Methyltransferase Ash1, histone methylation and their impact on Polycomb repression

(1)

Methyltransferase Ash1, histone methylation and their impact on Polycomb repression

Eshagh Dorafshan Esfahani

Department of Molecular Biology Umeå 2018

(2)

This work is protected by the Swedish Copyright Legislation (Act 1960:729) Dissertation for PhD

ISBN: 978-91-7601-932-0 ISSN: 0346-6612

New series 1986

Cover: The northern light monument at the Umeå University campus Electronic version available at: http://umu.diva-portal.org/

Printed by: Print & Media, Umeå University Umeå, Sweden 2017

(3)

To my parents and my beloved wife

(4)

i

List of manuscripts

I. Kahn, T. G., E. Dorafshan, D. Schultheis, A. Zare, P.

Stenberg, I. Reim, V. Pirrotta, and Y. B. Schwartz. 2016.

'Interdependence of PRC1 and PRC2 for recruitment to Polycomb Response Elements', Nucleic Acids Res, 44: 10132- II. Dorafshan, E., T. G. Kahn, and Y. B. Schwartz. 2017. 49.

'Hierarchical recruitment of Polycomb complexes revisited', Nucleus, 8: 496-505.

III. Eshagh Dorafshan, Tatyana G. Kahn, Alexander Glotov, Mikhail Savitsky, Matthias Walther, Gunter Reuter and Yuri B. Schwartz. 'Does Ash1 counteract Polycomb repression by methylating H3K36?'

IV. Eshagh Dorafshan, Tatyana G. Kahn, Alexander Glotov,

Mikhail Savitsky and Yuri B. Schwartz. 'Functional

dissection of Drosophila Ash1 domains'

(6)

iii

Abstract

Antagonistic interactions between Polycomb Group (PcG) and Trithorax Group (TrxG) proteins orchestrate the expression of key developmental genes. Distinct maternally loaded repressors establish the silenced state of these genes in cells where they should not be expressed and later PcG proteins sense whether a target gene is inactive and maintain the repression throughout multiple cell divisions. PcG proteins are targeted to genes by DNA elements called Polycomb Response Elements (PREs). The proteins form two major classes of complexes, namely Polycomb Repressive Complex 1 (PRC1) and Polycomb Repressive Complex 2 (PRC2). Mechanistic details of Polycomb repression are not fully understood, however, tri-methylation of Lysine 27 of histone H3 (H3K27me3) is essential for this process. Using Drosophila cell lines deficient for either PRC1 or PRC2, I investigated the role of H3K27 methylation and the interdependence of PRC1 complexes for their recruitment to PREs.

My results indicate that recruitment of PcG complexes to PREs proceed via multiple pathways and that H3K27 methylation is not needed for their targeting. However, the methylation is required to stabilize interactions of PRE-anchored PcG complexes with surrounding chromatin.

TrxG proteins prevent erroneous repression of Polycomb target genes where these genes need to be expressed. Ash1 is a TrxG protein which binds Polycomb target genes when they are transcriptionally active. It contains a SET domain which methylates Lysine 36 of histone H3 (H3K36). In vitro, histone H3 methylated at K36 is a poor substrate for H3K27 methylation by PRC2. This prompted a model where Ash1 counteracts Polycomb repression through H3K36 methylation. However, this model was never tested in vivo and does not consider several experimental observations.

First, in the ash1 mutant flies the bulk H3K36me2/H3K36me3

levels remain unchanged. Second, in Drosophila, there are two

other H3K36-specific histone methyltransferases, NSD and Set2,

which should be capable to inhibit PRC2. Third, Ash1 contains

(7)

iv

multiple evolutionary conserved domains whose roles have not been investigated. Therefore, I asked whether H3K36 methylation is critical for Ash1 to counteract Polycomb repression in vivo and whether NSD and Set2 proteins contribute to this process. I used flies lacking endogenous histone genes and complemented them with transgenic histone genes where Lysine 36 is replaced by Arginine. In these animals, I assayed erroneous repression of HOX genes as a readout for erroneous Polycomb repression. I used the same readout in the NSD or Set2 mutant flies. I also asked if other conserved domains of Ash1 are essential for its function. In addition to SET domain, Ash1 contains three AT hook motifs as well as BAH and PHD domains. I genetically complemented ash1 loss of function animals with transgenic Ash1 variants, in each, one domain of Ash1 is deleted. I found that Ash1 is the only H3K36-specific histone methyltransferase which counteracts Polycomb repression in Drosophila. My findings suggest that the model, where Ash1 counteracts PcG repression by inhibiting PRC2 via methylation of H3K36, has to be revised. I also showed that, in vivo, Ash1 acts as a multimer and requires SET, BAH and PHD domains to counteract Polycomb repression.

This work led to two main conclusions. First, trimethylation of H3K27 is not essential for targeting PcG proteins to PREs but acts afterwards to stabilize their interaction with the chromatin of the neighboring genes. Second, while SET domain is essential for Ash1 to oppose Polycomb repression, methylation of H3K36 does not play a central role in the process.

(8)

1

Introduction

Development of all metazoans requires elaborate transcriptional regulation of key developmental coordinators, such as HOX genes, that give identity to different body segments (Gellon and McGinnis 1998). Transcriptional patterns of such developmental coordinators, in turn, are established by maternal or early zygotic transcription factors at initial stages of embryogenesis. In later stages of development, however, these early transcription factors are absent but transcriptional pattern established by them needs to endure in a cell lineage-specific manner. This is achieved by antagonistic actions of ubiquitous Polycomb Group (PcG) and Trithorax Group (TrxG) proteins. Polycomb Group proteins sense the repressed state of a target gene and maintain this repression throughout multiple cell divisions even in the absence of the original early repressors. Trithorax Group proteins, on the other hand, ensure that the genes that are supposed to be expressed in a specific cell type, are not erroneously repressed by the action of Polycomb Group proteins. This, portraits the classical view on how antagonistic functions of Polycomb and Trithorax Group proteins orchestrate developmental programs in a cell/tissue-specific manner.

Analysis of genome-wide binding of Polycomb Group proteins to the chromatin reveals at least two classes of sites where Polycomb Group proteins bind to. The first class comprises specific DNA elements called Polycomb Response Elements (PRE) where the strongest bindings of Polycomb Group proteins are detected (Schwartz and Pirrotta 2007). PREs are bound by all Polycomb Group complexes and regulate transcriptional pattern of a subset of genes mainly involved in development. This represent the classical function of Polycomb Group proteins in epigenetic repression of their target genes. The second class surprisingly includes regions in the vicinity of most active genes (Jacob et al. 2011; Young et al. 2011;

Mousavi et al. 2012; Frangini et al. 2013; Schaaf et al. 2013;

(9)

2

Loubiere et al. 2016; Pherson et al. 2017). Binding of Polycomb Group proteins to these regions is considerably weaker than PREs and only one of the Polycomb Group complexes (PRC1) appears to bind here. Although the exact mechanisms underlying the function of Polycomb Group proteins near the active genes are not well understood, they are suggested to play a role in the transition from transcription initiation to elongation (Mousavi et al. 2012; Schaaf et al. 2013; Pherson et al. 2017).

In this thesis, I focus on the classical role of Polycomb/Trithorax Group proteins in epigenetic regulation of their target genes and do not address the second class described above. Drosophila melanogaster has been used, as the model organism, in all of the studies included in this thesis. However, since most of the proteins and complexes are evolutionary conserved in higher eukaryotes (Schwartz and Pirrotta 2013), the mammalian studies and examples are also discussed when relevant.

Polycomb Group (PcG) proteins and complexes

Polycomb Group genes were first discovered based on mutations

that led to the appearance of sex combs on the second and third legs

in mutant adult male flies, where they are normally absent (Slifer

1942; Kassis et al. 2017). Later, mutations leading to impaired HOX

gene repression were collectively called Polycomb Group, indicating

their phenotypic resemblance to mutations in Polycomb (Pc) (Gerd

1985). Pc was suggested as a general repressor of Bithorax Complex

(BX-C) genes, one of the two Drosophila HOX gene clusters (Lewis

1978). Bithorax Complex comprises Ubx, abd-A, and Abd-B genes

which have distinct expression patterns, and erroneous repression

or ectopic expression of these genes lead to visible phenotypes in

Drosophila embryos, larvae, and adult animals. Examples of such

phenotypes include transformation of embryonic abdominal

segments to more posterior fate as well as wing to haltere

transformation in adult flies. Because of these readily identifiable

phenotypes and perhaps due to their historical roles in

(10)

3

identification of Polycomb Group genes, Bithorax Complex was extensively used as a tool in genetic and transgenic studies attempting to unravel the function of Polycomb (and also Trithorax) Group proteins.

In Drosophila, Polycomb Group proteins form three major complexes, Polycomb Repressive Complex 1 (PRC1), Polycomb Repressive Complex 2 (PRC2), and Pho Repressive Complex (PhoRC).

Polycomb Repressive Complex 1

The Drosophila Polycomb Repressive Complex 1 (PRC1) core complex contains Ring1 (also known as dRING), Posterior sex comb (Psc) or its paralogue, Suppressor of zeste 2 (Su(z)2), Polycomb (Pc), and Polyhomeotic (Ph). In addition to the core components, PRC1 has a sub-stoichiometric subunit Sex combs on midleg (Scm) (Franke et al. 1992; Shao et al. 1999; Francis et al. 2001; Saurin et al. 2001; Lo et al. 2009). Mammalian RING1 and RING2 are homologues of Drosophila Ring1. Orthologues of Drosophila Psc in mammals include MEL18 (also known as PCGF2) and BMI1 (also known as PCGF4). The five chromodomain-containing proteins CBX2, CBX4, CBX6, CBX7, CBX8 are mammalian orthologues of Drosophila Pc. In mammals, SCMH1 and SCMH2 are orthologues of Drosophila Scm and PHC1, PHC2, and PHC3 are homologous to Drosophila Ph (Schwartz and Pirrotta 2013).

Tethering subunits of PRC1 to a reporter gene via DNA binding domain of an unrelated protein, represses the reporter gene, in vivo (Bunker and Kingston 1994; Muller 1995; Poux et al. 2001a; Poux et al. 2001b). While the mechanistic details of such repression is not completely clear, PRC1 or its components are reported to compact chromatin templates and inhibit nucleosome remodeling in vitro (Shao et al. 1999; Francis et al. 2001; Francis et al. 2004; King et al.

2005; Lo et al. 2009; Lo and Francis 2010).

(11)

4

Pc binds to H3K27me3 through Chromodomain and this domain is needed for Pc to bind chromatin (Messmer et al. 1992; Cao et al.

2002; Czermin et al. 2002). Drosophila Ring1, and its mammalian orthologue, have E3 ubiquitin ligase activity that mono- ubiquitylates histone H2A at Lys 118 in Drosophila or Lys 119 in mammals (Wang et al. 2004a; Gutierrez et al. 2012) and this E3- ligase activity is increased in Drosophila Ring1-Psc or murine Ring2-Bmi1 heterodimers (Buchwald et al. 2006; Lagarou et al.

2008). Ph contains a SAM domain through which it can interact with SAM domain of Scm (Peterson et al. 1997; Kim et al. 2005) and mutation of SAM domain of Scm disrupts its interaction with Ph and impairs the repression mediated by Polycomb Group proteins (Peterson et al. 2004). Although Scm was first co-purified with PRC1 components, later it was shown to also copurify with recombinant PRC2 subunits (Kang et al. 2015) and interacts with Sfmbt subunit of PhoRC (Frey et al. 2016) putting Scm at the intersection of the three protein complexes. In addition to SAM domain, Scm also has two MBT domains (Bornemann et al. 1996) which interact with mono-methylated lysine residues in histones H3 and H4 and deletion of these domains, too, leads to impaired Polycomb Group-mediated repression of HOX genes (Grimm et al.

2007).

Polycomb Repressive Complex 2

Polycomb Repressive Complex 2 (PRC2) core complex include Enhancer of zeste (E(z)), Suppressor of zeste 12 (Su(z)12), Extra sex comb (Esc) and the histone chaperon Caf1 (Ng et al. 2000; Czermin et al. 2002; Muller et al. 2002). The E(z) subunit of PRC2 has histone methyltransferase (HMTase) activity specific for Lys 27 of histone H3 (H3K27) and both Esc and Su(z)12 subunits are needed for this enzymatic activity (Czermin et al. 2002; Muller et al. 2002;

Ketel et al. 2005; Nekrasov et al. 2005). Histone methyltransferase

activity of E(z) is essential for repression of HOX genes (Muller et

al. 2002). Esc contains WD domains (Ng et al. 1997) and interacts

directly with E(z) (Jones et al. 1998; Tie et al. 1998). Su(z)12 also

(12)

5

interacts directly with E(z) and is needed for histone methyltransferase activity in vivo and in vitro, and repression of HOX genes in both mammals and Drosophila (Cao and Zhang 2004; Ketel et al. 2005; Jiao and Liu 2015; Lee et al. 2015). In mammals, EZH2, SUZ12, EED and RBAP48, the orthologues of Drosophila E(z), Su(z)12, Esc and Caf-1, respectively, also form a protein complex homologous to Drosophila PRC2 (Cao et al. 2002;

Kuzmichev et al. 2002; Cao and Zhang 2004). In addition to the four core subunits, PRC2 contains two mutually exclusive components, Jarid2 and Pcl. Pcl co-purifies with a fraction of PRC2 in Drosophila and is reported to bind Polycomb Group target genes (O'Connell et al. 2001; Tie et al. 2003; Papp and Muller 2006;

Nekrasov et al. 2007; Savla et al. 2008). Jarid2 was also co-purified with PRC2 in both mammals and Drosophila and was shown to enhance histone methyltransferase activity of PRC2 (Li et al. 2010;

Kang et al. 2015; Sanulli et al. 2015). While both Pcl and Jarid2, have been identified as components of PRC2, their incorporation in the complex seem to be mutually exclusive (Nekrasov et al. 2007;

Herz et al. 2012). Moreover, unlike Jarid2, Pcl is specifically enriched at PREs of the genes repressed by Polycomb Group proteins (Papp and Muller 2006; Nekrasov et al. 2007; Savla et al.

2008; Herz et al. 2012). Pcl is important for trimethylation of H3K27 around PREs (Nekrasov et al. 2007) and proper repression of HOX genes (Duncan 1982). Together, these observations suggest that there are two flavors of PRC2, that is Pcl-PRC2 and Jarid2- PRC2 with distinct functions.

Pho Repressive Complex

Pho Repressive Complex (PhoRC) was first identified as a

heterodimer between Sfmbt and either Pleiohomeotic (Pho) or its

closely related homologue Pleiohomeotic like (Phol) (Klymenko et

al. 2006). However, a later study showed that pull-down of Sfmbt

copurifies Rpd3, Mga, HP1b, and Nap1, in addition to Pho (Alfieri

et al. 2013). Pho and Phol are Drosophila orthologues of the

mammalian YY1 with similar DNA-binding activity (Brown et al.

(13)

6

1998; Brown et al. 2003). Both Pho and Phol contain a spacer domain which interacts with MBT domains of Sfmbt (Alfieri et al.

2013). MBT domains of Sfmbt also were shown to specifically interact with mono- and di- methylated H3K9 and H4K20 (Klymenko et al. 2006). In addition to MBT domains, Sfmbt also contains a SAM domain which can interact with SAM domain of Scm, providing a physical link between PRC1 and DNA-binding PhoRC complex (Frey et al. 2016).

Although Pho and Phol are partially redundant, corresponding mutant flies have distinct phenotypes (Brown et al. 2003). In line with this, genome-wide analysis of Pho and Phol reveals a difference in their binding preferences. While both Pho and Phol, together with Sfmbt, bind to PREs they also bind to the vicinity of Transcription Start Site (TSS) of a subset of active genes (Kahn et al. 2014). However, the strongest Pho binding sites, together with Sfmbt, are at PREs whereas Phol appears to preferentially bind near the TSS of active genes, suggesting that incorporation of Pho into PhoRC is preferred over Phol (Kahn et al. 2014). While Pho-specific DNA binding sites in PREs are required for binding and function of PhoRC in the repression of HOX genes (Fritsch et al. 1999; Busturia et al. 2001; Mishra et al. 2001; Fujioka et al. 2008), Sfmbt-mediated interaction with PRC1 is also needed to reinforce this binding (Kahn et al. 2014).

Polycomb Response Elements

Polycomb Response Elements (PREs) are DNA elements to which Polycomb and Trithorax Group proteins bind (Zink et al. 1991;

Fritsch et al. 1999; Rozovskaia et al. 1999; Tillib et al. 1999; Papp

and Muller 2006; Schwartz et al. 2006; Petruk et al. 2008). PREs

contain binding motifs for different DNA-binding proteins such as

Pho (Brown et al. 1998), GAGA factor (GAF), Pipsqueak (Psq),

Dorsal switch protein 1 (Dsp1), Grainyhead (Grh), Zeste (Z) and

Fs(1)h. However, neither of these motifs alone is sufficient to make

a PRE and combinations of these motifs vary between different

(14)

7

PREs (Kassis and Brown 2013). In addition to providing a binding platform for Polycomb Group proteins, PREs have three unique features that distinguish them from other regulatory DNA elements.

First, PREs can maintain the transcriptional pattern of a reporter gene, governed by certain enhancer elements usually regulating expression of HOX and other key developmental genes, in transgenic assays. This maintenance of repression strictly depends on the function of Polycomb Group proteins (Muller and Bienz 1991; Zink et al. 1991; Simon et al. 1993; Chan et al. 1994; Poux et al. 1996; Brown et al. 1998; Fritsch et al. 1999; Horard et al. 2000;

Americo et al. 2002; Park et al. 2012). The second unique feature of PREs is pairing sensitivity. This means that Polycomb Group mediated-repression of transgenic reporter genes harboring a PRE, is stronger if two copies of the transgenic construct is provided compared to only one copy of the transgene (Kassis et al. 1991;

Barges et al. 2000; Americo et al. 2002; Park et al. 2012; Kahn et al.

2016). The third feature is plasticity. PREs can be “switched” into Trithorax Response Elements (TRE) and maintain the derepressed state of a target gene (Cavalli and Paro 1999; Maurange and Paro 2002; Poux et al. 2002; Rank et al. 2002; Schmitt et al. 2005;

Fujioka et al. 2008; Erokhin et al. 2015). Switching PREs to TREs is triggered by transcriptional activation of the nearby target gene (Cavalli and Paro 1998; Cavalli and Paro 1999). Although this PRE to TRE transition was first reported to be both mitotically and meiotically inherited in the absence of the original trans-activator (Cavalli and Paro 1999), later studies questioned the meiotic inheritance of the TRE state (Poux et al. 2002; Erokhin et al. 2015).

Additionally, it was shown that switching a PRE to a TRE is accompanied by transcription through the PRE (Maurange and Paro 2002; Rank et al. 2002). However, Erokhin et al.

demonstrated that this transcription is not essential for the PRE-to- TRE switch. Interestingly, although such a functional switch leads to a reduction in Polycomb Group proteins binding at PRE/TRE, these proteins are not completely evicted (Schwartz et al. 2010;

Bowman et al. 2014; Erokhin et al. 2015; Elizar'ev et al. 2016).

(15)

8

Trithorax Group proteins

Trithorax Group genes were characterized in the genetic screens for mutations that either recapitulated the phenotypes of HOX genes loss of function or suppressed the phenotypes of Polycomb Group proteins mutations (Kassis et al. 2017). Although many genes were originally identified as members of trithorax Group (trxG), most of these genes such as brm, osa, mor, fs(1)h, skd, kto, vtd, ash2 and kis encode for proteins involved ATP-dependent chromatin remodeling and other general transcriptional regulatory functions.

It appears that participation of these proteins in the expression of HOX genes reflects their general role in transcriptional activation rather than specific involvement at loci regulated by Polycomb/Trithorax Group proteins (Kassis et al. 2017). Among all the reported Trithorax Group proteins, only Trithorax and Ash1 were shown to specifically counteract repression mediated by Polycomb Group proteins rather than having a general role in transcriptional activation of their target genes (Poux et al. 2002;

Klymenko and Muller 2004). In Drosophila, Ubx is repressed by Polycomb Group proteins in larval wing imaginal disc while it is derepressed, through the action of Trithorax Group proteins, in the haltere and third leg discs. Klymenko and Muller showed that in wing disc clones, mutant for Polycomb Group proteins, Ubx repression is lost. Expectedly, expression of Ubx in clones of haltere and third leg discs, mutant for either trx or ash1, is also lost. In clones doubly mutant for Polycomb Group proteins and either trx or ash1, however, Ubx is expressed in the wing, haltere and third leg discs. This demonstrates that neither Trx nor Ash1 are needed for activation of their target genes but rather are essential to counteract the repression mediated by Polycomb Group proteins (Klymenko and Muller 2004).

Trithorax

The first characterized allele of Drosophila trithorax (trx) was

originally named as lethal(3)bithorax

^variegated

, regulator of

bithorax, and eventually trithorax

¹

allele (Kassis et al. 2017).

(16)

9

Trithorax (Trx) co-localizes with Polycomb Group proteins (Chinwalla et al. 1995) and Ash1 (Rozovskaia et al. 1999) on larval polytene chromosomes and is involved in regulation of HOX genes expression (Ingham and Whittle 1980; Sato and Denell 1987; Mazo et al. 1990; Breen and Harte 1991; Breen et al. 1995). Trx contains a SET domain with histone methyltransferase activity specific to H3K4. Indeed, Trx was reported to mono- (Smith et al. 2004; Tie et al. 2014) and di-methylates H3K4 (Smith et al. 2004) and genome- wide binding profile of Trx is highly correlated with that of mono- (Tie et al. 2014; Rickels et al. 2016) and di-methylated H3K4 (Rickels et al. 2016). This histone methyltransferase activity of Trx was reported to be important to antagonize Polycomb Group- mediated repression (Tie et al. 2014).

Although Trx was first reported to co-purify with Ash1 from Drosophila embryonic nuclear extracts and their SET domain were shown to interact in Y2H (Rozovskaia et al. 1999), later it was co- purified with CBP and Sbf1 in a complex called TAC1, which acetylates core histones (Petruk et al. 2001). Recently, it was shown that mono-methyl H3K4 enhances acetylation of H3K27 and that mono-methylated H3K4 and acetylated H3K27 together with CBP and Trx are enriched at active enhancers. Moreover, CBP knockdown suppresses Trx overexpression phenotype (Tie et al.

2014) and CBP-related H3K27 acetylation is reduced in trx mutants (Tie et al. 2009) indicating functional interaction of the two proteins. However, in another study, Trx together with Ash2, Mennin-1, HCF1, Dpy30, Rbbp5, and Wds was reported to form a COMPASS-like complex ((Mohan et al. 2011). For more details on this complex and other COMPASS-like complexes in Drosophila see the section ‘Methylation of Lys 4 of histone H3’.

Drosophila Trithorax, like its mammalian orthologue MLL-1, is cleaved by Taspase-1 (Hsieh et al. 2003a; Hsieh et al. 2003b;

Capotosti et al. 2007). Although both N- and C-terminal moieties of

Trx are reported to remain associated in a complex with CBP (Tie et

al. 2014), they show distinct genome-wide binding profiles

(17)

10

(Schuettengruber et al. 2009; Schwartz et al. 2010). While both N- and C-terminal moieties of Trx bind to PREs regardless of repression state of the cognate gene, N-terminus of Trx forms a broad domain of binding over the gene body of a derepressed gene, together with Ash1 (Schwartz et al. 2010).

Ash1

ash1 was characterized based on mutations that caused homeotic transformations in thoracic and posterior abdominal segments in adult flies. These phenotypes include first and third legs to second leg and female genitalia to leg transformations (Shearn et al. 1987;

Shearn 1989), indicating the critical role of ash1 in proper expression of HOX genes (Shearn et al. 1987; Shearn 1989;

Tripoulas et al. 1994; LaJeunesse and Shearn 1995; Tripoulas et al.

1996). Ash1 is a unique Trithorax Group protein in that, unlike Trx, it only binds to the target genes if the gene is in derepressed state (Papp and Muller 2006; Schwartz et al. 2010).

Ash1 has a SET domain (Tripoulas et al. 1996) and although it was originally reported to methylate H3K4 and to a lesser extent H3K9 (Byrd and Shearn 2003; Gregory et al. 2007), later it was shown to specifically di-methylate H3K36 both in vitro and in vivo (Tanaka et al. 2007; Dorighi and Tamkun 2013; Huang et al. 2017;

Schmahling et al. 2018). Another study reports that Ash1 also mono-methylates H3K36 and in vitro it has great activity towards unmethylated and mono-methylated but not di-methylated H3K36 substrates (Yuan et al. 2011). Interestingly, several studies have shown that di- and tri-methylated H3K36 antagonizes enzymatic activity of PRC2 (Schmitges et al. 2011; Yuan et al. 2011; Voigt et al.

2012). This has inspired a model in which Ash1 counteracts Polycomb Group-mediated repression through methylation of H3K36 (see Figure 1).

Ash1 and Trx co-localize on polytene chromosomes and although

they were first reported to interact with each other via SET domain

and copurify from embryonic nuclear extract (Rozovskaia et al.

(18)

11

1999), several later studies suggested other biochemical partners for this protein. One of such works reported that Ash1 and Fsh interact biochemically and that genome-wide binding profile of both Ash and Fsh to chromatin are highly corelated (Kockmann et al. 2013).

Indeed Fsh and Ash1 were previously shown to interact genetically

(Shearn 1989). Ash1 was also reported to interact biochemically and

genetically with CBP (Bantignies et al. 2000). However, two recent

studies demonstrated that Ash1 forms a complex with Caf1 and

Mrg15 and this complex di-methylates H3K36 and

methyltransferase activity of Ash1 is important for regulation of

HOX genes (Huang et al. 2017; Schmahling et al. 2018).

(19)

12

Figure 1. Current model to explain how Ash1 antagonizes Polycomb Group-mediated repression. At the repressed state, PRE- bound PRC2 tri-methylates H3K27 leading to the repression of the target gene. At the derepressed state, Ash1 methylates Lys 36 of H3 (H3K36).

This inhibits enzymatic activity of PRC2 and counteracts repression of the target gene.

Histone modifications

As discussed above, both Trithorax Group and Polycomb Group

proteins have enzymatic activity that modifies histone molecules at

(20)

13

specific residues. Below, these histone modifications and their possible roles in antagonistic functions of Polycomb and Trithorax Group proteins are explored.

Methylation of Lys 4 of histone H3

Methylation of histone H3 at Lys 4 residue is evolutionary conserved from yeast to mammals. However, while in yeast all three levels of H3K4 methylation is done by COMPASS complex that contains Set1 (Miller et al. 2001; Roguev et al. 2001; Krogan et al.

2002; Nagy et al. 2002; Schneider et al. 2005), in Drosophila there are three H3K4-specific histone methyl transferases, namely Trithorax (Trx), Trithorax-related (Trr) and Set1. Each of these histone methyltransferases form a separate COMPASS-like complex analogous to that of yeast (Mohan et al. 2011). The Drosophila Set1-containing COMPASS-like complex di- and tri- methylates H3K4 (Ardehali et al. 2011; Mohan et al. 2011; Hallson et al. 2012). In mammals, Set1A and Set1B are two orthologues for Drosophila Set1 which form similar complexes to Drosophila Set1- containing COMPASS-like complex (Lee and Skalnik 2005; Lee et al. 2007a). There are two mammalian orthologues for Drosophila Trx, MLL-1 and MLL-2. These proteins, like Trx, have histone methyltransferase activity (Nakamura et al. 2002) and form COMPASS-like complexes analogous to that of Drosophila (Hughes et al. 2004; Yokoyama et al. 2004; Mohan et al. 2011). The third Drosophila COMPASS-like complex also have mammalian counterparts. MLL-3 and MLL-4, the orthologues of Drosophila Trr form distinct mammalian COMPASS-like complexes (Cho et al.

2007; Lee et al. 2007b; Mohan et al. 2011).

While di- and tri-methylated H3K4 are enriched in transcription start site (TSS) of active genes (Santos-Rosa et al. 2002; Ng et al.

2003; Kharchenko et al. 2011), H3K4 mono-methylation (together

with acetylated H3K27) is a mark of active enhancers (Creyghton et

al. 2010; Lee et al. 2013). Methylation of H3K4 can be by recognized

by the effector proteins involved in both activation and repression

(21)

14

of transcription. In addition, it appears that H3K4 methylation at active genes occurs after transcription initiation (Sims and Reinberg 2006; Ruthenburg et al. 2007). This makes it complicated to understand the role of this modification with regard to regulation of transcription.

In addition to the general activating and repressive roles in transcription, di- and tri-methylated H3K4 have also been shown to antagonize histone methyltransferase activity of PRC2 in vitro (Schmitges et al. 2011; Voigt et al. 2012). This is especially interesting since it can potentially explain how Trithorax, which has histone methyltransferase activity specific for H3K4, antagonizes Polycomb Group proteins. However, it was shown that replacement of Lys 4 of histone H3 with Arg or Ala does not lead to erroneous repression of Ubx or Abd-B in Drosophila larval imaginal discs (Hodl and Basler 2012), undermining this hypothesis.

Methylation of H3K36, like H3K4 methylation, is also evolutionary conserved from yeast to higher eukaryotes, and likewise, more enzymes have evolved in metazoans to undertake different levels of H3K36 methylation. In yeast, Set2 is the only H3K36-specific histone methyltransferase responsible for all three levels of H3K36 methylations (Strahl et al. 2002; Wagner and Carpenter 2012).

Drosophila and mammalian orthologues of the yeast Set2 are responsible for tri-methylation of H3K36 (Bell et al. 2007;

Edmunds et al. 2008), and in Drosophila two other H3K36-specific histone methyltransferases, NSD and Ash1, mono- and di- methylate H3K36 (Bell et al. 2007; Tanaka et al. 2007; Yuan et al.

2011; Dorighi and Tamkun 2013). In mammals, several enzymes are involved in mono- and di-methylation of H3K36 (Wagner and Carpenter 2012).

Although H3K36me3 was reported to regulate alternative splicing

(Luco et al. 2010; Pradeepa et al. 2012), replacement of endogenous

histone H3 with histones carrying Lys 36 to Arg mutation

(22)

15

(H3K36R) does not confirm the previous reports (Meers et al.

2017). H3K36 methylation also has been associated with DNA damage repair (Li et al. 2013; Jha and Strahl 2014; Pai et al. 2014) and dosage compensation (Larschan et al. 2007).

While di- and tri-methylated H3K36 are reported to recruit deacetylase complexes to suppress cryptic transcription initiation (Carrozza et al. 2005), they are usually associated with active transcription. Tri-methylated H3K36 is enriched at the body of active genes and is biased towards exons (Kolasinska-Zwierz et al.

2009) and it negatively correlates with tri-methylated H3K27, a histone modification generally linked to Polycomb repression (Kharchenko et al. 2011). H3K36 di-methyl is also enriched at active genes but is biased toward 5’ end of the genes (Bell et al. 2007).

Additionally, mono-methylated H3K36 is associated with active enhancers.

Interestingly, mass-spectrometry studies of histone H3 isolated from HeLa cells found no histone molecules containing both tri- methylated K27 and tri-methylated K36. (Yuan et al. 2011). As mentioned before, di- and trimethylated H3K36 were shown to antagonize the PRC2 histone methyltransferase activity, in vitro (Schmitges et al. 2011; Yuan et al. 2011; Voigt et al. 2012). As we discussed in Ash1 section, this prompted a model where di- methylation of H3K36 by Ash1 antagonizes Polycomb Group- mediated repression.

PRC2 is responsible for mono-, di-, and tri-methylation of H3K27

in Drosophila (Lee et al. 2015). While tri-methylated H3K27 is

highly enriched at Polycomb Group-mediated repressed genes

equipped with PREs, di-methylated H3K27 is much more broadly

distributed throughout the entire transcriptionally inactive

chromatin, except for the repressed Polycomb Group target genes,

mentioned above (Schwartz et al. 2006; Lee et al. 2015). Both di-

and tri-methylated H3K27 appear to have a repressive role in

(23)

16

transcription (Lee et al. 2015). Lee et al. showed that a dramatic reduction of di- and tri-methylated H3K27 levels, due to a temperature-sensitive mutation in E(z), is accompanied by a significant increase in transcription, not only at genes subjected to Polycomb Group-mediated repression, but also at other genes as well as intergenic regions. Also, regions enriched in active chromatin marks such as tri-methylated H3K4 and acetylated H3K27 are depleted of H3K27 tri-methyl (Schwartz et al. 2010;

Kharchenko et al. 2011). Mono-methyl H3K27, on the other hand, correlates positively with active transcription (Schwartz et al. 2006;

Kharchenko et al. 2011; Lee et al. 2015) which seems to be due to incomplete de-methylation of di- or tri-methylated H3K27 at these loci.

Although mechanistic details of tri-methylated H3K27 function is not clear, it is demonstrated to serve as an epigenetic memory in short term maintenance of Polycomb Group-mediated repression.

To establish a long-term full memory from this short-term memory, the functions of both PREs and Polycomb Group proteins are required (Coleman and Struhl 2017; Laprell et al. 2017). To further emphasize the role of H3K27 trimethylation in Polycomb Group- mediated repression, mutation of Lys 27 of histone H3 to Arg or Ala (H3K27R or H3K27A) lead to derepression of Ubx and Abd-B genes, essentially recapitulating PRC2 loss of function (Pengelly et al.

2013; McKay et al. 2015).

Ubiquitylation of Lys 118 of histone H2A

Ring1 is the E3 ubiquitin ligase responsible for all mono- ubiquitylation of H2AK118 in Drosophila (Gutierrez et al. 2012; Lee et al. 2015) and to efficiently ubiquitinylate H2AK118, it needs a PCGF protein as a cofactor (Buchwald et al. 2006; Lagarou et al.

2008). One of the Drosophila Ring1-containing complexes is PRC1

which uses Psc or its paralogue Su(z)2 as the PCGF subunit and has

been previously discussed in detail. The other Ring1 complex uses

the third Drosophila PCGF protein, L(3)73Ah , as the cofactor and

(24)

17

is responsible for the majority of H2A ubiquitylation in Drosophila (Lee et al. 2015). Although Ring1-L(3)73Ah complex has not been biochemically purified, two observations suggest that such a complex exists. First, knock-down of Ring1 by RNA interference (RNAi) shows nearly complete ablation of H2A mono- ubiquitylation indicating that Ring1 is the only enzyme responsible for mono-ubiquitylation of H2A. Second, depletion of L(3)73Ah by RNAi reduces the global level of mono-ubiquitin H2A by about 70%

(Lee et al. 2015). Since L(3)73Ah is a PCGF protein (Irminger- Finger and Nothiger 1995), these observations collectively imply that a complex containing both Ring1 and L(3)73Ah forms, in vivo, and is responsible for the majority of H2A ubiquitylation.

In addition to the complexes discussed above, a previous study reported another Ring1-containing complex called dRAF, in Drosophila (Lagarou et al. 2008). While like PRC1, Psc or Su(z)2 are the PCGF cofactors in dRAF, it is distinct from PRC1 in that it lacks Pc and Ph and instead contains Kdm2 and was reported to be responsible for the majority of H2A mono-ubiquitylation in Drosophila. (Lagarou et al. 2008). Loss of function of dRAF allegedly enhances the homeotic phenotype of Pc mutation and suppresses that of Trithorax Group proteins trx and ash1, implying the role of dRAF in repression of HOX genes mediated by Polycomb Group proteins (Lagarou et al. 2008). Kdm2 in dRAF complex was reported to demethylate H3K36me2 and stimulate ubiquitylation of H2A by Ring1 (Lagarou et al. 2008). Recently, however, several observations have seriously questioned the validity of the dRAF study. First, loss of Kdm2 function in Drosophila was shown to have no effect on the overall levels of H2A ubiquitylation (Zheng et al.

2018), nor does it have any adverse developmental effects (Shalaby

et al. 2017; Zheng et al. 2018). Additionally, as mentioned above,

the majority of H2A ubiquitylation in Drosophila is done by a

complex containing Ring1 and L(3)73Ah (Lee et al. 2015). These

findings together suggest that existence of dRAF or at least its

involvement in global H2A ubiquitylation and in Polycomb Group-

mediated repression should be revisited.

(25)

18

The role of H2A mono-ubiquitylation has been controversial. H2A ubiquitylation in mammals was reported to inhibit transcription elongation (Stock et al. 2007) and recruit PRC2 (Blackledge et al.

2014; Cooper et al. 2014). However, multiple lines of evidence suggest that, at least in Drosophila, H2A ubiquitylation does not play a critical role in Polycomb Group-mediated repression. First, HOX genes in Drosophila show very low levels of H2A ubiquitylation at repressed state and many genomic loci enriched at H2A ubiquitylation do not correspond to H3K27 tri-methylation domains nor are bound by PRC1 or PRC2 (Lee et al. 2015). Second, repression of HOX in Drosophila is not impaired in the absence of ubiquitylation of H2A (Pengelly et al. 2015). Third, H2AK118ub is removed by PR-DUB, a protein complex consisting of Asx and Calypso which is a ubiquitin C-terminal hydrolase that can specifically de-ubiquitinates H2A118ub, in vitro (Scheuermann et al. 2010). Strikingly, both Calypso (Gaytan de Ayala Alonso et al.

2007) and Asx (Gerd 1985; Sinclair et al. 1998; Gaytan de Ayala

Alonso et al. 2007) are required for proper maintenance of

repression by Polycomb Group proteins. Furthermore, binding of

both Calypso and Asx to the chromatin overlaps with a significant

fraction of Ph and Pho binding sites (Scheuermann et al. 2010). This

suggests that de-ubiquitylation of H2A is important for the proper

Polycomb Group-mediated repression.

(26)

19

Aims of the thesis

• To evaluate the interdependence of PRC1 and PRC2 recruitment and the role of histone modifications in this process.

• To investigate the importance of H3K36 methylation by Ash1 to counteract Polycomb Group-mediated repression.

• To study the role of Ash1 domains in its function.

(27)

20

Results and discussion Manuscripts I and II

The ability of Pho to directly bind DNA within some PREs, and Pc subunit of PRC1 to bind tri-methylated H3k27 deposited by PRC2, have prompted a model where Pho binds to the PREs, leading to the recruitment of PRC2 to the PREs. Once at the PREs, PRC2 tri- methylates H3K27 and this histone modification can be recognized and bound by Pc which leads to recruitment of PRC1 (Wang et al.

2004b). However, two experimental observations question this model. First, tri-methylation of H3K27 is not restricted to the PREs and is spread in broad domains around them while binding of Polycomb Group proteins peak at PREs (Schwartz et al. 2006). If the model, mentioned above, is correct then we must see broad domains of binding for Polycomb Group proteins as well. Second, PREs are devoid of nucleosomes (Kahn et al. 2006; Papp and Muller 2006) therefore histone modifications cannot be involved in recruitment of Polycomb Group proteins to these sites.

More recently an alternative hierarchical model was proposed in mammals. According to this model, a Ring1/2 complex, containing Kdm2b is recruited to unmethylated CpG islands, near the genes regulated by Polycomb Group proteins, by its Kdm2b subunit and ubiquitylates histone H2A. This Ring1/2 complex, called variant PRC1 at the time, is different from PRC1 in that it lacks Cbx and Phc proteins and, in addition to Kdm2b, includes Pcgf1 (instead of Pcgf2/4) and Rybp/Yaf2 (Farcas et al. 2012; Gao et al. 2012;

Tavares et al. 2012; Blackledge et al. 2014; Cooper et al. 2014).

Mono-ubiquitylation of histone H2A by this complex drives the

recruitment of PRC2 through interaction of Jarid2-Aebp2 with

ubiquitylated H2A (Kalb et al. 2014) which leads to tri-methylation

of H3K27 by PRC2 and consequently recruitment of PRC1

(Blackledge et al. 2014; Cooper et al. 2014). This model, even if

correct for mammalian Polycomb Group complexes, does not apply

to Drosophila since mono-ubiquitylation of H2AK118, in flies, was

(28)

21

shown to be dispensable for Polycomb Group-mediated repression of HOX genes (Pengelly et al. 2015).

To test the interdependence of PRC1 and PRC2 recruitment, proposed in the hierarchical models, my co-authors and I used two mutant cell lines, deficient for either PRC1 or PRC2. In Psc/Su(z)2 mutant cells (hereafter PRC1-deficient), neither Psc nor Su(z)2 protein was detectable in western blots, and the levels of Ring1 and Pc components of PRC1 were significantly reduced but the global levels of E(z) and Pcl components of PRC2 remained unchanged.

This suggests that PRC1 subuinits are unstable outside the complex and degrade upon disruption of PRC1. Similarly, in Su(z)12 mutant cells (hereafter PRC2-deficient), the loss of Su(z)12 was accompanied by a dramatic reduction in the global levels of E(z) and Pcl subunits of PRC2 whereas Psc, Pc, and Ring1 were unaffected.

Interestingly, while the loss of PRC2 function completely ablated di- and tri-methylation of H3K27, binding of PRC1 to PREs was only slightly reduced. This rejects the hierarchical model based on pivotal role of H3K27me3 in recruitment of PRC1. Comparing binding profile of Pc to the PREs with that of other PRC1 components reveals an outstanding difference. While all PRC1 subunits produce a sharp peak at the PREs, Pc binding also extends in gradually declining tails as a function of distance from PREs (Kahn et al. 2006; Papp and Muller 2006; Schwartz et al. 2006).

This may reflect the ability of Pc, anchored at the PRE, to interact with the surrounding chromatin, embedded in tri-methylated H3K27. Indeed, among PRC1 components tested, Pc was the most affected and especially tails of Pc binding to the chromatin flanking the PREs, was completely abrogated in the PRC2-deficient cell line.

This observation, together with a study that showed Polycomb

Group proteins at PREs loop out to interact with the surrounding

chromatin (Comet et al. 2011), let us to propose a new model for the

role of H3K27me3 in the context of Polycomb Group-mediated

repression. We suggest that tri-methylated H3K27 prolong the

interaction of PRE-anchored Polycomb Group proteins with their

surrounding chromatin through chromodomain of Pc. This could

(29)

22

deliver Polycomb Group proteins to the promoters or enhancers of the target genes, providing the opportunity to repress these genes.

We also observed that while the overall level of H2A mono- ubiquitylation was not affected in PRC2-deficient cells, the level of this modification, assayed by chromatin immunoprecipitation (ChIP) followed by quantitative real-time PCR (qPCR), was significantly reduced at regions surrounding the PREs, further bolstering our model.

PRC1 loss of function had diverse effects on binding of PRC2 to different PREs. While binding of E(z) and Pcl was dramatically reduced at majority of computationally defined PREs tested by us (hereafter PRC1-dependent PREs), about one third of PREs (hereafter PRC1-independent PREs) retained a significantly higher level of both E(z) and Pcl. It is important to note that PRC1- dependent and -independent PREs are not two discrete categories but rather they represent two extremes of a spectrum. This suggests that recruitment of PRC2 to the PREs is a function of multiple, not a single, mechanisms which are employed to different extents by different PREs. To make sure that this observation is not due to indirect effects, my co-authors and I asked if difference in non- coding transcription through PREs can explain the difference between the two classes of PREs. We observed that intergenic transcription through PREs did not increase in mutant cells compared to the control and that there is no difference between the two classes of PREs in this regard. We also asked whether differential derepression of the target genes, perhaps due to loss of PRC1 function, was responsible for such a difference between PREs.

Although the transcriptional activity of the target genes was generally increased in PRC1-deficient cells, we saw no correlation between the fold change in transcription and PRC2 loss at different PREs. Moreover, we did not observe domains of Ash1 binding, a hallmark of derepression, to most of the tested target genes.

All PREs tested in transgenic assays, to date, have been PRC1-

dependent. This prompted us to ask if PRC1-independent PREs can

(30)

23

autonomously recruit Polycomb Group proteins and repress a reporter gene. We observed no difference in the ability to repress a mini-white reporter gene as well as recruitment of Polycomb Group proteins and tri-methylation of H3K27, between PRC 1-dependent and -independent PREs.

Manuscript III

As mentioned earlier, di- and tri-methylation of H3K36 are reported to antagonize enzymatic activity of PRC2 (Schmitges et al.

2011; Yuan et al. 2011; Voigt et al. 2012). Ash1 has been also reported to methylate H3K36 (Tanaka et al. 2007; Yuan et al. 2011;

Dorighi and Tamkun 2013; Huang et al. 2017; Schmahling et al.

2018). This has led to a model where Ash1 methylates H3K36 and, through this histone methyltransferase activity, antagonizes the repression mediated by Polycomb Group proteins. However, in addition to Ash1, two other histone methyltransferases specific for H3K36, NSD and Set2, are encoded by Drosophila genome. First, I asked whether either or both proteins are involved in counteracting Polycomb Group-mediated repression. Throughout this study, I used expression of HOX genes Ubx and Abd-B and homeotic transformations caused by the erroneous repression of these genes as a readout to evaluate the functional integrity of Trithorax Group proteins. I compared NSD and Set2 loss of function mutants with ash1 mutant and observed that ash1 mutation leads to erroneous and stochastic repression of Ubx and Abd-B genes whereas neither NSD nor Set2 showed irregularities in expression of these HOX genes. This indicates that Ash1 is the only Drosophila H3K36- specific histone methyltransferase involved in Polycomb/Trithorax Group system.

If methylation of H3K36 is how Ash1 functions to antagonize

Polycomb Group-mediated repression, then the lack of Lys 36 of

histone H3 should recapitulate the ash1 mutation phenotype. To

test this hypothesis I used a fly line in which all 23 copies of histone

cluster, containing His1, His2B,His2A,His4, and His3.2 genes, were

(31)

24

deleted (Gunesdogan et al. 2010). I supplemented this fly line with transgenes containing 12 copies of either a wild-type histone cluster (hereafter ∆HisC;H3.2WT) or a histone cluster where Lys 36 of histone His3.2 is replaced with Arg (hereafter ∆HisC;H3.2K36R) (McKay et al. 2015). Surprisingly, while the fitness of

∆HisC;H3.2K36R flies was clearly lower than that of

∆HisC;H3.2WT animals, in neither of these flies did I observe any homeotic transformations, characteristic of ash1 mutation.

Although this observation might indicate that H3K36 methylation

is not relevant for the function of Ash1 in the context of Polycomb

Group-mediated repression, there are two possible alternative

explanations. First, maybe the role of H3K36 methylation is to

prevent another modification from happening at this position and

this presumptive modification is needed for Polycomb Group

proteins to function properly (placeholder hypothesis). In this

scenario, Ash1 methylates H3K36, preventing other modifications

to occur, and consequently counteracts the repression mediated by

Polycomb Group proteins. If that is the case, the phenotype of ash1

mutation is expected to be suppressed by simultaneous replacement

of Lys with Arg at position 36 of histone H3. Intriguingly,

combination of ash1 mutation with ∆HisC;H3.2K36R genetic

background (∆HisC;ash1

^-

,H3.2K36R) did not suppress the lethality

of ash1 mutation. This rejects the placeholder hypothesis. The

second possible explanation is that, in addition to His3.2 (also

known as canonical or replication-dependent histone H3) in

histone clusters, Drosophila genome contains two more His3 genes,

His3.3A and His3.3B, both of which are also known as variant or

replication-independent histone H3. It was shown that in the

absence of histone H3.2, replication-independent H3.3 histones, if

expressed in the right time and quantity, can compensate for the

lack of H3.2 and that in imaginal discs, clones expressing only H3.3

does not show mis-regulation of Ubx or Abd-B genes (Hodl and

Basler 2012). It is possible that in the absence of wild-type H3.2,

H3.3 histones replace transgenic H3.2K36R histones and are

methylated by Ash1 at Lys 36. To test this possibility, I introduced

(32)

25

the deletion of both His3.3A and His3.3B (Hodl and Basler 2009) to ∆HisC;H3.2K36R genetic background (hereafter

∆His3.3;∆HisC;H3.2K36R). Although unlike ∆HisC;H3.2K36R flies, the ∆His3.3;∆HisC;H3.2K36R animals died at late embryonic or first instar larval stage, immunostaining of embryonic CNS did not show erroneous repression of Abd-B. This indicates that Trithorax Group function of Ash1 is not impaired in the zygotic absence of Lys 36 of histone H3. These observations together, have left us with two possible hypotheses. Either methyltransferase activity of Ash1 is not essential to counteract the repression mediated by Polycomb Group proteins or the relevant target of this enzymatic activity is a non-histone substrate. The former seems improbable, since a point mutation in SET domain of Ash1 that abolishes histone methyltransferase activity of the protein, in vitro, was shown to recapitulate homeotic transformations caused by an ash1 strong hypomorphic mutation (Schmahling et al. 2018).

Moreover, I showed that a transgenic Ash1 lacking the SET domain, unlike the transgenic full-length Ash1, does not restore homeotic transformations of ash1 mutation.

This work refutes the probable scenarios in which methylation of H3K36 by Ash1 is required to antagonize the repression mediated by Polycomb Group proteins and, together with another study (Schmahling et al. 2018), strongly suggests that methyltransferase activity of Ash1 is still needed for this function. This leaves us with the hypothesis that Ash1-mediated methylation of yet unknown non-histone substrate(s) is required to counteract Polycomb repression.

Manuscript IV

In the previous study (manuscript III) I showed that methylation of

H3K36 is unlikely to be relevant for Ash1 to counteract Polycomb

repression. However, I observed that a transgenic Ash1, lacking SET

domain, was able to restore the viability, but did not suppress the

homeotic transformations, caused by ash1 mutation. This may pose

(33)

26

a question whether Ash1 has other roles distinct from counteracting Polycomb Group-mediated repression and if the lethality of ash1 mutants reflects this additional role. In addition to SET domain, Ash1 contains a BAH and a PHD domain as well as three AT hook motifs whose roles remains unexplored. Here I generated several transgenic variants of Ash1, each lacking a conserved domain. BAH, PHD, and SET domains were deleted in Ash1∆BAH, Ash1∆PHD, and Ash1∆SET transgenes, respectively. In Ash1∆AT, all three AT hooks were deleted and ASH1FL transgene contains the full-length Ash1 Open Reading Frame (ORF) as a control. I introduced each transgene to a trans-heterozygous ash1

⁹⁰¹¹

/ash1

²²

mutant background. ash1

⁹⁰¹¹

allele contains a large deletion, Df(3L)Exel9011 (Parks et al. 2004), which spans the entire gene region of ash1 as well as several other genes. ash1

²²

, which was reported as an amorphic allele (Tripoulas et al. 1994; Tripoulas et al. 1996), is a point mutation that introduces a premature stop codon early in the ash1 ORF. Both alleles contain additional unidentified recessive lethal mutations rendering them homozygous lethal, therefore they need to be used in trans- heterozygous combination. ash1

⁹⁰¹¹

/ash1

²²

mutants die at pupal stage but combination of trans-heterozygous ash1

²²

with a hypomorphic allele, ash1

²¹

(Tripoulas et al. 1994; Tripoulas et al.

1996), yields flies with visible homeotic transformations. When introduced to ash1

⁹⁰¹¹

/ash1

²²

mutant animals, Ash1∆AT as well as Ash1FL transgenes were able to both restore the viability and suppress the homeotic transformations of the mutant flies. This indicates that AT hook motifs are dispensable for Ash1 function at least in the context of Polycomb Group-mediated repression.

Surprisingly, while except for Ash1∆BAH, the remaining two

transgenic Ash1 variants, provided in two copies, restored the

viability of ash1

⁹⁰¹¹

/ash1

²²

mutant flies, none of the Ash1∆BAH,

Ash1∆PHD, or Ash1∆SET transgenes suppressed the homeotic

transformations of the mutants. This observation has two possible

implications. Either the lethality of ash1

⁹⁰¹¹

/ash1

²²

mutants and the

homeotic transformations, seen in ash1

²²

/ash1

²¹

combination,

(34)

27

results from two distinct functions of Ash1, or ash1

²²

is not a null allele and low levels of Ash1, below my detection threshold, is produced in this allele by translation through stop codon. If the former is the case, combination of ash1

⁹⁰¹¹

with another allele where SET domain activity is abolished must restore the viability, but not suppress the homeotic transformations of mutant flies supplemented with Ash1∆SET transgene. To test that, I set out to delete part of SET, containing conserved amino acids involved in histone methyltransferase activity, and the entire post-SET domain of Ash1 by Cas9-CRISPR system, to abolish enzymatic activity of Ash1. Two novel ash1 alleles were recovered in the process. The first allele, called ash1

^7F

, bears the expected deletion where the rest of the protein, after the deletion, remains in-frame. The second allele, ash1

^3M

, contains approximately the same deletion but with a frame- shift after the deletion which introduces multiple consecutive stop codons and the frame-shift persists until the end of the protein.

Therefore, ash1

^3M

only contains three AT hooks, AWS and almost half of SET domain but lacks post-SET, BAH and PHD domains and part of SET domain bearing multiple conserved amino acids.

When I supplemented ash1

⁹⁰¹¹

/ash1

^3M

flies with two copies of each

transgenic Ash1 variant, only Ash1∆AT and Ash1FL transgenes

restored the viability of mutant flies. This has two important

implications. First, it confirms that AT hooks, are dispensable for

Ash1 function while BAH and PHD domains, as well as SET domain,

are essential. Second, ash1

²²

cannot be a null allele. Interestingly,

when the same set of transgenes were used to complement

homozygous ash1

^7F

mutant, only Ash1∆SET failed to restore the

viability in the ash1 mutant background, once more indicating that

SET domain of Ash1 is essential for the viability of flies (for

summary of results see Figure 2).

(35)

28

Figure 2. Genetic complementation of ash1 mutation with transgenic Ash1 variants. Four different transgenic Ash1 mutant variants (illustrated in colors) as well as a transgenic full-length Ash1 (Ash1FL, illustrated in black) were all integrated in an identical landing site on the second chromosome (grey). (A) When introduced to ash1^7F genetic background (yellow), all Ash1 transgenic variants except for Ash1∆SET rescued the lethality of ash1^7F mutation. (B) In ash1^3M/ash1⁹⁰¹¹ background (yellow), however, only Ash1FL and Ash1∆AT were capable to compensate for the ash1 loss of function.

This observation, considering that both BAH and PHD domains are

essential for Ash1 function, suggests that Ash1 molecules

multimerize in vivo and one Ash1 molecule can use domains of the

other molecule in trans. To test this hypothesis further, I used

combinations of Ash1∆BAH, Ash1∆PHD, or Ash1∆SET transgenes

(36)

29

to supplement ash1

⁹⁰¹¹

/ash1

^3M

mutant flies. Intriguingly, while none of the combinations fully restored the fitness of of ash1 mutants, Ash1∆BAH/Ash1∆SET and Ash1∆PHD/Ash1∆SET combinations but not Ash1∆BAH/Ash1∆PHD were viable. This indicates that while trans-complementation of Ash1 domains, indeed, occur in vivo, both BAH and PHD must be on the same molecule of Ash1 to be functional (Figure 3).

Figure 3. Trans-complementation of Ash1 domains. Combination of different transgenic mutant variants of Ash1 in ash1 mutant background showed that SET domain can be provided in trans and does not have to be present in the same Ash1 molecule together with other domains (A and B). However, both PHD and BAH domains must be present in the same Ash1 molecule to be functional (C).

(37)

30

Conclusions

• Manuscripts I and II:

o Binding of PRC2 to the PREs relies on multiple mechanisms and different PREs employ these mechanisms to different extents.

o One role of methylated H3K27 in Polycomb Group-mediated repression is to reinforce the interaction of PRE-bound PRC1 with the surrounding chromatin.

• Manuscript III:

o SET domain is essential for Ash1 to counteract Polycomb Group-mediated repression.

o Ash1 counteracts Polycomb repression by mechanisms other than H3K36 methylation.

o H3K36-specific histone methyltransferases NSD and Set2 are not redundant with Ash1 and are not essential to counteract Polycomb Group-mediated repression.

• Manuscript IV:

o Both BAH and PHD domains are as essential as SET domain of Drosophila Ash1 while AT hooks are dispensable for Ash1 function in antagonizing Polycomb Group-mediated repression.

o Ash1 protein acts as a multimer where SET

domain of one molecule can combine with

BAH and PHD of the other molecule .

(38)

31

Acknowledgement

Here I would like to thank all the people who helped me during my years of research and study at the Umeå University. First and foremost, my sincere gratitude to my supervisor, Dr. Yuri

Schwartz, for his intellectual support and valuable scientific

inputs in both one-on-one and group discussions. He taught me, among all other things, critical thinking from which I will benefit in my future career as well as my personal life. I also thank all the current and former members of Yuri Schwartz group including Dr.

Tatiana Kan, Dr. Mikhail Savitsky, Dr. Alexander Glotov, Dr. Jana Smigova, Dr. Juan Ignado Barrasa Lopez, Dr.

Sarina Cameron and Moa Lundkvist for creating a nice and

friendly atmosphere to work in. I would like to especially acknowledge Dr. Tatiana Kan, whose help with all the ChIP experiments, presented in this thesis, has been invaluable; Dr.

Mikhail Savitsky for his help in characterization and

identification of phenotypes in mutant flies; and Dr. Alexander

Glotov who spent a lot of time preparing polytene chromosome

squashes and western blots for the manuscripts used in this thesis.

I also want to thank my co-supervisor Dr. Jan Larsson for all the help I received from him and for his critical comments on my work during yearly student seminars and biannual EpiCoN meetings. I also thank his group members especially Dr. Maria Kim and Dr.

Anna-Mia Johansson for all their help. They shared their

knowledge, material, and fly lines with me whenever I needed them.

In the end I would like to dedicate my work to my parents and my

beloved wife for their unconditional support which helped me

through difficult times and desperate moments.

(39)

32

References

Alfieri C, Gambetta MC, Matos R, Glatt S, Sehr P, Fraterman S, Wilm M, Muller J, Muller CW. 2013. Structural basis for targeting the chromatin repressor Sfmbt to Polycomb response elements. Genes Dev 27: 2367-2379.

Americo J, Whiteley M, Brown JL, Fujioka M, Jaynes JB, Kassis JA. 2002. A complex array of DNA-binding proteins required for pairing-sensitive silencing by a polycomb group response element from the Drosophila engrailed gene. Genetics 160: 1561-1571.

Ardehali MB, Mei A, Zobeck KL, Caron M, Lis JT, Kusch T. 2011. Drosophila Set1 is the major histone H3 lysine 4 trimethyltransferase with role in transcription. EMBO J 30: 2817-2828.

Bantignies F, Goodman RH, Smolik SM. 2000. Functional interaction between the coactivator Drosophila CREB-binding protein and ASH1, a member of the trithorax group of chromatin modifiers. Mol Cell Biol 20: 9317- 9330.

Barges S, Mihaly J, Galloni M, Hagstrom K, Muller M, Shanower G, Schedl P, Gyurkovics H, Karch F. 2000. The Fab-8 boundary defines the distal limit of the bithorax complex iab-7 domain and insulates iab-7 from initiation elements and a PRE in the adjacent iab-8 domain. Development 127:

779-790.

Bell O, Wirbelauer C, Hild M, Scharf AN, Schwaiger M, MacAlpine DM, Zilbermann F, van Leeuwen F, Bell SP, Imhof A et al. 2007. Localized H3K36 methylation states define histone H4K16 acetylation during transcriptional elongation in Drosophila. EMBO J 26: 4974-4984.

Blackledge NP, Farcas AM, Kondo T, King HW, McGouran JF, Hanssen LL, Ito S, Cooper S, Kondo K, Koseki Y et al. 2014. Variant PRC1 complex- dependent H2A ubiquitylation drives PRC2 recruitment and polycomb domain formation. Cell 157: 1445-1459.

Bornemann D, Miller E, Simon J. 1996. The Drosophila Polycomb group gene Sex comb on midleg (Scm) encodes a zinc finger protein with similarity to polyhomeotic protein. Development 122: 1621-1630.

Bowman SK, Deaton AM, Domingues H, Wang PI, Sadreyev RI, Kingston RE, Bender W. 2014. H3K27 modifications define segmental regulatory domains in the Drosophila bithorax complex. Elife 3: e02833.

Breen TR, Chinwalla V, Harte PJ. 1995. Trithorax is required to maintain engrailed expression in a subset of engrailed-expressing cells. Mech Dev 52: 89-98.

Breen TR, Harte PJ. 1991. Molecular characterization of the trithorax gene, a positive regulator of homeotic gene expression in Drosophila. Mech Dev 35: 113-127.

(40)

33

Brown JL, Fritsch C, Mueller J, Kassis JA. 2003. The Drosophila pho-like gene encodes a YY1-related DNA binding protein that is redundant with pleiohomeotic in homeotic gene silencing. Development 130: 285-294.

Brown JL, Mucci D, Whiteley M, Dirksen ML, Kassis JA. 1998. The Drosophila Polycomb group gene pleiohomeotic encodes a DNA binding protein with homology to the transcription factor YY1. Mol Cell 1: 1057-1064.

Buchwald G, van der Stoop P, Weichenrieder O, Perrakis A, van Lohuizen M, Sixma TK. 2006. Structure and E3-ligase activity of the Ring-Ring complex of polycomb proteins Bmi1 and Ring1b. EMBO J 25: 2465-2474.

Bunker CA, Kingston RE. 1994. Transcriptional repression by Drosophila and mammalian Polycomb group proteins in transfected mammalian cells.

Mol Cell Biol 14: 1721-1732.

Busturia A, Lloyd A, Bejarano F, Zavortink M, Xin H, Sakonju S. 2001. The MCP silencer of the Drosophila Abd-B gene requires both Pleiohomeotic and GAGA factor for the maintenance of repression. Development 128:

2163-2173.

Byrd KN, Shearn A. 2003. ASH1, a Drosophila trithorax group protein, is required for methylation of lysine 4 residues on histone H3. Proc Natl Acad Sci U S A 100: 11535-11540.

Cao R, Wang L, Wang H, Xia L, Erdjument-Bromage H, Tempst P, Jones RS, Zhang Y. 2002. Role of histone H3 lysine 27 methylation in Polycomb-group silencing. Science 298: 1039-1043.

Cao R, Zhang Y. 2004. SUZ12 is required for both the histone methyltransferase activity and the silencing function of the EED-EZH2 complex. Mol Cell 15:

57-67.

Capotosti F, Hsieh JJ, Herr W. 2007. Species selectivity of mixed-lineage leukemia/trithorax and HCF proteolytic maturation pathways. Mol Cell Biol 27: 7063-7072.

Carrozza MJ, Li B, Florens L, Suganuma T, Swanson SK, Lee KK, Shia WJ, Anderson S, Yates J, Washburn MP et al. 2005. Histone H3 methylation by Set2 directs deacetylation of coding regions by Rpd3S to suppress spurious intragenic transcription. Cell 123: 581-592.

Cavalli G, Paro R. 1998. The Drosophila Fab-7 chromosomal element conveys epigenetic inheritance during mitosis and meiosis. Cell 93: 505-518.

Cavalli G, Paro R. 1999. Epigenetic inheritance of active chromatin after removal of the main transactivator. Science 286: 955-958.

Chan CS, Rastelli L, Pirrotta V. 1994. A Polycomb response element in the Ubx gene that determines an epigenetically inherited state of repression.

EMBO J 13: 2553-2564.

Methyltransferase Ash1, histone methylation and their impact on Polycomb repression