INHIBITION OF GASCl

FIELD OF INVENTION

The present invention relates to a method for identifying compounds that are capable of acting as activity modulators of the JmjC domain containing proteins and which are useful for prevention and/or treatment of diseases in which genomic instability is involved in the pathogenesis.

BACKGROUND OF THE INVENTION

Methylation of lysine and arginine residues on histone tails constitutes important epigenetic marks delineating transcriptionally active and inactive chromatin. For instance, methylation of lysine 9 on histone H3 (H3-K9) is associated with epigenetically silenced chromatin ^1"3 .

In contrast to other histone modifications such as acetylation and phosphorylation, methylation has been regarded as irreversible because of the high thermodynamic stability of the N-CH3 bond. Recently however, Shi and coworkers ⁴ identified the protein LSDl, a nuclear amine oxidase homologue, as a histone demethylase highly specific for mono- and di-methylated H3- K4. This enzyme demethylates its substrates through an amine oxidase reaction. However,

LSDl is unable to demethylate tri-methylated lysine H3-K4, most likely due to the absence of a protonated nitrogen required for oxidation4. Just prior to the completion of the present study, Tsukada et al. demonstrated that the Jumonji protein FBXLIl (JHDMIa) can specifically demethylate di-methyl H3-K36 in a Fe(II) and α-ketoglutarate-dependent manner ⁵ . Although the reaction mechanism for FBXLl 1-mediated demethylation at least in theory could utilize tri- methyl H3-K36 as substrate, no such activity could be demonstrated ⁵ . Thus, although the identification of LSDl and FBXLIl as histone demethylases constituted important milestones for epigenetic research demonstrating the dynamic regulation of methyl marks, they have not resolved the question of the reversibility of tri-methylated lysine marks. As documented by studies of the SUV39H1 knockout mouse, loss of the tri-methyl variant of the H3-K9 mark results in chromosomal aberrations and predisposes to cancer. Hence enzymes capable of reversing this mark have long been sought, although their existence has been questioned. The latter view has been reinforced by the fact that tri-methylated H3-K9 is required for the establishment and maintenance of heterochromatin, a "very stable and heritable chromatin state". The identification of such enzymes and inhibitors of their activity would provide a novel approach to the prevention and treatment of cancers.

SUMMARY OF THE INVENTION

The present invention provides a method of testing the ability of a test compound to bind to and optionally modulate the activity of a protein of the JMJD2 subfamily of Jumonji proteins, said

method comprising incubating a test compound with a protein of the JMJD2 subfamily of Jumonji proteins, a co-factor of said protein and, optionally, a substrate for demethylation.

In a presently preferred embodiment of the invention, the method comprises the steps of: i) preparing a suspension of a substrate for demethylation in a solution containing iron and alpha-ketoglutarate;

Ii) adding said test compound to the suspension; iii) adding to said suspension in ii) a protein of the JMJD2 subfamily of Jumonji proteins ; iv) incubating said suspension and; and v) determining the extent of demethylation of said substrate.

DEFINITIONS

The term 'nucleic acid molecule' refers to an oligomer or polymer of ribonucleic acid (RNA) or deoxyribonucleic acid (DNA) or mimetics thereof. This term includes molecules composed of naturally-occurring nucleobases, sugars and covalent intemucleoside (backbone) linkages as well as molecules having non-naturally occurring nucleobases, sugars and covalent intemucleoside (backbone) linkages which function similarly or combinations thereof. Such modified or substituted nucleic acids are often preferred over native forms because of desirable properties such as, for example, enhanced cellular uptake, enhanced affinity for nucleic acid target and increased stability in the presence of nucleases and other enzymes, and are in the present context described by the terms "nucleic acid analogues" or "nucleic acid mimics". Preferred examples of nucleic acid mimetics are peptide nucleic acid (PNA-), Locked Nucleic Acid (LNA-), xylo-LNA-, phosphorothioate-, 2'-methoxy-, 2'-methoxyethoxy-, morpholino- and phosphoramidate- containing molecules or the like.

A "polynucleotide sequence" (e.g. a nucleic acid, polynucleotide, oligonucleotide, etc.) is a polymer of nucleotides comprising nucleotides A,C,T,U,G, or other naturally occurring nucleotides or artificial nucleotide analogues, or a character string representing a nucleic acid, depending on context. Either the given nucleic acid or the complementary nucleic acid can be determined from any specified polynucleotide sequence.

With respect to the present invention the term 'polypeptide' refers to an amino acid chain of any length, including a full-length protein, oligopeptides, short peptides and fragments thereof, wherein the amino acid residues are linked by covalent bonds. All polypeptide sequences in the present specification and claims are, also when not explicitly stated, written from the N-terminal to the C-terminal end in the conventional format.

Numbering of a given amino acid polymer or nucleotide polymer "corresponds to" or is "relative to" the numbering of a selected amino acid polymer or nucleic acid polymer when the position of any given polymer component (e.g. amino acid, nucleotide, also referred to generically as a "residue") is designated by reference to the same or an equivalent position in the selected amino acid or nucleotide polymer, rather than by the actual numerical position of the component in the given polymer. Thus, for example, the numbering of a given amino acid position in a given polypeptide sequence corresponds to the same or equivalent amino acid position in a selected polypeptide sequence used as a reference sequence.

A "variant" is a polypeptide comprising a sequence, which differs (by deletion of an amino acid, insertion of an amino acid, and/or substitution of an amino acid for a different amino acid) in one or more amino acid positions from that of a parent polypeptide sequence. The variant sequence may be a non-naturaliy occurring sequence, i.e. a sequence not found in nature.

In the present context, the term "synthetic peptide" refers to a peptide, including a short peptide that has been synthesized in vitro. The term further encompasses peptides or short peptides that have been modified by substitution with unusual or non-natural amino acids.

"Naturally occurring" as applied to an object refers to the fact that the object can be found in nature as distinct from being artificially produced by man. For example, a polypeptide or polynucleotide sequence that is present in an organism (including viruses, bacteria, protozoa, insects, plants or mammalian tissue) that can be isolated from a source in nature and which has not been intentionally modified by man in the laboratory is naturally occurring. "Non-naturally occurring" as applied to an object means that the object is not naturally-occurring i.e. the object cannot be found in nature as distinct from being artificially produced by man.

A "fragment" or "subsequence" refers to any portion of a given sequence. It is to be understood that a fragment or subsequence of a sequence will be shorter than the sequence itself by at least one amino acid or one nucleic acid residue. Thus, a fragment or subsequence refers to a sequence of amino acids or nucleic acids that comprises a part of a longer sequence of amino acids (e.g. polypeptide) or nucleic acids (e.g. polynucleotide) respectively.

The term 'sequence identity' indicates a quantitative measure of the degree of homology between two amino acid sequences or between two nucleic acid sequences of equal length. If the two sequences to be compared are not of equal length, they must be aligned to give the best possible fit, allowing the insertion of gaps or, alternatively, truncation at the ends of the polypeptide sequences or nucleotide sequences. The sequence identity can be calculated as

' " ^' ' . ', ^ώf — , wherein N _d i _f is the total number of non-identical residues in the two sequences when aligned and wherein N _ref is the number of residues in one of the sequences. Hence, the DNA se- quence AGTCAGTC will have a sequence identity of 75% with the sequence AATCAATC (N _djf =2

and N _ref =8). A gap is counted as non-identity of the specific residtιe(s), i.e. the DNA sequence AGTGTC will have a sequence identity of 75% with the DNA sequence AGTCAGTC (N _dif =2 and N _ref =8).

In all polypeptide or amino acid based embodiments of the invention, the percentage of sequence identity between one or more sequences is based on alignment of the respective sequences as performed by clustalW software (http:/www. ebi.ac.uk/clustalW/index. html) using the default settings of the program. These settings are as follows: Alignment=3Dfull, Gap Open 10.00, Gap Ext. 0.20, Gap separation Dist. 4, Protein weight matrix: Gonnet. With respect to the nucleotide-based embodiments of the invention, the percentage of sequence identity between one or more sequences is also based on alignments using the clustalW software with default settings. For nucleotide sequence alignments these settings are: Alignment=3Dfull, Gap Open 10.00, Gap Ext. 0.20, Gap separation Dist. 4, DNA weight matrix: identity (IUB).

In the present context "complementary sequence" refers to nucleotide sequences which will hybridise to a nucleic acid molecule of the invention under stringent conditions. The term "stringent conditions" refers to general conditions of high stringency. The term "stringency" is well known in the art and is used in reference to the conditions (temperature, ionic strength and the presence of other compounds such as organic solvents) under which nucleic acid hybridisations are conducted. With "high stringency conditions", nucleic acid base pairing will occur only between nucleic acid fragments that have a high frequency of complementary base sequences, as compared to conditions of "weak" or "low" stringency.

As an example, high stringency hybridisation conditions comprise (1) low ionic strength and high temperature for washing, such as 0.015 M NaCI/0.0015 M sodium citrate, pH 7.0

(0. IxSSC) with 0.1% sodium dodecyl sulfate (SDS) at 50 ⁰ C; (2) hybridisation in 50% (vol/vol) formamide with 5 x Denhardt's solution (0.1% (wt/vol)) highly purified bovine serum albumin/0.1% (wt/vol) Ficoll/0.1% (wt/vol) polyvinylpyrrolidone), 50 mM sodium phosphate buffer at pH 6.5 and 5 x SSC at 42°C; or (3) hybridisation in 50% formamide, 5 x SSC, 50 mM sodium phosphate (pH 6.8), 0.1% sodium pyrophosphate, 5 x Denhardt's solution, sonicated salmon sperm DNA (50 μg/ml), 0.1% SDS, and 10% dextran sulfate at 42°C with washes at 42°C in 0.2 x SSC and 0.1% SDS.

The term "isolated nucleic acid" may refer to a nucleic acid (e.g. DNA or RNA) that is not immediately contiguous with both of the coding sequences with which it is immediately contiguous (i.e. one at the 5' and one at the 3' end) in the naturally occurring genome of the organism from which the nucleic acid of the invention is derived. Thus, this term includes e.g. a cDNA or a genomic DNA fragment produced by polymerase chain reaction (PCR) or restriction endonuclease treatment, whether such cDNA or genomic DNA fragment is incorporated into a vector, integrated into the genome of the same or a different species than the organism, including e.g. a virus, from which it was originally derived, linked to an additional coding

sequence to form a hybrid gene encoding a chimeric polypeptide, or independent of any other DNA sequences. The DNA may be double-stranded or single-stranded, sense or anti-sense.

A "recombinant polynucleotide" or a "recombinant polypeptide" is a non-naturally occurring polynucleotide or polypeptide that includes nucleic acid or amino acid sequences, respectively, from more than one source nucleic acid or polypeptide, which source nucleic acid or polypeptide can be a naturally occurring nucleic acid or polypeptide, or can itself have been subjected to mutagenesis or other type of modification. A nucleic acid or polypeptide may be deemed "recombinant" when it is artificial or engineered, or derived from an artificial or engineered polypeptide or nucleic acid. A recombinant nucleic acid (e.g. DNA or RNA) can be made by the combination (e.g. artificial combination) of at least two segments of sequence that are not typically included together, not typically associated with one another, or are otherwise typically separated from one another. A recombinant nucleic acid can comprise a nucleic acid molecule formed by the joining together or combination of nucleic acid segments from different sources and/or artificially synthesized. A "recombinant polypeptide" (or "recombinant protein") often refers to a polypeptide (or protein) which results from a cloned or recombinant nucleic acid or gene. The source polynucleotides or polypeptides from which the different nucleic acid or amino acid sequences are derived are sometimes homologous (i.e. have, or encode a polypeptide that encodes, the same or a similar structure and/or function), and are often derived from different isolates, serotypes, strains, species, of organism or from different disease states, for example.

The term "recombinant" when used with reference, e.g. to a cell, nucleotide, vector, protein, or polypeptide typically indicates that the cell, nucleotide, or vector has been modified by the introduction of a heterologous (or foreign) nucleic acid or the alteration of a native nucleic acid, or that the protein or polypeptide has been modified by the introduction of a heterologous amino acid, or that the cell is derived from a cell so modified. Recombinant cells express nucleic acid sequences (e.g. genes) that are not found in the native (non-recombinant) form of the cell or express native nucleic acid sequences (e.g. genes) that would be abnormally expressed under-expressed, or not expressed at all. The term "recombinant" when used with reference to a cell indicates that the cell replicates a heterologous nucleic acid, or expresses a peptide or protein encoded by a heterologous nucleic acid. Recombinant cells can contain genes that are not found within the native (non-recombinant) form of the cell. Recombinant cells can also contain genes found in the native form of the cell wherein the genes are modified and reintroduced into the cell by artificial means. The term also encompasses cells that contain a nucleic acid endogenous to the cell that has been modified without removing the nucleic acid from the cell; such modifications include those obtained by gene replacement, site-specific mutation, and related techniques.

The term "recombinantly produced" refers to an artificial combination usually accomplished by either chemical synthesis means, recursive sequence recombination of nucleic acid segments or other diversity generation methods (such as, e.g. shuffling) of nucleotides, or manipulation of

isolated segments of nucleic acids, e.g. by genetic engineering techniques known to those of ordinary skill in the art. "Recombinantly expressed" typically refers to techniques for the production of a recombinant nucleic acid in vitro and transfer of the recombinant nucleic acid into cells in vivo, in vitro, or ex vivo, where it may be expressed or propagated.

"Conservative" as used herein means (i) that the alterations are as conformationally neutral as possible, that is, designed to produce minimal changes in the tertiary structure of the mutant peptide or polypeptides as compared to the native protein. Conformational neutrality is desirable for preserving biological activity. Rules exist which can guide those skilled in the art to make alterations that have high probabilities of being conformationally neutral, see e.g. (77) and (78). Some of the more important rules include (1) replacement of physicochemically similar, i.e. synonymous, residues is less likely to produce conformational changes because the replacing amino acid can play the same structural role as the replaced amino acid; and (2) alteration of evolutionarily conserved sequences is likely to produce deleterious conformational effects because evolutionary conservation suggests sequences may be functionally important.

Unless otherwise defined herein or below in the remainder of the specification, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the art to which the invention belongs.

DETAILED DESCRIPTION OF THE INVENTION

The present invention is based on the discovery that compounds capable of acting as activity modulators of the JmjC domain containing proteins are useful for prevention and/or treatment of diseases in which genomic instability is involved in the pathogenesis. In particular, modulators of the JmjC domain containing proteins are useful for prevention and/or treatment of cancer. The present invention further pertains to the identification of the Jumonji protein; gene amplified in squamous cell carcinoma 1 (GASCl) as a protein capable of interacting with di-methylated and tri- methylated lysine 9 on histone H3 (H3-K9). GASCl belongs to the JMJD2 subfamily of the Jumonji family. The inventors of the present invention have shown that the JMJD2 family of proteins are histone-demethylases demethylating tri- and di-methylated H3-K9. Furthermore, the inventors have demonstrated that ectopic expression of GASCl and JMJD2 members dramatically decreases tri-and di-methylated H3-K9, increases mono-methylated H3- K9, delocaϋses HPl and reduces heterochromatin in vivo. Thus, in addition to being the first identified histone tri-methyl demethylase, proteins of the JMJD subfamiliy of the Jumonji family are also involved in cancer development. Inhibitors of the catalytic activity of these enzymes are therefore important candidates for development of novel anti-cancer therapies.

The present invention thus pertains to a method of identifying, screening, characterising or designing a compound which is capable of modulating the activity of a protein of the JMJD2

subfamily of Jumonji proteins. The method of the invention can be used for screening large numbers of compounds to identify a group of compounds that are candidate compounds for clinical use for treatment of certain cancers especially prostate cancers. Other compounds that do not have activity in the screening assays can be eliminated from further consideration as candidate compounds. The method of the invention therefore has utility in the pharmaceutical industry.

According to the present invention a compound which inhibit the binding of a protein of the JMJD2 subfamily of Jumonji proteins to its substrates (histone peptides, histones or nucleosomes) or cofactors (a-ketoglutarate or iron) can be identified in competitive binding assays. Alternatively, the binding of a test compound to a protein of the JMJD2 subfamily of Jumonji proteins, or its substrates or cofactors can be measured directly. This latter type of assay is called a direct binding assay. Both direct binding assays and competitive binding assays can be used in a variety of different formats, similar to the formats used in immunoassays and receptor binding assays generally known in the art. For a description of different formats for binding assays, including competitive binding assays and direct binding assays, see Basic and Clinical Immunology 7th Edition (D. Stites and A. Terr ed.) 1991; Enzyme Immunoassay, E. T. Maggio, ed., CRC Press, Boca Raton, FIa. (1980); and "Practice and Theory of Enzyme Immunoassays," P. Tijssen, Laboratory Techniques in Biochemistry and Molecular Biology, Elsevier Science Publishers B. V. Amsterdam (1985), each of which is incorporated herein by reference.

In competitive binding assays, for example, a test compound compete with a labeled analyte for specific binding sites on a binding agent bound to a solid surface. In this type of format, the labeled analyte can be labeled histone peptides and the binding agent can be a protein of the JMJD2 subfamily of Jumonji proteins which is bound to a solid phase. Alternatively, the labeled analyte can be labeled JMJD2 protein and the binding agent can be a solid phase histone peptide. The concentration of labeled analyte bound to the capture agent is inversely proportional to the ability of a test compound to compete in the binding assay. The amount of inhibition of labeled analyte by the test compound depends on the binding assay conditions and on the concentrations of binding agent, labeled analyte, and test compound that are used. Under specified assay conditions, a test compound is said to be capable of inhibiting the binding of the substrate (i.e. histone peptides) or co-factor to a protein of the JMJD2 subfamily of Jumonji proteins in a competitive binding assay, if the amount of binding of the labeled analyte to the binding agent is decreased by 10% or more. When a direct binding assay format is used, a test compound is said to inhibit the binding the substrate (i.e. histone peptides) or co-factor (i.e. a-ketoglutarate) to JMJD2 enzyme when the signal measured is twice the background level or higher.

In a competitive binding assay, the sample compound competes with labeled protein for binding to a specific binding agent. As described above, the binding agent may be bound to a solid surface to effect separation of bound labelled protein from the unbound labelled protein. Alternately, the competitive binding assay may be conducted in liquid phase, and any of a variety of techniques known in the art may be used to separate the bound labeled protein from the unbound labeled protein. Following separation, the amount of bound labeled protein is determined. The amount of protein present in the sample is inversely proportional to the amount of labelled protein binding.

Alternatively, a homogenous binding assay may be performed in which a separation step is not needed. In these types of binding assays, the label on the protein is altered by the binding of the protein to its specific binding agent. This alteration in the labelled protein results in a decrease or increase in the signal emitted by label, so that measurement of the label at the end of the binding assay allows for detection or quantitation of the protein.

The binding assay formats described herein employ labeled assay components. The label can be in a variety of forms. The label may be coupled directly or indirectly to the desired component of the assay according to methods well known in the art. A wide variety of labels may be used. The component may be labeled by any one of several methods.

Traditionally, a radioactive label incorporating 3 H, 125 I, 35 S, 14 C, or 32 P is used. Nonradioactive labels include ligands which bind to labeled antibodies, fluorophores, chemiluminescent agents, enzymes, and antibodies which can serve as specific binding pair members for a labelled ligand. The choice of label depends on sensitivity required, ease of conjugation with the compound, stability requirements, and available instrumentation. For a review of various labelling or signal producing systems which may be used, see U.S. Pat. No. 4,391,904, which is incorporated herein by reference.

The terms "histone peptides" and "substrate" refer to those fragments of histones that bind to JMJD2 protein. Methods of production of JMJD2 and histone proteins, for use in screening assays are known to those of skill in the art.

In particular, the method may be a method of testing the ability of a test compound to bind to and optionally modulate the activity of a protein of the JMJD2 subfamily of Jumonji proteins, said method comprising incubating a test compound with a protein of the JMJD2 subfamily of Jumonji proteins, a co-factor of said protein and, optionally, a substrate for demethylation. Particularly, it is within the scope of the present invention to identify a compound which is capable of inhibiting the activity of a protein of the LMJD2 subfamily of Jumonji proteins.

Typically, the method utilise polypeptides of the JMJD2 subfamily of Jumonji proteins and typically a GASCl polypeptide. In particular, the method utilise a human GASCl polypeptide or a fragment of this polypeptide, or variants of these. Accordingly, in an important embodiment the invention provides a method, wherein said protein of the JMJD2 subfamily of Jumonji 5 proteins is selected from the group comprising: a) the amino acid sequence of SEQ ID NO: 1; b) a fragment of the amino acid sequence of SEQ ID NO: 1; c) an amino acid having at least 45% sequence identity to either the sequence in a) or the sequence in b), or both. 10

Cloned members of the JMJD2 subfamily of jumonji proteins include GASCl (Swis prot accession number Q9H3R0, and its homologues JMJD2a and JMJD2b (Swis prot accession numbers 075164 and 094953, respectively).

15 An equally important embodiment of the invention pertains to a method, wherein said protein of the JMJD2 subfamily of Jumonji proteins is selected from the group comprising: a) the amino acid sequences of SEQ ID NO: 1 (GASC-I (JMJ2Dc)), SEQ ID NO: 2 (JMJD2a), SEQ ID NO: 3 (JMJD2b). b) a fragment of any one of the amino acid sequences in a);

20 c) an amino acid sequence having at least 45% sequence identity to any one of the sequences in a) and/or any one of the sequences in b).

The skilled person will realize that the presently known members of the JMJD2 subfamily of Jumonji proteins are approximately 45% identical by sequence when the amino acid sequences

25 are aligned to a best fit. Therefore, for the present application the amino acid sequences used are at least 45% identical to the sequence of any of the members of the JMJD2 subfamily of Jumonji proteins. In particular, the sequence identity may be at least 45%, such as at least 46%, 47%, 48%, 49%, 50%, 52.5%, 55%, 57.5%, 60%, 62.5%, 65%, 67.5%, 70%, 72.5%, 75%, 77.5%, 80%, 82.5%, 85%, 87.5%, 90%, 92.5%, 95%, 97.5%, 98%, 98.5%, 99% or

30 99.5%.

In order to incorporate the functional features of the native polypeptide members of the JMJD2 subfamily of Jumonji proteins it is further preferred that the fragments are of at least 100 amino acids, such as at least 105, 110, 115, 120, 125, 130, 135, 140, 145, 150, 155, 160, 165, 170, 35 175, 180, 185, 190, 195, 200, 210, 220, 230, 240, 250, 260, 270, 280, 290, 300, 320, 340, 360, 380, 400, 420, 440, 460, 480, 500, 520, 540, 560, 580, 600, 620, 640, 660, 680, 700, 750, 800, 850, 900, 950, 1000, or at least 1050 amino acids.

Preferably, the fragments are of at least 150 amino acids, such as at least 151, 152, 153, 154, 40 155, 156, 157, 158, 159, 160, 162, 164, 166, 168, 170, 172, 174, 176, 178, 180, 182, 184, 186, 188, 190, 192, 194, 196, 198 or 200 amino acids.

The present invention further contemplates analogues of the amino acid sequences formed by conservative amino acid substitution. The principle behind conservative amino acid substitution is that certain amino acid pairs have compatible side chains such that, when one amino acid is substituted for the other, there will be only minimal changes in the tertiary structure of the peptide. Rules for conservative substitutions are explained In Bowie et al. Science 247(1990) 1306-1310. It is an object of the present invention to utilise polypeptides, fragments and variants that retain the ability of the JMJD2 subfamily proteins to bind substrate. In some preferred embodiments it is also preferred that the polypeptides, fragments and variants retain the ability of the JMJD2 subfamily proteins to bind co-factors, including α-ketoglutarate.

Where required, each of the polypeptides, fragments and variants, where required, may be provided either in purified or un-purified form, for example as cellular extracts or by purification of the relevant component from such extracts. Alternatively, the polypeptides, fragments and variants can be recombinantly produced by recombinant expression techniques, and purified for use in the assay. Alternatively, the polypeptides, fragments and variants can be expressed recombinantly in a cell for use in cell based assays.

Typically, a polynucleotide encoding the relevant component is provided within an expression vector. Such expression vectors are routinely constructed in the art and may for example involve the use of plasmid DNA and appropriate initiators, promoters, enhancers and other elements, such as for example polyadenylation signals which may be necessary and which are positioned in the correct orientation in order to allow full protein expression. Suitable vectors would be very readily apparent to those of skill in the art, such as those described in more detail in the examples of the present application. Promoter sequences may be inducible or constitutive promoters depending on the selected assay format.

There are a number of common structural features found in the JMJD2 subfamily of Jumonji proteins. These include the JmjN, JmjC, PHD and Tdr sequences, one or more of which are preferred features of the JMJD2 jumonji proteins. These functional domains are characterised by the presence of certain particular amino acid residues located at key positions in the amino acid sequence. While not readily distinguishable when merely inspecting the amino acid sequence of a protein by eye, these domains may be determined using the SMART program (http:/smart. embl-heidelberg.de/smart). The algoritm on which the SMART program is based are described in further details in Letunic et al. ⁵⁸ and Schultz et al. ⁵⁹

Accordingly, in a preferred embodiment of the invention the protein of the JMJD2 subfamily of jumonji proteins as well as the polypeptides, fragments and variants comprise one or more amino acid sequences selected from the group consisting of JmjN-, JmjC-, PHD and TUDOR domains from Jmjd2a: JmjN (AA 13-55):

RIMTFYPTMEEFRNFSRYIAYIESQGAHRAGLAKVVPPKEWKP (SEQ ID NO: 4)

JmjC (AA 142-308) :

EKHVDEWNIGRLRTILDLVEKESGITIEGVNTPYLYFGMWKTSFAWHTEDMDLYSIN YLHFGEPKSWYSV

PPEHGKRLERLAKGFFPGSAQSCEAFLRHKMTLISPLMLKKYGIPFDKVTQEAGEFM ITFPYGYHAGFNHG FNCAESTNFATRRWIEYGKQAVLCSC (SEQ ID NO: 5)

PHD (AA 709-767):

MCFTSTGCSTDINLSTPYLEEDGTSILVSCKKCSVRVHASCYGVPPAKASEDWMCSR CS (SEQ ID NO:

6)

PHD (AA 829-885) : KCIFCKKRRKRTAGCCVQCSHGRCPTAFHVSCAQAAGVMMQPDDWPFVVFITCFRHK (SEQ ID NO:

7)

TUDOR (AA 897-954):

QSITAGQKVISKHKNGRFYQCEVVRLTTETFYEVNFDDGSFSDNLYPEDIVSQDCLQ F (SEQ ID NO: 8)

TUDOR (AA 955-1011): GPPAEGEVVQVRWTDGQVYGAKFVASHPIQMYQVEFEDGSQLVVKRDDVYTLDEELP (SEQ ID NO:

9)

In an equally preferred embodiment of the invention the protein of the JMJD2 subfamily of jumonji proteins as well as the polypeptides, fragments and variants comprise one or more amino acid sequences selected from the group consisting of JmjN-, JmjC-, PHD and TUDOR domains from Jmjd2b:

JmjN (AA 14-56) :

KIMTFRPTMEEFKDFNKYVAYIESQGAHRAGLAKIIPPKEWKP (SEQ ID NO: 10)

JmJC (AA 143-309): DDDVAQWNIGSLRTILDMVERECGTIIEGVNTPYLYFGMWKTTFAWHTEDMDLYSINYLH FGEPKSWYA

IPPEHGKRLERLAIGFFPGSSQGCDAFLRHKMTLISPIILKKYGIPFSRITQEAGEF MITFPYGYHAGFNHGF

NCAESTNFATLRWIDYGKVATQCTC (SEQ ID NO: 11)

PHD (AA 731-789):

MCFTSGGENTEPLPANSYIGDDGTSPLIACGKCCLQVHASCYGIRPELVNEGWTCSR CA (SEQ ID NO: 12)

PHD (AA 851-907):

KCVYCRKRMKKVSGACIQCSYEHCSTSFHVTCAHAAGVLMEPDDWPYVVSITCLKHK (SEQ ID NO:

13)

TUDOR (AA 917-974): RAVSLGQVVITKNRNGLYYRCRVIGAASQTCYEVNFDDGSYSDNLYPESITSRDCVQL (SEQ ID NO:

14)

TUDOR (AA 975-1031):

GPPSEGELVELRWTDGNLYKAKFISSVTSHIYQVEFEDGSQLTVKRGDIFTLEEELP (SEQ ID NO: 15)

In embodiments of the invention which currently are the most preferred the protein of the JMJD2 subfamily of jumonji proteins as well as the polypeptides, fragments and variants

comprise one or more amino acid sequences selected from the group consisting of JmjN-, JmjC, PHD and TUDOR domains from Jmjd2 (GASCl) :

JmjN (AA 15-57) :

KIMTFRPSMEEFREFNKYLAYMESKGAHRAGLAKVIPPKEWKP (SEQ ID NO: 16) JmjC (AA 144-310) :

DEGVDEWNIARINTVLDVVEEECGISIEGVNTPYLYFGMWKTTFAWHTEDMDLYSIN YLHFGEPKSWYAI

PPEHGKRLERLAQGFFPSSSQGCDAFLRHKMTLISPSVLKKYGIPFDKITQEAGEFM ITFPYGYHAGFNHG

FNCAESTNFATVRWIDYGKVAKLCTC (SEQ ID NO: 17)

PHD (AA 689-747) : MCFIYSEENIEYSPPNAFLEEDGTSLLISCAKCCVRVHASCYGIPSHEICDGWLCARCK (SEQ ID NO:

18)

PHD (AA 809-865) :

KCIFCRHRVKRVSGACIQCSYGRCPASFHVTCAHAAGVLMEPDDWPYVVNITCFRHK (SEQ ID NO:

19) TUDOR (AA 877-934) :

KVISVGQTVITKHRNTRYYSCRVMAVTSQTFYEVMFDDGSFSRDTFPEDIVSRDCLK L (SEQ ID NO:

20)

TUDOR (AA 935-991) "

GPPAEGEVVQVKWPDGKLYGAKYFGSNIAHMYQVEFEDGSQIAMKREDIYTLDEELP (SEQ ID NO: 21)

It should be acknowledged that the amino acid sequences of these domains may be altered by substitution without loosing function. Comprised by the present invention therefore are protein of the JMJD2 subfamily of jumonji proteins as well as the polypeptides, fragments and variants comprise one or more amino acid sequences which are at least 75% identical to any one of the sequences above, such as at least 80%, 85%, 90%, 95%, 98%, 99%, or at least 99.5% identical to any of the above sequences.

It is to be understood that the protein of the JMJD2 subfamily of Jumonji proteins and a potential modulator of the activity may be incubated together under conditions which in the absence of any inhibitor of the enzyme activity provide for de-methylation of lysine 9 on histone H3 Since GASCl is a Fe(II)- and α-ketoglutarate dependent de-methylase, the protein of the JMJD2 subfamily of Jumonji proteins is typically contacted with a substrate in the presence of a co-substrate. Typically, α-ketoglutarate, Fe(II) and ascorbic acid are used as the co-substrate, but it is within the scope of the present invention to use other co-substrates.

Various approaches to determining the activity of a protein of the JMJD2 sub-family of jumonji proteins and of the polypeptides, fragments and variants described above are available to the skilled person, including measurements of substrate and/or co-substrate utilization as well as measurement of product and/or by-product formation. In particular embodiments of the invention, the method thus comprises the additional steps of:

a) monitoring in a test sample any of the following parameters: i) the methylation state of the substrate; Ii) the release of formaldehyde; iii) the carboxylation state of an α-ketoglutarate substrate co-factor; iv) the oxygen consumption b) comparing the values obtained for the test sample in step a) with values obtained for a control sample, thereby determining the ability of the test compound to modulate the activity of a protein of the JMJD2 subfamily of Jumonji proteins.

Additionally, the method of the invention may incorporate measurements of any downstream effects of the factors mentioned above. In particular, since GASC action may affect transcription of certain genes it is possible to envision the use of gene reporter assays for measuring GASC activity. Such assays may rely on the measurements of activation of transcription of genes that are naturally repressed by a protein of the JMJD2 subfamily of Jumonji proteins. In addition, one may contemplate in vitro assays relying on the use genetically engineered reporter constructs including a promoter that is subject to methylation and a transcriptionally linked reporter gene.

In particular embodiments these reporter assays may involve the use of a gene which encodes a fluorescent reporter; currently one of the most commonly used fluorescent reporters is Green Fluorescent Protein (GFP).

Determination may be quantitative or qualitative. Both quantitative and qualitative determinations may, as stipulated above, be carried out in the presence of a control. If a control is used it may, depending on which assay method is used, be provided by the following steps: a) Incubating a protein of the 3M3D2 subfamily of Jumonji proteins and a substrate for demethylation under conditions allowing demethylation b) monitoring any of the following parameters in a test sample: i) the methylation state of the substrate; ii) the release of formaldehyde in a test sample; iii) the carboxylation state of an α-ketoglutarate substrate co-factor; iv) the oxygen consumption

In some embodiments of the present invention the incubation time and methylation state of the substrate and/or the release of formaldehyde are used to determine a rate of demethylation, which is used for comparison.

As the natural substrate for GASCl is histone H3 it will be understood that in some embodiments of the invention the substrate for demethylation comprises a methylated site corresponding to lysine 9 on histone H3. In a particular embodiment of the invention the

substrate for demethylation is a peptide comprising the amino acid sequence of SEQ ID NO: 22 (TARKSTG).

The lysine residue in said amino acid sequence (SEQ ID NO: 22) corresponding to Lys9 in 5 Histone H3 is methylated. In most embodiments of the invention it is preferred that the lysine residue is tri-methylated, but assays may also be performed wherein said lysine residue is di- or mono-methylated.

Currently, as a substrate it is preferred to use a peptide, such as a synthetic peptide, 10 comprising N-terminal residues 6-12 of histone H3, such as residues 5-9, 4-10, 3-10, 2-11, 1-

12, 1-13, 1-14, 1-15, 1-20, 1-25, 1-30, 1-35, 1-40, 1-45, 1-50, 1-55, 1-60, 1-65, 1-70, 1-75, 1-80, 1-85, 1-90, 1-95 or 1-100 of histone H3. In the present application the full sequence of Histone H3 is provided in SEQ ID NO.: 23

15 In accordance with the above description of the substrate for demethylation, it is preferred that said substrate is a peptide of at least 6 amino acid residues, such as at least 7, 8, 9, 10, 11, 12,

13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 32, 34, 36, 38, 40, 45, 50, 60, 70, 80, 90, 100, 120, 140, 160, 180, 200, 250, 300 350, 400, 500, 600, 700, 800, 900 or such as least 1000 amino acid residues. A peptide of 80 amino acid residues or less is

20 conveniently obtained through chemical synthesis using commercially available services or apparatus.

The skilled artisan will realize that peptides which are at least 75% identical to the N-terminal residues of histone H3 as specified above, such as at least 77.5%, 80%, 82.5%, 85%, 87.5%, 25 90%, 92.5%, 95%, 97.5%, 98%, 98.5%, 99% or 99.5% identical to the N-terminal residues of histone H3. The peptides may incorporate substitutions as explained in detail above for the protein of the JMJD2 subfamily of Jumonji proteins. In particular, conservative substitutions may be preferred, following the same principles as described above.

30 In further preferred embodiments of the invention, as a minimum, the substrate comprises an amino acid sequence selected from the group consisting of:

In a currently most preferred embodiment, a synthetic peptide 43 amino acids long mimicking the N-terminal tail (1-40) of histone H3, ARTKQTARKSTGGKAPRKQLATKAARKSAPATGGVKKPHR 35 (SEQ ID NO: 24) is used as a substrate.

It may further be preferred to use a substrate selected from the group consisting of: bulk histones, synthetic peptides, and nucleosomes.

40 As mentioned above the method of the invention may be performed with a substrate, wherein the methylated site is mono-methylated, di-methylated or tri-methylated

As mentioned above, α-ketoglutarate serves as a co-factor of GASCl and is therefore be required in the method according to the invention. It is further within the scope of the invention to use as co-factor an analogue of α-ketoglutarate. Additionally, the GASCl catalytic acitivity benefits from the presence of ascorbic acid, and in addition to α-ketoglutarate, Fe(II) ions may be important as a co-factor. Accordingly, in a preferred embodiment of the invention the compound and said protein of the JMJD2 subfamily of Jumonji proteins are also incubated with Fe(II) ions, and/or ascorbic acid. It is to be understood that the presence of ascorbic acid is not essential, but ascorbic acid may be needed in order to obtain the full catalytic activity since it converts the iron present in the reaction to the appropriate form, Fe(II).

A cell-free system may be applied when testing compounds acting directly on a protein of the JMJD2 subfamily of Jumonji proteins. The test system is contacted with the test compound and the ability of the test compound to regulate activity of said protein of the JMJD2 subfamily of Jumonji proteins is determined by measuring substrate (nucleosomes, histones or histone- peptides) conversion utilizing an immunoassay.

The peptide substrate of the protein of the JMJD2 subfamily of Jumonji proteins, or the product of its catalytic activity, may for instance be immobilised, e.g. on a plate or bead, and the methylation state or the de-methylation of the appropriate methylated site detected using an antibody or other binding molecule which binds the peptide with different affinity depending on the methylation state. Accordingly, immunological binding partners, which recognize either tri-, di- or mono-methylated histone H3-K9 constitute a part of the test system and will enable a fast determination of a compounds effect on the activity. Such binding partners, including substrate- and product-specific antibodies, are commercially available.

Assay methods of the present invention may also take the form of an in vivo assay. The in vivo assay may be performed using cell based, organ based or whole animal assays. Preferably, the in vivo assay is performed in a cell based system, for instance using a cell line in which the relevant polypeptides or peptides are expressed from one or more vectors introduced in the cell. The cell may, for example, be a yeast cell or a cell of mammalian origin.

In addition, the in vivo assays used in relation to the present invention may rely on isolated cells/primary cells or tissue sections. In a particular interesting embodiment the in vivo assay involves the use of the KYSE-150 cell line, which is a human esophageal squamous cell carcinoma established from the poorly differentiated esophageal squamous cell carcinoma resected from upper (cervical) esophagus of a 49-year-old Japanese woman after receiving radiotherapy (tumor was invading contiguous structures); described as carrying amplified oncogenes, c-erb-B (8x) and cyclin Dl (4x) and producing tumors in nude mice confirmed as human with IEF of AST, LDH, MDH

Measurements of the methylation state may, as presently preferred, be combined with the release of formaldehyde- ³ H by fluorography. A similar technique for measuring formaldehyde release by a formaldehyde-dehydrogenase coupled assay is also contemplated in the present invention.

In preferred embodiments of the invention the methylation state of the substrate and/or the release of formaldehyde and/or the carboxylation state of an α-ketoglutarate substrate co- factor and/or the oxygen consumption are measured by immunoassays (ELISA, RIA, IRMA, TRIFMA), immunoprecipitation, western blotting, BIAcore, x-ray crystallography, in solution NMR, mass-spectrometry, spectroscopic or fluorescence techniques.

In a currently preferred embodiment of the invention, the measurements of substrate utilization and/or product formation is performed by immunofluorescence analyses, preferably by confocal immunostainings. In the presently most preferred embodiments of the invention the peptide substrate is linked to biotin as explained in the examples.

When measuring the methylation state in assays relying on substrate specific antibodies a decreased level of binding of the immunological binding partner, as compared to suitable controls, means a decrease in tπ- or di-methylated H3-K9 which correlate with an increase in the activity of a protein of the JMJD2 subfamily of Jumonji proteins. Binding of the immunological binding partner can be assessed by techniques generally know in the art, for example Western blot, ELISA, RIA, TRIFMA, immuno-precipitation or histology. Preferred techniques comprise high-throughput screening techniques as Fluorescence Polarisation, Time Resolved Fluorescence Resonance, energy Transfer Assay (TR-FRET), Scintillation Proximity Assay and "Fluorescence Quenching" Assay. The expression can be coupled to an easy detectable reporter protein, such as, but not limited to, β-galactosidase, chloramphenicol acetyltransferase (CAT), Green Fluorescent Protein, or luciferase

For competition measurements a synthetic or natural occurring histone peptide might be supplied either in a labelled or unlabelled form. The antibodies may be used with or without modifications. The antibodies may be labelled by joining them, either covalently or non- covalently, with a reporter molecule. Suitable reporter molecules or labels, which may be used for ease of detection, include radionucleotides, enzymes, fluorescent molecules, chemiluminescent, or chromogenic agents as well as substrates, cofactors, inhibitors, magnetic particles, and the like. Antibodies or synthetic peptides of the kit might be immobilised, preferably on a solid surface like a micro-titter plate, possibly by conjugation to a suitable protein carrier like BSA, thyroglobulm, ovalbumin or keyhole limpet hemocyanine.

In a presently preferred embodiment of the invention, the method comprises the steps of: ι) preparing a suspension of a substrate for demethylation as defined in any of claims 8-13 in a solution containing iron and alpha-ketoglutarate;

ii) adding said test compound to the suspension; iii) adding to said suspension in ii) a protein of the JMJD2 subfamily of Jumonji proteins as defined in any of claims 2-4; iv) incubating said suspension, preferably at a temperature of 35-38 ⁰ C for 10 minutes or more; and v) determining the extent of demethylation of said substrate.

It is further preferred that the extent of demethylation is assayed using an immunological binding partner specific for histone H3 methylated at lysine 9.

Also, it is preferred that said immunological binding partner is an anti trimethylated histone H3-K9 antibody. Equally preferred is the use of an anti dimethylated histone H3-K9 antibody as the immunological binding partner.

In a further preferred embodiment, the extent of demethylation is assayed using the release of formaldehyde. Further, the extent of demethylation may be assayed by measuring consumption of oxygen.

Alternatively, the extent of demethylation is assayed by measuring the carboxylation state of an α-ketoglutarate substrate co-factor.

As an example the buffer in i) comprises a final concentration of 50 mM Hepes pH 7.5, 5OmM KCI, 4mM MgCI ₂ , 1 mM α-ketoglutarate, 40μM FeSO ₄ , 2mM ascorbic acid. For practical reasons it is further preferred that the suspension has a total volume of 10-50 μl.

Finally, in addition, in order to minimize degradation of the protein components of the reaction it may be advantageous to add one or more protease inhibitors to a final concentration of 1-5 μg/μl. As an example, aprotinine and leupeptine may be included as protease inhibitors.

With respect to the above description of the various aspects of the present invention and of the specific embodiments of these aspects it should be understood that any feature and characteristic described or mentioned above in connection with one aspect and/or one embodiment of an aspect of the invention also apply by analogy to any or all other aspects and/or embodiments of the invention described.

When an object according to the present invention or one of its features or characteristics is referred to in singular this also refers to the object or its features or characteristics in plural. As an example, when referring to "a polypeptide" it is to be understood as referring to one or more polypeptides.

Throughout the present specification the word "comprise", or variations such as "comprises" or "comprising", will be understood to imply the inclusion of a stated element, integer or step, or group of elements, integers or steps, but not the exclusion of any other element, integer or step, or group of elements, integers or steps

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1-2: Jumonji histone demethylases GASCl, FBXLIl and their human homologues. a, Diagram of the GASCl, FBXLIl and their human homologues with their domains: JmjC-Jumonji C domain, JmjN-Jumonji N domain, Tdr-Tudor domain, PHD- Plant homeodomain CXXC- CxxC zinc finger domain, FBX-Fbox domain, LR- Leucine rich repeat. Functional domains are determined using the SMART program (http:/smart.embl- heidelberg.de/smart). b, Phylogenetic tree of the human JmjC-domain containing proteins. The phylogenetic tree is calculated based on the JmjC domains of the proteins using the Neighbour Joining (NJ) method The sequences used for this calculation are provided in Supplementary Figure S2.

Figure 3-6: GASCl demethylates tri- and di-methylated histone H3K9. a, Purification of recombinant GASCl by size exclusion chromatography. 0,5 ml fractions were collected and 10 μl of the material was incubated with bulk histones for 30 mm. at 37°C. Demethylation activity was assayed by blotting for tπ-methylated H3K9. b, Demethylation assay using selected fractions (F4, F6, F8, F9, FlO, FIl, F12, F14, F16) from the size-exclusion chromatography of GASCl. Fractions were assayed for demethylating activity by immunoblotting as indicated, c, Mass spectrometric analysis of demethylation of a H3K9me3 peptide (ARTKQTARKSTGGKAPRKQLATKAARKSAPATGGVKKPHRYC-Ttds-K(Bιotιn) -NH2) by purified GASCl. The peptide was analyzed by LC-MS on a LTQ-FT after treatment with GASCl (+GASC1) and without GASCl (-GASC1). The deviations of the measured, mono-isotopic masses from the calculated masses are given in brackets. The calculated neutral masses are H3K9mel : 5130.885 Da, H3K9me2: 5144.900 Da, H3K9me3: 5158.916 Da, mass difference as result of de-methylation: 14.016 da H3K9 (unmethylated) was not observed. The multiplicity of the individual peaks is caused by the natural occurrence of different isotopes as simulated by IsoPro 3.0 (http://members.aol.com/msmssoft/) for our peptide (molecular formula C223H390N75O61S2), charge state +10 and resolution 50,000 (simulated spectrum), d, Demethylation assay using recombinant GASCl and synthetic tri-methylated H3K9 peptides as substrates. Peptides are incubated with 6 μg His-GASCl in the presence of co-factors for 30 mm. at 37°C, one third of the material was analyzed by western blotting, the rest was used fo mass spectrometry analysis (Fig 2e). e, Alignment of the JmjC domains of the human Jumomi proteins GASCl, JMJD2A, FIH and FBXLIl and the fission yeast Jumomi protein Epel. Yellow highlighting indicates similar or identical ammo acids. Black, red asterisks indicate residues predicted to be involved in iron binding and α-ketoglutarate, respectively. A green asterisks indicates residues mutated in GASCl iron-binding mutants. Arrows indicate strands of the

double-stranded β-helix numbered corresponding to β-strands 8-15 of FIH, as defined by the PDB sum entry for Ih2k. f Predicted structure of GASCl (blue) in complex with the tπ- methylated H3 histone tail (1-11) (green) and its co-factors α-ketoglutarate (red) and iron (magenta), the lysine H3K9 methyl groups are marked in yellow. Left panel shows an overview, right panel shows a magnification of the iron-binding residues Hιsl90, Hιs288, Glul92, (white sticks), g, Demethylation assay using varying amounts of wild-type and mutant (H190G/E192A) GASCl, using bulk histones as substrate.

Figure 7-8: GASCl demethylation occurs through a hydroxylation reaction dependent on iron and α-ketoglutarate. a, Proposed reaction scheme for GASCl-mediated demethylation of tπ- and dι-methylated H3K9. 1. Fe(II) (magenta) is bound to GASCl by the facial triad of metal binding residues Hιsl90, Glul92 and Hιs288 (blue text). 2. α-ketoglutarate (red) is bound, probably through residues Thrl86 and Lys208 3. Iron (magenta) subsequently binds molecular oxygen 4. Next oxidative decarboxylation of α-ketoglutarate (red) produces carbon dioxide, succinate (red) and ferryl (FeIV=O, magenta). 5. Ferryl (magenta) subsequently hydroxylates a methyl group of lysine H3K9 (green), releasing formaldehyde. 6. Finally the hydroxylated histone tail (green) and succinate (red) can leave the GASCl molecule leading it vacant for new demethylation cycle, b, Demethylation assay using synthetic H3K9me3 peptide as substrates. Peptides were incubated with 20 μg His-GASCl in the presence of, or absence of its co-factors ascorbic acid, α-ketoglutarate and Fe2SO4, and EDTA, for 30 mm. at 37 ⁰ C, and analyzed by western blotting. The relative demethylation activity of GASCl was quantified using a Fujifilm darkbox LAS 3000 reader and normalized to the loading control (α-bιotιn). H3K9me3 peptide (lanel) was set to 100%. Lane 2 -7; H3K9me3 peptide incubated in demethylation buffer in the presence or absence of EDTA or co-factors as indicated, c, Demethylation assay using recombinant GASCl and bulk histones as substrates. Histones were incubated with His-GASCl in the presence of co-factors (FeSO ₄ and α-ketoglutarate) for 30 mm. at 37°C, in the presence or absence of varying amounts of the α-ketoglutarate analog N- oxalylglycme. The methylation status was evaluated by western blotting, d, Binding mode of 2- ketoglutarate and N-oxalylglycine at the Fe(II) active site, e, Formaldehyde release by GASCl mediated H3K9 _me3 demethylation. Formaldehyde release was measured using a demethylation- FDH-coupled assay ⁴ . The demethylation-FDH-coupled assays were carried out using fixed amount of the enzyme GASCl (40 μg) but varying amounts of the substrate H3K9 _me3 peptide. Formaldehyde production was monitored by measuring NADH production at OD 340 nm. 50 μg H3K9 peptide (unmethylated) as substrate was used as negative controls.

Figure 9-11: Ectopic expression of GASCl leads to loss of H3K9 di- and tri-methylation in vivo, a, Confocal microscopy of Tιg-3 cells transduced with pBabe-GASCl. Tri-methylated H3K9 is lost in Tιg-3 cells over-expressing GASCl. White arrows indicate cells expressing HA- tagged GASCl. b, Loss of the ability to demethylate tri-methylated H3K9 in U2OS cells expressing the GASCl H190G/E192A mutants, evaluated by immunoflourescence. White arrows indicate cells expressing HA-tagged GASCl or expressing the HA-tagged H190G/E192A GASCl

mutant, c, Loss of the ability to demethylate tπ-methylated H3K9 in U2OS cells transfected with H190G/E192A GASCl mutant, evaluated by western blotting, d, Loss of tπ- and di-methylated H3K9 and increase of mono-methylated H3K9 in U2OS cells over-expressing JMJD2a. e, Loss of tπ- and di-methylated H3K9 and increase of mono-methylated H3K9 in U2OS cells over- expressing JMJD2b. f, Confocal microscopy of 3T3 cells transfected with HA-tagged GASCl or the HA-tagged H190G/E192A GASCl mutant. White arrows indicate cells expressing HA-tagged proteins. HPlβ is delocalized in cells over-expressing GASCl. g, Western blotting of total lysate and chromatin bound fraction from HEK 293 cells with tetracycline- (Tet) inducible GASCl- expression (Myc-tagged). Addition (+) or omission (-) of Tet as indicated.

Figure 12-14: GASCl expression induces growth and is positively correlated with tumourigenicity. a. Transfection of KYSE-150 and KYSE-450 with siRNA to GASCl or scrambled siRNA control. The relative amount of GASCl messenger RNA was determined by real-time quantitative RT-PCR. b, Transfection of KYSE-150 with LMP-GASCl or empty LMP vector followed by selection for 3 days. The relative amount of GASCl messenger RNA was determined by real-time quantitative RT-PCR. c, Binding mode of quercetin at the Fe(II) active site of GASCl. d, Demethylation assay using recombinant GASCl and bulk histones as substrates. Histones were incubated with His-GASCl in the presence of co-factors (FeSO ₄ and α-ketoglutarate) for 30 mm. at 37 ^C C, in the presence or absence of varying amounts of quercetin. The methylation status was evaluated by western blotting, e, Effect of quercetin on the H3K9me3 methylation status of KYSE-150 and U2OS cells as evaluated by immunoflourescence. f, Model for the involvement of GASCl in cancer development.

Figure 15 (Sl): Histone peptide pulldown of HeLa nuclear proteins Pull-down assays were performed using HeLa cell nuclear extracts and histone H3 peptide tπ-methylated at lysine 9. The HeIa nuclear extract was pre-cleared with uncoupled streptavidin beads (precleaπng 1) and then with streptavidin beads coupled to unmodified H3 histone (precleaπng 2). The pre- cleared lysate was then incubated with H3 histone peptide tπmethylated at K9 (H3K9me3), precoupled to streptavidin beads. Beads were washed and bound proteins were subsequently eluted, resolved by SDS PAGE and visualized by silver staining. Protein bands enriched in the H3K9me3 pull-down were subse-quently identified by mass-spectrometry. One of the enriched bands was identified as the jumonji protein GASCl (see arrow). The HPl α-γ proteins also enriched in the pull-down are likewise indicated. Molecular weight markers are indicated on the left.

Figure 16-19 (S2): Alignment of the JmjC domain human Jumonji proteins. Proteins were aligned using Vector NTI applying a Clustal W algorithm. Conserved residues, blocks of similar residues and identical residues are shaded in grey. Numbers in parenthesis indicates residue number within the protein.

Figure 20 (S3): Purification of His-tagged recombinant GASCl, GASCl (H190G/E192A) mutant and JMJD2A and JMJD2b. Fractions 1-4 eluted from a cobalt-

affinity column, were subjected to SDS-PAGE and Coomassie stained. Fractions F2 were further purified by size exclusion chromatography to obtain highly pure protein preparations.

Figure 21-23 (S4): GASC demethylates H3K9me3 and H3K9me2. a, Demethylation assay using varying amounts of recombinant GASCl and oligonucleosomes or bulk histones as substrates, b, Demethylation assay using recombinant GASCl and bulk histones as substrates. Histones were incubated with varying amounts of His-GASCl as indicated, in the presence of its co-factors for 30 mm. at 37°C, and analyzed by western blot analysis. Heat inactivation (inact.), for 5 mm. at 6O ⁰ C, abrogated the demethylating activity of GASCl. c, Demethylation assay using recombinant GASCl and synthetic tπ-methylated H3K9 peptides as substrates. Peptides are incubated with 10 μg His-GASCl in the presence of co-factors for 30 mm. at 37°C. d, Demethylation assay using recombinant GASCl or JMJD2a and synthetic tπ- dι- and mono- methylated H3-K9 peptides as substrates. Peptides are incubated with 8 μg His-GASCl or His- JMJD2a in the presence of co-factors for 30 mm. at 37°C, and analyzed by western blot analysis.

Figure 24-25 (S5): JMJD2A and JMJD2B demethylate histones tri- and di- methylated at H3-K9. a, Purification of recombinant 3MJD2A by size exclusion chromatography. One ml fractions were collected and 10μl of the material was incubated with bulk histones for 30 mm. at 37°C. Demethylation activity was assayed for by blotting for tπ-methylated H3-K9. b, Demethylation assay using recombinant JMJD2A and bulk histones as substrates. Various fractions from size-exclusion chromatography (Fig. S5a) were analyzed by western blot analysis, c, Purification of recombinant JMJD2B by size exclusion chromato-graphy. One ml fractions were collected and 10μl of the material was incubated with bulk histones for 30 mm. at 37°C. Demethylation activity was assayed for by blotting for tπ-methylated H3-K9.

Figure 26 (S6): Size exclusion chromatography of wild-type and H190G/E192A mutant GASCl.

Figure 27 (S7): α-Ketoglutarate analogues and cobalt and nickel salts inhibit GASCl demethylation activity. Demethylation assay using recombinant GASCl and bulk histones as substrates. Histones were incubated with His-GASCl in the presence of co-factors for 30 mm. at 37°C, and in the presence or absence of either (a), CoCI ₂ or (b) NiSO ₄ and analyzed by western blot analysis.

Figure 28-29 (S8): Ectopic expression of GASCl does not affect H3K4 and H3K27 methylation. a, H3K4 tπmethylation status in U2OS cells overexpressing GASCl evaluated by immunoflourescence. b, H3K27 tπmethylation status in U2OS and Tιg3 cells overexpressing GASCl evaluated by immuno-flourescence.

Figure 30 (S9): Putative role of GASCl in heterochromatin modelling/plasticity, a,

Confocal microscopy of U2OS cells transfected with HA-tagged JMJD2a and JMJD2b. White

arrows indicate cells expressing HA-tagged GASCl. HPlβ is delocalized in cells over-expressing JMJD2a and JMJD2b. b, Model for GASCl's role in heterochromatin modelling and plasticity. Recruitment of H3K9 specific histone methyl-transferases (HMTs) as GLP, G9a and SUV39H1/H2 to chromatin will lead to the progressive methylation of the H3K9 mark. First H3K9 can be mono-methylated, by enzymes as GLP, G9a. Processive HMTs as SUV39H1/H2 can subsequently act to produce dι- and tri-methylated H3K9 permitting the binding of HPl. In turn, HPl-binding will lead to further recruitment of HMTs and propagation of heterochromatin. GASCl may counteract the spreading and formation of heterochromatin by removing the di-methyl and tri- methyl marks hampering HPl-binding. GASCl may work in concert with other demethylases as LSDl ⁴⁶ and JMJD2a, b to counter heterochromatin.

Figure 31-33 (SlO): H3K9me3 demethylation activity of various GASCl deletion mutants, a. Schematic representation of generated GASCl deletion mutants used in the present study, numbers in parentheses indicate deleted residues b, Demethylation activity of wildtype GASCl and GASC deletion mutants evaluated by immunoflourescence.

Figure (SIl): The expression of the JMJD2 family of histone demethylases is increased in prostate carcinoma, a. Data from Lapointe et al (2004) PNAS 101:811-816. The study includes 62 primary prostate tumors (PC, 61 adenocarcinomas and one adenoid cystic tumor), 41 matched normal prostate tissues (NP, from the noncancerous region of the prostate), and nine unmatched (i.e. different patient) pelvic lymph node metastases (LNM). Gene expression profiling was performed by using cDNA microarrays containing 26,260 different human genes (UniGene clusters). Additional details of the study including pathological and clinical data are available at Oncomine (www.oncomine.org) or on the PNAS web site (www.PNAS.org). Bars indicated medians. P-values Mann-Withney U-test (one-tailed) are provided, b. Data from Yu et al J CIm Oncol (2004) 22:2790-2799. The study comprises human samples from men of various ages, included prostate tumors (PC, completely free of normal prostate acinar cells) and normal prostate tissues (NP, adjacent to tumor free of tumor cells). Gene expression profiling was performed by using the Affymetπx (Santa Clara, CA) U95a, U95b, and U95c chip sets (37,777 genes and expression sequence tags). Additional details of the study are available at the Oncomine web site (www.oncomine.org). Bars indicated medians. P- values Mann-Withney U-test (one-tailed) are provided.

EXAMPLES

In an effort to identify proteins interacting with tri-methylated variants of H3K9, HeLa nuclear proteins were affinity purified using biotinylated H3 peptides immobilized on streptavidin- agarose. Synthetic 43-mers mimicking the whole N-terminal tail of histone HS _{1 40} , either un- methylated or tπ-methylated at lysine 9, were used in the in vitro binding experiments. Bound

proteins were eluted from the agarose-matπx, separated by SDS PAGE, silver stained and identified by mass spectrometry. Several proteins were specifically enriched on tπ-methylated H3K9 as compared to un-methylated H3K9, supplementary figure Sl.

Among the proteins identified was the Jumonji protein GASCl (Fig. Ia). Due to the presence of a Jumonji domain JmjN and a related JmjC domain the protein is also denoted JMJD2c ⁸ ' ⁹

(Fig. Ia). The Jumonji protein family comprises a diverse group of proteins that belong to the α- ketoglutarate-dependent oxygenase superfamily. These proteins regulate various cellular processes, comprising cell-cycle progression and transcriptional regulation ⁵ ' ^{10 14} . A widespread feature of this protein group is the ability to bind Fe(II) ions. In addition, many of these proteins have the capability to hydroxylate protein substrates using α-ketoglutarate as a co- factor ¹⁵ . Of special interest, JmjC domain proteins have recently been identified as possible candidates for histone demethylases ¹⁵ .

Since GASCl was found to be associated with tri-methylated (me3) H3K9, the idea that GASCl might be involved in demethylating this epigenetic mark was intriguing. To test this possibility, recombinant full-length His-tagged human GASCl was generated ad purified from bacculovirus- infected insect cells (Supplementary Fig. S3). The affinity purified GASCl was further purified by size exclusion chromatography, and the eluted highly pure GASCl was tested for demethylation activity, by incubation with bulk histones. As shown in Figure 3a, GASCl very efficiently reduced tπ-methylation at H3K9 as evaluated by immuno-blotting. It was also found that GASCl could demethylate oligonucleosomes, the relevant physiological template (supplementary figure S4a).

To further analyse the specificity of the demethylating activity of GASCl, various fraction from the size-exclusion chromatography of GASCl representing high to low amounts of GASCl were incubated with bulk histones and synthetic H3-peptιdes methylated at lysine K9 or K27, and the methylation status of various epigenetic marks was evaluated by immunoblotting (Fig. 3b and S4b). Here dι- and tri-methylated H3K9 from histones was completely removed in the fractions with the highest concentrations of GASCl (F9-12). A concomitant increase in mono- methylated (mel) H3K9 was also noted, consistent with GASCl converting tri-methyl H3K9 to di-methyl (me2) and further to mono-methyl H3K9 (mel).

In contrast, the levels of other tested epigenetic marks including di-methylated H3K4, tπ- methylated H3K4, tπ-methylated H3K27 and tπ-methylated H4K20 appeared to be unaffected by GASCl treatment (Fig. 3b and S4b).

Interestingly, di-methylated H3K9 appeared to be increased when histones were reacted with low to moderate amounts of GASCl (F6, F8, F14 and F16). This indicates that the reason for the apparent increase in the di-methylated H3K9 mark could be due to a steady state level of this mark being reached in the initial phase of the demethylation reaction. Thus, it can be envisioned that although some di-methyl H3K9 is lost through GASCl demethylation, additional

di-methyl H3K9 will be added to the global pool by GASCl demethylation of tπ-methyl lysine, at least as long as significant levels of tπ-methyl H3K9 are present.

Consistent with the results in Figure 3a and b, the demethylation activity of GASCl vary in a concentration dependent manner and can be abolished by heat-denaturation of GASCl, Fig. S4 b,c , suggesting that the reaction is an enzyme-reaction. An additional in vitro demethylation study was performed, where pure synthetic tπ-methylated H3K9 peptide was incubated with GASCl Following 30 minutes incubation of pure tri-methylated H3K9 with GASCl almost all tri- methyl H3K9 was converted to di- and mono-methyl H3K9 (Fig. S4 c,d).

To formally show that GASCl does indeed demethylate tri- and di-methylated H3-K9, mass spectrometry was performed on synthetic peptides that had been incubated alone or in the presence of GASCl. This analysis showed that the majority of H3K9me3 peptide (approximately 80%) was converted to di- and mono-methylated H3-K9 after incubation of with GASCl (Fig. 4c-6f).

Next it was considered whether the close GASCl-homologues JMJD2a and JMJD2b (Fig.l a, Fig. 2 b) could also demethylate H3K9 in vivo. To investigate this full-length human JMJD2a, JMJD2b were cloned and recombinant proteins were produced from bacculovirus infected cells. Both homologues could demethylate H3K9me3 and H3K9me2 from bulk histones (supplementary figure S5). Likewise, GASCl and its homolog JMJD2A appeared to have a comparable ability to demethylate synthetic H3K9 substrates (supplementary figure S4 d,e). Together, these results strongly suggest that both tri- and di-methylated H3K9 are substrates of GASCl and its homologues in vitro.

To obtain insights into the specificity of the demethylation reaction in vιvo,πs£ human GASCl was modelled onto the structure of FIH, the only Jumonji protein for which the three- dimensional structure has been resolved ¹¹ ' ¹⁵ ' ^{17 19} . In agreement with the alignment (Fig. 5 and supplementary Fig. S2), the modelling indicated that residues histidine H190, glutamic acid E192, and histidine H288 form an essential part of the iron-binding groove of GASCl (Fig. 6h, ι). Further, the in silico model also predicted that residues T187 and R208 are involved in binding of α-ketoglutarate and critical for the demethylation function of GASCl. Next, the tπ- methylated H3 histone tail was modelled onto the predicted GASCl structure. The H3 histone tail tri-methylated at K9 fitted well into the iron-binding groove of GASCl and the K9 methyl groups could be placed in close vicinity to the reactive iron without causing any steric clashes between the methylated histone tail and the GASCl structure (Fig. 6h). On the basis of this in silico model it was predicted that mutating H190 and E192 would be sufficient to abrogate the iron-binding ability of GASCl and thus also to inhibit its demethylation activity.

To test this hypothesis, performed demethylation assays were performed by incubating bulk histones with wild-type and mutant GASCl (H190G/E192A) in which histidine H190 and glutamic acid E192 had been replaced with glycine and alanine, respectively. While wild-type

GASCl had a robust demethylating activity, mutant GASCl had no detectable demethylation activity, indicating that the putative iron-binding residues are crucial to the demethylating activity of GASCl Size-exclusion chromatography profiles of wild-type and mutant GASCl were almost identical indicating that the mutations did not compromise the overall structure of the protein (supplementary figure S6).

Next, several experiments were performed to address the reaction mechanism for GASCl- mediated demethylation of the H3K9 tri and di-methyl marks. Various studies ¹¹ ' ^{17 24} of α- ketoglutarate-dependent oxygenases including the Factor Inhibiting HIFl (FIH) suggest the following reaction mechanism: first the enzyme binds iron through its metal-binding motif HXD/EX _n H the so-called facial triad ²⁰ (HXEX _n H in GASCl). Then, the Fe(II)-enzyme complex binds the co-factor α-ketoglutarate (αKG), and subsequently the substrate and oxygen. The binding of oxygen is followed by the oxidative decarboxylation of αKG to produce succinate, carbon dioxide and ferryl. The latter, is a highly reactive group and can potentially oxidise a C-H bond in a lysine methyl-group, forming an unstable carbinolamine that rapidly would break down leading to the release of formaldehyde and loss of a methyl group from lysine (Fig. 7a).

Dioxygenases belonging to the cupin superfamily are dependent upon Fe(II) and α- ketoglutarate. In addition, some cupin dioxygenases, including FBXLIl have the additional requirement of ascorbate for full catalytic activity ⁵ ' ²¹ ' ²³ . The mode of action of ascorbate is presently unclear but has been suggested to reduce Fe(III) to its active state Fe(II) or to function as a "surrogate reducing substrate" to 'rescue' the dioxygenase enzyme in the event of the uncoupled production of a ferryl (FeIV=O) intermediate ¹⁷ ' ²² .

To test whether GASC-I mediated demethylation would fit the reaction-mechanism described above and depicted on Fig. 7a, the importance of the putative co-factors for the demethylation reaction was tested.

Purified His-tagged GASCl was incubated with bulk histones as substrate in the presence or absence of its co-factors (Fig. 3b). In the presence of all its co-factors GASCl demethylates H3K9me3 in vitro, whereas GASCl-incubations in the presence of EDTA (chelating iron) or GASCl incubations in the absence of its co-factors was significantly reduced (Fig. 8b). The ability of the demethylation reaction to occur in spite of no addition of co-factors, is probably due to co-factors (iron and αKG) co-puπfying with the recombinant GASCl. For this reason we further sought to confirm the importance of these co-factors by testing the ability of N- oxalylglycine, quercetine, CoCI ₂ and NiSO ₄ to inhibit GASCl-mediated demethylation of H3K9me3. All these compounds were all able to inhibit the demethylation reaction effectively (Fig. 8c and data not shown). N-oxalylglyαne (Fig. 8d) and quercetine are αKG-analogues and presumably inhibit the activity of GASCl by displacing αKG from the iron-binding residues of GASCl. Analogously the inhibition of GASCl by CoCI ₂ and NiSO ₄ probably involves dislocation of Fe(II) from the iron-binding site of GASCl by competing cobalt(II) or nιckel(II) ions. The requirement of ascorbic acid, αKG and iron in the demethylation reaction and the inhibition of

the reaction by nickel and cobalt salts and αKG analogues strongly suggest that GASCl is indeed a dioxygenase.

Moreover, in the presence of its cofactors Fe(II) and αKG, recombinant His-tagged GASCl released formaldehyde (Fig. 8e) consistent with the proposed reaction scheme on figure 3a. Taken together, these results demonstrate that GASCl directly demethylates tπ- and dι- methylated lysine K9 on histone H3 in vitro through a hydroxylation reaction requiring iron and αKG and producing formaldehyde and mono-methylated H3K9.

Having established that GASCl can demethylate dι- and tπ-methylated H3K9 in vitro, it was considered whether GASCl could modulate heterochromatin formation /maintenance in vivo. To address this question it was first investigated whether enforced expression of GASCl could modulate H3K9 methylation in vivo. Human diploid fibroblasts were infected with a retrovirus expressing wild-type human GASCl and the methylation status at H3K9 was tested using confocal microscopy. GASCl localized to the nucleus, consistent with the presence of a putative nuclear localization signal in the protein. Infection with pBabe-HA-GASCl caused an efficient decrease of H3K9 tπ-methyl and an increase of mono-methylated H3K9 (Fig. 9a). In contrast, tri-methylated H3K4 and tri-methylated H3K27 were unaffected by GASCl over-expression, (supplementary figure S7).

Next, the human osteosarcoma cell line U2OS was transfectred with plasmids expressing wild- type and mutant GASCl (H190G/E192A) in which histidine H190 and glutamic acid E192 had been replaced with glycine and alanine, respectively. While ectopic expression of wild type GASCl efficiently abrogated the tri- and di-methylation of K9 on histone H3 in vivo, mutant GASCl was unable to do so (Fig. 9b).

Interestingly, ectopic expression of GASCl led to a significant increase in mono-methylated H3K9, suggesting that GASCl demethylates both tπ- and di-methylated H3K9 in vivo. The decrease in tπ-methylated H3K9 was also apparent in GASCl transfected U2OS cells, when evaluating the global levels of this mark by western blotting (Fig. 10c), further confirming the crucial importance of these predicted Fe(II)-coordιnatιng residues.

Next, it was tested whether the close GASCl-homologues JMJD2a and JMJD2b (Fig.l a, Fig. 2b) also could demethylate H3K9 in vivo, by over-expressing the proteins in U2OS cells. As for GASCl, ectopic expression of JMJD2a and JMJD2b led to a significant reduction of tπ- and dι- methylated H3K9, with a concomitant increase in mono-methylated H3K9 (Fig. 10 d, e). These results suggest that GASCl belongs to a family of histone H3K9 demethylases.

Heterochromatin formation and maintenance requires the presence of tri- and di-methylated H3K9 and HPl-binding ^{25 27} . Since the ectopic expression of GASCl leads to a global reduction of H3K9 dι- and tπ-methylation, it was tested whether HPl-binding and localization was affected by increased levels of GASCl. NIH 3T3 cells were transfected with a plasmid expressing HA-

tagged GASCl, and determined the endogenous HPlβ distribution in pre-extracted cells by confocal microscopy. As anticipated, HPlβ was delocalized in cells over-expressing GASCl, whereas no derealization occurred when the H190G/E192A mutant was transfected (Fig. llf). Likewise ectopic expression of the GASCl homologues JMJD2a and JMJD2b led to de-localization of HPlα (supplementary figure S8). These results were further validated by demonstrating decreased levels of HPlα and HPlγ associated with chromatin in GASCl over-expressing cells (Fig. llg). For this purpose, HEK293 cells with tetracycline-regulated expression of Myc-tagged GASCl were generated. Selected cells were incubated in the presence or absence of tetracycline overnight and as shown in Fig. l lg, the amount of chromatin-bound HPlα and HPlγ was reduced in GASCl over-expressing cells. Taken together these results suggest that GASCl could play a physiological important role in controlling heterochromatin formation and maintenance.

Until recently, tπ-methylated lysine has been considered as an irreversible covalent histone modification ²⁸ and the present report is the first to identify a histone tπ-methyl demethylase. Our results show that GASCl specifically demethylates methylated H3K9, but in contrast to LSDl and FBXLIl that are specific for mono- or di-methylated substrates ⁴ ' ⁵ , GASCl has the ability to remove both the dι- and tπ- methyl species of H3K9 in vitro and in vivo.

GASCl contains several domains, which potentially could serve to confer the protein with its exquisite specificity towards tπ- and di-methylated H3K9 (Fig. Ia). In addition to a Jumonji C (Jm]C) and Jumonji N domain (JmjN), the protein features two plant homeodomains (PHD) as well as a tandem Tudor domain (Tdr). The function of these domains is presently unclear but they have all been implicated in chromatin modulation ⁸ - ^{10/ 12/ 29f 30} . For example, the Tudor domain of 53BP1 has been reported to bind to the tπ-methylated form of lysine 79 on histone H3 ³¹ and the PHD domain has been shown to collaborate with Bromo domains in binding to nucleosomes ³⁰ . In order to gain insights into which domains are required for the demethylation function of GASCl and to resolve which domains determine its substrate specificity, mutational studies of the GASCl protein were performed. This was done by generating a series of GASCl expression constructs carrying deletions of the JmJC, JmjN and PHD domains. These results showed that the JmjC and JmjN domains are indispensable for the demethylating activity of GASCl (supplementary figure S9). In contrast the deletion of the PHD domains did not appear to affect the ability of GASCl to demethylate H3K9me3 in vivo (Fig. S9).

In light of the similar reaction mechanism of GASCl and FBXLIl, it is pertinent to ask why GASCl has the ability to demethylate the tri-methylated histone lysine substrate, whereas FBXLIl has not. A potential explanation could be that the iron/substrate-binding groove is smaller in FBXLIl than in GASCl. Although this explanation would require experimental testing, it is supported by in silico modelling. When comparing models of the two Jumonji proteins modelled with FIH as template, it was found that they both feature a loop region located at the edge of the iron-binding groove of the models. This loop region is significantly longer in the FBXLIl model opening the possibility that it could flip, making the iron/substrate binding

groove of this protein smaller and consequently inhibiting the accommodation of a tri-methyl lysine substrate.

It is predicted that the demethylation pathway described in the present report and summarized on supplementary figure S8b is evolutionary conserved. Of note in this context, the Schizosaccharomyces pombe Jumonji protein, Epel has been reported to antagonize heterochromatization ¹⁰ . Thus a recent report ¹⁰ has demonstrated that enforced expression of Epel causes a decrease in di-methylated H3K9 and antagonizes hetero-chromatin with a concomitant destabihzation of centromers and mating type loci ¹⁰ . This Epel-mediated phenotype was dependent of an intact JmjC domain and although it has not been shown that the capacity of Epel to antagonize heterochromatization is due to a demethylation activity, it is appealing to assume such a role in light of our present data.

The existence of several closely related human homologues (JMJD2a and JMJD2b, Fig. 2b), with similar enzymatic activity, as GASCl is indicative of functional redundancy and the necessity for a tight control of H3K9 tri-methyl demethylase activity. This is to be expected as the strict control of heterochromatin formation and maintenance is critical for both proper biological function and genomic integrity. For instance centromeric heterochromatin formation is essential for the correct segregation of chromosomes during mitosis ³² . Similarly it has been demonstrated that the deletion of Suv39Hl and Suv39H2 genes in mice, whose products are responsible for dι- and tri-methylation of H3K9 in heterochromatin, leads to chromosomal instability and collaborates with oncogenes in inducing mouse lymphomas ⁶ . Therefore aberrant activity of GASCl or its homologues, could potentially lead to genomic instability and consequently cancer. Indeed, GASCl was originally identified as a gene frequently amplified in oesophageal squamous cell (ESC) carcinoma ²² and GASCl is over-expressed in various cancer types containing chromosomal abberations ³³ ' ³⁴ .

To obtain further support for the involvement of GASCl in cancer, the Oncomine database was searched for differential GASCl expression in normal versus tumor tissue. The expression of GASCl and its homologues JMJD2a and JMJD2b were found to be significantly increased in prostate cancers relative to normal tissue (supplementary figure SIl). Previously GASCl has been shown to be amplified in several cell lines derived from ESC carcinomas ³³ . The methylation status of the H3K9 epigenetic mark was tested in an ESC cell line with GASCl gene amplification (KYSE-150) as well as in one cell line with more moderate expression (KYSE-450). The human osteosarcoma cell-line U2OS was included as controls. To test the importance of GASCl amplification we tested the effect of GASCl knockdown on the H3K9 methylation status in cells transfected with siRNA ohgos to GASCl and/or to its homologues. Cells were transfected with LMP-GASCl followed by three days selection with puromycin. The efficiency of GASCl knockdown was assessed by RT-QPCR (Fig 5a). H3K9me3 was significantly increased in KYSE- 150 cells in response to GASCl knockdown as evidenced by western blotting (data not shown).

Di- and tri-methylated H3K9 is required for HPl-binding, essential for heterochromatin formation and associated with transcriptional repression. Moreover these epigenetic marks are increased on specific genes in senescence, a key mechanism guarding cells against cancer ^{36 39} . It can therefore be speculated that amplification of GASCl may reduce the ability of cells to become senescent, or lead to de-repression of otherwise silenced oncogenes (Fig 6d).

Accordingly, inhibition of GASCl-demethylating activity could thus potentially constitute a new antineoplastic therapeutic modality.

Intπguingly, it was found that quercetin (Figure 12b), a plant flavanoid associated with antiproliferative and anti-cancer properties ^{40 42} was able to effectively inhibit GASCl -mediated demethylation of H3K9me3 in vitro (Fig. 12c). In vivo, overnight treatment of KYSE-150 cells with quercetin caused a staggering increase in H3K9 tπmethylation as evidenced by immunoflourescence and western blotting (Fig 13d and data not shown). The increase of H3K9me3 was accompanied with a marked decrease of growth rate and a senescent-like phenotype characterized by changed cell morphology, appearance of senescence associated heterochromatin foci (SAHFs) and β-gal staining (data not shown). Interestingly quercetin- treatment did not seem to affect the other cell-lines (expressing-low and moderate levels of GASCl) to the same extent. Thus U2OS, WI38 and KYSE-450 cells did not display the same increase in H3K9me3 nor the same senescent-like phenotype and growth inhibition. This result suggests that some cancers characterized by increased expression or activity of JMJD2 enzymes and subsequent decrease of H3K9me3/me2 may be especially susceptible to the growth- inhibitory effect of this compound. The biological effects of quercetin have previously been attributed to inhibition of topoisomerase I ⁴³ , modulation of protein and lipid kinase signalling pathways ⁴⁴ and scavenging of reactive oxygen species ⁴⁵ , In light of the present findings it is tempting to speculate that the anti-cancer effect of quercetin might at least partially by explained by its inhibitory effect on GASCl and other members of the JMJD2 subfamily.

In summary, using several independent lines of evidence it has been shown that GASCl directly demethylates the repressive histone dι- and tπ-methyl H3K9 marks both in vitro and in vivo. The present findings demonstrate that histone tπ-methylation is a reversible modification. This finding may potentially have far-reaching implications for human disease, notably cancer.

MATERIALS AND METHODS

Demethylation assay.

Bulk histones, oligonucleosomes or synthetic histone peptides were reacted with purified His- GASCl in demethylation buffer (50 mM Tris pH 8.0, 5OmM KCI, 1OmM MgCI ₂ , 1 mM α- ketoglutarate, 40μM FeSO ₄ , 2mM ascorbic acid, 0.01% (w/v) BSA and 5% (v/v) glycerol) at 37°C. In a typical reaction, either 6 μg bulk histones or 2 μg of modified histone peptides were reacted with 20 μg GASCl in a volume of 100 μl for 30 minutes. Reaction mixtures were

analyzed by either western blotting using specific antibodies, or by formaldehyde-release assays.

Formaldehyde release assay. Formaldehyde release assays were performed essentially as described ⁴ . All reactions were performed in a total volume of 200μl per reaction in a quartz cuvette. In short, recombinant GASCl (typically 40 μg) was incubated in 150 μl demethylation buffer (see above) in the presence of 2mM NAD+ and 0.2 U formaldehyde dehydrogenase (FDH) for 37°C for 5 minutes. Then the reaction was started by adding substrates (histone-peptides). The absorbance at 340 nm was measured with 0.5 mm intervals (15 mm total) using a Genesys lOUV Thermospectronic spectrophotometer at 37°C.

GASCl demethylation of H3K9me3 peptide for mass spectrometry analysis.

Six μg of recombinant GASCl, was incubated with 3μg H3K9me3 peptide in FDH buffer in a final volume of 90μl for 30 mm at 37°C. Urea was added to a final concentration of 4M and the mixture was incubated at 20 ⁰ C for 15 mm. An equal volume of 1% TFA was added and the sample loaded on a reversed phase mini C8 column packed in a lOOμl tip (column volume of 20 μl). After washing in 1% TFA, the bound peptide was eluted in 20 μl (30% methanol, 25% formic acid). One third of the eluted material was analysed by SDS page and Western blotting using first antι-H3K9me3, followed by anti-biotin antibody (Fig. 4c). The rest of the material was analysed by mass spectrometry (Fig. 4d).

Mass spectrometry (MS) analysis.

One third of the eluate was injected in 1% TFA using an Agilent 1100 Nano HPLC (Palo Alto, CA) onto a C18-column (Reprosil-Pur C18-AQ 3 μm; Dr. Maisch GmbH, Ammerbuch-Entπngen, germany) packed into a spray emitter (75 μm ID, 8μm opening, 70 mm length; New Objectives, USA). Peptides were eluted in a gradient from buffer A (5%acetonιtrιle and 0.5% acetic acid) to buffer B (acetonitrile and 0.5% acetic acid) going from 0 to 20 % in 10 mm at 300 nL/min. Spectra were recorded on a LTQ-FT mass spectrometer (Thermoelectron, Bremen, Germany).

Supplementary Methods

Materials.

Synthetic peptides 43 amino acids long mimicking the N-terminal tail (1-40) of histone H3

(ARTKQTARKSTGGKAPRKQLATKAARKSAPATGGVKKPHR-Tyr-Cys-(Ttds)- Lys-bιotιn). The peptides were synthesized with a C-terminal tyrosine and cysteine for coupling and linked to biotin through lysine and a Ttds spacer. Peptides used in experiments were either unmodified, tri-methylated at lysine 27 or mono-, dι-, or tπ-methylated at lysine K9 were purchased from Jeπni, GMBH Germany. Formaldehyde dehydrogenase (FDH, F1879) and nicotine amine and bulk histones (H9250) were purchased from Sigma. Antibodies used in the study were as follows: anti tπ-methylated H3-K9, (Upstate 07-523), anti di-methylated H3-K9 (Upstate 07- 212), anti mono-methylated H3-K9 (Abeam Ab9045-50), anti tπ-methyl H3-K27 (Upstate 07-

449), anti tri-methyl H4-K20 (Upstate 07-463), anti tri-methylated H3-K4, (Abeam ab8580-50), anti di-methylated H3-K4 (Upstate 07-030), anti histone H3 (Abeam Abl791-100), anti HPlα (Upstate 05-689), anti-HPlγ (upstate 05-690), anti biotin-HRP (Sigma A4541), anti-His (Upstate 05-531) and anti HA (CRP Inc. AFC-IOlP). 5

Cell lines and tissue culture.

Human esophageal squamous cell carcinoma cell lines KYSE-70 and 150 were obtained from the German Collection of Microorganisms and Cell Cultures, Braunschweig, Germany. Diploid human fibroblast (TIG-3) expressing hTert and U2OS cells expressing the murine ecotropic retrovirus

10 receptor EcoR were used for the experiments. HEK293 cells stably expressing tetracycline inducible amino ^: terminally myc-tagged GASCl was generated using the 293 TRex-flip-in cells essentially as described by the manufacturer (InVitrogen). KYSE cells were maintained in 49% RPMI 1640, 49% Ham's F12 supplemented 2% foetal calf serum (FCS), 5% CO ₂ . All other cells were maintained at 37°C in Dulbecco's modified Eagle's medium (DMEM) supplemented with

15 10% foetal calf serum (FCS), 5% CO ₂ .

Histone peptide pulldown of HeLa nuclear proteins Pull-down assays were performed using HeLa cell nuclear extracts and histone H3 peptide tri-methylated at lysine 9 (ARTKQTARK(me3)STGGKAPRKQLATKAA-RKS APATGGVKKPHR-Tyr-Cys-(Ttds)-Lys-biotin).

20 Nuclei were prepared from HeLa cells, lysed with lysis buffer (50 mM Tris, pH 7.2, 300 mM NaCI, 0.5% IGEPAL CA-630, 1 mM EDTA (pH 8.0), 1 mM PMSF containing protease inhibitors). The lysate was pre-cleared with uncoupled streptavidin beads (Fig. Sl, preclearing 1) and then with streptavidin beads coupled to unmodified H3 histone (Fig. Sl, preclearing 2). The pre- cleared lysate was then incubated with H3 histone peptide trimethylated at K9 (H3K9me3),

25 precoupled to streptavidin beads. Beads were washed avidly and bound proteins were subsequently eluted from the resin with 2x LSB buffer (100 mM Tris-HCI pH 6.8, 200 mM DTT, 4% SDS, 20% glycerol and 0.2% bromphenolblue), Figure Sl, H3K9me3. Proteins were resolved by SDS PAGE and visualized by silver staining. Protein bands enriched in the H3K9me3 pull-down were excised, in-gel digested with trypsin and analyzed by nanoelectrospray tandem

30 mass spectrometry as described by WiIm et al. (WiIm M. (1996) Nature 379, 466-469). Immunoflourescense Confocal immunostainings were done essentially as described by Sørensen, CS. et al., Nat. Cell. Biol. 7, 195-201 (2005).

Preparation of cell lysates.

35 Total lysates of cells were generated from the same number of cells lysed in urea buffer (1% SDS, 9M urea, 25 mM Tris-HCI, pH 6.8, 1 mM EDTA, 0.7 M β-mercaptoethanol), boiled for 5 min. and sonicated. For chromatin fractionation an equal number of cells were incubated 30 min. on ice in 200μl pre-extraction buffer (2OmM Hepes-KOH, pH 7.2 containing 0.5% IGEPAL- 630, 5OmM NaCI, 3mM MgCI2 and 30OmM sucrose). 100 μl was mixed with 2χLSB buffer (100

40 mM Tris-HCI pH 6.8, 200 mM DTT, 4% SDS, 20% glycerol and 0.2% bromphenolblue), boiled, sonicated and used as total lysate. The remaining 100 μl was spun down at 1300 x g for 10 min

at 4°C, washed in 1 ml pre-extraction buffer, re-suspended in 100 μl 2χLSB buffer, boiled, sonicated and used as chromatin fraction.

SiRNA and shRNA to GASCl and JMJD2a/b. Small interfering RNA (siRNA) oligonucleotides to GASCl and its homologs JMJD2a and JMJD2b were synthesized by Dharmacon Research, Inc.. Cells (lχl0 ⁵ /well) were transfected with siRNA oligos (0.3μg/well) in 6-well plates using Oligofectamine reagent (Invitrogen) following the manufacturer's protocol. SiRNA can be designed according to procedures known in the art for instance using the SiRNA retriever; (http://katahdin.cshl.org :9331/homepaαe/siRNA/RNAi.cqi?tvpe=siRNA^

ShRNA Constructs

Short hairpin RNA constructs were generated in the MSCV/LTRmiR30-PIG (LMP) vector (Dickens NG et al (2005) "Probing tumor phenotypes using stable and regulated synthetic microRNA precursors" Nature Genetics VoI 37, No 11 1289-1295.) according to the procedure described in Paddison et al. (2005) (Paddison PJ, Cleary M, Silva JM, Chang K, Sheth N, Sachidanandam R & Hannon GJ. Cloning of short hairpin RNAs for gene knockdown in mammalian cells Nature methods ppl63 - 167).

Three oligonucleotides were designed using the "shRNA retriever"

(http://katahdin.cshl.org :9331/homepaqe/siRNA/RNAi.cqi?tvpe=shRNA ^' ). Using the oligonucleotide TGCTGTTGACAGTGAGCGCGCCAGATAGCAGCAATGAAGATAG- TGAAGCCACAGATGTATCTTCATTGCTGCTATCTGGCTTGCCTACTGCCTCGGA as template a 150 bp PCR product was amplified, digested with Xhol and EcoRI and ligated into the digested LMP vector. The shRNA construct targets a 22 bp sequence at position 2117 in the GASCl coding sequence, underlined in the above sequence. The efficiency of the shRNAi construct to knockdown the GASCl mRNA was estimate as approximately 80% as assessed by real-time quantitative RT-PCR.

Real-time quantitative RT-PCR

Cells were transfected with the indicated shRNA constructs and subjected to a brief puromycin selection (2 μg/ml). Total RNA was made from transfected cells using the Quiagen RNAeasy mini kit according to the manufacturers instructions. cDNA was generated using the Taqman reverse transcription kit and poly dT primer according to the manufacturers instructions. The cDNA was used as template in real-time quantitative PCR reactions with GASCl specific primers on a Applied biosystems 7700. The reactions were prepared using 2xSYBR green reaction mix from Applied biosystems.

BrdU Incorporation. Thirty hours after transfection of siRNA, the cells were split into 4-well chamber slides and incubated with culture medium

containing BrdU forl5-30 min.

cDNA cloning of human GASCl.

The putative open reading frames of human GASCland JMJD2A were amplified by PCR from HeLa cDNA and cDNA from a human foetal brain cDNA library (Invitrogen, Carlsbad, CA). Primer sequences are available upon request. PCR products were gel purified, cloned into pCR8/GW (Invitrogen), and verified by DNA sequencing. Using these Gateway-compatible entry clones GASCl and JMJD2A were transferred into pCMV-HA, pCMV-myc and pBabepuro. A double mutant in the conserved iron-binding domain (HTE) of GASCl changing amino acids 190-192 from histidine, threonine and glutamic acid respectively to glycine, threonine and alanine was generated using standard mutagenesis methods.

Recombinant Proteins.

Full-length amino-terminally hexahistidine-tagged human GASCl baculovirus transfer vector was generated by Gateway-mediated transfer of GASCl cDNA from pCR8/GW and into a

Gateway-modified form of pAcHLT-A (Pharmingen). Recombinant baculoviruses were generated by cotransfection of baculovirus transfer vector containing the GASCl gene and Bsu36I linearized Bakpak6 baculovirus DNA essentially as previously described. Histidine-tagged GASCl and JMJD2A was expressed and purified by cobalt-affinity chromatography essentially as described previously (Christensen et al. Nucleic Acids Research 33, 5458-5470 (2005)). The eluted fractions were analyzed by SDS-PAGE and selected fractions were subjected to further purification by size exclusion chromatography (SEC). In short, SEC was performed on a Superose 12, 10/300 column (Pharmacϊa-Amersham) equilibrated with 25mM HEPES-KOH, pH 7.7, containing 5OmM NaCI and eluted with the same buffer at a flow-rate of 0.3 ml/min. Eluted material was collected in 0.5 ml fractions, flash frozen in liquid N ₂ , and stored at -80 ⁰ C. All procedures were performed on ice or at 4°C in the presence of complete EDTA-free protease inhibitor (Boehringer Mannheim, Germany).

Retrovirus transduction. To generate recombinant retroviruses expressing GASCl and its mutant H190G/E192A, the open reading frame of GASCl was transferred into pBabepuro by Gateway-mediated recombination generating pBabepuro-HA-GASCl, and pBabepuro-HA-H190G/E192A. High titers of retroviral particles were obtained 24-48 hours after transfection of the Phoenix-Eco 293 cell packing cell line. Transfections were done using the calcium phosphate method. Transduction of TIG3-hTert-EcoR cells or U2OS-EcoR cells was achieved by adding virus containing supernatants from the packaging cell line to the cell dishes four times within a 24 hour period. Transduced cells were selected for 2-5 days in the presence puromycin (1 μg/μl).

Molecular modelling The structure of FIH complexed with α-ketoglutarate, CAD peptide and iron was available in Protein Data Bank (PDB, Brookhaven National Library, Upton Ny, USA) with accession number

1H2K. GASC 1 ₂₂ - ₃₄₉ was homology-modelled with FIH as template using homology module INSIGHT II (2005) (Accelrys, San Diego). The histone tail (ARTKQTARKSTG) was modelled with the CAD peptide, as template. The program GRID (Version 22, Molecular Discovery Ltd., Oxford, UK) (Goodford, 1985) was used to compute the hydrophobic surface contours of the GASCl molecule. The histone peptide was subsequently docked onto the hydrophobic surface contours of GASCl using AUTODOCK (Ver. 3.0.5, Scripps Research Institute and Molecular Graphics Laboratory). Finally structural models of the complexes were drawn with Pymol (http://pymol.sourceforae.net/).

Supplementary Methods

Materials. Synthetic peptides 43 amino acids long mimicking the N-terminal tail (1-40) of histone H3 (ARTKQTARKSTGGKAPRKQLATKAARKSAPATGGVKKPHR-Tyr-Cys-(Ttds)-Lys -biotin). The peptides were synthesized with a C-terminal tyrosine and cysteine for coupling and linked to biotin through lysine and a Ttds spacer. Peptides used in experiments were either unmodified, tri-methylated at lysine 27 or mono-, di-, or tri-methylated at lysine K9 were purchased from Jerini, GMBH Germany. Formaldehyde dehydrogenase (FDH, F1879) and nicotine amine and bulk histones (H9250) were purchased from Sigma. Antibodies used in the study were as follows: anti tri-methylated H3-K9, (Upstate 07-523), anti di-methylated H3-K9 (Upstate 07-212), anti mono-methylated H3-K9 (Abeam Ab9045-50), anti tri-methyl H3-K27 (Upstate 07-449), anti tri-methyl H4-K20 (Upstate 07-463), anti tri-methylated H3-K4, (Abeam ab8580-50), anti di-methylated H3-K4 (Upstate 07-030), anti histone H3 (Abeam Abl791-100), anti HPlα (Upstate 05-689), anti-HPlγ (upstate 05-690), anti biotin-HRP (Sigma A4541), anti- His (Upstate 05-531) and anti HA (CRP Inc. AFC-IOlP).

Cell lines and tissue culture.

Human esophageal squamous cell carcinoma cell lines KYSE-70 and 150 were obtained from the German Collection of Microorganisms and Cell Cultures, Braunschweig, Germany. Diploid human fibroblast (TIG-3) expressing hTert and U2OS cells expressing the murine ecotropic retrovirus receptor EcoR were used for the experiments. HEK293 cells stably expressing tetracycline inducible amino-terminaHy myc-tagged GASCl was generated using the 293 TRex-flip-in cells essentially as described by the manufacturer (InVitrogen). KYSE cells were maintained in 49% RPMI 1640, 49% Ham's F12 supplemented 2% foetal calf serum (FCS), 5% CO ₂ . All other cells were maintained at 37°C in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% foetal calf serum (FCS), 5% CO ₂ .

Histone peptide pulldown of HeLa nuclear proteins Pull-down assays were performed using HeLa cell nuclear extracts and histone H3 peptide tri-methylated at lysine 9 (ARTKQTARK(me3)STGGKAPRKQLATKAA-RKS APATGGVKKPHR-Tyr-Cys-(Ttds)-Lys-bϊotin). Nuclei were prepared from HeLa cells, lysed with lysis buffer (50 mM Tris, pH 7.2, 300 mM NaCI, 0.5% IGEPAL CA-630, 1 mM EDTA (pH 8.0), 1 mM PMSF containing protease inhibitors). The lysate was pre-cleared with uncoupled streptavidin beads (Fig. Sl, preclearing 1) and then

with streptavidin beads coupled to unmodified H3 histone (Fig. Sl, preclearing 2). The pre- cleared lysate was then incubated with H3 histone peptide trimethylated at K9 (H3K9me3), precoupled to streptavidin beads. Beads were washed avidly and bound proteins were subsequently eluted from the resin with 2x LSB buffer (100 mM Tris-HCI pH 6.8, 200 mM DTT, 4% SDS, 20% glycerol and 0.2% bromphenolblue), Figure Sl, H3K9me3. Proteins were resolved by SDS PAGE and visualized by silver staining. Protein bands enriched in the H3K9me3 pull-down were excised, in-gel digested with trypsin and analyzed by nanoelectrospray tandem mass spectrometry as described by WiIm et al. (WiIm M. (1996) Nature 379, 466-469). Immunoflourescense Confocal immunostainings were done essentially as described by Sørensen, CS. et al., Nat. Cell. Biol. 7, 195-201 (2005).

Preparation of cell lysates.

Total lysates of cells were generated from the same number of cells lysed in urea buffer (1%

SDS, 9M urea, 25 mM Tris-HCI, pH 6.8, 1 mM EDTA, 0.7 M β-mercaptoethanol), boiled for 5 min. and sonicated. For chromatin fractionation an equal number of cells were incubated 30 min. on ice in 200μl pre-extraction buffer (2OmM Hepes-KOH, pH 7.2 containing 0.5% IGEPAL- 630, 5OmM NaCI, 3mM MgCI2 and 30OmM sucrose). 100 μl was mixed with 2χLSB buffer (100 mM Tris-HCI pH 6.8, 200 mM DTT, 4% SDS, 20% glycerol and 0.2% bromphenolblue), boiled, sonicated and used as total lysate. The remaining 100 μl was spun down at 1300 x g for 10 min at 4°C, washed in 1 ml pre-extraction buffer, re-suspended in 100 μl 2χLSB buffer, boiled, sonicated and used as chromatin fraction.

siRNA and shRNA to GASCl and JMJD2a/b.

Small interfering RNA(siRNA) oligonucleotides to GASCl and its homologs JMJD2a and JMJD2b were synthesized by Dharmacon Research, Inc.. Cells (lχlO ⁵ /well) were transfected with siRNA oligos (0.3μg/well) in 6-well plates using Oligofectamine reagent (Invitrogen) following the manufacturer's protocol. SiRNA can be designed according to procedures known in the art for instance using the SiRNA retriever;

(http://katahdin.cshl.org :9331/homepaqe/siRNA/RNAi.cqi?type=siRNA).

ShRNA Constructs

Short hairpin RNA constructs were generated in the MSCV/LTRmiR30-PIG (LMP) vector (Dickens

NG et al (2005) "Probing tumor phenotypes using stable and regulated synthetic microRNA precursors" Nature Genetics VoI 37, No 11 1289-1295.) according to the procedure described in Paddison et al. (2005) (Paddison PJ, Cleary M, Silva JM, Chang K, Sheth N, Sachidanandam R &

Hannon GJ. Cloning of short hairpin RNAs for gene knockdown in mammalian cells Nature methods ppl63 - 167).

Three oligonucleotides were designed using the "shRNA retriever" (http://katahdin.cshl.org :9331/homepaqe/siRNA/RNAi.cqi?tvpe=shRNA>. Using the oligonucleotide TGCTGTTGACAGTGAGCGCGCCAGATAGCAGCAATGAAGATAG-

TGAAGCCACAGATGTATCTTCATTGCTGCTATCTGGCTTGCCTACTGCCTCGGA as template a 150 bp PCR product was amplified, digested with Xhol and EcoRI and ligated into the digested LMP vector. The shRNA construct targets a 22 bp sequence at position 2117 in the GASCl coding sequence, underlined in the above sequence. The efficiency of the shRNAi construct to knock- down the GASCl mRNA was estimate as approximately 80% as assessed by real-time quantitative RT-PCR.

Real-time quantitative RT-PCR

Cells were transfected with the indicated shRNA constructs and subjected to a brief puromycin selection (2 μg/ml). Total RNA was made from transfected cells using the Quiagen RNAeasy mini kit according to the manufacturers instructions. cDNA was generated using the Taqman reverse transcription kit and poly dT primer according to the manufacturers instructions. The cDNA was used as template in real-time quantitative PCR reactions with GASCl specific primers on a Applied biosystems 7700. The reactions were prepared using 2xSYBR green reaction mix from Applied biosystems.

BrdU Incorporation.

Thirty h after transfection of siRNA, the cells were split into 4-well chamber slides and incubated with culture medium containing BrdU forl5-30 min.

cDNA cloning of human GASCl.

The putative open reading frames of human GASCland 3MJD2A were amplified by PCR from

HeLa cDNA and cDNA from a human foetal brain cDNA library (Invϊtrogen, Carlsbad, CA). Primer sequences are available upon request. PCR products were gel purified, cloned into pCR8/GW (Invitrogen), and verified by DNA sequencing. Using these Gateway-compatible entry clones GASCl and 3MJD2A were transferred into pCMV-HA, pCMV-myc and pBabepuro. A double mutant in the conserved iron-binding domain (HTE) of GASCl changing amino acids 190-192 from histidine, threonine and glutamic acid respectively to glycine, threonine and alanine was generated using standard mutagenesis methods.

Recombinant Proteins.

Full-length amino-terminally hexahistidine-tagged human GASCl baculovirus transfer vector was generated by Gateway-mediated transfer of GASCl cDNA from pCR8/GW and into a Gateway-modified form of pAcHLT-A (Pharmingen). Recombinant baculoviruses were generated by cotransfection of baculovirus transfer vector containing the GASCl gene and Bsu36I linearized Bakpakβ baculovirus DNA essentially as previously described. Histidine-tagged GASCl and 3M3D2A was expressed and purified by cobalt-affinity chromatography essentially as described previously (Christensen et al. Nucleic Acids Research 33, 5458-5470 (2005)). The eluted fractions were analyzed by SDS-PAGE and selected fractions were subjected to further purification by size exclusion chromatography (SEC). In short, SEC was performed on a

Superose 12, 10/300 column (Pharmacia-Amersham) equilibrated with 25mM HEPES-KOH, pH 7.7, containing 5OmM NaCI and eluted with the same buffer at a flow-rate of 0.3 ml/min. Eluted material was collected in 0.5 ml fractions, flash frozen in liquid N ₂ , and stored at -80 ⁰ C. All procedures were performed on ice or at 4°C in the presence of complete EDTA-free protease inhibitor (Boehringer Mannheim, Germany).

Retrovirus transduction.

To generate recombinant retroviruses expressing GASCl and its mutant H 190G/E192A, the open reading frame of GASCl was transferred into pBabepuro by Gateway-mediated recombination generating pBabepuro-HA-GASCl, and pBabepuro-HA-H190G/E192A. High titers of retroviral particles were obtained 24-48 hours after transfection of the Phoenix-Eco 293 cell packing cell line. Transfections were done using the calcium phosphate method. Transduction of TIG3-hTert-EcoR cells or U2OS-EcoR cells was achieved by adding virus containing supernatants from the packaging cell line to the cell dishes four times within a 24 hour period. Transduced cells were selected for 2-5 days in the presence puromycin (1 μg/μl).

Molecular modelling

The structure of FIH complexed with α-ketoglutarate, CAD peptide and iron was available in

Protein Data Bank (PDB, Brookhaven National Library, Upton Ny, USA) with accession number 1H2K. GASC 1 ₂₂ - ₃₄₉ was homology-modelled with FIH as template using homology module

INSIGHT II (2005) (Accelrys, San Diego). The histone tail (ARTKQTARKSTG) was modelled with the CAD peptide, as template. The program GRID (Version 22, Molecular Discovery Ltd., Oxford, UK) (Goodford, 1985) was used to compute the hydrophobic surface contours of the GASCl molecule. The histone peptide was subsequently docked onto the hydrophobic surface contours of GASCl using AUTODOCK (Ver. 3.0.5, Scripps Research Institute and Molecular Graphics Laboratory). Finally structural models of the complexes were drawn with Pymol (http://pymol.sourceforge.net/).

REFERENCES

1. Fischle, W., Wang, Y. & AIUs, C. D. Histone and chromatin cross-talk. Curr Opin Cell Biol 15, 172-83 (2003).

2. Margueron, R., Trojer, P. & Reinberg, D. The key to development: interpreting the histone code? Curr Opin Genet Dev 15, 163-76 (2005). 3. Peterson, C. L. & Laniel, M. A. Histones and histone modifications. Curr Biol 14, R546-51

(2004).

4. Shi, Y. et al. Histone demethylation mediated by the nuclear amine oxidase homolog LSDl. Cell 119, 941-53 (2004).

5. Tsukada, Y. et al. Histone demethylation by a family of JmjC domain-containing proteins. Nature 439, 811-6 (2006).

6. Peters, A. H. et al. Loss of the Suv39h histone methy (transferases impairs mammalian heterochromatin and genome stability. Cell 107, 323-37 (2001).

7. Bannister, A. J. & Kouzarides, T. Reversing histone methylation. Nature 436, 1103-6 (2005).

5 8. Clissold, P. M. & Ponting, C. P. JmjC: cupin metalloenzyme-like domains in jumonji, hairless and phospholipase A2beta. Trends Biochem Sci 26, 7-9 (2001).

9. Katoh, M. & Katoh, M. Identification and characterization of JMJD2 family genes in silico. Int J Oncol 24, 1623-8 (2004).

10. Ayoub, N. et al. A novel jmjC domain protein modulates heterochromatization in fission 10 yeast. MoI Cell Biol 23, 4356-70 (2003).

11. Dann, C. E., 3rd, Bruick, R. K. & Deisenhofer, J. Structure of factor-inhibiting hypoxia- inducible factor 1: An asparaginyl hydroxylase involved in the hypoxic response pathway. Proc Natl Acad Sci U S A 99, 15351-6 (2002).

12. Gray, S. G. et al. Functional characterization of JMJD2A, a histone deacetylase- and 15 retinoblastoma-binding protein. J Biol Chem 280, 28507-18 (2005).

13. Kim, T. G., Chen, J., Sadoshima, J. & Lee, Y. 3umonji represses atrial natriuretic factor gene expression by inhibiting transcriptional activities of cardiac transcription factors. MoI Cell Biol 24, 10151-60 (2004).

14. Kim, T. G., Kraus, J. C, Chen, J. & Lee, Y. JUMONJI, a critical factor for cardiac 20 development, functions as a transcriptional repressor. J Biol Chem 278, 42247-55

(2003).

15. Schofield, C. J. & Zhang, Z. Structural and mechanistic studies on 2-oxoglutarate- dependent oxygenases and related enzymes. Curr Opin Struct Biol 9, 722-31 (1999).

16. Trewick, S. C, McLaughlin, P. J. & Allshire, R. C. Methylation: lost in hydroxylation? 25 EMBO Rep 6, 315-20 (2005).

17. Knowles, H. J., Raval, R. R., Harris, A. L. & Ratcliffe, P. J. Effect of ascorbate on the activity of hypoxia-inducible factor in cancer cells. Cancer Res 63, 1764-8 (2003).

18. Lee, C, Kim, S. J., Jeong, D. G., Lee, S. M. & Ryu, S. E. Structure of human FIH-I reveals a unique active site pocket and interaction sites for HIF-I and von Hippel-

30 Lindau. J Biol Chem 278, 7558-63 (2003).

19. Elkins, J. M. et al. Structure of factor-inhibiting hypoxia-inducible factor (HIF) reveals mechanism of oxidative modification of HIF-I alpha. J Biol Chem 278, 1802-6 (2003).

20. Hegg, E. L. & Que, L., Jr. The 2-His-l-carboxylate facial triad— an emerging structural motif in mononuclear non-heme iron(II) enzymes. Eur J Biochem 250, 625-9 (1997).

35 21. Kivirikko, K. I. & Myllyharju, J. Prolyl 4-hydroxylases and their protein disulfide isomerase subunit. Matrix Biol 16, 357-68 (1998). 22. MyIIyIa, R., Majamaa, K., Gunzler, V., Hanauske-Abel, H. M. & Kivirikko, K. I. Ascorbate is consumed stoichiometrically in the uncoupled reactions catalyzed by prolyl 4- hydroxylase and lysyl hydroxylase. J Biol Chem 259, 5403-5 (1984). 40 23. Wilmouth, R. C. et al. Structure and mechanism of anthocyanidin synthase from

Arabidopsis thaliana. Structure 10, 93-103 (2002).

24. Zhou, J. et al. Spectroscopic studies of substrate interactions with clavaminate synthase 2, a multifunctional alpha-KG-dependent non-heme iron enzyme: correlation with mechanisms and reactivities. J Am Chem Soc 123, 7388-98 (2001).

25. Bannister, A. J. et al. Selective recognition of methylated lysine 9 on histone H3 by the 5 HPl chromo domain. Nature 410, 120-4 (2001).

26. Elgin, S. C. & Grewal, S. I. Heterochromatin: silence is golden. Curr Biol 13, R895-8 (2003).

27. Lachner, M., O'Carroll, D., Rea, S., Mechtler, K. & Jenuwein, T. Methylation of histone H3 lysine 9 creates a binding site for HPl proteins. Nature 410, 116-20 (2001).

10 28. Kubicek, S. & Jenuwein, T. A crack in histone lysine methylation. Cell 119, 903-6 (2004).

29. Maurer-Stroh, S. et al. The Tudor domain 'Royal Family': Tudor, plant Agenet, Chromo, PWWP and MBT domains. Trends Biochem Sci 28, 69-74 (2003).

30. Ragvin, A. et al. Nucleosome binding by the bromodomain and PHD finger of the 15 transcriptional cofactor p300. J MoI Biol 337, 773-88 (2004).

31. Huyen, Y. et al. Methylated lysine 79 of histone H3 targets 53BP1 to DNA double-strand breaks. Nature 432, 406-11 (2004).

32. Pidoux, A. L. & Allshire, R. C. The role of heterochromatin in centromere function. Philos Trans R Soc Lond B Biol Sci 360, 569-79 (2005).

20 33. Yang, Z. Q. et al. Identification of a novel gene, GASCl, within an amplicon at 9p23-24 frequently detected in esophageal cancer cell lines. Cancer Res 60, 4735-9 (2000). 34. Yang, Z. Q. et al. A novel amplicon at 9p23 - 24 in squamous cell carcinoma of the esophagus that lies proximal to GASCl and harbors NFIB. Jpn 3 Cancer Res 92, 423-8

(2001). 25 35. Soejima, H. et al. Silencing of imprinted CDKNlC gene expression is associated with loss of CpG and histone H3 lysine 9 methylation at DMR-LITl in esophageal cancer.

Oncogene 23, 4380-8 (2004).

36. Braig, M. et al. Oncogene-induced senescence as an initial barrier in lymphoma development. Nature 436, 660-5 (2005).

30 37. Chen, Z. et al. Crucial role of p53-dependent cellular senescence in suppression of Pten- deficient tumorigenesis. Nature 436, 725-30 (2005).

38. Collado, M. et al. Tumour biology: senescence in premalignant tumours. Nature 436,

642 (2005).

39. Michaloglou, C. et al. BRAFE600-associated senescence-like cell cycle arrest of human

35 naevi. Nature 436, 720-4 (2005).

40. Welford, R. W., Schlemminger, L, McNeil!, L. A., Hewitson, K. S. & Schofield, C. J. The selectivity and inhibition of AIkB. J Biol Chem 278, 10157-61 (2003).

41. Lee, T. J. et al. Quercetin arrests G2/M phase and induces caspase-dependent cell death in U937 cells. Cancer Lett (2005).

40 42. Yoshida, M., Yamamoto, M. & Nikaido, T. Quercetin arrests human leukemic T-cells in late Gl phase of the cell cycle. Cancer Res 52, 6676-81 (1992).

43. Das, B. B. et al. Differential induction of Leishmania donovani bi-subunit topoisomerase I-DNA cleavage complex by selected flavones and camptothecin: activity of flavones against camptothecϊn-resistant topoisomerase I. Nucleic Acids Res 34, 1121-32 (2006).

44. Spencer, J. P., Rice-Evans, C. & Williams, R. J. Modulation of pro-survival Akt/protein 5 kinase B and ERK1/2 signaling cascades by quercetin and its in vivo metabolites underlie their action on neuronal viability. J Biol Chem 278, 34783-93 (2003).

45. Chen, 3. & Kang, J. H. Quercetin and trichostatin A cooperatively kill human leukemia cells. Pharmazie 60, 856-60 (2005).

46. Metzger, E. et al. LSDl demethylates repressive histone marks to promote androgen- 10 receptor-dependent transcription. Nature 437, 436-9 (2005).

47. Remington's Pharmaceutical Sciences 16th edition, Osol, A. Ed. (1980)

48. Bowie et al. Science 247(1990) 1306-1310

49. Merrifield, R. B. (1963), J. Amer. Chem. Soc. 85, 2149-2154)

50. Kent, S.B.H. (1988), Annu. Rev. Biochem. 57, 957-989

15 51. Carpino, LA. and Han, G.Y. (1972), J. Org. Chem. 37, 3404-3409

52. Stewart, J. M. and Young, J. D. (1983), "Solid Phase Peptide Synthesis", Pierce Chemical Company, Rockford, Illinois

53. Atherton, E. and Sheppard, R. C. (1989), "Solid Phase Peptide Synthesis", IRL Press at Oxford University Press

20 54. Pennington, M. W. and Dunn, B. M. (eds.) (1994), "Peptide Synthesis Protocols", Humana

Press, Totowa, New Jersey 55. Pessi, A. et al. Totally solid phase synthesis of peptide(s) - containing retro-inverted peptide bond, using crosslinked sarcosinyl copolymer as support. European Patent

97994-B, 30 Sep. 1987 (8739). 25 56. Verdini, A. S. & Viscomi, G. C. (1985) Synthesis, resolution, and assignment of configuration of potent hypotensive retro-inverso bradykinin potentiating peptide

5a(BPP5a) analogues. 3. Chem. Soc. Perkin Trans. I, 697-701. 57. Bonelli, F. et al. (1984) Solid phase synthesis of retro-inverso peptide analogues. Int. J.

Peptide Protein Res., 24, 553-556. 30 58. Letunicl, Copley RR, PiIs B, Pinkert S, Schultz, J, Bork P SMART 5: domains in the contexct of genomes and networks. Nucleic acid Research (2006)3an 1 : 34: D257-60. 59. Schultz J, Milpetz F, Bork P, Ponting CP SMART, a simple modular architecture research tool: identification of signaling domains. Proc Natl Acad Sci U S A. 1998 May

26;95(ll):5857-64.