with cytoplasmic retention of GLI-1, remains an interesting question for future studies. The expression of Sufu in cells next to Shh- or Ihh-producing cells during mouse and human embryogenesis, coupled with the ability of Sufu to inhibit Gli- mediated transcriptional activation, indicates that an important function of Sufu may be to act in an intracellular negative feedback mechanism and to impose thresholds on the responsiveness of cells to Shh and Ihh. A similar role for D-Axin has been proposed as regards Wingless signalling in Drosophila [Hamada et al., Science 283:1739-1742 (1999)].

Finally, GLI-1 is an oncoprotein that is able to transform cells in vitro in corporation with El A [Ruppert et al., Mol. Cell Biol. 11:1724-1728 (1991)]; overexpression of GLI-1 induces epidermal tumours in frogs and consistent overexpression of GLI-1 is observed in BCCs [Dahmane et al., Nature 389:876-881 (1997)]. This possibility is now being investigated in both human and murine systems. We have also found two organ-specific expression variants of SU(FU) which might provide a further regulatory mechanism. Both these variants lack the ability of binding GLI1 thereby uncoupling the transcription factor from one of its cytoplasmic retention mechanisms. The sources of both variants, testis and leukocytes, are proliferating tissues. Since in both cases the variants contributed only partially to the total amount of SU(FU) expressed, it might be a particular subpopulation of cells which expresses them. Since we have found that SU(FU) is expressed in all adult tissues with comparable abundance, it is possible that in the adult the activity of SU(FU) is not primarily regulated by the level of SU(FU) transcription but rather by the exon composition. Another transcriptional variation has been described [Stone et al., J. Cell Science 112:4437-4448 (1999)] which is expressed dominantly in testis and has the N-terminal 432 amino acids in common with the standard SU(FU) but then a STOP codon after one further altered amino acid. According to our data this variant comprises exons 1- 10.

In the embryo, Su(fu) was expressed with tissue-specific preference. As stated above, the sites of expression grossly congrue with the sites of expression of other members of the hedgehog signalling pathway, in particular with the expression of Gli3. However, in detail this overlap is not found in all cases, and it is not always complete. In some cases the expression of Su(fu) is more restricted than the expression of Gli 1. For example in mouse skin, until dpc 15.5, Su(fu) is expressed only in the vibrissae follicles stronger than the general background while GUI is expressed in vibrissae and pelagic hair follicles from their onset [St-Jaques et al., Curr. Biol. 8:1058-1068 (1998)]. In other cases the expression of Gli 1 is more restricted than the expression of Su(fiι). For example, GUI [Hynes et al., Neuron 19:15-26 (1997)] and Gli2 [Matise et al., Development 125:2759-2770 (1998)] were reported to be expressed only in parts of the brain and the neural tube. In contrast, Su(fu) was expressed in the entire brain and spinal cord. Furthermore, we found examples where predominant Su(fu) and Gli expres- sion sites are adjacent to each other (GUI was shown to be expressed in the papillal epidermis of the tongue [Hall et al., J. Comp. Neurol. 406:143-155 (1999)], while

Su(fu) was expressed in the tongue mesenchyme). Since GUI activity is negatively controlled by Su(fu) and GUI expression is a positive target of GUI activity the expression of GUI could be modulated by the expression of Su(fu). By inducing the transc- ription both of Ptch 1 and of itself, GUI triggers a negative and a positive feedback cycle. We do not know which of these activities would dominate under physiological conditions. Therefore we cannot anticipate that Su(fu) should always have an inhibitory effect on hedgehog signalling. Furthermore, it is possible that GUI could influence the expression pattern of the other types of Gli at neighbouring sites. One such ex- ample was observed in the digits where GUI and Su(fu) are expressed in the condensing mesenchyme while Gli2 and Gli3 are expressed in the interdigital mesenchyme [Mo et al., Development 124:113-123 (1997)]. However, only in Gli2 and Gli3 mutant mice the development of the long and short bones of the limbs is affected [Mo et al., Development 124:113-123 (1997)] while in GUI mutant mice (deletion of the zinc finger region) development appears to be normal [Park et al., Development 127: 1593- 1605 (2000)].

We have investigated the possibility that the mammalian SU(FU) and Su(fu) would be the genes responsible for the SHFM3 and Dae disorders. During the course of these studies the Dae mutation was identified as an alteration in the previously unknown F- Box/WD40 gene dactylin [Sidow et al., Nat. Genet. 23: 104-107 (1999)]. At the same time, the human homolog for dactylin was also published [Ianakev et al., Biochem. Biophys. Res. Commun. 261:64-70 (1999)]. For the human Dactylin, however, a link to SHFM3 is not yet demonstrated.

A role of SU(FU) in SHFM3 still appears as an intriguing possibility not only because of a possible functional relation of dactylin and Su(fu) and because of the overlapping of their chromosomal loci but also because SU(FU) and other members of the hedgehog signalling pathway are expressed at the ossification zone of the digits in mice (this report) and humans. Moreover, other members of the hedgehog signalling pathway were shown to be involved in anomalies of finger development. Preaxial polydactyly in mouse has been shown to be caused by ectopic expression of Indian hedgehog in the limbs which results in ectopic GUI expression [Yang et al., Development 125:3123- 3132 (1998)]. A loss of function-mutation in GLI3, associated with an increased Sonic hedgehog expression, has been shown to cause Postaxial Polydactyly Type Al

(41)[Radhakrishna et al., Am. J. Hum. Genet. 56:597-604 (1997)] and Extratoes (Xt), a similar disorder, in the mouse [Hui et al., Nat. Genet. 3:241-246 (1993)]. Other hereditary syndroms with a polydactyly phenotype in which GLI3 is mutated are Greigs cephalopolysyndactyly [Vortkamp et al., Nature 352:539-540 (1991)] and Pallister- Hall syndrome [Jang et al., J. Med. Genet. 34:441-446 (1997)]. Thus, increased hedgehog signalling appears to correlate well with poly- and syndactyly. A decrease in digit number may conversely be associated with a repression of hedgehog signalling. Su(fu) is an antagonist of both GUI and Gli3 and gain of function mutations in Su(fu) would further reduce intracellular transduction of a hedgehog signal. As discussed by Ozen et al., for all other known genes which are located in the same region (FGF8, PAX2,

ZNF32, FGFR2, HOX11, HMX2, WNT8b, LDB1, LBX1, and PITX1) identity with the SHFM3 locus is unlikely for various reasons [Ozen et al., Am. J. Hum. Genet. 64:1646-1654 (1999)].

We have discovered a novel phenotype in the Dae mice which is the delay of ossification of that part of the skull that emerges under the membrane of the developing brain in direct vicinity to the neural tissue and which is similar to the human Adams-Oliver syndrome [Chitayat et al., Am. J. Med. Genet. 44:562-566 (1992)]. Ossification here begins in wild-type mice at dpc 16.5 and has at dpc 17.5 already further progressed than in the Dae homozygous mouse at dpn 1 [Kaufman, The atlas of mouse development, Acad. Press Ltd. London, GB (1992)]. At dpc 15.5, the latest stage that we have investigated, Su(fu) was expressed strongly in the brain as it was during the entire period, but very weakly in the skin surrounding the brain. It has been reported that Shh and Ptchl are expressed in the osteogenic fronts of the calvarial bones [Kim et al., Development 125: 1241-1251 (1998)]. Expression was confined to the osteogenic fronts and later to the sutures between the parietal and between the parietal and the occipital bones, which are the bones affected in the Dae mouse, but not in the coronar suture which lays between affected and unaffected bones. This cong- rues very well with the assumption that a component of the hedgehog signalling pathway might be involved in causing the Dae phenotype. However, since no such calvarial phenotype has been reported for SHFM3 patients, SHFM3 and Dae may be similar but not identical, or the lack of this phenotype in humans may reflect species dif- ferences.

The invention will be further described with reference to the enclosed figures and tables, in which:

Table 1 shows exon-intron boundaries of the human SU(FU) and mouse Su(fu) genes;

Figure 1 presents the predicated amino-acid sequence (SEQ.ID.NO.l) encoded by the

SUFUH gene and its expression in adult human tissues and mouse embryos. a, The SUFUH amino-acid sequence (lower lines) is 40% identical to the Drosophila melanogaster Su(fu) sequence (SEQ.ID.N0.9)(upper lines). In the human sequence, the N-terminal part of the protein consists of a series of 20 short (A, G, P) and short hydroxylated (S, T) amino acids only (amino acids 6-25), including a duplicated PGPTAPPA motif (underlined), which is not found in the fly sequence. The region between amino acids 31 and 151 of SUFUH is highly conserved from fly to humans, with the C-terminal region exhibiting somewhat less similarity. A potential PEST motif in the Drosophila sequence (amino acids 308-327) [Pham et al., Genetics 140:587- 598 (1995)] is not conserved in the human sequence. Vertical lines between the two sequences indicate identical amino acids; two dots indicate small and one dot indicates large evolutionary distance between the amino acids. b, A human multiple-tissue northern blot was probed with a fragment corresponding to base pairs, 1,182-1,630 of the human SUFUH cDNA. The sizes of the detected mRNAs are ~ 5.5-6 and 2.8 kilobases (kb), respectively. An extra hybridizing band of 4.4 kb was observed in the heart only. 1, Heart; 2, brain; 3, placenta; 4, lung; 5, liver; 6, skeletal muscle; 7, kidney; 8, pancreas. c, Sufu expression in the mouse embryo. Whole-mount in situ hybridization of an NMRI mouse embryo with a murine Sufu cDNA probe at d.p.c. 12.5; lateral view. The interfollicular tissue of the vibrissae field, the vertebral column and the interdigital mesenchyme are stained;

Figure 2 shows that SUFUH and PTCH1 are both highly expressed in osteoblasts in the perichondrium during human embryonic bone formation. Expression of the

SUFUH and PTCH1 mRNAs co-localize in osteoblasts in the perichondrium in the forming finger bones of a 12-week-old human embryo, as visualized by bright-field (a- d) and dark-field (e) photomicrography, a, An overview of the embryonic fingers counterstained with haematoxylin and eosin, b, A section of a finger hybridized with an antisense SUFUH mRNA probe, showing a strong autoradiographic signal (black grains) in the osteoblasts in the perichondrium (arrow) next to the chondrocytes (c). c, A serial section hybridized with a sense SUFUH probe showed no signal, d, e, Paired bright- and dark- field photomicrographs showing abdundant PTCH1 mRNA signal in the same cellular compartment as shown in b. Scale bar in a represents 230 μm and in b-e represents 24 μm;

Figure 3 discloses that SUFUH inhibits transcriptional and biological activity of GLI- 1. a-c, 293 or d-f, C3H10T1/2 cells were transiently transfected with a reporter construct and/or with indicated expression constructs, a, The reporter was a 4,3-kb frag- ment of the PTCH1 promoter connected to the luciferase reporter gene. Expression of SUFUH alone has no effect on reporter gene activity whereas transfection of GLI-1 results in a five- to sixfold increase in luciferase activity. This activation can be inhibited by co-transfection of SUFUH in a dose-dependent manner. Values shown in parentheses are concentrations of SLT LTH-expressing plasmid. b, Similar to the experi- ment shown in a, except that the reporter construct was a synthetic enhancer consisting of 12 copies of the GLI-binding consensus site [Unden et al., Cancer Res. 57:2336-

2340 (1997)]linked to the thymidine kinase (TK) basal promoter and the luciferase reporter gene. The results are qualitatively the same as those obtained with the PTCH1 promoter except that GLI-1 activates this construct much more (200-300-fold). A cor- rected haemagglutinin-tagged GLI- 1 construct (SEQ.ID.NO.11 )(WTGLI 1 ) activates the reporter even more strongly and this activation can be suppressed by coexpression of SUFUH but not by a construct containing the 385 N-terminal amino acids of SUFUH (SUFUHΔ)(SEQ.ID.NO10). c, SUFUH does not nonspecifically suppress transactivation by an unrelated transcription factor. A reporter containing two copies of a glucocorticoid response element connected to the TK basic promoter and luciferase reporter gene was used. The reporter is activated by a constitutively active glucocorticoid receptor (τlc) in which the DNA-binding domain is directly fused to the transactivation domain, d, Osteogenic differentiation of C3H10T1/2 cells transfected with GLI- 1, detected by alkaline phosphatase staining, e, Example of a field of alkaline- phosphatase-negative cells (in this case transfected with both GLI-1 and SUFUH). f, Quantification of the differentiation assays, based on three independent experiments. The number of alkaline-phosphatase-positive cells following the transfection of different expression constructs is indicated. GLI-1 strongly induces osteogenic differentiation; this can be inhibited by full-length but not by truncated SUFUH. The N-terminal fragment of SHH also induces differentiation of C3H10T1/2 cells; this can be blocked by coexpression of SUFUH. The y-axis denotes the number of positive cells per transfected 35-mm well;

Figure 4 shows that SUFUH and GLI1 interact physically. The expression constructs used in this experiment were epitope-tagged (SUFUH with Myc and GLI- 1 with hae- magglutinin (HA)) and transfections to 293 cells were done as indicated, a, Full-length GLI-1 and SUFUH interact. The anti-Myc antibody does not precipitate HA-tagged GLI-1 when only HA-GLI1 (SEQ.ID.NO.l l)(lane 1) or Myc-SUFUH (SEQ.ID.NO.12)(lane 2) is expressed in the cells. However, HA-GLI1 is already pre- cipitated by the anti-Myc antibody in cells co-transfected with bort constructs (lane 3). Conversely, Myc-SUFUH is not present in anti-HA precipitates from cells transfected with HA-GLIl (lane 4) or Myc-SUFUH (lane 5) alone but is present when cells are co- transfected with both constructs (lane 6). IP, immunoprecipitation. Relative molecular mass standards (in thousands) are indicated to the sides of the gels, b, A C-terminally truncated GLI- 1 construct containing the 407 N-terminal amino acids (GLIl(N407)(SEQ.ID.NO.13)) interacts with SUFUH, HA- GLIl (N407)(SEQ.ID.NO.13) co-precipitates with Myc-SUFUH (SEQ.ID.NO.12) only when they are co-expressed (lane 2). c, SUFUHΔ (SEQ.ID.NO.10) does not interact with GLI-1. Lane 1, an anti-Myc immunoprecipitation from cells expressing Myc- SUFUHΔ (SEQ.ID.NO.14) and HA-GLIl (SEQ.ID.NO.11); lane 2, lysate containing both proteins, d, Both SUFUH (SEQ.ID.NO.l) and SUFUHΔ (SEQ.ID.NO.10) are produced at comparable levels and with the expected relative molecular masses. Ly- sates from cells transfected with Myc-SUFUH (SEQ.ID.NO.12) or Myc-SUFUHΔ (SEQ.ID.NO.14) were blotted with anti-Myc antibodies, e, SUFUH does not associate nonspecifically with expression constructs for IκBα or anti-Myc antibodies as indicated;

Figure 5 reveals that SUFUH but not SUFUHΔ retains nuclear GLI-1 variants in the cytoplasm. Subcellular localization of SUFUH and GLI-1 constructs was determined after transfection of epitope-tagged constructs into 293 cells. In each case, 450 ng of specific plasmid DNA was used, resulting in a 1 :1 ratio in cotransfections. This ratio was shown previously (Fig. 3 a, b) to cause strong transcriptional inhibition, a, SUFUH (SEQ.ID.NO.l) localizes to the cytoplasm and the nucleus, as determined by immuno- fluorescent microscopy, b, C-terminally truncated SUFUHΔ (SEQ.ID.NO.10) is also present in both cytoplasm and nucleus, c, EGFP-tagged GLI-1 shows cytoplasmic localization, d, EGFP-tagged GLI1(N407) is constitutively nuclear, e, f, EGFP- GLI1(N407) is partially retained in the cytoplasm by coexpression of SUFUH(e)(SEQ.ID.N0.1) but not SUFUHΔ(f) (SEQ.ID.NO.10). SUFUH was detected by indirect immuno fluorescence; EGFP-GLI1 constructs were visualized di- rectly (green); nuclei were stained with Hoechst dye (blue). The overlap of the two colours appears turquoise;

Figure 6 shows that GLI-1 accumulates in the nucleus when nuclear export is inhibited. a, A deletion construct containing the 525 N-terminal amino acids of GLI-1

(SEQ.ID.NO.15) is located in both the nucleus and the cytoplasm, b, The sequence of the putative NES in GLI-1 is conserved in all vertebrate species; the position of the first leucine in the protein sequence is indicated. Ci contains a putative NES in the same region (starting at amino acid 774) as well as two extra ones, at least one of which is conserved in vertebrate GLI-2 and GLI-3. c, A mutated GLI1(N525)

(SEQ.ID.NO.16) protein, with leucines 495 and 501 mutated to alanines, localizes exclusively to the nucleus, d, Full-length GLI-1 (SEQ.ID.NO.5) translocates to the nucleus when nuclear export is inhibited by the addition of leptomycin B (LMB, 5 nM). e, Coexpression of SUFUH inhibits nuclear accumulation of GLI-1 in the presence of LMB. f, The C-terminally truncated SUFUHΔ (SEQ.ID.NO.10) construct has no effect on GLI-1 nuclear accumulation. The EGFP-GLI1 proteins appear green; the nuclei are stained blue with Hoechst dye; and the overlap between the two colours appears turquoise. For transfections, 405 ng specific plasmid DNA was used, resulting in a 1 : 1 ratio in co-transfections;

Figure 7 discloses that SUFUH can form a complex with constitutively nuclear GLI- 1 variants bound to DNA. a, Nuclear GLI1(N407) (SEQ.ID.NO.6) expressed in 293 cells forms a specific complex with the consensus (labelled) GLI-binding oligonucleotide (lane 1); addition of excess unlabelled GLI oligonucleotide outcompetes the labelled oligonucleotide (lane 2). Coexpression of SUFUH results in complexes that are more retarded (lane 3); these complexes, by supershifting with anti-tag antibodies (anti-Myc for SUFUH and anti-EGFP for GLI1(N407), were shown to contain both GLI1(N407) (lane 4) and SUFUH (lane 5). Analysis of nuclear extracts after transfection of SUFUH alone revealed no retarded complex (late 6). b, Coexpression of SUFUH causes a re- distribution of GLI 1 (N407) from the nuclear pellet to the nuclear exract and appear- ance of GLI1(N407) in the cytoplasmic fraction. The cytoplasmic and nuclear extracts were prepared in parallel using cells from three plates for each condition as described [Ausubel et al., Current Protocols in Molecular Biology (John Wiley, New York 1987)], using 1,2 M KCl for nuclear extraction. The remaining nuclear pellet was solubilized in SDS sample buffer and GLI1(N407) was detected by western blotting using anti-EGFP antibodies (Clontech). 15 μg protein from cytoplasmic and nuclear extracts or 5 μl solubilized nuclear pellet was loaded in each well. This represents 3%, 15% and 2,5% of the total cytoplasmic extract, nuclear extract and nuclear pellet, respectively. Relative molecular mass standards (in thousands) are indicated at the right;

Figure 8 discloses a model for the role of SUFUH in the nuclear-cytoplasmic shuttling of GLI 1. GLI-1 is normally kept in the cytoplasm by active nuclear export mediated by its NES. We propose that SUFUH acts to inhibit GLI 1 -mediated transcriptional activation and thereby modulate cellular responsiveness to SHH/IHH. This inhibition can oc- cur by preventing nuclear access of GLI- 1 , either by anchoring GLI- 1 in the cytoplasm or by masking its putative NLS. SUFUH may also interfere with GLI-1 function in the nucleus by binding to GLI-1 on the DNA;

Figure 9 relates to organization and structure of the mouse Suppressor of fused (Su(fu)) gene locus and cDNAs. A) The mouse Su(fu) gene as assembled from mapping and partial sequencing of BAC clones 17985 and 17986. Boxes indicate position of exons. Filled and unfilled boxes indicate coding and non-coding regions, respectively. Arrows indicate direction of the transcriptional unit for Su(fu) and the linked upstream gene for actin-related protein 1 a-isoform (Arpϊ) and the downstream gene^ør ADP-ribosyl- like protein 3 (Arl3). Only the exon containing the ATG translational start of Arpl and the exon after the ATG-containing exon of Arl3 have been mapped and sequenced. Double arrowheads indicate start and end of BAC clones. Restriction enzyme sites of a selected set that have been sequenced and/or mapped are shown. Not all sites of the indicated enzymes have been mapped. E, EcoRI; B, BamHI; H, Hindlll; K, Kpnl; Xb, Xbal; Sm, Smal. B) Structure of the fully sequenced mouse Su(fu) EST cDNAs 1195307 (mammary gland), 513730 (testis), 1224813 (heart), and 963990 (mammary gland), from the IMAGE consortium collection. Numbers 1 through 12 indicate the Su(fu) exons drawn to scale for a full-length mRNA transcript. ATG and TAG indicate the translational start and stop, respectively. The thick line represent the structural composition of each of the four EST clones positioned relative to the hypotetical full- length transcript;

Figure 10 shows transcriptional variations of human SU(FU). A. RT-PCR of a SU(FU)-fragment containing the exons 5 to 9 from various tissues. Tt and Lk indicate - the respective variations. The other minor bands in lane 3 and lane 4 represent PCR products which are not related to SU(FU). B. Same as in A, after EcoNI digestion.The lower band represents the EcoNI susceptible SU(FU)-XL variant (SEQ.ID.NO.17). Since the mobility of the EcoNI-digested fragment equals the mobility of the SU(FU)- Lk band (SEQ.ID.NO.18), lane 8 was omitted here. Lane 1, thymus; 2, prostate; 3, spleen; 4, testis; 5, colon; 6, ovary; 7, small intestine; 8, peripheral leukocytes. C. Sequence of the intron 8a in SU(FU)-Tt (SEQ.ID.NO.19). The letters in italics signify the flanking exons 8 and 9;

Figure 11 discloses that Myc-SU(FU)-Lk (SEQ.ID.NO.21) and myc-SU(FU)-Tt (SEQ.ID.NO.22) cannot coprecipitate HA-GLIl (SEQ.ID.NO.11). Each 10 μg of HA- GLIl (SEQ.ID.NO.i l) and myc-SU(FU)-Lk (SEQ.ID.N0.21)(lane 1), myc-SU(FU)- Tt (SEQ.ID.N0.22)(lane 2), myc-SU(FU)-XL (SEQ.ID.NO.20)(lane 3) and myc- SU(FU) (SEQ.ID.NO.12)(lane 4) were transfected into 293 cells. Lysates were precipitated with anti-myc (A) and anti-HA (B) antibodies and blotted with anti-myc antise- rum. The numbers to the left indicate the positions of the molecular weight marker

(kD) for each blot. The arrowhead next to blot B indicates the position of the copreci- pitated myc-SU(FU)-XL (SEQ.ID.NO.20) and myc-SU(FU) (SEQ.ID.NO.12);

Figure 12 shows whole mount in situ hybridization of mouse embryos with a Su(fiι) antisense probe. A. dpc 8.5. Somites and the neural tube hybridize with the probe. B. dpc 9.5 lateral view, f, forelimb; h, heart; m, mandibular arch, s, somites; *, the mesenchyme surrounding the prospective mouth cavity. C. dpc 9.5, lateral view, prior to clearance with glycerol. nt, neural tube. D. dpc 10.5, dorsal view, b, brain vesicles. E. dpc 10.5, dorsal view of the tail after removal of the skin, drg, dorsal root ganglia. F. dpc 9.5, sagittal view. Besides the heart, the dorsal aorta (da) and the intersegmentary arteries (is) are stained. G. dpc 13.5, sagittal view. Most prominent is the staining of the spinal cord (sc) and the brain (b). Mandible (m), tongue (t), genital tubercle (g), and the mesenchyme close to the vertebrae (v) are also stained. H-J Staining of the mesenchyme adjcent to the ossification zone of the digits in the hindlimb at dpc 13.5 (H), 14.5 (I), 15.5 (J). The lines point to the Su(fu) expressing areas which moves distally with time. K. dpc 14.5 sagittal section of the snout. At this point, mainly the tongue (t) expresses Su(fu). Less staining was detected at the distalmost part of the snout (arrow) and of the mandible. L. dpc 15.5, sagittal view of the dorsal part of the thorax. The lines point to the Su(fu) expressing tracheoles; and

Figure 13 relates to Alizarin- Alcian Blue staining of the skull of a dpn 0 Dac/Dac (A) and wild type (B) mouse. While the occipital (O) and the frontal bones (F) are ossified in both cases (red staining), the supraoccipital (S), intraparietal (I), and parietal (P) bones are ossified in wildtype only, while in the Dac/Dac mouse there is poorly segmen- ted cartilagenous mass (blue). Note also that the curvature of the skull is much flatter in the Dac/Dac mouse.

The invention will now be further disclosed in the following experimental part.

Experimental part

General Methods relating to Examples 1 - 8:

Cloning of mammalian SUFUH ESTs for the human and mouse Su(fu) homologues were obtained from Genome Systems. The nucleotide sequence of the ESTs were determined and a full-length cDNA of the human Su(fu) homologue, referred to as SUFUH, was obtained by RACE, essentially as described [Zaphiropolous et al., DNA Cell. Biol. 15:1049-1056 (1996)].

Northern blot hybridization

Multiple tissue northern blots with human RNAs were obtained from Clontech. Fragments of the human SUFUH clone (SEQ.ID.NO.23) corresponding to base pairs 1,182- 1,630 and to the3' untranslated region were labelled with 32p using the Megaprime

DNA-labelling kit (Amersham) and hybridized to these blots. After stripping the blots with 1% SDS, the amount and quality of the RNA was monitored with a probe specific for β-actin.

In situ hybridization

Digoxin-labelled riboprobes for murine Sufu were prepared using the DIG RNA- labelling kit (Boehringer Mannheim) using the mouse Sufu EST as a template and Sp6 and T7 poymerase to make sense and antisense probes, respectively. Timely mated NMRI mice were killed and the embryos collected, and, after fixation with 4% para- formaldehyde, whole-mount in situ hybridization was done according to standard procedures [Wilkinson, In Situ Hybridization: A Practical Approach (IRL, Oxford 1992). The hybridizing probe was detected using the DIG nucleic acid detection kit (Boehringer Mannheim). Unspecific binding of the probe was momtored by comparison of samples hybridized with sense and antisense probes.

Radioactive in situ hybridization was done as described [Unden et al., Cancer Res. 57:2336-2340 (1997)]. Briefly, a human SUFUH cDNA clone (SEQ.ID.NO.23) containing the complete coding region and a human PTCH1 cDNA fragment (base pairs 190-628) were used to generate 35s-labelled RNA probes. Sections were treated with proteinase K and washed in 0,1 M triethanolamine buffer containing 0,25% acetic anhydride. Sections were hybridized with 2,5 x 10^ c.p.m. of each of the labelled anti- sense and sense probes at 55°C. Autoradiographic exposure lasted for 2 days for SUFUH and 2 weeks for PTCH1. Formalin- fixed sections from a 12-week-old human embryo were obtained from the Department of Pediatric Pathology, Karolinska Hospital, Sweden.

Epitope-tagged expression constructs

Wild-type GLI-1 was generated by PCR mutagenesis using oligonucleotides leading to a sequence change of G to A position 2,876 of the GLI-1 cDNA (SEQ.ID.NO.27) [Hy- nes, Neuron 19:15-26 (1997)]. Using the full-length cDNA clones encoding SUFUH, a variant of SUFUH truncated after amino acid 385 (SUFUHΔ)(SEQ.ID.NO.10) and human wild-type GLI-1 (SEQ.ID.NO.5) as templates, we generated plasmids coding for SUFUH (SEQ.ID.NO.1 ) and SUFUHΔ (SEQ.ID.NO.10) with an N-terminal Myc- epitope tag [Evan et al., Mol. Cell. Biol. 5:3610-3616 (1985)], and plasmid encoding human GLI-1 with an N-terminal haemagglutinin tag (SEQ.ID.NO.l 1) [Pate, Gene 114:285-288 (1992)], after ligation into the CMV-5 expression vector. GLI-1 was subcloned into the EGFP-C3 vector (Clontech) for localization studies. GLIl(N407)(SEQ.ID.NO.6) was constructed by digesting the GLI-1 expression vector with c l and M and GLI1(N525)(SEQ.ID.N0.15) by digesting the GLI-1 expression vector with Kpnl andXbαl restriction enzymes followed by religation after blunt- ending. Leucines 495 and 501 were mutated to alanine in GLI(N525)(SEQ.ID.N0.16) by PCR-mediated mutagenesis (the first two nucleotides in codons 495 and 501 were mutated from CT to GC, resulting in the indicated amino-acid change). All PCR- general constructs were verified by sequencing.

Transient transfections Human 293 cells (transformed kidney epithelial cell line) and C3H 10T1/2 cells (murine fibroblast cell line) were obtained from ATCC and grown according to the supplier's specifications. Cells were passaged to plates the day before the experiments and transfections were carried out with DNA complexed to the Superfect transfection rea- gent (Qiagen) according to the manufacturer's instruction. Expression was analysed 24 h, and differentiation 96 h, after transfections.

Reporter-gene and differentiation assays

The PTCH-luc construct [Shin et al., Proc. Natl. Acad. Sci. USA 96:2880-2884 (1999)] and the 2Gluc-RE-TKO-luc construct activable by a constitutively active glucocorti- coid receptor, τlc [Almlof et al., Mol. Cell. Biol. 17:934-945 (1997)] have been described. The 12GLI-RE-TKO-luc construct was generated by ligating annealed oligonucleotides corresponding to GLI consensus sequence into the TKO-luc vector. Luciferase assays were carried out as described construct [Shin et al., Proc. Natl. Acad. Sci. USA 96:2880-2884 (1999)]. The results from at least four experiments from two independent transfections were compiled. The differentiation assays were done as described [Murone et al., curr. Biol. 9:76-84 (1999)]. Briefly, 4 days after transfection of the indicated constructs, cells were fixed and stained for alkaline phosphatase activity with X-phos and pNBT (SIGMA) according to the manufacturer's instructions. The positive cells in each well were counted and at least three independent experiments were compiled for quantitative results.

Immunoprecipitations and western blotting

Lysates were prepared in immunoprecipitation buffer (PBS, 1% Nonidet-P40, protease inhibitors); immunoprecipitations were performed using 0,4 μg rabbit anti- haemagglutinin antibody and 20 μl packed protein G plus protein A agarose or 50 μl agarose-conjugated anti-Myc beads and separated along with the lysates on 7,5% SDS- PAGE, transferred to poly(vinylidene)fluoride (PVDF) membrane (Millipore), and in- cubated with the indicated primary (Santa Cruz Biotechnology) and corresponding secondary horseradish-peroxidase-conjugated antibodies (Jackson Immunoresearch). Following washes, enhanced chemiluminescence detection (Pierce) was performed according to the manufacturer's instructions.

Localization studies

239 cells were transfected with appropriate expression constructs and, where indicated, 5 mM leptomycin B was added to the cells for the last 6 h. The cells were fixed for 10 - min in 2% formaldehyde, 0,2% glutaraldehyde and PBS, then permeabilized in PBSTX (PBS, 0,05% Triton-XlOO). Myc-SUFUH (SEQ.ID.NO.12) was detected by indirect immunofluorescent staining whereas EGFP constructs were detected directly by microscopy. Nuclei were stained with the Hoechst dye (1 μg ml^{- 1}). At least 50 transfected cells were inspected in every case, all of which showed a similar staining pattern. For fluorescent microscopy we used a Leitz DMR confocal microscope (Leica); the images were captured with an ORBIS, cooled, chargecoupled-device camera (Spectrasource Instruments) and analysed using Slidebook software (Intelligent Imaging Innovations).

Subcellular fractionation and EMSAs

Nuclear and cytoplasmic extracts were prepared from cells transfected with either SUFUH or GLIl(n407) or both constructs, using standard protocols [Ausubel et al., Current Protocols in Molecular Biology (John Wiley, New York 1987)]. The nuclear extracts were incubated with end-labelled GLI-RE double-stranded oligonucleotides [Pham et al., Genetics 140:587-598 (1995)] in the presence or absence of excess unla- belled oligonucleotides or anti-tag antibodies. The extracts were prepared and the complexes formed were resolved and detected as described [Ausubel et al., Cinrent Protocols in Molecular Biology (John Wiley, New York 1987)].

Example 1 : Cloning of human SUFUH. By searching the expressed sequence tag (EST) database for genes homologous to the Drosophila melanogaster Su(fu) ? zm. et al., Genetics 140:587-598 (1995)], we found two clones encoding partial human and murine homologoues. Starting with the human EST, we cloned the complete coding region of the gene named SUFUH by rapid amplification of cDNA ends (RACE) technology. The resulting cDNA encompasses 2,239 base pairs and encodes a protein of 484 amino acids (SEQ.ID.NO.l). The SUFUH amino-acid sequence shows no striking homology to any other protein described so far, and shares 40% identity and 61% similarity with the Drosophila sequence (Fig. 1 a)(SEQ.ID.N0.9). The region between amino acids 31 and 124 of SUFUH is highly conserved, being 66% identical to the Drosophila sequence, whereas the remaining parts have an identity of 32%. The partial murine Sufu amino-acid sequence is 98% identical to the human sequence.

Example 2: Expression of SUFUH in adult and embryonic tissues.

SUFUH (SEQ.ID.NO.l) was expressed in all adult human tissues tested (Fig. lb). We detected two messenger RNAs, a minor, 2.8-kilobase, band and a major, 5.5-6- kilobase, band. Similar hybridization patterns were seen with probes corresponding to the coding region and to the 3 '-untranslated region, indicating that the two mRNAs may be processing variants. The use of an alternative polyadenylation signals is also suggested by the identification of other mouse ESTs containing additional 3'- untranslated regions (our unpublished observations).

To test whether vertebrate Sufu is expressed in a pattern consistent with a potential role in mediating Shh signalling during embryogenesis, we used whole-mount in situ hybridization to analyse Sufu expression in mouse embryos at days 8.5 to 15.5 post- coitum (d.p.c). Throughout the entire period we observed signals in the neural tube and, at the later stages, in its derivatives, brain and spinal cord (data not shown). The somites expressed Sufu at all stages; the vibrissae field stained positively for Sufu from d.p.c. 12.5 and onwards, with the vibrissae themselves being spared. The Sufu expression pattern during limb-bud development appeared to be separated into two distinct phases, with strong homogeneous staining all over the limb buds being observed from their emergence at d.p.c. 9.5, whereas at d.p.c. 12.5 only the interdigital mesenchyme of the limbs stained positively (Fig. lc). This expression pattern partially overlaps with the expression of Ptch and the Ci homologues Gli 1-3, and is compatible with a conserved role for Sufu in Shh signalling.

To substantiate this observation in more detail and in the human system, we analysed the expression of SUFUH and PTCHl in the developing limb of a 12-week-old human embryo by radioactive in situ hybridization (Fig. 2a-e). The results show marked SUFUH expression in the osteoblasts of the perichondrium, where PTCHl is also highly expressed. These findings are consistent with earlier observations in the avian and murine systems, in which Ptchl and Glil are highly expressed in the same type of cells in response to Ihh secretion by prehypertrophic chondrocytes [Vortkamp et al., Science 273:613-622 (1996); Vortkamp et al., Mech. Dev. 71:65-76 (1998)]. Taken together, these results show that SUFUH is preferentially expressed in cells that receive a Hedgehog signal, and indicate that, during embryogenesis, SUFUH may be co-regulated with PTCHl and GLIL However, further studies will be required to demonstrate this relationship unequivocally, as well as to show a direct regulation of SUFUH by Hedgehog signalling.

Example 3: SUFUH inhibits transcriptional activity of GLI-1.

In a more direct attemption to confirm a role for SUFUH in SHH-PTCH-GLI signalling, we used reporter gene assays to analyse the effect of co-transfection of SUFUH and GLI-1 on GLI 1 -activated transcription and biological responses. As a reporter we used the firefly luciferase gene, controlled either by the human PTCHl promoter containing two well-conserved GLI -binding consensus sites (Fig. 3 a, PTCH-luc) or by a synthetic promoter consisting of 12 repeated GLI consensus sites fused to the thymidi- ne kinase basic promoter (Fig. 3b, 12GLI-RETKO-luc). Both reporters were activated by GLI-1, whereas SUFUH had no effect by itself. However, co-traήsfection of SUFUH efficiently suppressed activation of these reporter gene constructs by GLI-1. As the originally isolated GLI-1 gene contained a potentially activating mutation [Hynes et al., Neuron 19: 15-26 (1997)], we introduced a reverse mutation into the GLIl cDNA (SEQ.ID.NO.27) [Kinzler et al., Nature 332:371-374 (1988)]. This wild-type GLI-1 construct also contains an N-terminal haemagglutinin epitope tag and lacks the 3'- untranslated region; it activates the 12GLI-RETKO-luc reporter very strongly (Fig. 3b). This activation can again be suppressed by co-transfection of SUFUH but not by a car- boxy-terminally truncated SUFUH lacking amino acids 386-484 (SU-

FUHΔ)(SEQ.ID.NO.10). The wild-type GLI-1 construct, or a similar construct encoding the enhanced green fluorescent protein (EGFP) tag, was used in all subsequent experiments. Specificity of transcriptional inhibition by SUFUH was confirmed by the fact that SUFUH did not interfere with transcriptional activation of a reporter gene in- duced by co-transfection of a constitutively active glucocorticoid receptor (Fig. 3 c).

Example 4: SUFUH inhibits differentiation in response to SHH signalling.

To determine whether SUFUH was also able to inhibit a relevant cellular response to SHH or to one of its downstream effectors, GLI-1, we used the C3H10Tl/2-cell osteo- genic-differentiation assay [Nakamura et al., Biochem. Biophys. Res. Commun. 237:465-469 (1997); Murone et al., Curr. Biol. 9:76-84 (1999)]. Transfection of GLI-1 induced differentiation (Fig. 3d-f), as measured by alkaline phosphatase staining. Co- transfection of SUFUH but not of SUFUHΔ (Fig. 3f) inhibited this response. SUFUH also inhibited the differentiation response in cells transfected (Fig. 3f) with the N- terminal active fragment of SHH.

Example 5: SUFUH and GLI-1 are present in an intracellular complex. These results indicated that the functions of GLI-1 and SUFUH may also be closely linked in the mammalian signal-transduction pathway and raised the possibility that the two proteins might associate physically, as reported for the fly counterparts [Monnier et al., Curr. Biol. 8:583-586 (1998)]. To test this possibility, we transiently expressed epitope-tagged GLI-1 and SUFUH, either alone or in combination, in 293 cells. Hae- magglutinin-tagged GLI-1 (SEQ.ID.NO.l 1) co-precipitated with Myc-tagged SUFUH when the latter protein was immunoprecipitated with an anti-Myc antibody; conversely, when haemagglutinin-tagged GLI-1 was immunoprecipitated with an anti- haemagglutinin antibody, Myc-SUFUH (SEQ.ID.NO.12) was also present in the complex (Fig. 4a). A deletion construct containing the 407 N-terminal amino acids of GLI-1 (SEQ.ID.NO.6) also co-precipitated with SUFUH (Fig. 4b)(SEQ.ID.N0.1), whereas haemagglutinin-GLIl did not co-precipitate with the truncated SUFUHΔ protein (SEQ.ID.NO.10)(Fig. 4c). The truncated and full-length proteins were made at similar levels (Fig. 4d). As a further control, we showed that an unrelated overexpressed protein, IκBα, does not interact with SUFUH (Fig. 4e). We conclude that the role of

SUFUH as a negative regulator of GLIl -activated transcription is most likely mediated by a direct physical interaction between SUFUH and GLI-1. Our results show that the C-terminal domain of SUFUH (SEQ.ID.NO.2, SEQ.ID.NO.3) and N-terminal part of GLI-1 (SEQ.ID.NOs.6-8) are involved in protein-protein interactions, indicating that these domains may contain new types of interacting motif. The possibility that this interaction is mediated by an unidentified cellular protein appears unlikely, because all of the haemagglutinin-tagged GLI-1 could be immunoprecipitated by anti-Myc antibody from double-transfected cells, and proteins translated in vitro are also able to interact (data not shown). However, whether this interactions is direct or indirect remains to be confirmed.

Example 6: SUFUH can retain nuclear GLI-1 variants in the cytoplasm.

Using confocal microscopy, we determined the subcellular localization of transfected epitope-tagged GLI-1, SUFUH and SUFUHΔ in 293 cells; in each case, we studied at least 50 positive cells, all of which showed a similar pattern. SUFUH was present both in the cytoplasm and in the nucleus (Fig. 5a). SUFUHΔ showed a very similar subcellular localization (Fig. 5b). To study GLI-1 localization, we generated an expression construct containing an N-terminal EGFP tag. Using this construct, we showed that GLI-1 , like Ci, was cytoplasmic (Fig. 5c). Previous studies were not able to demonstrate the presence of full-length Ci [Aza-Blanc et al., Cell 89:1043-1053 (1997)] in the nucleus under any conditions, a fact assumed to be caused by the presence of a cytop- lasmic-retention domain. However, truncation of Ci [Aza-Blanc et al., Cell 89:1043- 1053 (1997)], GLI-1 or GLI-3 [Shin et al., Proc. Natl. Acad. Sci. USA 96:2880-2884 - (1999); Ruiz i Altaba et al., Development 126:3205-3216 ( 1999)] after the DNA- binding domain leads to nuclear localization of the protein. Like these proteins, the truncated GLIl (N407) (SEQ.ID.NO.6) construct was constitutively nuclear (Fig. 5d). This protein became partially cytoplasmic when coexpressed with SUFUH (SEQ.ID.NO. l)(Fig. 5e) but remained nuclear when coexpressed with SUFUHΔ (SEQ.ID.NO.10)(Fig. 5f).

Example 7: GLI-1 localization is regulated by nucleocytoplasmic shuttling.

We used another construct, containing the 525 N-terminal amino acids of GLI-1 (SEQ.ID.NO.15), in an attempt to map the region that mediates the cytoplasmic localization of GLI-1. This construct, like full-length GLI-1, was predominantly cytoplasmic (Fig. 6a). By inspecting the region between amino acids 407 and 525 (corresponding to the presumed cytoplasmic-retention domain) for specific sequence motifs, we found a perfect match to a leucine-rich nuclear-export signal (NES), starting at amino acid 495 (Fig. 6b). This sequence was fully conserved in other vertebrate species. In Ci we found a potential NES in the same region, as well as two extra potential NESs closer to the C terminus. The last two, but not the first, potential NESs were conserved in the mammalian Ci homologues Gli-2 and Gli-3 (Fig. 6b). To determine whether this region is indeed a functional NES, we mutated leucines 495 and 501 into alanines in this construct (SEQ.ID.NO.16) by polymerase chain reaction (PCR)-mediated mutagenesis. The resulting protein was exclusively nuclear, indicating that this motif in GLI-1 indeed functions as an NES (Fig. 6c).

Leucine-rich NESs mediate nuclear export of proteins that is dependent on the NES- receptor protein CRMl [Bogerd et al., J. Virol. 72:8627-8635 (1998)] which can be specifically inhibited by the drug leptomycin B (LMB) [Kudo et al., Exp. Cell. Res. 242:540-547 (1998)]. To test whether full-length GLI-1 is a target for active nuclear export, we studied the effects of LMB treatment (5 nM for 6 h) on the subcellular localization of GLI-1. Consistent with an important function for nuclear export in regu- lating the subcellular localization of full-length GLI- 1 , GLI- 1 was localized to the nucleus when LMB was added (Fig. 6d). Co-transfection of SUFUH (Fig. 6e), but not SUFUHΔ (Fig. 6f), inhibited this LMB-dependent nuclear accumulation of GLI-1. It has turned out that SUFUH interacts with a subsequence from amino acid 381 to amino acid 407 of GLI-1. This interaction blocks shuttling of GLI-1.

Example 8: SUFUH can interact with GLI-1 bound to DNA.

The presence of SUFUH in the nucleus raises the possibility it might interact with GLI-1 in this cellular compartment too, potentially affecting DNA binding by GLI-1 and/or its contacts with other transcription factors. To assess this possibility, we used the constitutively nuclear GLI1(N407) variant (SEQ.ID.NO.6) in combination with SUFUH and incubated nuclear extracts from either singly or doubly transfected cells with a radiolabelled double-stranded oligonucleotide (GLI-RE) corresponding to the GLIl -binding consensus DNA sequence. We then separated the complexes by native acrylamide gel electrophoresis. GLI1(N407) specifically retards the electrophoretic mobility of the GLI-RE oligonucleotide (Fig. 7a, lane 1), as this retardation can be blocked by addition of excess unlabelled GLI-RE (lane 2). Co-transfection of SUFUH resulted in even stronger complex formation and in the appearance of a new, more retarded complex, indicating a SUFUH-GLI1 protein-protein interaction (Fig. 7a, lane 3). Antibody supershifting experiments using antibodies against the respective tags fu- sed to GLI-1 (Fig. 7a, lane 4) and SUFUH (lane 5) sequences showed the presence of

SUFUH in both retarded complexes and GLI-1 in at least the more quickly migrating complex. In support of a direct interaction between GLI-1 and SUFUH that is independent of other nuclear proteins, GLI1(N407) could be co-immunoprecipitated with SU- FUH (Fig. 4b) and similar electrophoretic mobility-shift assay (EMSA) results were obtained with in v/vø-translated proteins (data not shown). Analysis of nuclear extracts from cells transfected with SUFUH alone revealed no retarded complexes (Fig. 7a, lane 6).

The increase in DNA binding seen after co-transfection of SUFUH (SEQ.ID.NO.1 ) correlated with increased levels of soluble GLI1(N407) (SEQ.ID.NO.6) in the nuclear extract (Fig. 7b, lower panels). In contrast, the levels of GLI1(N407) in the nuclear pellet decreased on co-transfection with SUFUH and, as most GLI1(N407) resides in the nuclear pellet, the total nuclear level of GLI1(N407) is reduced, consistent with the immunofluorescence data (Fig. 5). GLI1(N407) protein appeared in the cytoplasm only when SUFUH was co-transfected (Fig. 7b, upper panel). Our results indicate that SUFUH can act both by preventing access of GLI-1 to the nucleus and by binding to GLI- 1 when this factor is already bound to DNA.

General Methods relating to Examples 9 - 12:

Biological materials

Timely-mated NMRI mice were obtained from the Charles River Laboratories. The Dac/Dac mice and their wild-type SMNZB congenic counterparts and DNA samples from both were obtained from Jackson Laboratories. The partially overlapping BAC clones RPCI-11-170J3 containing the 5' end and 243114 containing the 3' of the human SU(FU) gene were obtained from Research Genetics. The mouse Su(fu) BAC clones 17985 and 17986 were obtained from Genomic Systems, Inc. The mouse Su(fu) EST clones 513730, 1195307, 1224813, and 963990 from the IMAGE consortium were obtained from the UK HGMP Resource Centre, Hinxton, Cambridge.

Chromosomal localization, mapping, and sequencing of the human SU(FU) and mouse Su(fu) genes

The chromosomal localization of SU(FU) was determined by Radiation Hybrid Mapping using the primer pair 5'cagttgtgtcaacgagatctcc (SEQ.ID.NO.28) and 5'ctgtggcctgtgcatggcac (SEQ.ID.NO.29) and the primer pair 5'ccgctggctaagccttgtgac (SEQ.ID.NO.30) and 5 'cagggtccagagtcaaacctca (SEQ.ID.NO.3 l)(both primer pairs are derived from the 3' untranslated region) on the Genbridge G4 map (Research Genetics). The intron-exon borders were determined by sequencing BACs with primers oriented in both directions for each exon at least two times. The sequences of these primers are available on request. Restriction enzyme digests of BAC DNA and subcloned DNA fragments were transfered to Hybond N⁺ filters (Amersham). The filters were hybridized with radiolabelled oligonucleotide probes for individual exons to assemble the Su(fu) locus map.

Screening for mRNA variants

cDNAs from normal human tissues (Human MTC panel II) or from xenografted human tumors (Human tumor MTC panel) were obtained from Clontech. Mouse embryo cDNA: Organs were homogenized and RNA was extracted using RNAzol B [Chomc- zynski et al., Anal. Biochem. 162:156-159(1987)]. Subsequently, cDNA was prepared using o(dT)ι 5 and reverse transcriptase from mouse mammary tumor virus (Clontech) according to standard procedures. Primary PCR products were generated containing the entire coding region of SU(FU)/Su(fu) and subsequently nested PCR products containing the complete exons 2-3, 4-5, 6-7, 8-9, 10-11 respectively and flanking sequences of the neighbouring exons were obtained. After gel electrophoresis, variant transcripts were detected by size difference. Whole mount in situ hybridization

Mouse embryos were cut sagitally and after fixation in 4% paraformaldehyde hybridized according to standard procedures (22) [Wilkinson, In situ Hybridization: A practical approach, IRL Press, Oxford 1992]. Digoxin-labelled probes were prepared using the DIG RNA labelling kit and the label was detected using the DIG Nucleic Acid Detection Kit (both Boebringer Mannheim). Specificity of the reaction was monitored by comparison to samples hybridized to sense probes.

Epitope-tagged constructs

Myc-SU(FU) and HA-GLIl, both in pCMV-5, have been described above. The myc- tagged constructs for the transcription variants were generated by cloning cDNAs into the pGEM-T vector (Promega), verifying the sequence of the PCR product and exchanging a fragment between two appropriate restriction sites in the myc-SUFUH. In detail, a BstEII (nt 327) -Sail (behind the STOP codon) fragment from near full length SU(FU)-XL was cloned into myc-SU(FU) and EcoNI (nt 751)-SalI fragments of SU(FU)-Tt and SU(FU)-Ly were ligated into myc-SU(FU)-XL.

Transient transfections and Western Blotting

10 μg each of the tagged GLI and SU(FU) constructs were transfected into a near confluent 15 cm culture dish with 293 kidney carcinoma cells using 75 μg of polyethyle- neimine (Sigma Chemicals) in 12 ml serum- free medium as a transfecting agent. After 36 h cells were harvested and lysed in 1% Triton X-100. The lysates were precleared one time over protein A/G-agorse preincubated with goat serum and then precipitated with mouse-anti myc monoclonal antibody 9E10 or rabbit anti mouse-antiserum. After electrophoretic separation on a reducing 10% PAGE-SDS gel and blotting onto PVDF membrane, myc- tagged proteins were detected by rabbit-anti myc antiserum, followed by incubation with a peroxidase-conjugated second antibody.

Alizarin- Alcian Blue staining

Newborn and 1 day old pups were dissected removing their skin, muscles and inner organs. Then the carcasses were stained as described [Inoue, Cong. Anom. 16:171- 173].

Sequencing of the coding regions of SU(FU) in SHFM3 patients and in the Dae mouse

mRNA was obtained from transformed patient B-cell lines or from lung of an adult homozygous Dae mouse as decribed (see above). cDNA and subsequently full length (near full length in case of Dae) PCR products of the coding regions were generated. Each two overlapping nested PCR products were generated and sequenced by standard procedures from both ends (in case of Dae the initial 49 nucleotides of the coding region are missing). At least 13 clones from each patient sample and 6 clones from the Dae mouse were included. All exons from the Dae mouse and also exon 1 from the patients were sequenced from genomic DNA using primers with intron sequences and sequen- ces of flanking untranslated regions.

Example 9: Genomic organization of the human SU(FU) and of the mouse Su(fu) genes

The full length coding sequences of SU(FU) and Su(fu), the human and the murine homologs of the Drosophila signal transducer molecule Su(fu) were indentified above. By radiation hybrid mapping the human SU(FU) gene was localized to chromosome 10q24 in a region syntenic to chromosome 19 in the mouse. We identified artificial bacterial chromosomes (BACs) containing both the human SU(FU) and the murine Su(fu) genes and determined the intron-exon organization by sequencing using primers from the coding region.

Both genes contain twelve exons with exon-intron boundaries at corresponding posi- tions (Table 1). When mapping the overlapping BAC clones 17985 and 17986, the mouse Su(fu) gene was found to span approximately 100 kb (Fig. 9). The promoter region contained no consensus TATA- or CAAT-box immediately upstream of the putative transcriptional start site. However, using the Transcription Start Site Wingender database (TSSW) at UK HGMP Resource Centre (www.hgmp.mrc.ac.uk) a promoter in this region with an excellent prediction value (LDF=20.80) was found. When sequencing upstream both the human and murine promoters, the ATG-containing exon for the gene actin-related protein 1 (ARPl and Arpl in human and mouse respectively) was detected in opposite transcriptional orientation separated by only 1516 bp (in human) between the initial methionine codons. The ubiquitously expressed ARPl [Clark et al., Mol. Biol. Cell 5:1301-1310 (1994)] and Arpl [Schroer et al., J. Cell Biol.

115:1309-1319 (1991)] are members of the microtubule-associated dynactin complex which is essential for spindle formation [Merdes et al., J. Cell Biol. 149:851-862 (2000)] and organelle transport [Xiang et al., Curr. Biol. 18:603-606 (2000)], for a review see [Schafer et al., Annu. Rev. Cell. Dev. Biol. 15:341-363 (1999)] (Fig. 9). In addition, sequencing off the end of BAC 17986 revealed an exon for the ADP- ribosylation factor-like protein 3 (Arl3) downstream of the mouse Su(fu) gene (Fig. 9). Mining the GenBank mouse EST database, several Su(fu) EST clones originating from different tissues including adult testis, heart, kidney, B cell, mammary gland, day post natum (dpn) 6 head, dpn 10 skin, fertilized egg, embryos from dpc 8, 10-11, and 13 were found, demonstrating expression in a wide range of tissues. Four mouse ESTs were sequenced and their structure is shown in Fig. 9B. The combined EST cDNAs extend over 4400 bp, which fits well with the mRNA size seen on Northern blot (4.5 kb). The predicted AUG codon of the translational initiation has a good Kozak consensus sequence. The clones 1224813 and 963990 that extends the most 3' as well as a number of other human and mouse ESTs in the GenBank end at the same position. In- stead of the consensus polyadenylation signal AAUAAA we found an AUUAAA sequence about 15 bp upstream of the end of the longest ESTs in both human and mouse.

Example 10: SU(FU) mRNA isoforms

We screened a commercially available panel of cDNAs from various tissues of apparently healthy donors for expression of variants of SU(FU). One variation which we found was a trinucleotide insertion after nucleotide 756 of the coding sequence resulting in an extra-glutamine after amino acid 252. This insertion created a novel EcoNI cleavage site which allowed easy detection of this variant which we called SU(FU)-XL (SEQ.ID.NO.17). SU(FU)-XL was found to be expressed in five of seven tissues tested with a relative abundance varying from roughly 10% to 50% of total SU(FU) (Fig. 10). The explanation for this variation was found in the sequence of the downstream end of intron 6 which was determined as ...TTTTC AAGC AG-exon 7... (Table 1). Probably, both underlined AG dinucleotides could function as alternative splice acceptor sites.

We found two additional splice variants (Fig. 10), one of them, SU(FU)-LK (SEQ.ID.NO.18), was expressed in peripheral leukocytes, the other one, SU(FU)-TT (SEQ.ID.NO.19), in testes. By sequencing it was found that SU(FU)-LK is lacking exon 10, and that the downstream sequence has a shifted reading frame resulting in a STOP codon after two amino acids. SU(FU)-TT has an extra exon after exon 8 (exon 8a), which encodes 19 amino acids followed by a STOP codon.

We also screened a commercially available panel of tumor cell xenograft cDNAs (two breast, two lung, two colon, one prostate and one pancreatic carcinoma sample) and cDNA samples from a panel of dpc 13.5 mouse embryonic tissues (lung, intestine, skeletal muscle, heart, liver, eye, brain, spinal cord, tongue, kidney, whole limb, skin) and we detected the normal SU(FU)/Su(fu) transcripts but no variants except SU(FU)- XL. There may be additional isoforms, since our method did not allow detection of iso- forms, where the first or the last exon of SU(FU) is missing. Functional analysis of these SU(FU) variants compared with the standard SU(FU) (as we call the originally cloned variant) was performed by assessing their potential to co- precipitate GLI-1, one of the physiological binding partners of SU(FU). N-terminally myc-tagged versions of the coding sequences of all SU(FU) variants and a hemagluti- nin-tagged version of GLI-1 (HA-GLI) were transfected into 293 cells. With an antibody against HA we could coprecipitate SU(FU) when cotransfected with myc- SU(FU) or myc-SU(FU)-XL, but not with myc-SU(FU)-LK or myc-SU(FU)-TT as detected by an anti-myc antibody after Western Blotting (Fig. 11).

Example 11: Expression of SU(FU) during mouse embryogenesis

We analyzed mouse embryos in the time range between dpc 8.5 and dpc 15.5 for expression of SU(FU) by whole mount in situ hybridization. Two other reports have described similar experiments [Stone et al., J. Cell Science 112:4437-4448 (1999); Pearse et al., Developmental Biology 212:323-336 (1999)]. Here we will focus on those results not described before.

Above, SU(FU) has been shown to be expressed ubiquitously in the adult. A similar result has been obtained for the mouse embryo [Stone et al., J. Cell Science 112:4437-4448

(1999); Pearse et al., Developmental Biology 212:323-336 (1999)] and our studies confirm these results in principle. In particular at dpc 10.5 and 11.5 we observed an intense all over reactivity of which only the neural tube and the forming CNS emerged as even stronger and only the liver appeared to be negative. For the other stages however, we found that certain tissues expressed Su(fu) with considerably stronger intensity than the all over reactivity of the embryos. At dpc 8.5 and 9.5 we detected Su(fu) in the neural tube and in the somites (Fig. 12A-C). At dpc 9.5 Su(fu) was expressed further in the heart and the dorsal artery as well as in the intersegmentary arteries (Fig. 12C, F), in the bud of the forelimb (Fig. 12B, C), in the branchial arches (Fig. 12B, F), and in the mesenchyme of the trunk and of the head surrounding the prospective mouth cavity (Fig. 12B). Su(fu) expres- sion in the neural tissues persisted through the entire observed period as previously reported [Stone et al., J. Cell Science 112:4437-4448 (1999); Pearse et al., Developmental Biology 212:323-336 (1999)]. At dpc 10.5 Su(fu) was expressed in all brain vesicles (Fig. 12D) and in the dorsal root ganglia (Fig. 12E) as detected after removal of the skin. At all later stages, the brain and the spinal cord were stained with high intensity (Fig. 12G). The expression of Su(fu) in the somites persisted as well, however from dpc 10.5 only the sclerotome and the sclerotome-derived tissues were stained with the probe. The mesenchyme surrounding the ossifying part of the vertebrae expressed Su(fu) from dpc 12.5 (Fig. 12G) and of the ribs from dpc 13.5 [Pearse et al., Developmental Biology 212:323-336 (1999)]. Like in the somite-derived skeletal parts, even in the digits the expression of Su(fu) was observed to parallel the progress of ossification as the area which expresses Su(fu) was migrating distally during the course of development (Fig. 12H-J). The expression of Su(fu) in the branchial arches was continuing within the maxilla, mandible and in the tongue and disappearing from dpc 12.5 onwards when the mouth cavity was closed, with at dpc 15.5 only tongue being stained (Fig. 12K). Further, through the entire observed period Su(fu) was expressed in the genital tubercle (Fig. 12G) and from dpc 13.5 in the mesenchyme surrounding the developing tracheoles of the lung (Fig. 12L).

Example 12: Examination of RNA from SHFM3 patients and DNA from Dae mice for mutations in the Su(fu) gene

We have located the SU(FU) gene in man between the markers AFM183XB12 and D10S192, with a distance of 0.10 cR from AFM183XB12 (lod>3.0). This narrow region also contains the locus for the genetic disorder Split Hand Split Foot-Malformation- 3 (SHFM3). Members of the Hedgehog signalling pathway have been shown to be involved in the process of digit formation as will be discussed below.

To investigate the hypothesis that SU(FU) would be the gene involved in SHFM3, we sequenced a number of cDNA clones of SU(FU) from B-cell lines derived from two unrelated heterozygous SHFM3 patients (since we met difficulties in reverse transcrip- tion of the first exon we sequenced the first exon from the genomic DNA of these cell lines.). However, in the entire coding sequence no differences from the normal sequence were observed. With our approach mutations in the promotor or in introns which affect the transcriptional regulation or splicing of SU(FU) can not be excluded.

In the mouse, the Dae mutant [Chai, J. Hered. 72:234-237 (1981)] with a similar phenotype [Crackover et al., Dev. Biol. 201:78-89 (1998); Seto et al., Teratology 56:161- 270 (1997)], is linked to chromosome 19 in a region syntenic to the one on human chromosome 10 containing the locus for the SHFM3 gene. It has been suggested that Dae might be the murine counterpart of SHFM3 [Johnson et al., Genomics 29:457-464 (1995)]. We amplified and sequenced all Su(fiι) exons from the genomic DNA of a homozygous Dae mouse. Additionally, the coding region of the cDNA from nucleotide 50 after the initial AUG codon downstream was sequenced. No mutations in the Su(fu) gene were observed and the presence of a normal mRNA sequence makes splice muta- tions very unlikely.

During our studies on Dae mice we observed abnormalities not only in the digit formation but also in cranial development. The ossification of the skull bones of the Dae mouse in the perinatal period was delayed compared to the wildtype. The intramemb- raneously developing parietal, interparietal and supraoccipital bones have not been ossified at all even at dpn 1 (two mice examined at dpn 0, one at dpn 1), while they have been ossified completely at dpn 0 in the wild-type control mice. The endochondrally developing frontal, exoccipital and temporal bones, as well as the basal processes and the parts of the jaw, however, were developed in the Dae mouse as well as in wild-type (Fig. 13). Further, in the newborn Dac/Dac mouse the skull was flat and less curved than in wild-type. Table 1. Exon-intron boundaries of the human SU(FU) and mouse Su(fu) genes

Exon Exon Intron (length) number

1 1-ATGGCG...GTACTG-182 gtatgc. .ttgcag

1-ATGGCG...GTACTG-182 gtctgα. .ttgcag (3.7 kb)^f 2 183-GTTGGG...CCATGA-317 gtgaag. ..tttcag

183-GTTGGG;..CCATGA-317 gtgagt.. .tttcag (24 kb)^1"

3 318-GTTTAC,„AGTCAG-454 gtagga. ..ccacag

318-GTTTAC...AGTCAG-454 gjaata.. .ccacag (22 kb)^f

4 455-AGAACA...CTCCAG-597 gtgagg. ..cctcag

455-AGAACA...CTCCAG-597 gjgagg. ..ccacag (607 bp)^§

5 598-ATCGTT...GCCTAT-683 gtgagt. .ccacag

598-ATTGTT...GCCCAT-683 gtgagt., .ccacag (226 bp)^§

6 684-TGCTGG...CTGCAA-756 gtatgt., aagcag™

684-TGCTGG...CTGCAA-756 gtatgt. aagcag™ (2.3 kb)^f

7 757-GAGAGA...GCAAAG-910 gtggga ..ttgcag

757-GAGAGA...GCAAAG-910 . gttgag. .ttgcag (1419 bp)^§

8 911-ACACAG...GGCCCC-1022 gtaagt. .tcacag

911-ACACAG...GGCTCC-1022 gtaagt. .tcacag (22 kb)^1-

9 1023-GAGCCG...CCTAAG-1157 gtgagc. ..ctccag

1023-GAGCCG...CCTAAG-1157

10 1158-GGGCAG...TTACAA-1296 gtgaga ..tcacag

1158-GGGCAG...TTACAA-1296 gtgaga ..tcccag (9.9 kb)^f

11 1297-ATTCTGTTGA...GAGGAA- gtaagc...ccacag¹¹

1365 gtgagc...tcacag (1.9 kb)^f 1297-ATTCTG...GAGGAA-1365

12 1366-TTCAAA... 1366-TTTAAA... CTGCAG-4309

Note: The upper and lower row in each pair refers to the human and mouse sequence, respectively. The translational start site is numbered as nucleotide 1 since the transcriptional start site is not exactly defined. Exon 1 contains approximately an additional 150 bp 5^'of the translational start site.

^§Fully sequenced. ^tPartially sequenced.

™The dotted underlined ag.is the intron border in SU(FU)-XL

'The double-underlined TGA is the STOP codon in SU(FU)-LK

Claims

Claims

1. A peptide consisting of one or more multiples of a protein fragment comprising 5 - 104 amino acid residues, preferably 10 - 50 amino acid residues, derived from the amino acid sequence of SEQ.ID.NO.2, said protein fragment showing a homology of at least 70 %, preferably at least 90 %, with the amino acid sequence of SEQ.ID.NO.2, said peptide being able to spercifically bind to a protein having the sequence of SEQ.ID.NO.5.

2. A peptide according to claim 1, characterised in that it consists of one or more multiples of a protein fragment consisting of 10 - 97 consecutive amino acid residues from the amino acid sequence of SEQ.ID.NO.3.

3. A peptide according to claim 2, characterised in that said protein fragment has the amino acid sequence of SEQ.ID.NO.4.

4. A peptide consisting of one or more multiples of a protein fragment comprising 5 - 407 amino acid residues, preferably 5 - 50 amino acid residues, derived from the ami- no acid sequence of SEQ.ID.NO.6, said protein fragment showing a homology of at least 70%, preferably at least 90%, with the amino acid sequence of SEQ.ID.NO.6, said peptide being able to specifically bind to a protein having the sequence of SEQ.ID.NO.l .

5. A peptide according to claim 4, characterised in that it consists of one or more multiples of a protein fragment consisting of 5 - 26 consecutive amino acid residues from the amino acid sequence of SEQ.ID.NO.7, and which fragment contains the ami- no acid subsequence lys-arg-his-arg.

6. A peptide according to claim 4, characterised in that it consists of one or more multiples of a protein fragment consisting of 5 - 20 consecutive amino acid residues from the amino acid sequence of SEQ.ID.NO.8.

7. A DNA sequence encoding a peptide according to anyone of claims 1 - 6.

8. A monoclonal antibody or an antibody fragment directed against a peptide according to anyone of claims 1 - 6.

9. A peptide according to anyone of claims 1 - 6, for medical use.

10. Use of a peptide according to anyone of claims 1 - 6, for preparing a pharmaceutical composition for treating cancer.

11. A monoclonal antibody or an antibody fragment according to claim 8, for medical use.

12. Use of a monoclonal antibody according to claim 8, for preparing a pharmaceutical composition for treating cancer.

13. A pharmaceutical composition comprising a peptide according to anyone of claims 1 - 6 and/or a monoclonal antibody or an antibody fragment according to claim 8, together with a pharmaceutically acceptable carrier, excipient or diluent.