Protein Design of the Mammalian DNA Methyltransferase Dnmt3a

Protein design of the mammalian DNA methyltransferase Dnmt3a

by Abu Nasar Siddique

A thesis submitted in partial fulfillment of the requirements for the degree of

Doctor of Philosophy in Molecular Biotechnology

Approved, Thesis Committee

______Prof. Dr. Albert Jeltsch

______Prof. Dr. Sebastian Springer

______Prof. Dr. Marianne G. Rots

Date of Defense: May 25, 2011 School of Engineering and Science Jacobs University Bremen

I. ACKNOWLEDGMENTS I would like to express my deepest gratitude to my supervisor Prof. Dr Albert Jeltsch for his excellent guidance into the Science of Biochemistry, for his encouragement and his enthusiasm, which has always been the source of my inspiration and motivation. I am thankful to Prof. Sebastian Springer for being the co-referee of my PhD thesis, and especially to Prof. Marianne G. Rots, who has traveled a long way to be a part of my thesis committee. I would like to express my appreciation to my Laboratory supervisor Dr. Tomasz Jurkowski for his constructive discussions, excellent commentaries and suggestions. I am grateful to all my lab fellows, especially to Renata, Arumugam, Sanjay, Arun as well as to our neighbor Abhishek for creating a friendly working atmosphere and many useful advices. I would also like to thank my friends: Farhan, Amna, Farhat, Nawab, Nadia, Sadiq, Noorshad, Aasim, Wakeel, Guftaar, Qazi, Imran, Ihteram, Ali, Tariq, Noor Muhammad, Zia, Naveed, Shoaib, Tahir and Amir, whom I will not enumerate due to the lack of space. Thank you for bringing so much fun, support and distraction into my life. I am also thankful to the Higher Education Commission (HEC) in Pakistan and Deutscher Akademischer Austauschdienst (DAAD) in Germany, for providing financial support and help in every aspect during PhD studies. Lastly, but not at all least, I am thankful to my family members especially my wife, brother, sisters and my mother for giving me moral support, when my frustration overbalanced my passion.

Declaration

The work described in this thesis is my own work, unless otherwise stated or mentioned in the references. The thesis was written by me and nobody else.

Abu Nasar Siddique

Bremen, May 2011.

II. LIST OF PUBLICATIONS

Jurkowska RZ*, Siddique AN*, Jurkowski TP, Jeltsch A. Approaches to Enzyme and Substrate Design of the Murine Dnmt3a DNA Methyltransferase.

ChembioChem. 2011, in press

* Both authors contributed equally to this work

(In this paper I performed all the experiment related to the directed evolution of

Dnmt3a)

Siddique AN, Jurkowska RZ, Jurkowski TP, Jeltsch A. Auto-methylation of the mouse DNA-(cytosine C5)-methyltransferase Dnmt3a at its active site cysteine residue.

FEBS J. 2011, in press

(In this paper I performed all the experiments)

TABLE OF CONTENTS I. ACKNOWLEDGEMENT------II II. LIST OF PUBLICATIONS------4 III. TABLE OF CONTENTS------5 IV. ABSTRACT------6 2. INTRODUCTION------7 2.1. Epigenetics------7 2.1.1 DNA methylation in mammals------7

2.1.2 Importance of DNA methylation in mammals------8 2.2. Types of mammalian DNA methyltransferases------9 2.2.1 Catalytic mechanism of DNA methylation in mammals------11 2.2.2 Dnmt3 family------13 2.2.3 Structure of Dnmt3a/3a-3L------14 2.3. Aim of the project - to increase the activity of Dnmt3a------15 2.3.1 Protein Engineering------17 2.3.2 Directed enzyme evolution------17 3. RESULTS AND DISCUSSION------19 Directed evolution of the mammalian DNA methyltransferase Dnmt3a towards higher activity------22 Auto-methylation of the mouse DNA methyltransferase Dnmt3a-C------25 4. REFERENCES------28 Supplement 1 Supplement 2 Supplement 3

IV. Abstract

The catalytic domain of the Dnmt3a DNA-(cytosine C5)-methyltransferase

(Dnmt3a-C) is active in isolated form but like full-length Dnmt3a it shows only weak DNA methylation activity. To improve this activity by directed evolution, we set up a selection system in which Dnmt3a-C methylate its own expression plasmid in E. coli and protect it from cleavage with methylation specific restriction enzymes. However, despite screening about 400 clones which were selected in 3 rounds from a library of 60000 clones, we were not able to isolate a variant with improved activity, most likely because of a background of uncleaved plasmids and plasmids which have lost the restriction site.

We also showed that the catalytic domain of mouse Dnmt3a DNA methyltransferase is able to transfer the methyl group from S-adenosyl-L- methionine (AdoMet) to a cysteine residue in its catalytic centre. This reaction is reversible and slow. The yield of auto-methylation is increased by addition of

Dnmt3L, which functions as a stimulator of Dnmt3a AdoMet complexes. In the presence of CpG containing double stranded DNA, the transfer of the methyl group from AdoMet to the flipped target base was preferred and auto-methylation was not detected. This reaction might constitute a regulatory mechanism which could inactivate unused Dnmt3a in the cell.

2. Introduction

2.1. Epigenetics

In multicellular organism all cells are derived from the zygote and they carry identical genetic information but they follow different developmental pathways. The phenotype and cellular fate is determined by epigenetic regulatory mechanisms that are heritable through cell divisions and function without changing the DNA sequence [Allis et al., 2007; Watson et al., 2008].

Epigenetic mechanisms include DNA methylation, histone protein covalent modifications (such as acetylation, phosphorylation and methylation) and expression of noncoding RNAs that lead to gene silencing [reviewed in Goldberg et al., 2007]. These epigenetic signals control the proper timing of gene expression during cellular development and differentiation. Out of all epigenetic mechanisms, the mechanism and inheritance of DNA methylation is best understood at a molecular level.

2.1.1 DNA methylation in mammals

Methylated bases can be found in almost all living organisms, ranging from bacteria and fungi, to plants and mammals. DNA methylation is the best known and the most abundant kind of DNA modifications which adds additional information to the genetic code and thus methylated bases can be considered as additional letters of the genetic alphabet [Jeltsch, 2002]. In mammals, DNA methylation occurs at the C5 position of cytosine residues, predominantly in the

CG dinucleotides and only occasionally at non-CG sites. However, only certain

CG sites are methylated, resulting in the generation of a tissue and cell-type- specific pattern of methylation [Jurkowska et al., 2011].

Although methylation of the bases in DNA does not influence the Watson-

Crick base pairing, the methyl group, which is placed in the major groove of the

DNA, can be recognized by DNA interacting proteins and influence the specificity, kinetic properties or the thermodynamics of their interactions with DNA [Auriol et al., 2005]. There are 56 million CG sites in the human genome, approximately

60-80% of them are modified (4-6% of all cytosines) [Lister et al., 2009; Laurent et al., 2010].

2.1.2 Importance of DNA methylation in mammals

DNA methylation plays central regulatory roles in the control of the cellular physiology. It is involved in embryonic development [Li et al., 1992], cellular reprogramming, [Yang et al., 2007], control of gene expression [Lande-Diner et al., 2007] stabilization of X chromosome inactivation, [Heard, 2004; Chang et al.,

2006; Yen et al., 2007], brain function and behavior [Sweatt, 2009; Zhang et al.,

2010], regulation of parental imprinting [Barlow, 1995; Delaval & Feil, 2004] and maintenance of the genome integrity through protection against endogenous retroviruses and transposons [Lengauer et al., 1997; Howard et al., 2008].

Aberrant methylation causes various human diseases like psychiatric and immune system diseases [Robertson, 2005; Feinberg, 2007], ICF

(immunodeficiency, centromeric region instability and facial anomalies) syndrome

[Hansen et al., 1999; Xu et al., 1999] and contributes to the development of

8 cancer [Jones & Baylin, 2007; Jones & Baylin, 2002] and ageing [Richardson,

2003].

2.2. Types of mammalian DNA methyltransferases (MTases)

In mammals, 5 different DNA methyltransferases or methyltransferase like proteins have been identified [Jeltsch, 2002]. These include maintenance methyltransferase which is Dnmt1 and de novo methyltransferases that are

Dnmt3a and Dnmt3b and an allosteric activator of the de novo methyltransferases – the Dnmt3L protein (Fig. 1). Additionally, the Dnmt2 enzyme which was initially predicted to be a DNA methyltransferase turned out to be involved in methylation of tRNAAsp [Goll et al., 2006].

Mammalian DNA methyltransferases contain two parts, a large multidomain N-terminal part, which is of variable size and has regulatory functions, and a catalytic part at the C-terminus (Fig. 1). The bigger N-terminal part guides the nuclear localization of the enzymes [Bestor & Verdine, 1995] and mediates their interactions with other proteins, DNA and chromatin [Chuang et al.,

1997; Fuks et al., 2000; Robertson et al., 2000]. The smaller C-terminal part is the active centre of the enzyme and contains ten amino acids motifs diagnostic for all DNA C5 cytosine MTases [Cheng, 1995].

All DNA MTases share a common core structure in the catalytic domains, known as “AdoMet-dependent methyltransferase fold”, which consists of a mixed seven-stranded β-sheet, formed by six parallel β strands and a seventh strand in an antiparallel orientation, inserted into the sheet between strands 5 and 6. Six

helices are folded around the central β-sheet [Cheng & Blumenthal, 2008]. This

domain is involved both in cofactor binding (motifs I and X) and catalysis (motifs

IV, VI and VIII). Target recognition domain (TRD) is the nonconserved region

between motifs VIII and IX which is involved in DNA recognition and specificity

[Cheng, 1995; Jeltsch, 2002].

N-terminal part C-terminal part P C N DNA-(cytosine-C5)-MTase motifs Charge rich N L DNA replication foci Cys – rich Pb-region region S targeting sequence CXXC I IV VI VIII IX X A Dnmt1 (1616 aa) I IV VI VIII IX X Dnmt2 (415 aa)

PWWP PHD I IV VI VIII IX X Dnmt3a (912 aa)

PWWP PHD I IV VI VIII IX X Dnmt3b (853 aa)

PHD I IV VI VIII Dnmt3L (387 aa)

Figure 1. Schematic representation of mammalian DNA methyltransferases [Jeltsch, 2002].

Besides the conservation of the structure, DNA MTases share an

important mechanistic similarity: in order to get access to the substrate base,

which is buried inside the DNA helix, DNA methyltransferase rotate it out of the

helix (this phenomenon is called base flipping) and bind the flipped base in a

hydrophobic pocket of the enzyme [O'Gara et al., 1998].

The DNA methyltransferase enzymes catalyze the transfer of the methyl

group from a cofactor molecule S-adenosyl-L-methionine (AdoMet or SAM) to the

C5 position of the cytosine residues. As a result, 5-methylcytosine is created, and

S-adenosyl-L-homocysteine (AdoHcy, SAH) is released from the enzyme [Jeltsch,

2002].

2.2.1 Catalytic mechanism of DNA methylation in mammals

In mammals, methylation of DNA at the C5-position of cytosine residues is the only known covalent modification of DNA. The DNA C5 methyltransferases share a similar catalytic mechanism [reviewed in Cheng, 1995; Jeltsch, 2002].

Cytosine SAM NH2

H3C O S Adenine O N O NH3 HO OH O N

Deoxyribose DNA 5-methylcytosine methyltransferase

NH2 SAH

CH3 O S Adenine N O

O NH3 O N HO OH

Deoxyribose

Figure2. Transfer of the methyl group from SAM to cytosine residue is catalysed by a DNA methyltransferase, creating 5-methylcytosine and SAH.

These enzymes use S-adenosyl-L-methionine as a donor of an activated methyl group, which is transferred to the 5th position of the cytosine base. The reaction requires a number of highly conserved amino acid residues but the key

11 step in the catalysis is the nucleophilic attack of the enzyme on the 6th position of the cytosine and a formation of a covalent bond between the enzyme and the substrate base [Wu & Santi, 1987; Kumar et al., 1994; O´Gara et al., 1996]. The catalytic cysteine residue which is located in PCQ motif (motif IV) performs this step. The negative charge density at the C5 atom of the cytosine is increased by the formation of the covalent cysteine-cytosine bond which then attacks the methyl group bound to S-adenosyl-L-methionine. The nucleophilic attack of the cysteine might be facilitated by a transient protonation of the cytosine ring at the endocyclic nitrogen atom (N3) by an enzyme derived acid [Chen, 1993]. This protonation is mediated by conserved glutamate residue from the ENV motif

(motif VI) with the help of arginine residue from the RXR motif (motif VIII) in case of M. HhaI methyltransferase [O´Gara et al., 1996]. Position 5 of the cytosine is strongly activated and attacks the methyl group. The covalent enzyme-DNA complex is resolved by deprotonation at position 5 which leads to the elimination of the cysteine SH group and reestablishes aromaticity [Jeltsch, 2002].

Figure3. Catalytic mechanism of C-MTases [Jeltsch, 2002].

2.2.2 Dnmt3 family

The Dnmt3 MTase family has been discovered in 1998 [Okano et al.,

1998]. It has three members, called as Dnmt3a, Dnmt3b and Dnmt3L. Dnmt3a and Dnmt3b are highly expressed in embryonic tissues [Okano et al., 1998] and establish the methylation pattern in the early development [Chen et al., 2003] and germ cells of mammals, whereas only a low expression is observed in fully differentiated cells [Robertson et al., 1999]. Dnmt3a and Dnmt3b do not display any significant preference between hemimethylated and unmethylated DNA, due to this reason they are considered as bona fide de novo methyltransferases

[Okano et al., 1998; Gowher & Jeltsch, 2001]. Both are closely related to each other and share 85% of sequence identity within their catalytic domains (and

36% overall). Although Dnmt3a and Dnm3b function primarily as de novo methyltransferases, they were also shown to co-operate with Dnmt1 in maintaining of DNA methylation patterns [Chen et al., 2003]. Targeted disruption of the any of their genes in mice resulted in lethal phenotype [Okano et al., 1998].

Dnmt3b knockout mouse embryos die in uterus whereas the Dnmt3a knockout animals develop to term, but become runts and die shortly after birth [Okano et al., 1999]. Dnmt3L knockout mice do not show clear morphological abnormalities and are viable [Hata et al., 2002] but male Dnmt3L knockout mice are sterile

[Bourc’his & Bestor, 2004] and female are fertile but the developing embryos die as a result of defects in the development of the neural tubes [Hata et al., 2002;

Webster et al., 2005].

Dnmt3L [Aapola et al., 2000] is catalytically inactive, unable to bind to the cofactor SAM and binds only very weakly to DNA [Gowher et al., 2005; Kareta et al., 2006]. It is expressed specifically in germ cells during gametogenesis and embryonic stages and is essential for establishment of a subset of methylation patterns in both male and female germ cells [Hata et al., 2002; Bourc’his &

Bestor, 2004]. It was shown that it directly interacts with Dnmt3a and Dnmt3b proteins and facilitates the catalysis by stimulating the binding of AdoMet and

DNA to these proteins [Gowher et al., 2005]. Dnmt3L is therefore considered as an activator of de novo MTases. Dnmt3L shows similarity to the Dnmt3a and

Dnmt3b proteins in both the N- and C- terminal domains. In spite of the clear similarity to Dnmt3a and Dnmt3b proteins, Dnmt3L lacks the conserved residues known to be involved in DNA methyltransferase activity and is inactive on its own

[Hata et al., 2002].

2.2.3 Structure of Dnmt3a/3a-3L

The Dnmt3a protein comprises of an N-terminal regulatory part with targeting function and a C-terminal catalytic part [Gowher & Jeltsch, 2002] that closely resembles the prokaryotic DNA cytosine-C5-methyltransferases. The N- terminal part of Dnmt3a contains two functional regions called ADD (ATRX-

DNMT3-DNMT3L) domain or PHD (plant homeodomain) domain that interacts with the histone H3 tail and PWWP domain that target the enzyme to heterochromatin [Chen et al., 2004]. The PWWP domain of Dnmt3a recognizes the H3K36 trimethylation mark and this interaction increases the activity of

Dnmt3a for methylation of nucleosomal DNA, [Dhayalan et al., 2010]. In Dnmt3L,

PWWP domain is absent and C terminal part lacks the MTase motifs IX and X and all important catalytic residues in its C-terminal domain. Dnmt3a and Dnmt3L form a linear heterotetramer complex which consists of two Dnmt3L (at the edges of the tetramer) and two Dnmt3a molecules (in the centre) [Jia et al., 2007] (Fig.

4). The complexed C-terminal domains of Dnmt3a and Dnmt3L show dimerization through Dnmt3a–Dnmt3a interaction, forming a tetrameric complex with two active sites [Jia et al., 2007].

Figure 4. Surface representation of the Dnmt3a-Dnmt3L heterotetramer complex. b. Dnmt3a-

Dnmt3L tetramer with one contiguous curve DNA molecule [Jia et al., 2007].

2.3. Aim of the project - to increase the activity of Dnmt3a

Dnmt3a is highly expressed in embryonic tissues [Okano et al., 1998] and establish the methylation pattern in the early development [Chen et al., 2003].

During the embryonic development, the chromosomal methylation pattern is set by the Dnmt3a and Dnmt3b methyltransferases. This process is effectively

15 accelerated by the direct interaction of these proteins with Dnmt3L. Interaction of

Dnmt3a and Dnmt3L increases the binding of the coenzyme AdoMet to Dnmt3a.

Moreover, Dnmt3L induces a conformational change of Dnmt3a that opens the active site of the enzyme and promotes binding of DNA and the AdoMet [Gowher et al., 2005]. It was shown, that up to 20 fold stimulatory effect of Dnmt3L is caused by allosteric activation of the catalytic domains of Dnmt3a. Until now, the molecular basis of this stimulation by Dnmt3L is not yet understood. It was our aim to increase the catalytic activity of Dnmt3a-C so that it could be used for the targeted DNA methylation.

Early studies that measured the global content of 5-methylcytosine of tumors showed that hypomethylation was a common feature of carcinogenesis, leading to abnormal chromosomal instability and transcriptional regulation [Eden et al., 2003; Pogribny & Beland, 2009]. Targeted DNA methylation is a new approach for stable silencing of gene expression by epigenetic mechanisms

[Jeltsch, 2006; Jurkowska & Jeltsch, 2010]. In targeted methylation, a methyltransferase is fused to a specific DNA binding domain, usually a designed zinc finger binding domain, and then this chimeric methyltransferase is delivered to the promoter region of the target gene to methylate it. This will lead to the repression of its expression. The stable silencing of gene expression by targeted

DNA methylation is a powerful technology for specific gene silencing [Li et al.,

2007], but it depends on the catalytic activity of the DNA methyltransferase. The stimulation of Dnmt3a by Dnmt3L shows that Dnmt3a has a potential for increased catalytic activity. Here, we propose to use the methods of directed

16 evolution to evolve Dnmt3a protein variants with higher activity which overcomes the dependence on Dnmt3L stimulation or another strategy which is described in the supplement 3, only available to reviewers.

2.3.1 Protein Engineering

The aim of the protein engineering is to improve or evolve an enzyme or protein in a way that it can be used for a broad range of applications. There are two general strategies for protein engineering: directed evolution [Farinas, et al.,

2000] and rational design [Bornscheuer & Pohl, 2001; Bornscheuer, 2002].

2.3.2 Directed enzyme evolution

Directed evolution is a powerful tool in protein engineering that could be applied to bring about the desired changes in an enzyme by generating a random library of the enzyme and then selection of the desired variant from random library. Dnmt3a is prone to in vitro evolution approaches, because it methylates the DNA. A library of Dnmt3a variants could be generated through error prone

PCR and the methylation activity of the highly active variants could be detected on the DNA by digesting it with methylation sensitive restriction enzymes.

Whenever the plasmid DNA is methylated within such a recognition site, the methylation sensitive restriction enzyme is unable to digest it and the methylated plasmid survives the selection and can be retrieved for the next round of directed evolution. The blueprints of the enzyme activity and specificity can be detected directly on the DNA coding for the protein. This gives the coupling of genotype

(DNA sequence) and phenotype (enzymatic properties) on the individual DNA molecule (Fig. 5).

Expression plasmid with MTase gene

Random mutagenesis

Library of MTase expression vectors with mutations

MTase expression in the cells

Plasmid DNA isolation

Library of MTase expression plasmids carrying the methylation pattern imprinted by the MTase variant encoded by each plasmid MTase methylates the DNA inside the cell

Figure 5. Directed evolution of Dnmt3a, ideally suited for evolutionary protein design (Figure

provided by Dr. Jurkowski).

An additional part of this thesis is available at supplement 3, for reviewers only.

Results and Discussions

One striking feature of DNA methyltransferases is that they can mark DNA with methyl groups at specific sites inside the cell. In mammals this methyl mark occurs at CG sites on both strands of the DNA in cell-type specific pattern.

Aberrant DNA methylation contributes to the development of many diseases, as it leads to loss of transcriptional control [reviews, Egger et al., 2004; Feinberg &

Tycko, 2004; Rodenhiser & Mann, 2006]. In most cancer cells it has been found that either the CG sites were less methylated [Esteller, 2005] or an abnormal

DNA methylation patterns have been found [Feinberg & Vogelstein, 1983].

Aberrant methylation of the promoter region of tumour suppressor genes is associated with transcriptional silencing of such genes [Esteller & Herman, 2002;

Jones & Baylin, 2002; Herman & Baylin, 2003]. Several genes like p15,

MDR1and HIC1 have been shown to be inactivated by methylation [Melki et al.,

1999; Toyota et al., 2001; Galm et al., 2005]. Hypomethylation of DNA on one hand can cause activation of oncogenes [Nakayama et al., 1998; Nishigaki et al.,

2005] and on the other hand it has other negative effects, like loss of imprinting

[Sakatani et al., 2005], generation of chromosomal instability [Eden et al., 2003;

Gaudet et al., 2003] and re-activation of transposons.

From the above discussion it is clear that the expression of any gene could be down-regulated by introducing DNA methylation in a CpG island in the promoter region of that gene.

Targeted DNA methylation is an approach which is used for the silencing of the gene expression by methylation of its promoter region [Jeltsch, 2006;

Jurkowska, 2010]. Targeted methylation is achieved by delivering a DNA methyltransferase specifically to a CpG island in the promoter region of the target gene. To this end the DNA methyltransferase is fused to specific DNA binding domain, often a designed zinc finger domain, which specifically binds to DNA in or near the promoter region, and a DNA methyltransferase domain that methylates CpG island near the zinc finger binding region in the promoter area.

The advantage of this technology is that methyltransferase-zinc finger fusion proteins could be generated at genetic level and the methyltransferase-zinc finger fusion protein then can be transfected into cell such that fusion protein is produced inside the cell. The aberrant expression of genes could be repressed or down-regulated by targeted DNA methylation. Principally, this technology could be used to suppress the expression of any gene where there is aberrant methylation or hypomethylation if there is a CpG island in the promoter region.

Active Dnmt3a-C

Zinc Zinc Zinc Finger Finger Finger

CG CG CG

Zinc-Finger binding site Target gene promoter

Figure 6. Schematic drawing of the chimeric DNA methyltransferase fused with targeting zinc finger [adapted from Jeltsch, 2006].

Targeted DNA methylation and the use of different DNA methyltransferases have been recently discussed [Jeltsch, 2006; Jurkowska &

Jeltsch, 2010] where the catalytic domain of Dnmt3a was identified as a promising candidate to be used in targeted DNA methylation. In principal, the use of the catalytic domain of Dnmt3a was successful in targeted DNA methylation [Li et al., 2007] but the limiting factor in this work was the low catalytic activity of

Dnmt3a. For more efficient targeted DNA methylation, we used two approaches to increase the catalytic activity of Dnmt3a. The first approach was directed evolution of the catalytic domain of Dnmt3a towards higher activity and the second approach to increase the catalytic activity of Dnmt3a was by protein design as described in the confidential supplemental document.

Directed evolution of the mammalian DNA methyltransferase

Dnmt3a towards higher activity

It has already been shown that Dnmt3a-C has relatively weak catalytic activities in vitro [Gowher & Jeltsch, 2002; Gowher & Jeltsch, 2001; Yokochi &

Robertson, 2002] when expressed in E. coli cells and the activity of Dnmt3a could be stimulated by Dnmt3L both in vivo and in vitro [Gowher et. al., 2005;

Hata et. al., 2002; Suetake et. al., 2004]. The reason for this low catalytic activity could be that the active centre of Dnmt3a is not in an optimal conformation to support efficient catalysis. To increase the catalytic activity of Dnmt3a, we applied the approach of directed evolution which is a powerful approach for enzyme design [Turner, 2009; Gerlt, P C Babbitt, 2009]. That had been successfully applied for prokaryotic DNA methyltransferases, which in general have a high catalytic activity [Chahar et. al., 2010; Gerasimaite et. al., 2009]. We generated random libraries of the catalytic domain of Dnmt3a with low, medium and high rate of mutagenesis, but despite of screening about 400 clones which were selected in 3 rounds from a library of 60000 clones, we were not able to isolate a variant with improved DNA methylation activity. Our selection system proved to be working because after a couple of rounds of selection, the surviving clones were found to have mutations within the restriction enzyme recognition site. We select for the absence of cleavage in the enzyme restriction site which can be due to methylation or mutation. Although, not part of the mutational target region, by chance sometime mutations are also introduced in the restriction site.

Such events are expected to happen at very low frequency. The enrichment of

22 corresponding clones after selection, therefore, illustrates that the selection system was working. There could be many reasons for the failure of selecting a highly active mutant which will be discussed in the coming section.

We mutagenised the whole catalytic domain of Dnmt3a which is more than 300 amino acids long. The chance of getting an activating mutant becomes very low when the target region for mutagenesis is large. In addition, we expected that more than one mutation will be needed for the activation of

Dnmt3a. Therefore, a high level of randomization was applied. However, under these conditions one mutation could be beneficial but there could be many mutations which could disrupt the protein folding such that the activating mutation would not be found. Having this point in mind, we also generated a random library of the catalytic domain of Dnmt3a with lower rate of mutations, but still we could not get active mutant.

A similar procedure was applied by one of our colleagues, quite successful to change the recognition sequence of EcoDam [Chahar et. al., 2010].

However, in the directed evolution of EcoDam, the design strategy was different and the mutagenesis was limited to few small loops which were selected rationally. Therefore the mutational space was much smaller and the likelihood of finding positive hits in the technical background was much higher. We also tried to follow a similar procedure as discussed above and generated a random library of Dnmt3a where mutagenesis was limited to that region of Dnmt3a which is interacting with Dnmt3L. This region was smaller but still we could not get an active mutant. Another problem in the Dnmt3a design was that selection of about

0.1-1% of all plasmids were not cleaved by methylation sensitive restriction enzymes even without methylation (incomplete cleavage). However, the catalytic activity of Dnmt3a is very low leading to protection of a similar degree. Therefore, if this low catalytic activity is stimulated a little bit in a mutant, still it is close to the background of incomplete cleavage. Then, protected plasmids encoding variants with improved activities cannot be easily enriched.

Auto-methylation of the mouse DNA methyltransferase Dnmt3a

We showed that the catalytic domain of mouse Dnmt3a DNA methyltransferase is able to transfer the methyl group from S-adenosyl-L- methionine (AdoMet) to a cysteine residue in its catalytic center. We confirmed this reaction using the alanine exchange mutant C120A of Dnmt3a [Reither et. al.,

2003] which did not get labeled. The transfer of the methyl group from AdoMet to cysteine residues of protein was already observed for some bacterial methyltransferases, including the DNA (cytosine C5) methyltransferases as M.

BspRI [Szilak et. al., 1994; Szilak et. al., 1995], Dcm [Hanck et. al., 1993] and the

DNA-(adenine N6) methyltransferase M. EcoPI [Hornby et. al., 1987]. For

M.BspRI, it was suggested that the methylation of cysteine residue of the protein results in the inactivation of the enzyme [Szilak et. al., 1994]. Due to limitation of time, I could not perform some experiments which we think could be interesting.

One question is whether auto-methylation of Dnmt3a is happening inside mammalian cells? To observe this phenomenon in vivo, wild type Dnmt3a protein and alanine exchange mutant C120A of Dnmt3a could be expressed in mammalian cells. Then the protein could be immunoprecipitated with an antibody.

After detecting the right size on an SDS gel, one can cut the exact size of protein band from the gel and do in gel trypsin digestion. The sample can then be analyzed on MALDI-TOF to detect the peptide fragment containing the methylated C120 residue, similarly as we did after in vitro auto-methylation. The experiment on one hand would confirm the occurrence of auto-methylation of the

Dnmt3a inside mammalian cells and on the other hand it may allow to estimate the extent of auto-methylation.

Another question is if auto-methylation of Dnmt3a-C has a biological role.

Due to auto-methylation, Dnmt3a may become inactived, and such inactivation might protect the genome against aberrant methylation when Dnmt3a is in idle state in the cell. For example outside of S-phase when not much methyltransferase activity is needed. The inactivation of auto-methylated Dnmt3a could be checked by purifying the modified protein by chromatographic procedures and then check the catalytic activity of methylated versus unmethylated protein. The reason of low catalytic activity of Dnmt3a, when purified from E. coli could also be its auto methylation.

In contrast to non-covalent AdoMet binding which happens with in minutes

[Gowher et. al., 2006], the radioactive signal of the auto-methylation of Dnmt3a-C increased slowly over the course of hours. We incubated the auto-methylated

Dnmt3a with S-Adenosyl-L-homocysteine (AdoHcy) and observed that the reaction stopped and there was no further incorporation of radioactivity. Though we observed that this reaction continues till overnight but inside living cell it may stop after sometime, due to accumulation of AdoHcy which is the end product of the reaction. The slow auto-methylation leading to inactivation could be a system to inactivate the enzyme if it doesn't get a substrate for methylation for a longer period of time. The methyl group may also recruit the auto-methylated Dnmt3a to a proteosomal complex which could have a methyl recognition domain. Once recognised by proteosomal complex or proteases, the degradation of inactive

26 auto-methylated Dnmt3a may start. The inactive auto-methylated Dnmt3a could also have a structural role rather than catalytic role. Methylated Dnmt3a might be targeted to the heterochromatin where it may stabilize the heterochromatin structure while remaining inactive inside the cell.

Recently, Jumonji C (JmJC)-domain-containing proteins has been identified as the enzymes that directly remove the methyl marks from lysines in histones and other proteins. It is an oxidative mechanism where methyl group is released as formaldehyde [Klose et al., 2006]. It may also happen that the methyl group from Dnmt3a protein is removed by demethylases enzymes following the same mechanism and the methyltransferase become active inside the cell. The demethylation of methylated Dnmt3a could be checked if methylated Dnmt3a protein is incubated with the cell extract for some time. Then the protein could be isolated and the extent of methylation checked by mass spectroscopy.

4. References

Aapola U, Kawasaki K, Scott HS, Ollila J, Vihinen M, Heino M, Shintani A,

Kawasaki K, Minoshima S, Krohn K, Antonarakis SE, Shimizu N, Kudoh J

& Peterson P. 2000. Isolation and initial characterization of a novel zinc

finger gene, DNMT3L, on 21q22.3, related to the cytosine-5-

methyltransferase 3 gene family. Genomics. 65:293-298.

Allis CD, Jenuwein T & Reinberg D. 2007. Epigenetics. Cold Spring Harbor, N.

Y.: Cold Spring Harbor Laboratory Press.

Auriol E, Billard LM, Magdinier F & Dante R. 2005. Specific binding of the methyl

binding domain protein 2 at the BRCA1-NBR2 locus. Nucleic acids

research. 33:4243-4254.

Barlow DP. 1995. Gametic imprinting in mammals. Science (New York, NY

270:1610-1613.

Bestor TH & Verdine GL.1995. DNA methyltransferases. Curr Opin Cell Biol.

6(3):380-9.

Bornscheuer UT & Pohl M. 2001. Improved biocatalysts by directed evolution

and rational protein design. Curr. Opin. Chem. Biol. 5:137-143.

Bornscheuer UT. 2002. Methods to increase enantioselectivity of lipases and

esterases. Curr. Opin. Biotechnol. 13:543-547.

Bourc'his D & Bestor TH. 2004. Meiotic catastrophe and retrotransposon

reactivation in male germ cells lacking Dnmt3L. Nature. 431:96-99.

Chahar S, Elsawy H, Ragozin S & Jeltsch A. 2010. Changing the DNA

recognition specificity of the EcoDam DNA-(adenine-N6)-

methyltransferase by directed evolution. J Mol Biol. 395(1):79-88.

Chang SC, Tucker T, Thorogood NP & Brown CJ. 2006. Mechanisms of X-

chromosome inactivation. Front Biosci. 11:852-866.

Chen L, MacMillan AM & Verdine GL. 1993. Mutational separation of DNA

binding from catalysis in a DNA cytosine methyltransferase. J Am Chem

Soc. 115:5318-5319.

Chen L, Paulsen DB, Scruggs DW, Banes MM, Reeks BY & Lawrence ML. 2003.

Alteration of DNA adenine methylase (Dam) activity in Pasteurella

multocida causes increased spontaneous mutation frequency and

attenuation in mice. Microbiology. 149:2283-2290.

Chen T, Tsujimoto N & Li E. 2004. The PWWP domain of Dnmt3a and Dnmt3b

is required for directing DNA methylation to the major satellite repeats at

pericentric heterochromatin. Mol Cell Biol. 24(20):9048-58.

Chen T, Ueda Y, Dodge JE, Wang Z & Li E. 2003. Establishment and

maintenance of genomic methylation patterns in mouse embryonic stem

cells by Dnmt3a and Dnmt3b. Molecular and cellular biology. 23:5594-

5605.

Cheng X & Blumenthal RM. 2008. Mammalian DNA methyltransferases: a

structural perspective. Structure. 16:341-350.

Cheng X. 1995. Structure and function of DNA methyltransferases. Annual

review of biophysics and biomolecular structure. 24:293-318.

Chuang LS, lan HI, Koh TW, Ng HH, Xu G & Li BF. 1997. Human DNA-(cytosine-

5) methyltransferase-PCNA complex as a target for p21WAF1. Science.

277:1996-2000.

Delaval K & Feil R. 2004. Epigenetic regulation of mammalian genomic

imprinting. Current opinion in genetics & development. 14:188-195.

Dhayalan A, Rajavelu A, Rathert P, Tamas R, Jurkowska RZ, Ragozin S &

Jeltsch A. 2010. The Dnmt3a PWWP domain reads histone 3 lysine 36

trimethylation and guides DNA methylation. J Biol Chem. 285(34):26114-

20.

Eden A, Gaudet F, Waghmare A & Jaenisch R. 2003. Chromosomal instability

and tumors promoted by DNA hypomethylation. Science. 300(5618):455.

Egger G, Liang G, Aparicio A & Jones PA. 2004. Epigenetics in human disease

and prospects for epigenetic therapy. Nature. 429(6990):457-63.

Esteller M & Herman JG. 2002. Cancer as an epigenetic disease: DNA

methylation and chromatin alterations in human tumours. J Pathol.

196(1):1-7.

Esteller M. 2005. Aberrant DNA methylation as a cancer-inducing mechanism.

Annu Rev Pharmacol Toxicol. 45:629-56.

Farinas E, Schwaneberg U, Anton G & Arnold FH. 2001. Directed evolution of a

cytochrome P450 monooxygenase for alkane oxidation. Adv. Synth. Catal.

343:601- 606.

Feinberg AP & Tycko B. 2004. The history of cancer epigenetics. Nat Rev

Cancer. 4(2):143-53.

Feinberg AP & Vogelstein B. 1983. Hypomethylation distinguishes genes of

some human cancers from their normal counterparts. Nature.

301(5895):89-92.

Feinberg AP. 2007. Phenotypic plasticity and the epigenetics of human disease.

Nature. 447:433–440.

Fuks F, Burgers WA, Brehm A, Hughes-Davies L & Kouzarides T. 2000. DNA

methyltransferase Dnmt1 associates with histone deacetylase activity.

Nature genetics 24:88-91.

Galm O, Wilop S, Lüders C, Jost E, Gehbauer G, Herman JG & Osieka R. 2005.

Clinical implications of aberrant DNA methylation patterns in acute

myelogenous leukemia. Ann Hematol. 1:39-46.

Gerasimaite R, Vilkaitis G & Klimasauskas S. 2009. A directed evolution design

of a GCG-specific DNA hemimethylase. Nucleic Acids Res. 37(21):7332-

41.

Gerlt JA & Babbitt PC. 2009. Enzyme (re)design: lessons from natural evolution

and computation. Curr Opin Chem Biol. 13(1):10-8.

Goldberg AD, Allis CD & Bernstein E. 2007. Epigenetics: a landscape takes

shape. Cell. 128:635-638.

Goll MG, Kirpekar F, Maggert KA, Yoder JA, Hsieh CL, Zhang X, Golic KG,

Jacobsen SE & Bestor TH. 2006. Methylation of tRNAAsp by the DNA

methyltransferase homolog Dnmt2. Science. 311:395-398.

Gowher H & Jeltsch A. 2001. Enzymatic properties of recombinant Dnmt3a DNA

methyltransferase from mouse: the enzyme modifies DNA in a non-

processive manner and also methylates non-CpG [correction of non-CpA]

sites. Journal of molecular biology. 309:1201-1208.

Gowher H & Jeltsch A. 2002. Molecular enzymology of the catalytic domains of

the Dnmt3a and Dnmt3b DNA methyltransferases. The Journal of

biological chemistry. 277:20409-20414.

Gowher H, Liebert K, Hermann A, Xu G & Jeltsch A. 2005. Mechanism of

stimulation of catalytic activity of Dnmt3A and Dnmt3B DNA-(cytosine-C5)-

methyltransferases by Dnmt3L. The Journal of biological chemistry.

280:13341-13348.

Gowher H, Loutchanwoot P, Vorobjeva O, Handa V, Jurkowska RZ, Jurkowski

TP & Jeltsch A. 2006. Mutational analysis of the catalytic domain of the

murine Dnmt3a DNA-(cytosine C5)-methyltransferase. Journal of

molecular biology. 357:928-941.

Hanck T, Schmidt S & Fritz HJ. 1993. Sequence-specific and mechanism-based

crosslinking of Dcm DNA cytosine-C5 methyltransferase of E. coli K-12 to

synthetic oligonucleotides containing 5-fluoro-2'-deoxycytidine. Nucleic

Acids Res. 21, 303-309.

Hansen RS, Wijmenga C, Luo P, Stanek AM, Canfield TK, Weemaes CM &

Gartler SM. 1999. The DNMT3B DNA methyltransferase gene is mutated

in the ICF immunodeficiency syndrome. Proceedings of the National

Academy of Sciences of the United States of America. 96:14412-14417.

Hata K, Okano M, Lei H & Li E. 2002. Dnmt3L cooperates with the Dnmt3 family

of de novo DNA methyltransferases to establish maternal imprints in mice.

Development (Cambridge, England) 129:1983-1993.

Heard E. 2004. Recent advances in X-chromosome inactivation. Current opinion

in cell biology. 16:247-255.

Herman JG & Baylin SB. 2003. Gene silencing in cancer in association with

promoter hypermethylation. N Engl J Med. 349(21):2042-54.

Hornby DP, Muller M & Bickle TA. 1987. High level expression of the EcoP1

modification methylase gene and characterisation of the gene product.

Gene. 54, 239-245.

Howard G, Eiges R, Gaudet F, Jaenisch R & Eden A. 2008. Activation and

transposition of endogenous retroviral elements in hypomethylation

induced tumors in mice. Oncogene. 27:404–408.

Jeltsch A. 2006. Mutational analysis of the catalytic domain of the murine

Dnmt3a DNA- (cytosine C5)-methyltransferase. J Mol Biol. 357:928-941.

Jeltsch A. 2006. On the enzymatic properties of Dnmt1: specificity, processivity,

mechanism of linear diffusion and allosteric regulation of the enzyme.

Epigenetics. 1(2):63-6.

Jestsch A. 2002. Beyond Watson and Crick: DNA methylation and molecular

enzymology of DNA methyltransferases. Chembiochem. 3:274-293.

Jia D, Jurkowska RZ, Zhang X, Jeltsch A & Cheng X. 2007. Structure of Dnmt3a

bound to Dnmt3L suggests a model for de novo DNA methylation. Nature.

449:248-251.

Jones PA & Baylin SB. 2002. The fundamental role of epigenetic events in

cancer. Nature reviews. 3:415-428.

Jones PA & Baylin SB. 2007. The epigenomics of cancer. Cell. 128:683-692.

Jurkowska RZ & Jeltsch A. 2010. Silencing of gene expression by targeted DNA

methylation: concepts and approaches. Methods Mol Biol. 649:149-61.

Jurkowska RZ, Jurkowski TP & Jeltsch A. 2011. Structure and Function of

Mammalian DNA Methyltransferases. ChemBioChem. 12(2):206-22.

Jurkowska RZ, Siddique AN, Jurkowski TP & Jeltsch A. 2011. Approaches to

Enzyme and Substrate Design of the Murine Dnmt3a DNA

Methyltransferase. Chembiochem. In press.

Kareta MS, Botello ZM, Ennis JJ, Chou C & Chédin F. 2006. Reconstitution and

mechanism of the stimulation of de novo methylation by human DNMT3L.

J Biol Chem. 281(36):25893-902.

Klose RJ, Kallin EM & Zhang Y. 2006. JmjC-domain-containing proteins and

histone demethylation. Nat Rev Genet. 7(9):715-27.

Kumar S, Cheng X, Klimasauskas S, Mi S, Posfai J, Roberts RJ & Wilson GG.

1994. The DNA (cytosine-5) methyltransferases. Nucleic acids research.

22:1-10.

Lande-Diner L, Zhang J, Ben-Porath I, Amariglio N, Keshet I, Hecht M, Azuara V,

Fisher AG, Rechavi G & Cedar H. 2007. Role of DNA methylation in stable

gene repression J. Biol. Chem. 282:12194–12200.

Laurent L, Wong E, Li G, Huynh T, Tsirigos A, Ong CT, Low HM, Kin Sung KW,

Rigoutsos I, Loring J & Wei CL. 2010. Dynamic changes in the human

methylome during differentiation. Genome Res. 20:320-331.

Lengauer C, Kinzler KW & Vogelstein B. 1997. DNA methylation and genetic

instability in colorectal cancer cells. Proceedings of the National Academy

of Sciences of the United States of America. 94:2545-2550.

Li E, Bestor TH & Jaenisch R. 1992. Targeted mutation of the DNA

methyltransferase gene results in embryonic lethality. Cell. 69:915-926.

Li F, Papworth M, Minczuk M, Rohde C, Zhang Y, Ragozin S & Jeltsch A. 2007.

Chimeric DNA methyltransferases target DNA methylation to specific DNA

sequences and repress expression of target genes. Nucleic Acids Res.

35(1):100-12.

Li JY, Pu MT, Hirasawa R, Li BZ, Huang YN, Zeng R, Jing NH, Chen T, Li E,

Sasaki H & Xu GL. 2007. Synergistic function of DNA methyltransferases

Dnmt3a and Dnmt3b in the methylation of Oct4 and Nanog. Molecular and

cellular biology. 27:8748-8759.

Lister R, Pelizzola M, Dowen RH, Hawkins RD, Hon G, Tonti-Filippini J, Nery JR,

Lee L, Ye Z, Ngo QM, Edsall L, Antosiewicz-Bourget J, Stewart R, Ruotti

V, Millar AH, Thomson JA, Ren B & Ecker JR. 2009. Human DNA

methylomes at base resolution show widespread epigenomic differences.

Nature. 462:315–322.

Melki JR, Vincent PC & Clark SJ. 1999 Cancer-specific region of

hypermethylation identified within the HIC1 putative tumour suppressor

gene in acute myeloid leukaemia. Leukemia. 13(6):877-83.

Nishigaki M, Aoyagi K, Danjoh I, Fukaya M, Yanagihara K, Sakamoto H, Yoshida

T & Sasaki H. 2005. Discovery of aberrant expression of R-RAS by

cancer-linked DNA hypomethylation in gastric cancer using microarrays.

Cancer Res. 65(6):2115-24.

O'Gara M, Horton JR, Roberts RJ & Cheng X. 1998. Structures of HhaI

methyltransferase complexed with substrates containing mismatches at

the target base. Nature structural biology. 5:872-877.

Okano M, Bell DW, Haber DA & Li E. 1999. DNA methyltransferases Dnmt3a

and Dnmt3b are essential for de novo methylation and mammalian

development. Cell. 99:247-257.

Okano M, Xie S & Li E. 1998. Cloning and characterization of a family of novel

mammalian DNA (cytosine-5) methyltransferases. Nature genetics.

19:219-220.

Pogribny IP & Beland FA. 2009. DNA hypomethylation in the origin and

pathogenesis of human diseases. Cell Mol Life Sci. 66(14):2249-61.

Reither S, Li F, Gowher H & Jeltsch A. 2003. Catalytic mechanism of DNA-

(cytosine- C5)-methyltransferases revisited: covalent intermediate

formation is not essential for methyl group transfer by the murine Dnmt3a

enzyme. Journal of molecular biology. 329:675-684.

Richardson B. 2003. Impact of aging on DNA methylation. Ageing research

reviews. 2:245-261.

Robertson KD, Ait-Si-Ali S, Yokochi T, Wade PA, Jones PL & Wolffe AP. 2000.

DNMT1 forms a complex with Rb, E2F1 and HDAC1 and represses

transcription from E2F-responsive promoters. Nature genetics. 25:338-

342.

Robertson KD, Uzvolgyi E, Liang G, Talmadge C, Sumegi J, Gonzales FA &

Jones PA. 1999. The human DNA methyltransferases (DNMTs) 1, 3a and

3b: coordinate mRNA expression in normal tissues and overexpression in

tumors. Nucleic acids research. 27:2291-2298.

Robertson KD. 2005. DNA methylation and human disease. Nature reviews

6:597-610.

Rodenhiser D & Mann M. 2006. Epigenetics and human disease: translating

basic biology into clinical applications. CMAJ. 174(3):341-8.

Sakatani T, Kaneda A, Iacobuzio-Donahue CA, Carter MG, de Boom Witzel S,

Okano H, Ko MS, Ohlsson R, Longo DL, Feinberg AP. 2005 Loss of

imprinting of Igf2 alters intestinal maturation and tumorigenesis in mice.

Science. 307(5717):1976-8.

Suetake I, Shinozaki F, Miyagawa J, Takeshima H & Tajima S. 2004. DNMT3L

stimulates the DNA methylation activity of Dnmt3a and Dnmt3b through a

direct interaction. The Journal of biological chemistry. 279:27816-27823.

Sweatt JD. 2009. Experience-dependent epigenetic modifications in the central

nervous system. Biol. Psychiatry. 65:191–197.

Szilak L, Finta C, Patthy A, Venetianer P & Kiss A. 1994. Self-methylation of

BspRI DNA methyltransferase. Nucleic Acids Res. 22:2876-2881.

Szilak L, Finta C, Patthy A, Venetianer P & Kiss A. 1995. Self-methylation of the

M.BspRI methyltransferase. Gene. 157:105.

Toyota M, Kopecky KJ, Toyota MO, Jair KW, Willman CL & Issa JP. 2001

Methylation profiling in acute myeloid leukemia. Blood. 97(9):2823-9.

Turner NJ. 2009. Directed evolution drives the next generation of biocatalysts.

Nat Chem Biol. 5(8):567-73.

Watson JD, Baker TA, Bell SP, Gann A, Levine M & Losick R. 2008. Molecular

Biology of the Gene, Cold Spring Harbor Laboratory Press.

Webster KE, O'Bryan MK, Fletcher S, Crewther PE, Aapola U, Craig J, Harrison

DK, Aung H, Phutikanit N, Lyle R, Meachem SJ, Antonarakis SE, de

Kretser DM, Hedger MP, Peterson P, Carroll BJ & Scott HS. 2005. Meiotic

and epigenetic defects in Dnmt3L-knockout mouse spermatogenesis. Proc

Natl Acad Sci U S A. 102(11):4068-73.

Wu JC, Santi DV. 1987. Kinetic and catalytic mechanism of HhaI

methyltransferase. The Journal of biological chemistry. 262:4778-4786.

Xu GL, Bestor TH, Bourc'his D, Hsieh CL, Tommerup N, Bugge M, Hulten M, Qu

X, Russo JJ & Viegas-Pequignot E. 1999. Chromosome instability and

immunodeficiency syndrome caused by mutations in a DNA

methyltransferase gene. Nature. 402:187-191.

Yang X, Smith SL, Tian XC, Lewin HA, Renard JP & Wakayama T. 2007.

Nuclear reprogramming of cloned embryos and its implications for

therapeutic cloning. Nat Genet. 39(3):295-302.

Yen ZC, Meyer IM, Karalic S & Brown CJ. 2007. A cross-species comparison of

X-chromosome inactivation in Eutheria. Genomics. 90(4):453-63.

Yokochi T & Robertson KD. 2002. Preferential methylation of unmethylated DNA

by Mammalian de novo DNA methyltransferase Dnmt3a. J Biol Chem.

277(14):11735-45.

Zhang Y, Jurkowska R, Soeroes S, Rajavelu A, Dhayalan A, Bock I, Rathert P,

Brandt O, Reinhardt R, Fischle W & Jeltsch A. 2010. Chromatin

methylation activity of Dnmt3a and Dnmt3a/3L is guided by interaction of

the ADD domain with the histone H3 tail. Nucleic Acids Res. 38(13):4246-

53.

Supplement (1)

Approaches to Enzyme and Substrate Design of the Murine Dnmt3a

DNA Methyltransferase

Jurkowska RZ*, Siddique AN*, Jurkowski TP, Jeltsch A. Approaches to Enzyme and Substrate Design of the Murine Dnmt3a DNA Methyltransferase.

ChembioChem. 2011, in press

* Both authors contributed equally to this work doi: 10.1002/cbic.201000673.

DOI: 10.1002/cbic.201000673 Approaches to Enzyme and Substrate Design of the Murine Dnmt3a DNA Methyltransferase Renata Z. Jurkowska, Abu Nasar Siddique, Tomasz P. Jurkowski, and Albert Jeltsch*[a]

Dnmt3a-C, the catalytic domain of the Dnmt3a DNA-(cytosine- of individual clones. Based on the enrichment and depletion of C5)-methyltransferase, is active in an isolated form but, like the certain bases in the positions flanking >1300 methylated CpG full-length Dnmt3a, shows only weak DNA methylation activity. sites, we were able to define a sequence-preference profile for To improve this activity by directed evolution, we set up a se- Dnmt3a-C from the À6 to the +6 position of the flanking selection system in which Dnmt3a-C methylated its own expres- quence. This revealed preferences for T over a purine at posi- sion plasmid in E. coli, and protected it from cleavage by meth- tion À2, A over G at À1, a pyrimidine at +1, and A and T over ylation-sensitive restriction enzymes. However, despite screen- Gat+3. We designed one “good” substrate optimized for ing about 400 clones that were selected in three rounds from methylation and one “bad” substrate designed not to be effi- a random mutagenesis library of 60000 clones, we were not ciently methylated, and showed that the optimized substrate able to isolate a variant with improved activity, most likely be- is methylated >20 times more rapidly at its central CpG site. cause of a background of uncleaved plasmids and plasmids The optimized Dnmt3a-C substrate can be applied in enzymat- that had lost the restriction sites. To improve the catalytic ic high-throughput assays with Dnmt3a-C (e.g., for inhibitor activity of Dnmt3a-C by optimization of the sequence of the screening), because the increased activity provides an im- DNA substrate, we analyzed its flanking-sequence preference proved dynamic range and better signal/noise ratio. in detail by bisulfite DNA-methylation analysis and sequencing

Introduction

DNA methylation occurs at the N6 position of adenine resi- and methylation of CpG sites in the promoter regions of genes dues, and at the N4 and C5 positions of cytosine residues; only usually leads to a reduction of gene expression. This regulatory the last type is observed in higher eukaryotes, including mam- function contributes to major biological processes such as the mals.[1,2] These methylated bases are natural components of epigenetic regulation of gene expression, genomic imprinting, DNA, distinct from the large variety of chemically modified x-chromosome inactivation, protection against selfish genomic bases formed during DNA damage. Physiological DNA-methyl- elements, and maintenance of genome stability.[5–6] ation positions the methyl groups in the major groove of the Here, we applied enzyme and substrate design on the C-ter- DNA, where they do not interfere with the Watson–Crick base minal catalytic domain of the Dnmt3a enzyme (Dnmt3a-C), pairing properties of the nucleotides, but the presence of a which contains all the characteristic DNA-(cytosine C5)-MTase methyl group can be detected by proteins that interact with amino acid motifs and is active in an isolated form.[7–9] How- the DNA, and have biological effects. In prokaryotes, DNA ever, we and other groups have found that Dnmt3a and methylation has important biological roles, such as the distinc- Dnmt3b, expressed in E. coli or insect cells, only show relatively tion between self and nonself DNA in restriction-modification weak catalytic activity in vitro (with turnover numbers in the systems, the marking of the parental DNA strand in post-repli- range of hours).[7,10–12] Most likely, the reason for this low cata- cative mismatch-repair, the control of DNA replication and the lytic activity is that the active center of the enzyme is not in an cell cycle, and the regulation of gene expression.[1,3] In addi- optimal conformation to support catalysis. This conclusion is tion, adenine-N6 methylation has been shown to be involved further supported by the finding that the activity of Dnmt3a in the pathogenicity of different bacteria,[4] including Bordetella can be stimulated by Dnmt3L, both in vivo and in vitro.[13–18] pertussis, Escherichia coli, Salmonella thyphimurium and Neisse- The aim of this project was to increase the catalytic activity of ria meningitidis. the C-terminal domain of Dnmt3a in the absence of activators In mammals, DNA methylation mainly occurs at CpG dinu- like Dnmt3L, by designing an optimized substrate, and by cleotide sequences. However, only certain CpG sites are methylated, and this results in the establishment of a tissue- and cell- [a] R. Z. Jurkowska,+ A. N. Siddique,+ T. P. Jurkowski, Prof. Dr. A. Jeltsch type specific pattern of methylation (modified and unmodified Biochemistry Laboratory, School of Engineering and Science CpG sites). The DNA methylation pattern in the organism is set Jacobs University Bremen Campus Ring 1, 28759 Bremen (Germany) during early embryogenesis by the Dnmt3a and Dnmt3b Fax: (+ 49)421-200-3249 methyltransferases (MTases) with the help of Dnmt3L.[5,6] Meth- E-mail: [email protected] ylation of CpG sites in mammals is involved in gene regulation, [+] These authors contributed equally to the work.

ChemBioChem 0000, 00, 1 – 7 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim &1& These are not the final page numbers! ÞÞ A. Jeltsch et al. applying directed evolution on the enzyme. DNA methyltrans- PaeR7I (CTCGAG), SmaI (CCCGGG), AatII (GACGTC), AvaI ferases are particularly suited to in vitro evolution approaches (CYCGRG), BsaHI (GRCGYC), NruI (TCGCGA), AclI (AACGTT), because these enzymes modify DNA at specific sites. Thus, BsaAI (YACGTR), BsiEI (CGRYCG), NaeI (GCCGGC), NarI after MTase expression in E. coli, its expression plasmid carries (GGCGCC), HaeII (RGCGCY), HhaI (GCGC), HpaII (CCGG) and a blueprint of the enzyme’s activity and specificity that can be NotI (GCGGCCGC). All these enzymes are inhibited by cytosine detected in the DNA coding for the protein. This provides a methylation of the CG sites within their recognition sequences. unique coupling of genotype (DNA sequence of the MTase The digested plasmids were analyzed by gel electrophoresis to gene) and phenotype (enzymatic properties of the MTase pro- determine the extent of DNA cleavage. Initial results for most tein). This facilitates directed-evolution experiments enormous- enzymes indicated very low levels of plasmid protection (data ly, because screening can be performed with a library of DNA not shown); this was not surprising, as Dnmt3-C is known to molecules, each of which codes for a different MTase variant have only weak enzymatic activity in bacteria.[10] As the NotI, and, consequently, might exhibit a different methylation pat- MluI and NruI recognition sites contain two CpG sites, these tern. If a DNA molecule that carries the desired methylation enzymes were expected to be the most sensitive to CpG meth- profile is identified, the DNA MTase variant that produced this ylation, and we focused on them. As shown in Figure 1B, pattern can be retrieved. So far, this approach had been suc- some in vivo methylation activity of Dnmt3a-C in E. coli was cessfully applied to prokaryotic DNA MTases, which in general detectable at NotI and NruI sites, as indicated by weak protec- have high catalytic activity.[19–23] Here we have attempted to tion of the plasmid against cleavage; this was not observed produce Dnmt3a-C variants with improved catalytic activity by after expression of a catalytically inactive Dnmt3a-C variant directed evolution. In addition, we investigated the flanking- that was used as control in these experiments. In the case of sequence preferences of Dnmt3a-C for methylation of DNA. MluI the protected plasmid co-migrated with the cleavage The contributions of bases at each of the different flanking po- product of pLacI (which is also present), so the protected plas- sitions were used to design two specific substrates: one opti- mid could not be observed. However, after additional digestion mized for methylation, and the other designed to be a bad of pLacI, no protection of the Dnmt3a plasmid at the MluI site substrate (not efficiently methylated). was observed. Typically, the same experiment performed with

Results and Discussion Setting up a system for directed evolution of Dnmt3a-C In order to set up a selection system based on the catalytic activity of Dnmt3a-C in E. coli,we expressed the murine Dnmt3a C- terminal part (608–908) in GT116 (DE3) E. coli cells, which do not contain endogenous DNA-(cytosine C5)-MTases. This strain was generated by introducing the T7 RNA polymerase gene (DE3) into the GT116 strain (Invivogen) by using the lDE3 lysogenization kit (Novagen). Expression of Dnmt3a-C in the lysogenized strain was confirmed after induction of the enzyme for 2 h by disruption of the cells, and west- Figure 1. Directed evolution of Dnmt3a towards higher catalytic activity. A) Induction of Dnmt3a-C in GT116 (DE3) ern blotting of the soluble frac- E. coli cells with 1 mm IPTG for 1, 2, and 3 h. Crude lysate on SDS polyacrylamide gels stained with Coomassie G- tion (Figure 1A). To set up a se- 250 (left panel), and His6-tagged Dnmt3a-C detected by western blot with anti-His6 antibody (right panel). The lection system for Dnmt3a-C ac- purified Dnmt3a-C runs at an apparent molecular weight of 37 kDa. BI: before induction; MW: molecular weight tivity, the Dnmt3a-C expression marker (15, 25, 35, 40, 55, 70, 100, 130, and 170 kDa; Fermentas); Protein: Dnmt3a-C purified from BL21 (DE3) pLysS cells as a reference. B) Partial methylation of the Dnmt3a expression plasmid after induction of Dnmt3a-C. plasmid was isolated after pro- NotI, MluI and NruI restriction sites contain two CpG sites, methylation of which blocks digestion. The Dnmt3a-C tein expression. Plasmids were expression plasmid has one cleavage site for NotI and MluI and two NruI sites. Weak methylation of the plasmid then digested with different re- led to a partial protection of the plasmid against NotI and NruI cleavage as indicated by the residual amount of striction enzymes: BspDI supercoiled plasmid (sc) after NotI and linear plasmid (lin) after NruI digestion (indicated with the arrows). Expres- sion of the catalytically inactive Dnmt3a-C S48E variant did not lead to plasmid protection. The plasmid prepara- (ATCGAT), ClaI (ATCGAT), FspI tions also contain the pLacI plasmid, which has one MluI site and does not contain any NotI or NruI sites. UD: (TGCGCA), MluI (ACGCGT), undigested.

&2& www.chembiochem.org 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 0000, 00, 1 – 7 ÝÝ These are not the final page numbers! Enzyme and Substrate Design of Dnmt3a a prokaryotic DNA MTase would lead to complete methylation smaller, and the likelihood of finding positive hits in the back- of all MTase target sites. This underscores the low level of ground was much higher. This interpretation agrees with the activity of Dnmt3a-C at the NotI and NruI sites. observation that most successful enzyme-design projects now To select for Dnmt3a-C variants with higher catalytic activity, combine sophisticated computing and modeling with screen- we carried out whole-gene randomization by error-prone PCR ing and selection methods.[26–29] of the Dnmt3a-C gene, and transformed the library into E. coli. The successful introduction of mutations was confirmed by Design of an optimized substrate for Dnmt3a sequencing of the Dnmt3a-C genes from ten randomly picked colonies: we observed 42 mutations in the 5570 bp that were In the light of the problems with the directed evolution ap- sequenced (0.76 mutations per 100 bp). The distribution of the proach, we turned to the design of a substrate that would mutations is shown in Table 1. After Dnmt3a-C expression, a exhibit improved catalytic turnover of Dnmt3a-C. To this end, we determined an optimized flanking-sequence context of the CpG site for methylation by Dnmt3a-C, and used it to design Table 1. Mutational spectrum observed in the Dnmt3a-C random libra- improved DNA substrates. We have shown previously that ries. Number of cases observed in 5500 nucleotides that were sequenced changes in the flanking sequence can influence the catalytic after subcloning. activity of Dnmt3a significantly.[9,30] Mutation Number of cases Mutation Number of cases To identify an optimized flanking sequence for Dnmt3a, we GtoA 5 TtoG 0 methylated two long DNA substrates, one from l-phage GtoT 7 TtoA 7 (520 bp, 40 CpGs) and the other from the CpG island upstream GtoC 2 TtoC 4 of the human SUHW1 gene (420 bp, 56 CpGs). The purified AtoG 3 CtoG 0 PCR products were methylated in vitro with 2.5 mm Dnmt3a-C AtoT 8 CtoA 4 AtoC 2 CtoT 0 and the DNA methylation patterns were investigated by bisulfite sequencing of 119 individual clones (5640 CpG sites). A total of 1354 methylation events were identified at CpG sites. Next, we collected the flanking bases at each methylated CpG pool of mutated plasmids was extracted, cleaved with NotI, site and computed the occurrence of each nucleotide at each and the superhelical DNA (i.e., protected against cleavage) was flanking position from À6to+6. Expected numbers of occur- isolated from agarose gels. It was expected that this plasmid rences were calculated for both substrates on the basis of the pool would be enriched with Dnmt3a-C variants that showed nucleotide composition of the substrates and the number of increased catalytic activity. However, after screening about 400 clones sequenced in both experiments. After summing ob- colonies that were obtained from four repetitions of up to served and expected numbers for both substrates, the ob- three cycles of selection, we were not able to retrieve such served/expected ratios were calculated (Figure 2A). The signifi- candidates. Unfortunately, after a couple of rounds of muta- cance of the overrepresentation or depletion of individual genesis and selection, all the surviving clones were found to bases at each position was calculated by exact binomial test- have mutations within the restriction-enzyme recognition sites, ing (Figure 2A). We observed strong and highly significant de- despite the fact that the cleavage site was outside the muta- viations from randomness at many sites, including a preference genized region. This indicated that the selection system was for T over purines at position À2, a preference for A over G at working in principle, because the rare plasmids that had ac- À1, a preference for a pyrimidine at +1 and a preference for A quired a mutation within the restriction site were enriched and and T over G at +3. were selected. Our results agree with, and extend previous studies. In an in- Directed evolution is a powerful approach for enzyme itial study of the flanking-sequence preferences of Dnmt3a, Lin design.[24,25] For example, we were recently able to change the et al. identified the same pattern of preferences at the À2 and recognition sequence of EcoDam by following a procedure +1 sites as we describe here.[31] Later, Handa et al. used in vitro very similar to the one applied here.[22] The failure to improve methylation of oligonucleotides differing at the À1and+1 the catalytic activity of Dnmt3a-C illustrates one critical param- sites, and derived a similar pattern of preferences at the +1 eter in directed evolution experiments: the ratio of improved site.[30] The À1 preferences could not be properly resolved in variants to background. When our experiment was carried out that work, because of the use of unmethylated substrates that with wild-type Dnmt3a-C, about 0.1–1 % of plasmids were not could be methylated on both strands. Recently, a flanking-se- cleaved by the restriction enzyme, because of their protection quence preferences study of Dnmt3a was conducted by exam- by methylation. If more-active variants were being generated ining the methylation of episomal DNA in a human cell line.[32] at a much lower frequency, the pool of the isolated plasmids In that work, a preference for T and C at À2 and +2 sites, and would still be dominated by the wild-type enzyme genes, such an aversion for A at À2 were also observed. However, the that even after several cycles no enrichment for improved other highly significant over- and under-representations ob- variants could be expected. In the example of the successful served here were not detected, presumably because in the in development of EcoDam specificity (above), mutagenesis was vivo setting the preferences of the Dnmt1 enzyme for main- limited to a few small loops of the protein that were selected taining the methylation overlap with the Dnmt3a specificity. In on a rational basis. Therefore, the mutational space was much addition, for fully methylated sites, the initial methylation of

ChemBioChem 0000, 00, 1 – 7 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chembiochem.org &3& These are not the final page numbers! ÞÞ A. Jeltsch et al.

Figure 2. Design of an optimized Dnmt3a substrate. A) Flanking-sequence preferences of Dnmt3a-C. Observed/expected ratios of occurrences of bases flanking the 1354 methylated CpG sites. Significantly over- or under-represented bases are shaded green and red, respectively. B) Left Pane: Methylation rates for the designed “good” (SG) and “bad” (SB) substrates, each carrying either a hemimethylated central CpG site or a fully methylated CpG site (reference). Relative methylation is the difference between hemimethylated and fully methylated substrates. Right pane: The methylation rate of the target CpG in the good substrate relative to the that of the bad substrate (rate= 1); the error bar represents the deviation from two independent analyses.

one DNA strand by Dnmt3a might be followed by Dnmt1- Using two substrates for the experimental analysis of the mediated methylation of the hemimethylated site. Hence, it is methylation at one specific cytosine residue was an important not possible to identify the sequence context of the initial improvement in the experimental design; this technique has Dnmt3a-mediated methylation event with confidence. not hitherto been applied to the analysis of Dnmt3a flanking- On the basis of our results for the flanking-sequence prefer- sequence preferences. This point is illustrated by the fact that ences of Dnmt3a, we designed one good substrate (SG: GAA the overall rates of methylation of good, and bad substrates GCT GGA CAG TAC GTC AAG AGA GTG CAA) and one bad sub- correlated poorly with the methylation at the central CpG, be- strate (SB: GAA GCT GGT ATA GGC GAA GGT CGA GTG CAA) by cause the number and flanking sequence of additional cyto- always using the best or worst nucleotide at each flanking sine residues varied between the two substrates. We conclude position, respectively. The substrates were synthesized with that we have successfully designed an optimized DNA sub- the lower strand pre-methylated and their methylation was strate for methylation with recombinantly expressed Dnmt3a- analyzed by measuring the transfer of radioactively labeled C. It remains to be seen whether the N-terminal part of the methyl groups from S-adenosyl-l-methionine (AdoMet) to the enzyme (or possible post-translational modification of Dnmt3a DNA (Figure 2B). However, Dnmt3a is also known to methylate in the mammalian cell) might modulate its flanking-sequence cytosine residues in a non-CpG context.[10] Hence, in order to preferences. The substrate developed here has already been determine the methylation of the target CpG cytosine, we successfully applied in high-throughput screening campaigns used a second pair of control substrates that were methylated that searched for Dnmt3a inhibitors.[33–34] In such assays high on both strands at the target CpG site (fully methylated). The and robust enzyme activity is of advantage and, therefore, methylation of the target cytosine in the upper strand at the using an optimized substrate led to an increased dynamic CpG site in each substrate was then given by the difference of range of the assay, and reduced scatter. the methylation of hemimethylated and fully methylated forms (Figure 2B). The results of the methylation reactions showed that the bad substrate is almost resistant to methylation at the Experimental Section central CpG site. In contrast, the central site is the most impor- Random mutagenesis and library creation: Whole-gene randomi- tant methylation site of the good substrate. On the basis of zation of the Dnmt3a-C gene was performed by using error-prone the weak residual methylation of the bad substrate, we esti- PCR (epPCR) that employed biased nucleotide composition, high mated the preference for methylation of the optimized sub- Mg2+ concentration, and the addition of Mn2+ ions.[35] As the mu- strate, and found that it was modified 21.7 (Æ0.5) times faster tational load depends on the number of cycles of the epPCR and than the bad substrate in two independent experiments. the amount of template used for the amplification reaction, we

&4& www.chembiochem.org 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 0000, 00, 1 – 7 ÝÝ These are not the final page numbers! Enzyme and Substrate Design of Dnmt3a prepared reactions with increasing numbers of epPCR cycles and 56 CpGs). Purified PCR product (50–100 nm) was incubated with decreasing amounts of DNA template (1 pg, 45 cycles; 5 pg, Dnmt3a-C (2.5 mm, 3 h). The methylation reaction was stopped by

40 cycles; 10 pg, 40 cycles; 50 pg, 30 cycles; and 100 pg: 30 cycles). freezing in liquid N2 and the product was digested with proteinase The epPCR reaction mixture (50 mL) contained Tris/HCl (10 mm, K (New England Biolabs). Afterwards the DNA was treated with m) m) m) [37] pH8.3), KCl (50 m , MgCl2 (7 m , MnCl2 (0.5 m , dATP and sodium bisulfate essentially as described, and cloned into a dGTP (0.2 mm each), dCTP and dTTP (1 mm each), oligonucleotide Topo-TA vector (Invitrogen). A total of 119 individual clones were primers (25 pmol each), template DNA (1–100 pg), betaine (0.5m) sequenced to determine their methylation pattern. and Taq polymerase (5 U, NEB). The forward primer was RM_ Dnmt3a-C kinetics using designed DNA oligonucleotide substrates: BT7Pro_for (CAG CAA CCG CAC CTG TGG CG) and the reverse was The following oligonucleotide substrates were used to measure RM_BT7Ter_rev (GCT TCC TTT CGG GCT TTG TTA GCA GC). epPCR the Dnmt3a-C catalytic activity (the target site is shown in bold product was obtained after 30–45 cycles (958C, 3 min; 30–45 and underlined; mC: 5-methylcytosine): [958C, 30 s; 678C, 30 s; 728C, 5 min]; 728C 12 min). The epPCR was re-amplified by normal nested PCR with epPCR as the template SG hemimet: 5’-GAA GCT GGA CAG TACGTC AAG AGA GTG CAA-3’, (50 ng), primers (25 pmol each), dNTPs (0.2 mm), ThermoPol buffer 5’-Bt-TTG CAC TCT CTT GAmCGTA CTG TCC AGC TTC-3’ (NEB, 1), and Taq DNA polymerase (2.5 U) in a 50 mL reaction SG fully met: 5’-GAA GCT GGA CAG TAmCGTC AAG AGA GTG CAA- volume. The forward primer was D3a_NdeI_for (GCG CGG CAG 3’,5’-Bt-TTG CAC TCT CTT GAmCGTA CTG TCC AGC TTC-3’ CCA TAT GAA CCA TGA CC) and the reverse was D3a_HindIII_rev (GTG CGG CCG CAA GCT TCA CAC AAG C). PCR conditions were: SB hemimet: 5’-GAA GCT GGT ATA GGCGAA GGT CGA GTG CAA-3’, 958C, 3 min; 30[958C, 30 s; 658C, 30 s; 728C 2 min]; 728C, 5’-TTG CAC TCG ACC TTmCGCC TAT ACC AGC TTC-3’ 10 min. SB fully met: 5’-GAA GCT GGT ATA GGmCGAA GGT CGA GTG CAA- The product of the nested PCR was cloned into a pET28a+ vector 3’,5’-TTG CAC TCG ACC TTmCGCC TAT ACC AGC TTC-3’ at the NdeI and HindIII sites, and transformed into XL-1 Blue MRF’ Oligonucleotides (Thermo Fisher) were obtained in HPLC-purified electrocompetent cells that were plated on LB-agar plates supple- form. Concentrations of oligonucleotides were determined spectro- mented with tetracycline and kanamycin. For selection, all colonies scopically by using e260 nm values provided by the supplier. To pre- grown on a plate were pooled, and plasmid DNA was extracted by pare double-stranded substrates, equal amounts of both strands using the NucleoSpin plasmid preparation kit (Macherey–Nagel). were mixed in 1methylation buffer comprising HEPES (20 mm, For screening, colonies were isolated, and DNA was prepared from pH 7.2), EDTA (1 mm), KCl (50 mm), and bovine serum albumin individual clones. (BSA, 25 mgmLÀ1), heated to 908C, and slowly cooled to ambient Methylation-sensitive restriction enzyme-based selection assay: temperature. Oligodeoxynucleotides were stored at À208C. Expression of the wild-type and mutagenized Dnmt3a-C was per- Time courses of DNA methylation were measured by the incorpo- formed in GT116 (DE3) pLacI cells. For this, wild-type Dnmt3a-C, ration of titrated methyl groups from labeled S-[methyl-3H]-adeno- S48E (its catalytically inactive variant) and the Dnmt3a-C random li- syl-methionine (Perkin–Elmer) into biotinylated oligonucleotides by brary were transformed into electro-competent GT116 (DE3) pLacI using the avidin-biotin methylation kinetic assay as described.[38,39] cells and grown on LB-agar plates supplemented with kanamycin The reactions were carried out in methylation buffer (as above) at and chloramphenicol overnight at 378C. For wild-type and S48E 378C, with substrate DNA (1 mm), AdoMet (0.76 mm), and Dnmt3a- Dnmt3a-C constructs, single colonies were used for expression of C (2.5 mm). To determine the initial slope, the data were fitted by the protein; whereas for the library all the colonies were pooled, linear regression of the initial part of the reaction progress curves. resuspended in LB medium with appropriate antibiotics, and used to inoculate 1 L culture. The cultures were grown to OD600 =0.5 and protein expression was induced with IPTG (1 mm), the cells Acknowledgements were grown (2 h), and harvested. Cell pellets were washed with 1 STE buffer (100 mm NaCl, 10 mm Tris/HCl pH 8.0, 1 mm EDTA) and plasmid DNA was isolated by using a medium scale DNA prepara- This work was supported by DFG (JE 252-5, JE 252-6) and DAAD. tion kit (Invitrogen).

For the restriction-protection analysis, DNA isolated from induced Keywords: directed evolution · DNA methylation · DNA GT116 (DE3) pLacI cells transformed with wild-type, S48E or recognition · enzyme specificity · protein design randomized library Dnmt3a-C plasmids, was digested with a CpG methylation-sensitive restriction enzyme. All restriction digests [1] A. Jeltsch, ChemBioChem 2002, 3, 274. were performed under conditions recommended by the supplier [2] R. J. Jurkowska, A. Jeltsch, The Chemical Biology of Nucleic Acids (Ed.: G. of the restriction enzyme. Typically 1–10 mg of plasmid DNA was di- Mayer), Wiley-VCH, Weinheim 2010. [3] D. Wion, J. Casadesﬄs, Nat. Rev. Microbiol. 2006, 4, 183. gested for 2–8 h in 10–100 mL reaction volume with 5–20 units of [4] G. Heusipp, S. Fälker, M. A. Schmidt, Int. J. Med. Microbiol. 2007, 297,1. restriction enzyme. The digested DNA was further purified and [5] R. J. Klose, A. P. Bird, Trends Biochem. Sci. 2006, 31, 89. treated with l DNA exonuclease (NEB) for 1 h in manufacturer-rec- [6] R. Z. Jurkowska, T. P. Jurkowski, A. Jeltsch, ChemBioChem 2011, 12, 206. ommended buffer again, gel purified, and transformed into XL-1 [7] H. Gowher, A. Jeltsch, J. Biol. Chem. 2002, 277, 20409. BLUE MRF’ electrocompetent cells. [8] S. Reither, F. Li, H. Gowher, A. Jeltsch, J. Mol. Biol. 2003, 329, 675. [9] H. Gowher, P. Loutchanwoot, O. Vorobjeva, V. Handa, R. Z. Jurkowska, Protein expression, purification, and DNA methylation kinetics: T. P. Jurkowski, A. Jeltsch, J. Mol. Biol. 2006, 357, 928. Wild-type Dnmt3a catalytic domain and its catalytically inactive [10] H. Gowher, A. Jeltsch, J. Mol. Biol. 2001, 309, 1201. S48E variant were expressed and purified as described.[13,36] [11] M. Okano, S. Xie, E. Li, Nat. Genet. 1998, 19, 219. [12] T. Yokochi, K. D. Robertson, J. Biol. Chem. 2002, 277, 11735. Dnmt3a-C activity on long DNA substrates: Two DNA fragments [13] H. Gowher, K. Liebert, A. Hermann, G. Xu, A. Jeltsch, J. Biol. Chem. 2005, were used, one from l-phage (520 bp, 40 CpGs), and the other 280, 13341. from the CpG island upstream of the human SUHW1 gene (420 bp, [14] K. Hata, M. Okano, H. Lei, E. Li, Development 2002, 129, 1983.

ChemBioChem 0000, 00, 1 – 7 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chembiochem.org &5& These are not the final page numbers! ÞÞ A. Jeltsch et al.

[15] F. Chedin, M. R. Lieber, C. L. Hsieh, Proc. Natl. Acad. Sci. USA 2002, 99, [29] M. V. Golynskiy, B. Seelig, Trends Biotechnol. 2010, 28, 340. 16916. [30] V. Handa, A. Jeltsch, J. Mol. Biol. 2005, 348, 1103. [16] Z. X. Chen, J. R. Mann, C. L. Hsieh, A. D. Riggs, F. ChØdin, J. Cell. Biochem. [31] I. G. Lin, L. Han, A. Taghva, L. E. O’Brien, C. L. Hsieh, Mol. Cell. Biol. 2002, 2005, 95, 902. 22, 704. [17] I. Suetake, F. Shinozaki, J. Miyagawa, H. Takeshima, S. Tajima, J. Biol. [32] B. L. Wienholz, M. S. Kareta, A. H. Moarefi, C. A. Gordon, P. A. Ginno, F. Chem. 2004, 279, 27816. ChØdin, PLoS Genet. 2010, 6, e1001106. [18] M. S. Kareta, Z. M. Botello, J. J. Ennis, C. Chou, F. ChØdin, J. Biol. Chem. [33] P. B. Arimondo, D. Guianvarc’h, A. Ceccaldi, C. SØnamaud-Beaufort, N. 2006, 281, 25893. Gagey, D. Dauzonne, A. Jeltsch, R. J. Jurkowska, V. Dumontet, Flavones [19] A. Jeltsch, T. Sobotta, A. Pingoud, Protein Eng. 1996, 9, 413. and flavanones derivates as DNA methyltransferases inhibitors, 2009, [20] D. S. Tawfik, A. D. Griffiths, Nat. Biotechnol. 1998, 16, 652. European patent applicaton: EPA09305840. [21] E. Timµr, G. Groma, A. Kiss, P. Venetianer, Nucleic Acids Res. 2004, 32, [34] A. Ceccaldi, A. Rajavelu, C. Champion, C. Rampon, R. J. Jurkowska, G. 3898. Jankevicius, C. SØnamaud-Beaufort, L. Ponger, N. Gagey, H. Dali Ali, J. [22] S. Chahar, H. Elsawy, S. Ragozin, A. Jeltsch, J. Mol. Biol. 2010, 395, 79. Tost, S. Vriz, S. Ros, D. Dauzonne, A. Jeltsch, D. Guianvarc’h, P. B. Arimon- [23] R. Gerasimaite, G. Vilkaitis, S. Klimasauskas, Nucleic Acids Res. 2009, 37, do, 2011, unpublished results. 7332. [35] R. C. Cadwell, G. F. Joyce, PCR Methods Appl. 1992, 2, 28. [24] N. J. Turner, Nat. Chem. Biol. 2009, 5, 567. [36] R. Z. Jurkowska, N. Anspach, C. Urbanke, D. Jia, R. Reinhardt, W. Nellen, [25] J. A. Gerlt, P. C. Babbitt, Curr. Opin. Chem. Biol. 2009, 13, 10. X. Cheng, A. Jeltsch, Nucleic Acids Res. 2008, 36, 6656. [26] D. Rçthlisberger, O. Khersonsky, A. M. Wollacott, L. Jiang, J. DeChancie, [37] Y. Zhang, C. Rohde, S. Tierling, H. Stamerjohanns, R. Reinhardt, J. Walter, J. Betker, J. L. Gallaher, E. A. Althoff, A. Zanghellini, O. Dym, S. Albeck, A. Jeltsch, Methods Mol. Biol. 2009, 507, 177. K. N. Houk, D. S. Tawfik, D. Baker, Nature 2008, 453, 190. [38] M. Roth, A. Jeltsch, Biol. Chem. 2000, 381, 269. [27] P. M. Murphy, J. M. Bolduc, J. L. Gallaher, B. L. Stoddard, D. Baker, Proc. [39] K. Liebert, A. Jeltsch, Methods Mol. Biol. 2008, 418, 149. Natl. Acad. Sci. USA 2009, 106, 9215. [28] J. B. Siegel, A. Zanghellini, H. M. Lovick, G. Kiss, A. R. Lambert, J. L. St Clair, J. L. Gallaher, D. Hilvert, M. H. Gelb, B. L. Stoddard, K. N. Houk, F. E. Received: November 9, 2010 Michael, D. Baker, Science 2010, 329, 309. Published online on && &&, 0000

&6& www.chembiochem.org 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 0000, 00, 1 – 7 ÝÝ These are not the final page numbers! FULL PAPERS

Better methylation: We describe a R. Z. Jurkowska, A. N. Siddique, system for the directed evolution of T. P. Jurkowski, A. Jeltsch* Dnmt3a to improve the enzymatic activ- && – && ity of this DNA methyltransferase. In addition, we determined the flanking-se- Approaches to Enzyme and Substrate quence preferences of Dnmt3a, and de- Design of the Murine Dnmt3a DNA signed a DNA substrate optimized for Methyltransferase methylation by Dnmt3a. The substrate can be applied in enzyme high-throughput assays with this enzyme.

ChemBioChem 0000, 00, 1 – 7 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chembiochem.org &7& These are not the final page numbers! ÞÞ

Supplement (2)

Auto-methylation of the mouse DNA-(cytosine C5)-methyltransferase

Dnmt3a at its active site cysteine residue

Siddique AN, Jurkowska RZ, Jurkowski TP, Jeltsch A. Auto-methylation of the mouse DNA-(cytosine C5)-methyltransferase Dnmt3a at its active site cysteine residue.

FEBS J. 2011, in press doi: 10.1111/j.1742-4658.2011.08121.x.

Auto-methylation of the mouse DNA-(cytosine C5)-methyltransferase Dnmt3a at its active site cysteine residue

Abu Nasar Siddique, Renata Z. Jurkowska, Tomasz P. Jurkowski* and Albert Jeltsch*

Biochemistry Laboratory, School of Engineering and Science, Jacobs University Bremen, Campus Ring 1, 28759 Bremen, Germany

*Corresponding authors Dr. Tomasz P. Jurkowski, email: [email protected] Prof. Albert Jeltsch, email: [email protected]

Running title: Auto-methylation of the Dnmt3a methyltransferase Abbreviations: AdoMet, S-adenosyl-L-methionine; AdoHcy, S-adenosyl-L-homocysteine, DNA methyltransferase, MTase Enzymes: Dnmt3a is a DNA-(cytosine C5)-methyltransferase, EC 2.1.1.37 Keywords: DNA methyltransferase, auto-methylation, enzyme mechanism, enzyme regulation, protein methylation Subdivision: Enzymes and catalysis Received Date : 16-Dec-2010 Revised Date : 28-Mar-2011 Accepted Date : 06-Apr-2011 Article type : Original Article

This is an Accepted Article that has been peer-reviewed and approved for publication in the FEBS Journal, but has yet to undergo copy-editing and proof correction. Please cite this article as an “Accepted Article”; doi: 10.1111/j.1742-4658.2011.08121.x

Summary The Dnmt3a DNA methyltransferase is responsible for establishing DNA methylation patterns during mammalian development. We show here, that the mouse Dnmt3a DNA methyltransferase is able to transfer the methyl group from S-adenosyl-L-methionine (AdoMet) to a cysteine residue in its catalytic center. This reaction is irreversible and relatively slow. The yield of auto-methylation is increased by addition of Dnmt3L, which functions as a stimulator of Dnmt3a and enhances its AdoMet binding. Auto-methylation was observed in binary Dnmt3a AdoMet complexes. In the presence of CpG containing dsDNA, which is the natural substrate for Dnmt3a, the transfer of the methyl group from AdoMet to the flipped target base was preferred and auto-methylation was not detected. Therefore, this reaction might constitute a regulatory mechanism which could inactivate unused DNA MTases in the cell, or it could simply be an aberrant side reaction caused by the high methyl group transfer potential of AdoMet.

Structured digital abstract:

Dnmt3a methylates Dnmt3a by methyltransferase assay (View interaction) Dnmt3a and DNMT3L methylate by methyltransferase assay (View interaction)

Introduction Methylation of biomolecules including proteins, DNA, RNA and small molecules plays important and diverse roles in biology [1]. For these reactions, S-adenosyl-L-methionine (AdoMet) is by far the most commonly used methyl group donor. It contains the methyl group bound to a positively charged sulfonium center, consequently the methyl group is highly activated towards a nucleophilic attack and AdoMet is a highly reactive compound with high methylation capacity. Overall, following ATP, AdoMet is the second most commonly used coenzyme in nature [2] and it has been estimated that about 3% of all enzymes listed in the EC nomenclature represent AdoMet dependent methyltransferases [3]. Methylation substrates range in size from small compounds like catechol to biopolymers like proteins, RNA and DNA, the target atoms for methylation can be carbon, oxygen, nitrogen, sulfur or even halides [4].

DNA methylation is common to almost all living organisms. In bacteria, three kinds of methylated bases are present: 5-methylcytosine, 4-methylcytosine and 6-methyladenine, whereas only 5-methylcytosine is found in higher eukaryotes [5]. In mammals, DNA methylation is restricted mostly to CpG dinucleotides. The cell and tissue specific DNA methylation pattern is set early during embryonic development by the action of the Dnmt3a and Dnmt3b de novo DNA methyltransferases (MTases). Once established, the methylation pattern is further maintained during each DNA replication and cell division by the maintenance MTase Dnmt1 [6- 7]. DNA methylation contributes to major biological processes, like epigenetic regulation of gene expression, genomic imprinting, x-chromosome inactivation, protection against selfish genomic elements and maintenance of genomic stability [6-7].

The Dnmt3a MTase comprises a large N-terminal regulatory domain and a C-terminal catalytic domain, which is active in an isolated form [8]. The catalytic domain shares a well conserved structure with all DNA MTases, called “AdoMet-dependent MTase fold”, which consists of a mixed seven-stranded β-sheet, formed by six parallel and the 7th anti-parallel β strands, inserted in between the 5th and 6th β strand. This central β-sheet is sandwiched between six alpha-helices [9-10]. Because the target base is buried in the DNA helix and not readily accessible for catalysis, DNA MTases flip out the target base and insert it in the hydrophobic pocket in the active center [11]. The catalytic mechanism used by Dnmt3a is characteristic for the DNA-(cytosine C5)-MTases and it is mainly involved in activation of the substrate by increasing its nucleophilicity [5]. For that purpose, DNA-(cytosine C5)-MTases use a catalytic cysteine residue to perform a nucleophilic attack on the sixth position of the cytosine, which leads to the formation of a covalent bond between the enzyme and the substrate base. The formation of the cysteine-cytosine bond increases the negative charge density at the C5 atom of the cytosine, which then attacks the methyl group bound to AdoMet. Base flipping and the nucleophilic attack of the cysteine are facilitated by a contact of a conserved glutamate residue to the exocyclic amino group and the N3 ring nitrogen atom. In addition, a conserved

arginine residue plays a role in base flipping and catalysis. Exchange of any of these residues led to a reduction or complete loss of the catalytic activity of Dnmt3a [12-13].

Results

Detection of auto-methylation of Dnmt3a-C The methylation of lysine and arginine residues of histones is an important post- translation modification involved in regulation of gene expression and chromatin biology [6, 14- 15]. However, recently the regulatory function of lysine methylation of non-histone proteins has moved into the focus of research [16-17]. To look into the possible regulation of Dnmt3a-C by lysine methylation, we investigated the potential lysine methylation of the Dnmt3a-C enzyme by several mammalian protein lysine methyltransferases. To this end, purified Dnmt3a- C was incubated with different protein lysine methyltransferases in the presence of S-[methyl- 3H]-adenosyl-L-methionine (AdoMet) with radioactively labeled methyl group in order to detect the transfer of radioactivity to Dnmt3a-C. Afterwards samples were analyzed by SDS-PAGE electrophoresis and autoradiography. However, after incubation of Dnmt3a-C with radioactively labeled AdoMet for longer periods of time, we detected the transfer of radioactivity to the Dnmt3a-C protein even without addition of a protein methyltransferase (Fig. 1). This modification was resistant to heat (95 °C for 5 minutes in the presence of 2% SDS and 5 mM DTT), therefore, it seemed to be of covalent nature and it most probably resulted from an intrinsic auto-methylation activity of the enzyme. A similar observation was also made with full-length Dnmt3a2 (Fig. 1C), which is the predominant isoform of Dnmt3a in embryonic stem cells and embryonal carcinoma cells [18].

Since we suspected this covalent labeling of Dnmt3a-C would inhibit the enzyme and it could have a regulatory role, we decided to study this phenomenon in more detail. Literature searches uncovered similar observations already made for some bacterial MTases, including the DNA-(cytosine C5)-MTases M.BspRI [19-20] and Dcm [21] and the DNA-(adenine N6)-MTase

M.EcoPI [22]. For M.BspRI, it was suggested that the methyl group can be directly transferred from the AdoMet to a cysteine residue of the protein, leading to the formation of a chemically stable S-methylcysteine and resulting in the inactivation of the enzyme [19].

Kinetics and irreversible nature of the auto-methylation To follow the time course of auto-methylation, we have incubated Dnmt3a-C with radioactively labeled AdoMet and removed aliquots from the reaction mixture at different time points. Reactions were stopped by the addition of SDS to a final concentration of 2 mM, followed by heat denaturation of the protein and SDS-PAGE electrophoresis. The extent of radioactivity bound to the protein was visualized by autoradiography and quantified by densitometry. As shown in Fig. 1A and B, the radioactive signal is increasing slowly over the course of hours with a roughly linear increase for the first 4 hours of the reaction. Fitting of the reaction progress curve to a single exponential rate equation gave an estimate of 0.1 h-1 for the auto-methylation rate constant, indicating that auto-methylation is a slow process in contrast to the non-covalent AdoMet binding or exchange which happens within minutes [12]. Furthermore, non-covalently bound AdoMet will not co-migrate with Dnmt3a-C in denaturing gel electrophoresis. Hence, we conclude that indeed an auto-methylation of Dnmt3a-C occurs.

To confirm the irreversible nature of the labeling of Dnmt3a-C, reactions were quenched with unlabeled AdoMet and AdoMet analogues. An initial auto-methylation reaction was performed for 1 hour, allowing the formation of some auto-methylated Dnmt3a-C, then large excess of unlabeled AdoMet or AdoHcy was added to the reaction and samples were taken in 1 hour intervals and analyzed by autoradiography. As expected, the addition of either unlabeled AdoMet or AdoHcy inhibited the further incorporation of radioactivity into the Dnmt3a-C protein (Fig. 2A). However, already incorporated radioactivity remained, indicating that the modification is stable and irreversible under in vitro conditions.

Auto-methylation occurs at the catalytic cysteine Taking into account the catalytic mechanism of Dnmt3a-C, it seemed very likely that the catalytic cysteine residue was the methyl group acceptor because it lies in close proximity to the methyl group of AdoMet and is the most reactive residue in the catalytic center of the enzyme. To test whether the catalytic cysteine is the target for auto-methylation, we have purified the alanine exchange mutant C120A of Dnmt3a-C [13] and incubated it with radioactively labeled AdoMet. As expected, the C120A mutant Dnmt3a did not get labeled (Fig. 2B); strongly suggesting that the active site cysteine is the target of modification.

To confirm that automethylation occurs at Cys120, Dnmt3a-C was incubated with unlabelled AdoMet and subjected to tryptic digestion and MALDI-TOF mass spectrometric analysis. As shown in Fig. 2C, the peak corresponding to the unmethlyated peptide containing Cys120 as well as the peak corresponding to the methylated peptide were clearly detectable. This peptide does not contain another amino acid residue that could function as nucelophile (Suppl. Fig. S1). The methylated peak was not detected with a control sample that was not pre- incubated with AdoMet (Fig. 2C). As an additional control, the C120A variant was incubated with AdoMet and subjected to mass spectrometric analysis. In this experiment the peptide containing the C120A mutation was detectable, but neither a peak corresponding to the methylated C120A peptide, nor a peak corresponding to the methylated C120 peptide was observed (Suppl. Fig. S2).

Our identification of the active site cysteine residue as the target for automethylation parallels literature findings with other enzymes. In the case of M.BspRI, Szilak and colleagues have identified two cysteine residues which were the targets for auto-methylation. One of them (C156) is the catalytic cysteine in M.BspRI, the other one (C181) is not conserved among DNA-(cytosine-C5)-MTases [19]. In the case of Dcm only the catalytic cysteine residue was found to get modified [21].

Extent of auto-methylation In order to estimate the fraction of Dnmt3a-C which gets self-methylated, we have compared the radioactivity signal generated by Dnmt3a-C after 16 h incubation under our standard reaction conditions with the signal coming from histone H3.1 monomethylated at lysine 4 by recombinant SET7/9 histone lysine MTase [23-25]. As shown in Fig. 3, the autoradiography signal of Dnmt3a-C protein is faint in comparison to the signal of histone H3. Taking into consideration the relative strength of the autoradiography signal and the total protein amounts of Dnmt3a-C and H3.1 loaded on the gel, we have estimated that about 2.6 % of the Dnmt3a-C got modified during this 16 h incubation. It is interesting that the extent of automethylation observed in the mass spectrometric analysis (Fig. 2C) was much higher than the extent of methylation observed in Fig. 3. Although this observation needs to be interpreted carefully, since mass spectroscopy is not a fully quantitative method, the higher methylation may be related to the fact, that the concentration of AdoMet was much higher in the mass spectrometric experiment (1 mM in the assay) than in the methylation with radioactively labeled AdoMet (0.76 µM in the assay).

Effect of DNA and Dnmt3L on auto-methylation of Dnmt3a It is known that Dnmt3L, an activator of Dnmt3a and Dnmt3b, stimulates the DNA methylation reaction catalyzed by these enzymes [26]. As shown in Fig. 4, larger amounts of radioactivity were transferred to the Dnmt3a-C protein after adding Dnmt3L-C to the auto- methylation reaction when compared to the reaction mixture with Dnmt3a alone. This result can be explained, because Dnmt3L stabilizes the conformation of the active site loop of Dnmt3a-C and it increases AdoMet binding [9, 26], which in turn will lead to increased formation of self-methylated Dnmt3a-C.

To test the effect of DNA on the auto-methylation reaction of Dnmt3a-C, we have added a 20 bp double-stranded DNA containing a single CG target site and followed the incorporation of radioactivity from AdoMet into DNA and Dnmt3a-C. As can be seen in figure 4, addition of a double stranded DNA substrate abolished the tritium incorporation into the MTase but at the

same time the DNA got efficiently methylated. This result illustrates that after binding both substrates (DNA and AdoMet) the enzyme has a high specificity for the transfer of the methyl group to the DNA and it efficiently avoids auto-methylation.

Discussion

We show here that the mammalian Dnmt3a enzyme undergoes auto-methylation in vitro at its catalytic cysteine by transferring the methyl group from its natural cofactor AdoMet to the cysteine residue. Analogous reactions were already observed for the bacterial DNA- (adenine N6)-MTase M.EcoPI [22] and the two DNA-(cytosine C5)-MTases Dcm [21] and M.BsuRI [19-20], but not for a mammalian DNA MTase. This observation highlights one interesting detail in the catalytic mechanism of DNA-(cytosine C5)-MTases. On one hand, these enzymes employ AdoMet as the donor for methyl groups, which is a highly activated coenzyme with very large methyl group transfer potential. On the other hand, they harbor a cysteine residue in their active centers that is activated towards performing a nucleophilic attack. This cysteine residue could easily react with AdoMet because the ΔG° for the transfer of the methyl group from AdoMet to cysteine is in the order of -70 kJ/mol [2, 4]. Therefore, it is essential for the enzyme that a close approximation of these two groups is avoided to prevent automethylatoin and inactivation of the enzyme. Indeed, in the Dnmt3a-C structure with AdoHcy, the sulfhydryl and AdoHcy sulfur atoms are separated by 7.66 Å, which would correspond to a distance of about 6 Å between the sulfhydryl sulfur and the methyl group of AdoMet, if AdoMet would replace AdoHcy without conformational change (Fig. 5). This suggests that a conformational change of about 3 Å has to occur before automethylation can happen, which may explain, why the process of automethylation is slow. We show here that the conformation of Dnmt3a prevents auto-methylation efficiently but not entirely, similarly as observed with the bacterial enzymes mentioned above. However, if DNA is bound, no auto- methylation is happening, suggesting that the reaction occurs in binary Dnmt3a-C AdoMet complexes, but in ternary complexes the transfer of the methyl group to the flipped target base is much preferred.

It is unclear whether this slow auto-methylation of Dnmt3a could have a biological function in cells. Since the methylcysteine is chemically stable, auto-methylation would inactivate the enzyme by causing steric constraints and interference with the reaction mechanism of DNA-(cytosine C5)-MTases. It is possible that in cells, Dnmt3a which is in an idle state may lose its activity via auto-methylation, thereby protecting the genome against aberrant methylation. In this respect it is important to note that we observed much higher levels of automethylation after incubation of the Dnmt3a-C at higher concentration of AdoMet. The intracellular concentrations of AdoMet have been reported in the range of 50-250 µM [27- 30], which is much higher than the AdoMet concentration used here in the radioactive methylation (0.76 µM) and may allow for more efficient automethylation in the cell. There also might be additional factors in the cells, which could stimulate Dnmt3a for auto-methylation and therefore inactivation. On the other hand, the auto-methylation of Dnmt3a and other MTases simply may be a side reaction caused by the high methyl group transfer potential of AdoMet.

Experimental procedures

The His6-tagged full length Dnmt3a2 and the catalytic domain of mouse Dnmt3a wild- type and catalytically inactive C120A mutant (corresponding to C706A in full-length Dnmt3a) and the His6-tagged fusion of the C-terminal part of human DNMT3L were expressed and purified as described [26, 31]. The human SET7/9 protein lysine methyltransferase (PKMT) which among other sites monomethylates histone H3 at Lys4 [24-25] was purified as described [32].

Auto-methylation reactions mixtures contained 2-5 µM Dnmt3a-C protein, 0.76 µM [methyl-3H]-AdoMet (PerkinElmer) in methylation buffer (20 mM HEPES pH 7.2, 1 mM EDTA, 50 mM KCl) in a 20 µl reaction volume. Reactions were incubated at room temperature for various time intervals, stopped by addition of 20 µl of Laemmli sample buffer (130 mM TRIS HCl pH=6.8, 20% glycerol, 4% SDS, 10 mM DTT, 0.02% bromophenol blue) and heating to 95°C for 5 min. Afterwards, the samples were analyzed on a 15% SDS-PAGE gel and either stained with colloidal Coomassie (Dyballa and Metzger 2009) or fixed with 10% methanol/10% acetic acid,

immersed in Amplify solution (Amersham) for 1 hour at room temperature with shaking, dried on a 3 mm Whatman paper and exposed to a X-ray film for 3-7 days. Signal intensities were analyzed by densitometry (AIDA v4, raytest Germany). Automethylation reactions with unlabelled AdoMet were carried out in the same buffer but in the presence of 1 mM AdoMet (Sigma-Aldrich) for 16 hours. Unlabelled AdoMet was dissolved in 10 mM sulfuric acid, stored in aliquots at -20°C and used only once after thawing.

Quenching reactions were prepared essentially like auto-methylation reaction, however the auto-methylation reactions were incubated for 1 h allowing the formation of some initial auto-methylated species and then quenched by the addition of either 1 mM non-radioactive AdoMet (Sigma-Aldrich), 1 mM AdoHcy (Sigma-Aldrich) or 20 µM of double stranded CG-DNA (GAA GCT GGG ACT TCC GGA GGA GAG TGC AA). The samples were collected at various time points and analyzed as described above. To study the effect of Dnmt3L on the auto-methylation reaction of Dnmt3a-C, auto-methylation reactions were supplemented with 7 µM recombinant Dnmt3L and analyzed like described above.

To calibrate the extent of auto-methylation of Dnmt3a, human recombinant histone H3.1 (NEB) was methylated with 1.22 µM recombinant SET7/9. H3 methylation was performed in methylation buffer for SET7/9 (50 mM TRIS/HCl pH 9.0, 5 mM MgCl2, 4 mM DTT) using 0.76 µM 3H-AdoMet. The reaction was incubated for 12 h to run to completion and different amounts of the methylated H3 methylation reaction mixtures were separated on a 15% SDS- PAGE gel together with the Dnmt3a-C after overnight incubation with labeled AdoMet. The amount of radioactivity incorporated in the protein bands was determined from scanned autoradiography pictures by densitometry.

Acknowledgements

This work was supported by DFG (JE 252-6) and DAAD.

References

1. Cheng X & Roberts RJ (2001) AdoMet-dependent methylation, DNA methyltransferases and base flipping. Nucleic Acids Res 29, 3784-3795 2. Cantoni GL (1975) Biological methylation: selected aspects. Annu Rev Biochem 44, 435- 451 3. Kagan RM & Clarke S (1994) Widespread occurrence of three sequence motifs in diverse S-adenosylmethionine-dependent methyltransferases suggests a common structure for these enzymes. Arch Biochem Biophys 310, 417-427 4. Schubert HL, Blumenthal RM & Cheng X (2003) Many paths to methyltransfer: a chronicle of convergence. Trends Biochem Sci 28, 329-335 5. Jeltsch A (2002) Beyond Watson and Crick: DNA methylation and molecular enzymology of DNA methyltransferases. Chembiochem 3, 274-293 6. Bonasio R, Tu S & Reinberg D (2010) Molecular signals of epigenetic states. Science 330, 612-616 7. Jurkowska RZ, Jurkowski TP & Jeltsch A (2011) Structure and Function of Mammalian DNA Methyltransferases. ChemBioChem, in press 8. Gowher H & Jeltsch A (2002) Molecular enzymology of the catalytic domains of the Dnmt3a and Dnmt3b DNA methyltransferases. J Biol Chem 277, 20409-20414 9. Jia D, Jurkowska RZ, Zhang X, Jeltsch A & Cheng X (2007) Structure of Dnmt3a bound to Dnmt3L suggests a model for de novo DNA methylation. Nature 449, 248-251 10. Cheng X & Blumenthal RM (2008) Mammalian DNA methyltransferases: a structural perspective. Structure 16, 341-350 11. Roberts RJ & Cheng X (1998) Base flipping. Annu Rev Biochem 67, 181-198 12. Gowher H, Loutchanwoot P, Vorobjeva O, Handa V, Jurkowska RZ, Jurkowski TP & Jeltsch A (2006) Mutational analysis of the catalytic domain of the murine Dnmt3a DNA- (cytosine C5)-methyltransferase. J Mol Biol 357, 928-941 13. Reither S, Li F, Gowher H & Jeltsch A (2003) Catalytic mechanism of DNA-(cytosine- C5)-methyltransferases revisited: covalent intermediate formation is not essential for methyl group transfer by the murine Dnmt3a enzyme. J Mol Biol 329, 675-684 14. Berger SL (2007) The complex language of chromatin regulation during transcription. Nature 447, 407-412 15. Goldberg AD, Allis CD & Bernstein E (2007) Epigenetics: a landscape takes shape. Cell 128, 635-638 16. Huang J & Berger SL (2008) The emerging field of dynamic lysine methylation of non- histone proteins. Curr Opin Genet Dev 18, 152-158

17. Rathert P, Dhayalan A, Ma H & Jeltsch A (2008) Specificity of protein lysine methyltransferases and methods for detection of lysine methylation of non-histone proteins. Mol Biosyst 4, 1186-1190 18. Chen T, Ueda Y, Xie S & Li E (2002) A novel Dnmt3a isoform produced from an alternative promoter localizes to euchromatin and its expression correlates with active de novo methylation. J Biol Chem 277, 38746-38754 19. Szilak L, Finta C, Patthy A, Venetianer P & Kiss A (1994) Self-methylation of BspRI DNA-methyltransferase. Nucleic Acids Res 22, 2876-2881 20. Szilak L, Finta C, Patthy A, Venetianer P & Kiss A (1995) Self-methylation of the M.BspRI methyltransferase. Gene 157, 105 21. Hanck T, Schmidt S & Fritz HJ (1993) Sequence-specific and mechanism-based crosslinking of Dcm DNA cytosine-C5 methyltransferase of E. coli K-12 to synthetic oligonucleotides containing 5-fluoro-2'-deoxycytidine. Nucleic Acids Res 21, 303-309 22. Hornby DP, Muller M & Bickle TA (1987) High level expression of the EcoP1 modification methylase gene and characterisation of the gene product. Gene 54, 239-245 23. Wang H, Cao R, Xia L, Erdjument-Bromage H, Borchers C, Tempst P & Zhang Y (2001) Purification and functional characterization of a histone H3-lysine 4-specific methyltransferase. Mol Cell 8, 1207-1217 24. Xiao B, Jing C, Wilson JR, Walker PA, Vasisht N, Kelly G, Howell S, Taylor IA, Blackburn GM & Gamblin SJ (2003) Structure and catalytic mechanism of the human histone methyltransferase SET7/9. Nature 421, 652-656 25. Wilson JR, Jing C, Walker PA, Martin SR, Howell SA, Blackburn GM, Gamblin SJ & Xiao B (2002) Crystal structure and functional analysis of the histone methyltransferase SET7/9. Cell 111, 105-115 26. Gowher H, Liebert K, Hermann A, Xu G & Jeltsch A (2005) Mechanism of stimulation of catalytic activity of Dnmt3A and Dnmt3B DNA-(cytosine-C5)-methyltransferases by Dnmt3L. J Biol Chem 280, 13341-13348 27. Chiang EP, Wang YC, Chen WW & Tang FY (2009) Effects of insulin and glucose on cellular metabolic fluxes in homocysteine transsulfuration, remethylation, S- adenosylmethionine synthesis, and global deoxyribonucleic acid methylation. J Clin Endocrinol Metab 94, 1017-1025 28. Hermes M, Osswald H, Riehle R, Piesch C & Kloor D (2008) S-Adenosylhomocysteine hydrolase overexpression in HEK-293 cells: effect on intracellular adenosine levels, cell viability, and DNA methylation. Cell Physiol Biochem 22, 223-236 29. Castro R, Rivera I, Martins C, Struys EA, Jansen EE, Clode N, Graca LM, Blom HJ, Jakobs C & de Almeida IT (2005) Intracellular S-adenosylhomocysteine increased levels are associated with DNA hypomethylation in HUVEC. J Mol Med 83, 831-836 30. Hoffman DR, Marion DW, Cornatzer WE & Duerre JA (1980) S-Adenosylmethionine and S-adenosylhomocystein metabolism in isolated rat liver. Effects of L-methionine, L- homocystein, and adenosine. J Biol Chem 255, 10822-10827

31. Jurkowska RZ, Anspach N, Urbanke C, Jia D, Reinhardt R, Nellen W, Cheng X & Jeltsch A (2008) Formation of nucleoprotein filaments by mammalian DNA methyltransferase Dnmt3a in complex with regulator Dnmt3L. Nucleic Acids Res 36, 6656-6663 32. Dhayalan A, Kudithipudi S, Rathert P & Jeltsch A (2011) Specificity analysis based identification of new methylation targets of the SET7/9 protein lysine methyltransferase. Chemistry & Biology, in press

Supporting information

Supplemental Fig. S1: Sequence of the tryptic peptide containing Cys120 and its position in the Dnmt3a structure.

Supplemental Fig. S2: Absence of automethylation with the Dnmt3a-C C120A variant.

Figures and figure legends

Figure 1: Auto-methylation of the Dnmt3-C DNA methyltransferase.

A) Dnmt3-C protein was incubated with radioactively labeled AdoMet in the standard reaction buffer for indicated time periods. Reactions were stopped and samples split into equal parts and both run on 15% SDS polyacrylamid gels. The first gel was fixed, sensitized, dried, and then exposed to X-ray films for 5 days (labeled Autoradiography). The second gel was stained with colloidal Coomassie to serve as loading control (labeled Coomassie). Dnmt3a-C migrates in the gel with an apparent mass of ~37 kDa.

B) Quantification of the autoradiography signal coming from the Dnmt3a-C protein band. The exposed and developed X-ray films were scanned and the strength of radioactivity signal was estimated using densitometry and the intensity values [a.u.] were plotted as a function of time and fitted to a single exponential rate equation.

C) Automethylation of full length Dnmt3a. The methylation was performed for 14 hours. Nine, six or three µg of the protein were loaded on polyacrylamide gels (lanes 1-3) and subjected to autoradiography or Coomassie staining.

Figure 2: Dnmt3a auto-methylation reaction is irreversible, dependent on AdoMet and occurs at the catalytic cysteine residue.

A) An auto-methylation reaction was incubated for 1 hour to allow for creation of auto- methylated species. Then, the reaction was quenched by addition of 1000-fold molar excess of either unlabeled AdoMet or AdoHcy. For reference the auto-methylation reaction was continued without addition of a quencher. Aliquots from the reactions were taken at 2h, 3h, 5h and o/n (12-14 h) after the addition of quencher, run on a 15% SDS polyacrylamide gel and the amount of incorporated 3H-methylgroups was checked by autoradiography.

B) Purified C120A and wild-type Dnmt3a-C were incubated with 3H-AdoMet for 0, 2, 4 and o/n (12-14 h) in the reaction buffer. Two aliquots from the reaction were taken at each time point and run separately on two 15% SDS-PAGE gels, from which one was stained with colloidal Coomassie G-250 and served as loading control; and the other was used for autoradiography.

C) Mass spectroscopic analysis of automethylation of Dnmt3a-C with and without incubation with unlabelled AdoMet (1 mM). The tryptic fragment containing the active site Cys120 has a

mass of 2934.3 Da (theoretical mass 2934.4 Da). After incubation with AdoMet an additional peak appears at 2948.3 Da corresponding to 2934.3 Da plus the mass of a methyl group (14 Da).

Figure 3: Quantification of the extent of auto-methylation.

A) Comparison of the radioactivity signal from auto-methylated Dnmt3a-C (labeled D3a-C) after over-night (14 hours) incubation with 3H-AdoMet in the standard reaction conditions with signal from different amounts of SET7/9 methylated histone H3.1 (NEB). The total protein amounts loaded on the gel are indicated.

B) The band intensities from the autoradiography picture were extracted by densitometry and background normalized.

Figure 4: Effect of Dnmt3L and DNA on the auto-methylation reaction.

Comparison of efficiencies of auto-methylation reactions carried out in the presence of 7 µM Dnmt3L C-terminal domain or 20 µM CG containing dsDNA with the standard auto-methylation reaction. Whereas Dnmt3L increases the level of auto-methylation of Dnmt3a-C, the methylation of the dsDNA substrate is efficiently competing with the Dnmt3a-C auto- methylation. Note, that oligonucleotiodes are not fully denatured in SDS gels and run as a mixture of single stranded and double stranded form. In addition, binding of Dnmt3a-C to the DNA causes the appearance of an additional retarded oligonucleotide band.

Figure 5. Positioning of AdoHcy and the catalytic cysteine in the active center of Dnmt3a-C.

In the crystal structure of Dnmt3a/Dnmt3L complex with AdoHcy [9] (PDB:2QRV). AdoHcy is shown in orange, Dnmt3a is colored by atom type. The distance between the sulfhydryl atom of the catalytic cysteine side chain and sulfur atom of AdoHcy (7.66 Å) is indicated.

Protein design of the mammalian DNA methyltransferase Dnmt3a

by Abu Nasar Siddique

A thesis submitted in partial fulfillment of the requirements for the degree of

Doctor of Philosophy in Molecular Biotechnology

Supplement (3)

Confidential document

Only available to reviewers

School of Engineering and Science Jacobs University Bremen