A cell’s identity depends on which of its genes are active. One way for cells to control this process is to change how accessible their genes are to the molecular machinery that switches them on and off. Special proteins called histones determine how accessible genes are by altering how loosely or tightly DNA is packed together.

Histones can be modified by enzymes, which are proteins that add or remove specific chemical ‘tags’. These tags regulate how accessible genes are and provide cells with a memory of gene activity. For example, a protein found in yeast called Epe1 helps reactivate large groups of genes after cell division, effectively ‘re-setting’ the yeast’s genome and eliminating past memories of the genes being inactive.

For a long time, Epe1 was thought to do this by removing methyl groups, a ‘tag’ that indicates a gene is inactive, from histones – that is, by acting like an enzyme. However, no direct evidence to support this hypothesis has been found. Raiymbek et al. therefore set out to determine exactly how Epe1 worked, and whether or not it did indeed behave like an enzyme.

Initial experiments testing mutant versions of Epe1 in yeast cells showed that the changes expected to stop Epe1 from removing methyl groups instead prevented the protein from ‘homing’ to the sections of DNA it normally activates. Detailed microscope imaging, using live yeast cells engineered to produce proteins with fluorescent markers, revealed that this inability to ‘home’ was due to a loss of interaction with Epe1’s main partner, a protein called Swi6. This protein recognizes and binds histones that have methyl tags. Swi6 also acts as a docking site for proteins involved in deactivating genes in close proximity to these histones.

Further biochemical studies revealed how the interaction between Epe1 and Swi6 can help in gene reactivation. The methyl tag on histones in inactive regions of the genome inadvertently helps Epe1 interact more efficiently with Swi6. Then, Epe1 can simply block every other protein that binds to Swi6 from participating in gene deactivation. This observation contrasts with the prevailing view where the active removal of methyl tags by proteins such as Epe1 switches genes from an inactive to an active state.

This work shows for the first time that Epe1 influences the state of the genome through a process that does not involve enzyme activity. In other words, although the protein may ‘moonlight’ as an enzyme, its main job uses a completely different mechanism. More broadly, these results increase the understanding of the many different ways that gene activity, and ultimately cell identity, can be controlled.

The covalent and reversible modification of histones allows cells to establish stable and heritable patterns of gene expression without any changes to their genetic blueprint. Histone H3 lysine nine methylation (H3K9me) is a conserved mark of transcriptional silencing that is associated with the formation of specialized domains called heterochromatin. Heterochromatin formation is critical for centromere and telomere function, silencing of transposons and repetitive DNA elements, and for maintaining lineage-specific patterns of gene expression (Grewal and Jia, 2007; Nicetto et al., 2019). It is widely assumed that a delicate interplay between histone modification readers, writers, and erasers regulates the establishment and maintenance of epigenetic states (Allis and Jenuwein, 2016). In fission yeast, the establishment of H3K9 methylation requires the enzymatic activity of a conserved methyltransferase, Clr4^Suv39h (Rea et al., 2000). H3K9 methylation acts as a platform to recruit chromatin effector proteins with reader domains. Two HP1 homologs, Swi6^HP1 and Chp2^HP1, are involved in H3K9 methylation binding and have distinct, non-overlapping functions during transcriptional silencing (Motamedi et al., 2008). Ultimately, factors such as Epe1, a putative histone demethylase, promote epigenetic erasure, thus reversing H3K9 methylation and transcriptional gene silencing (Ayoub et al., 2003; Trewick et al., 2005).

The sequence-specific recruitment of Clr4^Suv39h restricts heterochromatin establishment to distinct sites in the genome, such as the centromeres, telomeres, and the mating-type locus (Bayne et al., 2010; Hall et al., 2002; Verdel et al., 2004; Volpe et al., 2002). Once established, H3K9 methylation spreads to silence genes that are distant from heterochromatin nucleation centers. H3K9me spreading depends on the read-write activity of Clr4^Suv39h (Al-Sady et al., 2013; Zhang et al., 2008). Two conserved, structural attributes of Clr4^Suv39h mediate this process: a conserved chromodomain that recognizes and binds to H3K9me and an enzymatic SET domain that is involved in catalysis (Ivanova et al., 1998). The ability of Clr4^Suv39h to bind to the product of its enzymatic activity enables H3K9 methylated histones to serve as carriers of epigenetic information (Audergon et al., 2015; Ragunathan et al., 2015). Following DNA replication, H3K9 methylated histones that are partitioned between daughter DNA strands serve as templates to mark newly deposited histones.

H3K9 methylation acts as a multivalent platform that recruits the HP1 homolog, Swi6^HP1 to sites of constitutive heterochromatin (Ekwall et al., 1995; Ekwall et al., 1996). HP1 proteins have a conserved architecture consisting of a chromodomain (CD) that is involved in H3K9 methylation binding and a dimerization domain called the chromoshadow domain (CSD) that mediates protein-protein interactions (Bannister et al., 2001; Cowieson et al., 2000). The oligomerization of HP1 proteins promotes formation of higher-order chromatin complexes that exhibit phase separated, liquid-like properties in vitro and in cells (Larson et al., 2017; Sanulli et al., 2018; Strom et al., 2017). Swi6^HP1 interacts with a broad spectrum of agonists and antagonists that influence epigenetic silencing. Most notably, recruitment of Epe1, a putative histone demethylase that opposes heterochromatin formation, is dependent on Swi6^HP1 (Ayoub et al., 2003; Zofall and Grewal, 2006). Loss of Epe1 leads to increased heterochromatin spreading beyond normal boundary sequences, inheritance of H3K9 methylation via sequence-independent pathways, increased cell to cell variation in H3K9 methylation patterns, and acquisition of adaptive epigenetic traits (Audergon et al., 2015; Ayoub et al., 2003; Ragunathan et al., 2015; Sorida et al., 2019; Trewick et al., 2007; Wang et al., 2015; Zofall and Grewal, 2006; Zofall et al., 2012). Despite its critical role in heterochromatin regulation, how Epe1 exerts its anti-silencing function in cells remains mysterious (Trewick et al., 2007).

Histone demethylases have prominent roles in regulating the reversibility of epigenetic states (Iwase et al., 2007; Whetstine et al., 2006). Changes in their expression lead to widespread chromatin reorganization, which alters both the prognosis and treatment of diseases such as cancer (Liau et al., 2017). Epe1 most closely resembles JumonjiC (JmjC) domain containing proteins that use Fe (II) and α-ketoglutarate as co-factors to catalyze histone demethylation (Tsukada et al., 2006). Unlike active histone demethylases, where an HxD/E….H motif is involved in Fe (II) coordination, Epe1 harbors a non-canonical HXE….Y motif (Tsukada et al., 2006). Although Epe1 shares conserved features with other histone demethylases at the amino acid level, there is no biochemical evidence to support the notion that Epe1 has any in vitro enzymatic activity. Epe1 purified from fission yeast cells, or a recombinant source (insect cells), exhibits no H3K9 demethylase activity in vitro (Tsukada et al., 2006; Zofall and Grewal, 2006). However, point mutations of amino acid residues involved in Fe (II) or α-ketoglutarate binding affect Epe1 activity in cells and lead to heterochromatin spreading beyond normal boundary sequences (Trewick et al., 2007). These conflicting lines of biochemical and genetic data have prompted several alternative explanations for how Epe1 might fulfill its anti-silencing role. These models include the possibility that Epe1 acts as a protein hydroxylase which targets non-histone proteins such as Swi6^HP1, regulates the activity of the multi-subunit H3K9 methyltransferase CLRC complex, or functions as an H3K9 demethylase when in complex with Swi6^HP1 (Aygün et al., 2013; Iglesias et al., 2018; Trewick et al., 2007; Zofall and Grewal, 2006). Although these models represent attractive possibilities for how Epe1 regulates heterochromatin, there is no direct evidence to suggest that any of these proteins represent bonafide enzymatic targets. An alternative hypothesis is that Epe1 has a non-enzymatic function that regulates heterochromatin spreading and epigenetic inheritance. In support of this hypothesis, the overexpression of Fe (II) and α-ketoglutarate binding mutants of Epe1 suppresses heterochromatin spreading defects observed in epe1Δ strains (Trewick et al., 2007; Zofall and Grewal, 2006). Hence, co-factor binding mutants of Epe1 can act as multi-copy suppressors of epigenetic silencing despite the presumptive loss of enzymatic activity.

In this study, we discovered that the putative catalytic JmjC domain of Epe1 is, at least in part, dispensable for its anti-silencing function in cells. The C-terminus of Epe1 directly interacts with Swi6^HP1 and its interaction in the context of full-length Epe1 is regulated by H3K9 methylation. Expressing the Epe1 C-terminus alone is sufficient to reverse heterochromatin establishment and attenuate epigenetic inheritance. We propose that a cis interaction between the Epe1 N- and C-terminus inhibits Swi6^HP1 binding. H3K9 methylation binding attenuates this intramolecular interaction and promotes Swi6^HP1 binding. A requirement for H3K9 methylation to stabilize a complex comprising Epe1 and Swi6^HP1 restricts their interaction to a heterochromatin-specific context. Our work highlights the versatile, non-canonical ways in which histone demethylases can oppose establishment and maintenance of epigenetic states.

The establishment and maintenance of epigenetic states is primarily thought to depend on a balance of enzymatic activities between readers, writers, and erasers of histone modifications (Allis and Jenuwein, 2016). However, the replication-dependent and independent turnover of histones serve as a major mechanism that shapes genome-wide patterns of histone methylation (Chory et al., 2019). In Drosophila, modified histones are turned over more than once during each cell-cycle , which limits their capacity to serve as carriers of epigenetic information (Coleman and Struhl, 2017; Deal et al., 2010; Laprell et al., 2017). In principle, passive genome-wide nucleosome exchange can compete with histone modification-dependent read-write mechanisms to oppose epigenetic inheritance.

Our data support the notion that enzymatic erasure is at least in part, dispensable for regulating the inheritance of transcriptionally silent epigenetic states in fission yeast. Our observations are fully consistent with earlier work demonstrating an antagonistic relationship between Epe1, which enhances nucleosome turnover at sites of heterochromatin formation, and Clr3, which suppresses this process (Aygün et al., 2013). In addition, Epe1 also has newly defined roles in recruiting members of the SAGA complex which mediate histone acetylation, a chromatin feature that is also associated with increased nucleosome turnover (Bao et al., 2019). Recent work demonstrated a surprising and unexpected function for the N-terminus of Epe1 in transcriptional activation, which suppresses stochastic, epigenetic silencing (Sorida et al., 2019). Our findings are complementary to these recent studies and expand the gamut of anti-silencing functions associated with Epe1.

Auto-inhibition is a widely used strategy that proteins use to activate their latent functions in response to specific cellular or environmental signals (Pufall and Graves, 2002). The enzymatic and non-enzymatic properties of proteins can be regulated by internal control mechanisms that act in cis. For example, in the case of the chromatin remodeler ALC1, poly ADP ribosylation in response to DNA damage releases the macrodomain of the protein from an auto-inhibited state and activates its latent ATPase-dependent nucleosome remodeling function (Lehmann et al., 2017; Singh et al., 2017). The non-catalytic heterochromatin associated protein Swi6^HP1 is auto-inhibited by a histone H3 mimic sequence (Canzio et al., 2013). Release from auto-inhibition switches the protein to a spreading competent state.

We discovered that a weak interaction between Epe1 and Swi6^HP1 can be bolstered by H3K9 methylation. We propose that a cis interaction between the N- and C-terminal halves of Epe1 opposes Swi6^HP1 binding. These results suggest that Epe1 may be preserved in an auto-inhibited state until it interacts with H3K9 methylation (Figure 8). In the absence of H3K9 methylation, the two proteins fail to form a stable complex in vivo despite their ability to interact directly with each other in vitro. The ability of Epe1 to bind to H3K9 methylation raises the possibility that it competes with Swi6^HP1 for a shared binding site. However, heterochromatin consists of dense networks of H3K9 methylated nucleosomes. It is also possible that Epe1 while interacting with Swi6^HP1, can also be stimulated via an interaction with an adjacent or a distal nucleosome. Our data do not lend direct support to a structural change that occurs within Epe1 upon H3K9me3 peptide binding but strongly suggest that modified histones can influence the stability of chromatin associated complexes in living cells.

Figure 8

Download asset Open asset

Swi6^HP1 is dynamic and undergoes rapid exchange on the millisecond timescale between the free and H3K9 methylation-bound state (Cheutin et al., 2004; Cheutin et al., 2003). Given the relative abundance of Swi6^HP1 in cells, a simple protein-protein interaction between Epe1 and Swi6^HP1 would effectively titrate Epe1 away from sites of heterochromatin formation. By enforcing an H3K9 methylation-dependent mode of interaction, Epe1 selectively targets a sub-population of Swi6^HP1 molecules that are bound to sites of H3K9 methylation and have a causal role in heterochromatin formation. Therefore, one outcome of this H3K9 methylation-dependent mode of interaction is that Epe1 selectively interacts with heterochromatin-bound Swi6^HP1 molecules as opposed to freely diffusing Swi6^HP1 proteins. It is likely that this balance of protein-protein interactions is tunable via post-translational modifications. For example, Swi6^HP1 phosphorylation compromises Epe1 binding and promotes histone deacetylase recruitment to sites of heterochromatin formation (Shimada et al., 2009). Our studies demonstrate that altering the balance of Swi6^HP1-dependent protein-protein interactions profoundly affects the stability and heritability of silent epigenetic states.

Tethering Clr3 at an ectopic site renders heterochromatin refractory to the anti-silencing effects of Epe1. Our observations are in part, similar to previous findings relating to the fission yeast mating type locus where DNA binding proteins, Atf1 and Pcr1 recruit Clr3 to maintain epigenetic silencing following an RNAi-dependent initiation mechanism (Wang and Moazed, 2017; Yamada et al., 2005). We also recapitulate a unique requirement for the sequence-dependent recruitment of HDAC proteins in cis to facilitate the inheritance of silent epigenetic states despite the presence of anti-silencing factors such as Epe1. We favor a model where histone deacetylases such as Clr3 create a chromatin environment that promotes the read-write enzymatic function of Clr4^Suv39h. In essence, Epe1 prevents Clr3 mediated histone hypoacetylation by forming an inhibitory complex with Swi6^HP1 within a heterochromatin restricted genomic context.

The question of whether Epe1 harbors any histone demethylase activity remains unanswered. Inspired by earlier studies on the Drosophila dKDM4A protein where HP1 binding stimulates demethylase activity, we attempted to reconstitute the enzymatic function of Epe1 in the presence of Swi6^HP1 (Lin et al., 2008). However, our studies are unable to detect Epe1 mediated enzymatic demethylation in the presence or absence of Swi6^HP1 (Figure 2—figure supplement 2A). It is noteworthy that a histone demethylase like protein in Neurospora, DMM-1 shares many similarities with Epe1 and also surprisingly lacks any in vitro enzymatic activity (Honda et al., 2010). DMM-1 also interacts with the HP1 homolog in Neurospora leading us to speculate that this surprising mode of heterochromatin regulation we uncovered in the case of Epe1 might also extend to other fungal systems. Given that expressing the C-terminus of Epe1 only partially restores wild-type levels of Epe1 activity (Figure 3C,D), we speculate that there could potentially be unique substrates or conditions required to reconstitute the enzymatic activity of Epe1. We favor a model in which heterochromatin associated proteins could act as positive or negative allosteric regulators of Epe1. For example, Bdf2, a BRD4 homolog that localizes to heterochromatin boundaries along with Epe1, is a possible candidate that may activate the latent enzymatic functions of Epe1 as H3K9 methylation levels are negligible at the heterochromatin-euchromatin boundary (Wang et al., 2013).

Our observations add to an expanding list of proteins that mimic histone modifiers but have built-in non-enzymatic functions that regulate the establishment and maintenance of epigenetic states. The Drosophila protein dKDM4A is a prominent example of a histone demethylase with enzymatic and non-enzymatic functions (Colmenares et al., 2017). More extreme examples of histone-modifying enzyme mimicry are observed in the case of JARID2, a subunit of the PRC2 complex that shares several characteristics with JmjC domain-containing proteins but has a structural role in regulating PRC2 complex assembly (Kasinath et al., 2018; Son et al., 2013). Proteins involved in the Arabidopsis RNA-dependent DNA methylation pathway, SUVH2 and SUVH9, resemble SET domain methyltransferases but lack enzymatic activity. Instead, both proteins recognize methylated DNA and are involved in RNA pol V recruitment in plants to establish epigenetic silencing (Johnson et al., 2008; Law et al., 2013). Although Epe1 has a JmjC domain, it is possible that its putative catalytic function has been repurposed to regulate its interaction with Swi6^HP1. Our data suggest that H3K9 methylation promotes the cooperative assembly of complexes between Epe1 and Swi6^HP1 at sites of heterochromatin formation.

We speculate that the enzymatic and non-enzymatic functions of Epe1 are likely to oppose heterochromatin assembly on different timescales. As a first approximation, the stabilization of Epe1 at sites of heterochromatin formation is sufficient to displace histone deacetylases and hasten the loss of transcriptional silencing. Hence, the non-enzymatic function of Epe1 serves as the first line of opposition to heterochromatin establishment. This transient inhibition of heterochromatin assembly is likely to be followed by a slower enzymatic step where Epe1 might demethylate H3K9 methylated histones or activate transcription (Sorida et al., 2019). Our work provides a biochemical basis for how chromatin associated factors that are thought to purely function as enzymes have non-enzymatic properties that regulate heterochromatin assembly and epigenetic inheritance.

All data generated or analysed during this study are included in the manuscript and supporting files.

1. 1. Al-Sady B
   2. Madhani HD
   3. Narlikar GJ

   (2013) Division of labor between the chromodomains of HP1 and Suv39 methylase enables coordination of heterochromatin spread

   Molecular Cell 51:80–91.

   https://doi.org/10.1016/j.molcel.2013.06.013

   • PubMed
   • Google Scholar
2. 1. Allis CD
   2. Jenuwein T

   (2016) The molecular hallmarks of epigenetic control

   Nature Reviews Genetics 17:487–500.

   https://doi.org/10.1038/nrg.2016.59

   • PubMed
   • Google Scholar
3. 1. Audergon PNCB
   2. Catania S
   3. Kagansky A
   4. Tong P
   5. Shukla M
   6. Pidoux AL
   7. Allshire RC

   (2015) Restricted epigenetic inheritance of H3K9 methylation

   Science 348:132–135.

   https://doi.org/10.1126/science.1260638

   • Google Scholar
4. 1. Aygün O
   2. Mehta S
   3. Grewal SI

   (2013) HDAC-mediated suppression of histone turnover promotes epigenetic stability of heterochromatin

   Nature Structural & Molecular Biology 20:547–554.

   https://doi.org/10.1038/nsmb.2565

   • PubMed
   • Google Scholar
5. 1. Ayoub N
   2. Noma K
   3. Isaac S
   4. Kahan T
   5. Grewal SI
   6. Cohen A

   (2003) A novel jmjC domain protein modulates heterochromatization in fission yeast

   Molecular and Cellular Biology 23:4356–4370.

   https://doi.org/10.1128/MCB.23.12.4356-4370.2003

   • PubMed
   • Google Scholar
6. 1. Bähler J
   2. Wu JQ
   3. Longtine MS
   4. Shah NG
   5. McKenzie A
   6. Steever AB
   7. Wach A
   8. Philippsen P
   9. Pringle JR

   (1998) Heterologous modules for efficient and versatile PCR-based gene targeting in Schizosaccharomyces pombe

   Yeast 14:943–951.

   https://doi.org/10.1002/(SICI)1097-0061(199807)14:10<943::AID-YEA292>3.0.CO;2-Y

   • PubMed
   • Google Scholar
7. 1. Bannister AJ
   2. Zegerman P
   3. Partridge JF
   4. Miska EA
   5. Thomas JO
   6. Allshire RC
   7. Kouzarides T

   (2001) Selective recognition of methylated lysine 9 on histone H3 by the HP1 chromo domain

   Nature 410:120–124.

   https://doi.org/10.1038/35065138

   • PubMed
   • Google Scholar
8. 1. Bao K
   2. Shan CM
   3. Moresco J
   4. Yates J
   5. Jia S

   (2019) Anti-silencing factor Epe1 associates with SAGA to regulate transcription within heterochromatin

   Genes & Development 33:116–126.

   https://doi.org/10.1101/gad.318030.118

   • PubMed
   • Google Scholar
9. 1. Bayne EH
   2. White SA
   3. Kagansky A
   4. Bijos DA
   5. Sanchez-Pulido L
   6. Hoe KL
   7. Kim DU
   8. Park HO
   9. Ponting CP
   10. Rappsilber J
   11. Allshire RC

   (2010) Stc1: a critical link between RNAi and chromatin modification required for heterochromatin integrity

   Cell 140:666–677.

   https://doi.org/10.1016/j.cell.2010.01.038

   • PubMed
   • Google Scholar
10. 1. Canzio D
    2. Liao M
    3. Naber N
    4. Pate E
    5. Larson A
    6. Wu S
    7. Marina DB
    8. Garcia JF
    9. Madhani HD
    10. Cooke R
    11. Schuck P
    12. Cheng Y
    13. Narlikar GJ

    (2013) A conformational switch in HP1 releases auto-inhibition to drive heterochromatin assembly

    Nature 496:377–381.

    https://doi.org/10.1038/nature12032

    • PubMed
    • Google Scholar
11. 1. Cheutin T
    2. McNairn AJ
    3. Jenuwein T
    4. Gilbert DM
    5. Singh PB
    6. Misteli T

    (2003) Maintenance of stable heterochromatin domains by dynamic HP1 binding

    Science 299:721–725.

    https://doi.org/10.1126/science.1078572

    • PubMed
    • Google Scholar
12. 1. Cheutin T
    2. Gorski SA
    3. May KM
    4. Singh PB
    5. Misteli T

    (2004) In vivo dynamics of Swi6 in yeast: evidence for a stochastic model of heterochromatin

    Molecular and Cellular Biology 24:3157–3167.

    https://doi.org/10.1128/MCB.24.8.3157-3167.2004

    • PubMed
    • Google Scholar
13. 1. Chory EJ
    2. Calarco JP
    3. Hathaway NA
    4. Bell O
    5. Neel DS
    6. Crabtree GR

    (2019) Nucleosome turnover regulates histone methylation patterns over the genome

    Molecular Cell 73:61–72.

    https://doi.org/10.1016/j.molcel.2018.10.028

    • Google Scholar
14. 1. Coleman RT
    2. Struhl G

    (2017) Causal role for inheritance of H3K27me3 in maintaining the OFF state of a Drosophila HOX gene

    Science 356:eaai8236.

    https://doi.org/10.1126/science.aai8236

    • PubMed
    • Google Scholar
15. 1. Colmenares SU
    2. Swenson JM
    3. Langley SA
    4. Kennedy C
    5. Costes SV
    6. Karpen GH

    (2017) Drosophila histone demethylase KDM4A has enzymatic and Non-enzymatic roles in controlling heterochromatin integrity

    Developmental Cell 42:156–169.

    https://doi.org/10.1016/j.devcel.2017.06.014

    • PubMed
    • Google Scholar
16. 1. Cowieson NP
    2. Partridge JF
    3. Allshire RC
    4. McLaughlin PJ

    (2000) Dimerisation of a chromo shadow domain and distinctions from the chromodomain as revealed by structural analysis

    Current Biology 10:517–525.

    https://doi.org/10.1016/S0960-9822(00)00467-X

    • PubMed
    • Google Scholar
17. 1. Deal RB
    2. Henikoff JG
    3. Henikoff S

    (2010) Genome-wide kinetics of nucleosome turnover determined by metabolic labeling of histones

    Science 328:1161–1164.

    https://doi.org/10.1126/science.1186777

    • PubMed
    • Google Scholar
18. 1. Ekwall K
    2. Javerzat JP
    3. Lorentz A
    4. Schmidt H
    5. Cranston G
    6. Allshire R

    (1995) The chromodomain protein Swi6: a key component at fission yeast centromeres

    Science 269:1429–1431.

    https://doi.org/10.1126/science.7660126

    • PubMed
    • Google Scholar
19. 1. Ekwall K
    2. Nimmo ER
    3. Javerzat JP
    4. Borgstrøm B
    5. Egel R
    6. Cranston G
    7. Allshire R

    (1996)

    Mutations in the fission yeast silencing factors clr4+ and rik1+ disrupt the localisation of the chromo domain protein Swi6p and impair centromere function

    Journal of Cell Science 109 ( Pt 11:2637–2648.

    • PubMed
    • Google Scholar
20. 1. Grewal SI
    2. Jia S

    (2007) Heterochromatin revisited

    Nature Reviews Genetics 8:35–46.

    https://doi.org/10.1038/nrg2008

    • PubMed
    • Google Scholar
21. 1. Hall IM
    2. Shankaranarayana GD
    3. Noma K
    4. Ayoub N
    5. Cohen A
    6. Grewal SI

    (2002) Establishment and maintenance of a heterochromatin domain

    Science 297:2232–2237.

    https://doi.org/10.1126/science.1076466

    • PubMed
    • Google Scholar
22. 1. Honda S
    2. Lewis ZA
    3. Huarte M
    4. Cho LY
    5. David LL
    6. Shi Y
    7. Selker EU

    (2010) The DMM complex prevents spreading of DNA methylation from transposons to nearby genes in Neurospora crassa

    Genes & Development 24:443–454.

    https://doi.org/10.1101/gad.1893210

    • PubMed
    • Google Scholar
23. 1. Horton JR
    2. Upadhyay AK
    3. Hashimoto H
    4. Zhang X
    5. Cheng X

    (2011) Structural Basis for Human PHF2 Jumonji Domain Interaction with Metal Ions

    Journal of Molecular Biology 406:1–8.

    https://doi.org/10.1016/j.jmb.2010.12.013

    • Google Scholar
24. 1. Iglesias N
    2. Currie MA
    3. Jih G
    4. Paulo JA
    5. Siuti N
    6. Kalocsay M
    7. Gygi SP
    8. Moazed D

    (2018) Automethylation-induced conformational switch in Clr4 (Suv39h) maintains epigenetic stability

    Nature 560:504–508.

    https://doi.org/10.1038/s41586-018-0398-2

    • Google Scholar
25. 1. Ivanova AV
    2. Bonaduce MJ
    3. Ivanov SV
    4. Klar AJ

    (1998) The chromo and SET domains of the Clr4 protein are essential for silencing in fission yeast

    Nature Genetics 19:192–195.

    https://doi.org/10.1038/566

    • PubMed
    • Google Scholar
26. 1. Iwase S
    2. Lan F
    3. Bayliss P
    4. de la Torre-Ubieta L
    5. Huarte M
    6. Qi HH
    7. Whetstine JR
    8. Bonni A
    9. Roberts TM
    10. Shi Y

    (2007) The X-linked mental retardation gene SMCX/JARID1C defines a family of histone H3 lysine 4 demethylases

    Cell 128:1077–1088.

    https://doi.org/10.1016/j.cell.2007.02.017

    • PubMed
    • Google Scholar
27. 1. Jacobs SA
    2. Khorasanizadeh S

    (2002) Structure of HP1 chromodomain bound to a lysine 9-methylated histone H3 tail

    Science 295:2080–2083.

    https://doi.org/10.1126/science.1069473

    • PubMed
    • Google Scholar
28. 1. Johnson LM
    2. Law JA
    3. Khattar A
    4. Henderson IR
    5. Jacobsen SE

    (2008) SRA-domain proteins required for DRM2-mediated de novo DNA methylation

    PLOS Genetics 4:e1000280.

    https://doi.org/10.1371/journal.pgen.1000280

    • PubMed
    • Google Scholar
29. 1. Kasinath V
    2. Faini M
    3. Poepsel S
    4. Reif D
    5. Feng XA
    6. Stjepanovic G
    7. Aebersold R
    8. Nogales E

    (2018) Structures of human PRC2 with its cofactors AEBP2 and JARID2

    Science 359:940–944.

    https://doi.org/10.1126/science.aar5700

    • PubMed
    • Google Scholar
30. 1. Laprell F
    2. Finkl K
    3. Müller J

    (2017) Propagation of Polycomb-repressed chromatin requires sequence-specific recruitment to DNA

    Science 356:85–88.

    https://doi.org/10.1126/science.aai8266

    • PubMed
    • Google Scholar
31. 1. Larson AG
    2. Elnatan D
    3. Keenen MM
    4. Trnka MJ
    5. Johnston JB
    6. Burlingame AL
    7. Agard DA
    8. Redding S
    9. Narlikar GJ

    (2017) Liquid droplet formation by HP1α suggests a role for phase separation in heterochromatin

    Nature 547:236–240.

    https://doi.org/10.1038/nature22822

    • PubMed
    • Google Scholar
32. 1. Law JA
    2. Du J
    3. Hale CJ
    4. Feng S
    5. Krajewski K
    6. Palanca AM
    7. Strahl BD
    8. Patel DJ
    9. Jacobsen SE

    (2013) Polymerase IV occupancy at RNA-directed DNA methylation sites requires SHH1

    Nature 498:385–389.

    https://doi.org/10.1038/nature12178

    • PubMed
    • Google Scholar
33. 1. Lehmann LC
    2. Hewitt G
    3. Aibara S
    4. Leitner A
    5. Marklund E
    6. Maslen SL
    7. Maturi V
    8. Chen Y
    9. van der Spoel D
    10. Skehel JM
    11. Moustakas A
    12. Boulton SJ
    13. Deindl S

    (2017) Mechanistic insights into autoinhibition of the oncogenic chromatin remodeler ALC1

    Molecular Cell 68:847–859.

    https://doi.org/10.1016/j.molcel.2017.10.017

    • Google Scholar
34. 1. Liau BB
    2. Sievers C
    3. Donohue LK
    4. Gillespie SM
    5. Flavahan WA
    6. Miller TE
    7. Venteicher AS
    8. Hebert CH
    9. Carey CD
    10. Rodig SJ
    11. Shareef SJ
    12. Najm FJ
    13. van Galen P
    14. Wakimoto H
    15. Cahill DP
    16. Rich JN
    17. Aster JC
    18. Suvà ML
    19. Patel AP
    20. Bernstein BE

    (2017) Adaptive chromatin remodeling drives glioblastoma stem cell plasticity and drug tolerance

    Cell Stem Cell 20:233–246.

    https://doi.org/10.1016/j.stem.2016.11.003

    • PubMed
    • Google Scholar
35. 1. Lin CH
    2. Li B
    3. Swanson S
    4. Zhang Y
    5. Florens L
    6. Washburn MP
    7. Abmayr SM
    8. Workman JL

    (2008) Heterochromatin protein 1a stimulates histone H3 lysine 36 demethylation by the Drosophila KDM4A demethylase

    Molecular Cell 32:696–706.

    https://doi.org/10.1016/j.molcel.2008.11.008

    • PubMed
    • Google Scholar
36. 1. Matsuyama A
    2. Shirai A
    3. Yashiroda Y
    4. Kamata A
    5. Horinouchi S
    6. Yoshida M

    (2004) pDUAL, a multipurpose, multicopy vector capable of chromosomal integration in fission yeast

    Yeast 21:1289–1305.

    https://doi.org/10.1002/yea.1181

    • PubMed
    • Google Scholar
37. 1. Motamedi MR
    2. Hong EJ
    3. Li X
    4. Gerber S
    5. Denison C
    6. Gygi S
    7. Moazed D

    (2008) HP1 proteins form distinct complexes and mediate heterochromatic gene silencing by nonoverlapping mechanisms

    Molecular Cell 32:778–790.

    https://doi.org/10.1016/j.molcel.2008.10.026

    • PubMed
    • Google Scholar
38. 1. Nicetto D
    2. Donahue G
    3. Jain T
    4. Peng T
    5. Sidoli S
    6. Sheng L
    7. Montavon T
    8. Becker JS
    9. Grindheim JM
    10. Blahnik K
    11. Garcia BA
    12. Tan K
    13. Bonasio R
    14. Jenuwein T
    15. Zaret KS

    (2019) H3K9me3-heterochromatin loss at protein-coding genes enables developmental lineage specification

    Science 363:294–297.

    https://doi.org/10.1126/science.aau0583

    • PubMed
    • Google Scholar
39. 1. Pufall MA
    2. Graves BJ

    (2002) Autoinhibitory domains: modular effectors of cellular regulation

    Annual Review of Cell and Developmental Biology 18:421–462.

    https://doi.org/10.1146/annurev.cellbio.18.031502.133614

    • PubMed
    • Google Scholar
40. 1. Ragunathan K
    2. Jih G
    3. Moazed D

    (2015) Epigenetics. Epigenetic inheritance uncoupled from sequence-specific recruitment

    Science 348:1258699.

    https://doi.org/10.1126/science.1258699

    • PubMed
    • Google Scholar
41. 1. Rea S
    2. Eisenhaber F
    3. O'Carroll D
    4. Strahl BD
    5. Sun ZW
    6. Schmid M
    7. Opravil S
    8. Mechtler K
    9. Ponting CP
    10. Allis CD
    11. Jenuwein T

    (2000) Regulation of chromatin structure by site-specific histone H3 methyltransferases

    Nature 406:593–599.

    https://doi.org/10.1038/35020506

    • PubMed
    • Google Scholar
42. 1. Reddy BD
    2. Wang Y
    3. Niu L
    4. Higuchi EC
    5. Marguerat SB
    6. Bähler J
    7. Smith GR
    8. Jia S

    (2011) Elimination of a specific histone H3K14 acetyltransferase complex bypasses the RNAi pathway to regulate pericentric heterochromatin functions

    Genes & Development 25:214–219.

    https://doi.org/10.1101/gad.1993611

    • PubMed
    • Google Scholar
43. 1. Sadaie M
    2. Kawaguchi R
    3. Ohtani Y
    4. Arisaka F
    5. Tanaka K
    6. Shirahige K
    7. Nakayama J

    (2008) Balance between distinct HP1 family proteins controls heterochromatin assembly in fission yeast

    Molecular and Cellular Biology 28:6973–6988.

    https://doi.org/10.1128/MCB.00791-08

    • PubMed
    • Google Scholar
44. Preprint

    1. Sanulli S
    2. Trnka M
    3. Dharmarajan V
    4. Tibble R
    5. Pascal B
    6. Burlingame A
    7. Griffin P
    8. Gross J
    9. Narlikar G

    (2018) Disorganization of the histone core promotes organization of heterochromatin into phase-separated droplets

    bioRxiv.

    https://doi.org/10.1101/473132

    • Google Scholar
45. 1. Shimada A
    2. Dohke K
    3. Sadaie M
    4. Shinmyozu K
    5. Nakayama J
    6. Urano T
    7. Murakami Y

    (2009) Phosphorylation of Swi6/HP1 regulates transcriptional gene silencing at Heterochromatin

    Genes & Development 23:18–23.

    https://doi.org/10.1101/gad.1708009

    • PubMed
    • Google Scholar
46. 1. Singh HR
    2. Nardozza AP
    3. Möller IR
    4. Knobloch G
    5. Kistemaker HAV
    6. Hassler M
    7. Harrer N
    8. Blessing C
    9. Eustermann S
    10. Kotthoff C
    11. Huet S
    12. Mueller-Planitz F
    13. Filippov DV
    14. Timinszky G
    15. Rand KD
    16. Ladurner AG

    (2017) A Poly-ADP-Ribose trigger releases the Auto-Inhibition of a chromatin remodeling oncogene

    Molecular Cell 68:860–871.

    https://doi.org/10.1016/j.molcel.2017.11.019

    • PubMed
    • Google Scholar
47. 1. Son J
    2. Shen SS
    3. Margueron R
    4. Reinberg D

    (2013) Nucleosome-binding activities within JARID2 and EZH1 regulate the function of PRC2 on chromatin

    Genes & Development 27:2663–2677.

    https://doi.org/10.1101/gad.225888.113

    • PubMed
    • Google Scholar
48. 1. Sorida M
    2. Hirauchi T
    3. Ishizaki H
    4. Kaito W
    5. Shimada A
    6. Mori C
    7. Chikashige Y
    8. Hiraoka Y
    9. Suzuki Y
    10. Ohkawa Y
    11. Kato H
    12. Takahata S
    13. Murakami Y

    (2019) Regulation of ectopic heterochromatin-mediated epigenetic diversification by the JmjC family protein Epe1

    PLOS Genetics 15:e1008129.

    https://doi.org/10.1371/journal.pgen.1008129

    • PubMed
    • Google Scholar
49. 1. Strom AR
    2. Emelyanov AV
    3. Mir M
    4. Fyodorov DV
    5. Darzacq X
    6. Karpen GH

    (2017) Phase separation drives heterochromatin domain formation

    Nature 547:241–245.

    https://doi.org/10.1038/nature22989

    • Google Scholar
50. 1. Trewick SC
    2. McLaughlin PJ
    3. Allshire RC

    (2005) Methylation: lost in Hydroxylation?

    EMBO Reports 6:315–320.

    https://doi.org/10.1038/sj.embor.7400379

    • PubMed
    • Google Scholar
51. 1. Trewick SC
    2. Minc E
    3. Antonelli R
    4. Urano T
    5. Allshire RC

    (2007) The JmjC domain protein Epe1 prevents unregulated assembly and disassembly of heterochromatin

    The EMBO Journal 26:4670–4682.

    https://doi.org/10.1038/sj.emboj.7601892

    • PubMed
    • Google Scholar
52. 1. Tsukada Y
    2. Fang J
    3. Erdjument-Bromage H
    4. Warren ME
    5. Borchers CH
    6. Tempst P
    7. Zhang Y

    (2006) Histone demethylation by a family of JmjC domain-containing proteins

    Nature 439:811–816.

    https://doi.org/10.1038/nature04433

    • PubMed
    • Google Scholar
53. 1. Tsukada Y
    2. Nakayama KI

    (2010) In vitro histone demethylase assay

    Cold Spring Harbor Protocols 2010:pdb.prot5512.

    https://doi.org/10.1101/pdb.prot5512

    • PubMed
    • Google Scholar
54. 1. Verdel A
    2. Jia S
    3. Gerber S
    4. Sugiyama T
    5. Gygi S
    6. Grewal SI
    7. Moazed D

    (2004) RNAi-mediated targeting of heterochromatin by the RITS complex

    Science 303:672–676.

    https://doi.org/10.1126/science.1093686

    • PubMed
    • Google Scholar
55. 1. Volpe TA
    2. Kidner C
    3. Hall IM
    4. Teng G
    5. Grewal SI
    6. Martienssen RA

    (2002) Regulation of heterochromatic silencing and histone H3 lysine-9 methylation by RNAi

    Science 297:1833–1837.

    https://doi.org/10.1126/science.1074973

    • PubMed
    • Google Scholar
56. 1. Wang J
    2. Tadeo X
    3. Hou H
    4. Tu PG
    5. Thompson J
    6. Yates JR
    7. Jia S

    (2013) Epe1 recruits BET family bromodomain protein Bdf2 to establish heterochromatin boundaries

    Genes & Development 27:1886–1902.

    https://doi.org/10.1101/gad.221010.113

    • PubMed
    • Google Scholar
57. 1. Wang J
    2. Reddy BD
    3. Jia S

    (2015) Rapid epigenetic adaptation to uncontrolled heterochromatin spreading

    eLife 4:e06179.

    https://doi.org/10.7554/eLife.06179

    • Google Scholar
58. 1. Wang X
    2. Moazed D

    (2017) DNA sequence-dependent epigenetic inheritance of gene silencing and histone H3K9 methylation

    Science 356:88–91.

    https://doi.org/10.1126/science.aaj2114

    • PubMed
    • Google Scholar
59. 1. Whetstine JR
    2. Nottke A
    3. Lan F
    4. Huarte M
    5. Smolikov S
    6. Chen Z
    7. Spooner E
    8. Li E
    9. Zhang G
    10. Colaiacovo M
    11. Shi Y

    (2006) Reversal of histone lysine trimethylation by the JMJD2 family of histone demethylases

    Cell 125:467–481.

    https://doi.org/10.1016/j.cell.2006.03.028

    • PubMed
    • Google Scholar
60. 1. Yamada T
    2. Fischle W
    3. Sugiyama T
    4. Allis CD
    5. Grewal SI

    (2005) The nucleation and maintenance of heterochromatin by a histone deacetylase in fission yeast

    Molecular Cell 20:173–185.

    https://doi.org/10.1016/j.molcel.2005.10.002

    • PubMed
    • Google Scholar
61. 1. Zhang K
    2. Mosch K
    3. Fischle W
    4. Grewal SI

    (2008) Roles of the Clr4 methyltransferase complex in nucleation, spreading and maintenance of heterochromatin

    Nature Structural & Molecular Biology 15:381–388.

    https://doi.org/10.1038/nsmb.1406

    • PubMed
    • Google Scholar
62. 1. Zofall M
    2. Yamanaka S
    3. Reyes-Turcu FE
    4. Zhang K
    5. Rubin C
    6. Grewal SIS

    (2012) RNA elimination machinery targeting meiotic mRNAs promotes facultative heterochromatin formation

    Science 335:96–100.

    https://doi.org/10.1126/science.1211651

    • Google Scholar
63. 1. Zofall M
    2. Grewal SI

    (2006) Swi6/HP1 recruits a JmjC domain protein to facilitate transcription of heterochromatic repeats

    Molecular Cell 22:681–692.

    https://doi.org/10.1016/j.molcel.2006.05.010

    • PubMed
    • Google Scholar

Acceptance summary:

This paper identifies a non-enzymatic function in counteracting heterochromatin for the putative histone demethylase Epe1. The presented data suggest that binding of Epe1 to histone H3 methylated at lysine 9 – a hallmark of heterochromatin – enables Epe1 to interact with the key heterochromatin protein Swi6 by releasing an autoinhibitory conformation. The results highlight the possibility that putative enzymes may have non-catalytic modes of action that are as important or more important than catalytic ones.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "An auto-inhibitory mechanism regulates the non-enzymatic functions of a histone demethylase" for consideration by eLife. Your article has been reviewed by four peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Yota Murakami (Reviewer #3).

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work in its present form will not be considered further for publication in eLife.

The reviewers agreed that the mechanism by which Epe1 counteracts heterochromatic silencing is of great interest. The reviewers were also intrigued by the proposal that Epe1 autoinhibition is relieved by interaction with H3K9me, which enables Epe1 to bind Swi6, thereby displacing heterochromatin-promoting factors such as Clr3. However, there was strong consensus that key aspects of the model – Epe1 autoinhibition, and the release of autoinhibition by direct interaction with H3K9me – are not strongly supported by the data. The proposed Epe1 noncatalytic mechanism was also not strongly supported and the overall model of Epe1 action has important unstated implication that should be considered and discussed. We would be willing to consider a resubmitted manuscript that convincingly demonstrates how H3K9me affects Epe1 conformation and function and presents a clear, well-supported model of Epe1 anti-silencing activity.

Reviewer #1:

This manuscript investigates how Epe1, a putative histone demethylase for which catalytic activity has not been demonstrated, counteracts epigenetic silencing in S. pombe. The authors show that mutations in residues predicted to be involved in cofactor binding disrupt Epe1 activity, prevent its localization to heterochromatin, and disrupt its interaction with Swi6. Loss of Swi6 or H3K9me prevents Epe1 from localizing to heterochromatin. The authors show that Epe1 interacts with Swi6 via a region adjacent to the JmjC domain, that this domain interacts more strongly with Swi6 in isolation than in the context of the entire Epe1 protein, and that expression of this domain (lacking JmjC) can counteract heterochromatic silencing. The authors also show that the N- and C-terminal halves of Epe1 interact in vitro. Interaction between Epe1 and Swi6 is stimulated by H3K9me in vitro, and a mutation that reduces Swi6 affinity for H3K9me does not eliminate this effect. Finally, the authors show that tethering the histone deacetylase Clr3 to heterochromatin can overcome the anti-silencing function of Epe1. The authors propose that Epe1 exists in an auto-inhibited conformation which is relieved by interaction with H3K9me, allowing Epe1 to interact with Swi6, which in turn counteracts silencing via a non-enzymatic mechanism that involves displacing silencing proteins, such as Clr3, that are recruited by Swi6.

This manuscript contains a great deal of interesting results, but I feel that two of the key conclusions are supported by weak or ambiguous data.

First, the authors propose that Epe1 has a noncatalytic mode of anti-silencing. This is based on (1) the ability of a Swi6-interactin domain (lacking JmjC) of Epe1 to counteract heterochromatic silencing, (2) the ability of cofactor binding mutations to disrupt Epe1 localization, and (3) the demonstration that tethering the histone deacetylase Clr3 to heterochromatin can overcome the anti-silencing function of Epe1. However, the first result is quite artificial and does not show that the whole Epe1 protein, expressed at endogenous levels, has a noncatalytic anti-silencing mode. The authors don't know how their mutations interfere with Swi6 binding, and therefore the second result doesn't to me argue for a noncatalytic activity. Same of Clr3 tethering – I find it equally plausible that Clr3 and Epe1 have antagonistic catalytic activities, and tethering Clr3 tips the balance in favour of heterochromatin. I don't find the claim for a noncatalytic function of Epe1 to be strongly supported.

Second, the authors propose that binding of H3K9me to Epe1 relieves Epe1 autoinhibition and enables Epe1 to interact with Swi6. This is supported by the observations that (1) a domain of Epe1 interacts more strongly than the entire protein with Swi6 in vitro, that (2) the N- and C-terminal halves of Epe1 interact in vitro, that (3) the interaction between Epe1 and Swi6 is stimulated by H3K9me in vitro, that (4) a mutation that reduces Swi6 affinity for H3K9me does not eliminate this effect, and that (5) several mutations disrupt Epe1 localization to heterochromatin and binding to Swi6. However, the first two lines of evidence do not demonstrate autoinhibition, they just show in vitro effects that are consistent with autoinhibition and may not be relevant in vivo. The evidence that H3K9me stimulates the interaction with Swi6 via binding to Epe1 is weak and indirect. The authors do not demonstrate H3K9me binding to Epe1 and do not identify the domain responsible for this. The authors also do not know how the cofactor binding mutations disrupt Epe1 interaction with Swi6. The conclusion that binding of H3K9me to Epe1 relieves Epe1 autoinhibition and enables Epe1 to interact with Swi6 remains a hypothesis.

I am particularly concerned by the weak evidence because I find the overall model proposed by the authors counterintuitive (or at least nonintuitive). A simple model would be that the interaction between H3K9me and Swi6 increases the local concentration of Swi6 at heterochromatin (facilitated by the protein's phase transition properties), which allows Epe1 to bind. But the authors add a major complication – interaction between H3K9me and Epe1 is also required. In this model, Epe1 must interact with H3K9me and Swi6, but it is also competing with Swi6 for H3K9me binding. Is this competition important? Can tethering of Swi6 to chromatin recruit Epe1 in the absence of H3K9me? I feel that the authors' model needs substantially more support than is currently provided.

Reviewer #2:

In fission yeast, Epe1 is a putative histone demethylase that plays important roles in regulating the heterochromatin landscape. However, the mechanism by which Epe1 functions is not clear.

In this manuscript, the authors found that point mutations that abolish co-factor binding in the JmjC domain can disrupt Epe1's binding with Swi6 and also reduces Epe1 localization at heterochromatin. in vitro binding assays confirmed that one Epe1 mutant (H297A) has significantly reduced binding with Swi6. Truncation analysis mapped the Swi6 interaction interface to Epe1 433-630, which is outside of the JmjC domain. Moreover, Epe1 433-630 interacts with Swi6 stronger than full length Epe1, which led the authors to hypothesize that full-length Epe1 might adopt an auto-inhibitive conformation to prevent its binding with Swi6. Indeed, the authors found that Epe1's N-terminal region can bind to its C-terminal region. Authors further demonstrated that H3K9me3 is required for Swi6-Epe1 interaction in vivo and that H3K9me3 peptide enhanced Swi6-Epe1 interaction in vitro, but not Epe1-H29&A. These results suggest a mechanism for Epe1 to only interact with Swi6 at heterochromatin regions. Lastly, the authors showed that Epe1 blocks Swi6's interaction with Clr3, an HDAC important for heterochromatin assembly.

Overall, the data quality is high and the proposed model is very interesting. The findings will be of high interest to the heterochromatin field. I would recommend its publication after addressing several major questions.

1) The model suggests that the JmjC domain of Epe1 binds to H3K9me3 and that the H297A mutation affects this binding. Since the authors have recombinant proteins at hand, they should test these interactions.

2) The model also suggests that in the presence of H3K9me3, interaction between Epe1's N- and C-terminal regions will be reduced. This should also be tested.

3) In Figure 4, there appears to be very weak binding of WT Epe1 with Swi6 without H3K9me3 peptide. This is very different from results shown in Figures 2 and 3. What's the reason for this discrepancy?

4) Clr3∆ or Sir2∆ has different effects on H3K9me at centromeres, mating type region, and telomeres. If the hypothesis is correct, it is expected that Epe1 localization will change differently at different locations. ChIP analysis of Epe1 will complement Figure 4B and strengthen the hypothesis.

Reviewer #3:

The putative histone demethylase Epe1 in fission yeast has been shown to localize heterochromatin via interaction with a HP1 homologue Swi6 and antagonize heterochromatin formation probably through its JmjC-domain dependent demethylation activity, though the enzymatic activity has not been detected in vitro. In this manuscript, authors made several interesting observations. (1) JmjC-domain mutants of Epe1 loose interaction with Swi6 and heterochromatin localization. (2) Epe1 interacts with Swi6 via its C-terminal half that does not include JmjC domain. (3) Epe1 N-terminal half including JmjC-domain interacts with C-terminal half in trans. (5) H3K9me peptide enhance interaction between Swi6 and Epe1. (4) Expression of C-terminal half Epe1 antagonize maintenance of heterochromatic silencing with non-enzymatic activity. (6) Mutation of JmjC domain strengthen the interaction between Swi6 and Clr3, a Swi6-binding silencing factor. (7) Artificial recruitment of Clr3 supports maintenance of heterochromatin in the presence of wild type Epe1. Taken together authors claimed that Epe1-Swi6 binding is auto-inhibited by cis-interaction between N-terminal half and C-terminal half of Epe1. The auto-inhibition can be released by H3K9 methylation and Epe1 compete with Clr3 for Swi6-binding resulting in removal of heterochromatin.

Author's model is novel and interesting in terms of regulatory mechanism of heterochromatin by Epe1. However, their model is not fully supported by their experimental data. There is no direct evidence showing Epe1 N-terminal half and C-terminal half interacts in cis. Their result can be also explained by trans-interaction. Moreover, they did not present direct evidence showing this interaction is inhibitory to Swi6 binding. This could be tested by adding N-terminal half in the in vitro binding assay of Epe1^434-948 and Swi6 (Figure 3A). In addition, If H3K9me peptide release the auto inhibition through Epe1 as author claimed, addition of H3K9me peptide in the in vitro binding assay would disrupt interaction between N- and C-terminal half of Epe1. Furthermore, I think one of the main point of their model is that competition between Epe1 and Clr3 for Swi6, by which Epe1 inhibits stable maintenance of artificially formed heterochromatin. However, the evidence for this mechanism is not enough. At least, they should check localization of Clr3 at artificial heterochromatin by ChIP analysis. If their model is correct, amount of Clr3 at artificial heterochromatin (-Tet) would increase in epe1∆ or epe1-H297A cells and decreased in the presence of Epe1 or Epe1^434-948.

Recent paper (Sorida et al., 2019), showed that N-terminal domain that does not include jmjC domain has transcription activation activity and deletion of this domain caused severe loss of Swi6 binding and heterochromatin localization of Epe1. In addition, they showed H297A mutation also caused reduction of Swi6 binding. Probably, this paper published after the submission of this manuscript, but authors should mention and discuss about this paper in the revised manuscript.

In conclusion, this manuscript presents novel data and their model will attract interests of many readers, but the points I described above should be clarified before publication.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for submitting your article "An auto-inhibitory mechanism regulates the non-enzymatic functions of a histone demethylase" for consideration by eLife. Your article has been reviewed by four peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Kevin Struhl as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Songtao Jia (Reviewer #2); Yota Murakami (Reviewer #4).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

Since the discovery of the JmjC domain as the catalytic region of histone demethylases, the molecular function of Epe1 in fission yeast has been under intense debate, and no demethylase activity has been shown. Interest in this protein intensified with the finding that H3K9me can be inherited without initiation signals, but only in an epe1 mutant background. The assumption has been that Epe1 is a H3K9 demethylase, with the supporting evidence that epe1 mutants expected to abolish demethylase activity also abolished Epe1 function. This manuscript provides detailed characterization of these mutations, challenging the hypothesis that they abolish demethylase activity. Instead, this paper identifies a non-enzymatic function for Epe1 – binding to Swi6 – that is regulated by H3K9me. The authors' data also suggest, but do not prove yet, that binding of Epe1 to H3K9me enables Epe1 to interact with Swi6 by releasing an autoinhibitory conformation. The results highlight the possibility that proteins that appear to be primarily enzymes may have non-catalytic modes of action that are as important or more important than catalytic ones.

Essential revisions:

1) The most important outstanding concern is about the interaction between H3K9me3 and Epe1. This interaction is central to the authors' hypothesis yet is tested only with the entire Epe1 protein and reported in two supplementary panels. One concern is that the model presented by the authors – that Swi6 and Epe1 may not compete for H3K9me tails – depends on the binding affinities of these proteins for H3K9me. There are significantly fewer H3K9me tails (~2000 nucleosomes) in fission yeast than Swi6 molecules (~20,000, Sadaie, 2008), which would argue that only molecules at the affinity of Swi6 for H3K9me2/3, or tighter, could access tails. Co-occupancy of Swi6 and Epe1 of adjacent tails therefore is possible only in a circumstance where the affinity of Epe1 is at least as high as that of Swi6 (which is not terribly high, measured at ~2-15 µM depending on the group), given the substantial Swi6 excess. The easiest way to show that H3K9me binding by Epe1 directs its assembly on chromatin is to mutate the region or residues required for this recognition and test its impact.

Given the many available truncation constructs, the authors should be able to identify the domain that interacts with H3K9me and determine the affinity of the interaction. As previously suggested, the JmjC domain is a good candidate for H3K9me binding. These data could be used to generate a mutant that abolishes binding, and the mutant phenotypes should be tested in vivo. These experiments would be very important to support the authors' conclusions.

2) The presented data clearly indicate that an H3K9me peptide can disrupt the interaction between the N-terminal and C-terminal domains of Epe1. However, the authors have not tested if the addition of the N-terminal half inhibits the interaction between the C-terminus and Swi6. This would provide support for the authors' autoinhibition model.

3) The presented experiments strongly suggest, but do not prove, an autoinhibitory mechanism, as there are no experiments that directly address Epe1 conformation. Therefore, claims regarding such a mechanism should be softened. The title could be changed to "Interaction with histone H3 methylated at lysine 9 regulates the non-enzymatic functions of the putative histone demethylase Epe1" or something similar, and strong claims like "An auto-inhibitory conformation regulates the non-enzymatic properties of Epe1 through its interaction with Swi6" in the Abstract should be amended or removed.

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

The reviewers agreed that the mechanism by which Epe1 counteracts heterochromatic silencing is of great interest. The reviewers were also intrigued by the proposal that Epe1 autoinhibition is relieved by interaction with H3K9me, which enables Epe1 to bind Swi6, thereby displacing heterochromatin-promoting factors such as Clr3.

However, there was strong consensus that key aspects of the model – Epe1 autoinhibition, and the release of autoinhibition by direct interaction with H3K9me – are not strongly supported by the data. The proposed Epe1 noncatalytic mechanism was also not strongly supported and the overall model of Epe1 action has important unstated implication that should be considered and discussed. We would be willing to consider a resubmitted manuscript that convincingly demonstrates how H3K9me affects Epe1 conformation and function and presents a clear, well-supported model of Epe1 anti-silencing activity.

Since our initial submission, we have added experimental data that addresses some of the key issues highlighted by reviewers and editors, including, support for a model based on auto-inhibition, H3K9 methylation dependent release from auto-inhibition and additional evidence for a non-catalytic function associated with Epe1. We have also edited our Discussion section in light of recent work from the Murakami lab outlining a non-enzymatic function for the Epe1 N-terminus (Sorida et al., 2019). We have considered the implications of our work and how this affects current models of epigenetic inheritance in a revised Discussion section. The result of our efforts is twelve new figure panels, substantial edits to the manuscript and a critical assessment of reviewer feedback.

This significant update to the manuscript was made possible by four key experiments (along with several minor/additional revisions) that we have summarized here:

1) We detected a direct binding interaction between Epe1 and an H3K9me3 peptide and H3K9 methylated histones. Epe1 also exhibits specificity towards H3K9 methylated as opposed to H3K4 methylated histones (Figure 2—figure supplement 1H-I).

2) We also directly demonstrate that the interaction between the Epe1 N- and C-terminus in trans can be disrupted upon addition of an H3K9me3 peptide. These results lend strong support to a model where Epe1 is preserved in an auto-inhibited state with H3K9 methylation reversing this inhibition (new Figure 5A).

3) We identified and purified a C-terminal truncation mutant of Epe1, (Epe1-ΔC) which exhibits a substantial increase in its interaction with Swi6 relative to the full-length protein even in the absence of a modified peptide. Furthermore, Epe1-ΔC does not undergo an H3K9me3 dependent enhancement in its interaction with Swi6 unlike its full-length counterpart.

Collectively, these results clearly support our notion that the C-terminus of Epe1 plays a critical role in enforcing an H3K9 methylation dependent mode of interaction with Swi6 (new Figure 5B-E).

We designed a completely novel ectopic silencing approach in fission yeast to provide additional evidence for a non-catalytic function associated with the Epe1 C-terminus. Since Epe1 interacts with Swi6 and excludes other effector molecules, we hypothesized that Swi6 mediated heterochromatin initiation would be highly sensitive to Epe1 activity (enzymatic or nonenzymatic). We expressed the C-terminus of Epe1 (434-948) which lacks the JmjC domain in cells where a TetR-Swi6 CSD fusion has been tethered. The C-terminal fragment expression now completely blocks heterochromatin formation at the ectopic site. These results clearly support a model where the non-enzymatic function of Epe1 relies on its Swi6 interaction (Figure 3E and Figure 3C).

Reviewer #1:

[…] This manuscript contains a great deal of interesting results, but I feel that two of the key conclusions are supported by weak or ambiguous data.

We thank the reviewer for his/her positive comments about our findings. We have sought to address concerns raised about the ambiguity of our data and our interpretations through additional experiments. Our new experiments lend strong support to our proposed model and we hope that our new findings will alleviate these concerns.

First, the authors propose that Epe1 has a noncatalytic mode of anti-silencing.

The premise of our studies is based on prior work revealing a non-enzymatic basis for Epe1 in resetting epigenetic silencing (Trewick et al., 2005; Zofall et al., 2006). The overexpression of Epe1 point mutants that lack any catalytic functions completely restores normal levels of reporter gene silencing in epe1Δ cells. Recent studies have further substantiated these early findings and suggest that Epe1 might have a critical role in activating transcription within pericentromeric repeats (Bao et al., 2019; Sorida et al., 2019). Our studies, viewed in this broader context, adds a strong mechanistic perspective to this urgent question through stringent biochemical analysis.

This is based on (1) the ability of a Swi6-interactin domain (lacking JmjC) of Epe1 to counteract heterochromatic silencing, (2) the ability of cofactor binding mutations to disrupt Epe1 localization, and (3) the demonstration that tethering the histone deacetylase Clr3 to heterochromatin can overcome the anti-silencing function of Epe1. However, the first result is quite artificial and does not show that the whole Epe1 protein, expressed at endogenous levels, has a noncatalytic anti-silencing mode. The authors don't know how their mutations interfere with Swi6 binding, and therefore the second result doesn't to me argue for a noncatalytic activity.

We thank the reviewer for this important critique. Although we do not know the mechanism that leads to the loss of interaction between Epe1 JmjC mutants and Swi6, we can unambiguously catalog the molecular consequence of the loss of their interaction. We demonstrate that a loss of interaction between the full-length Epe1 JmjC mutant (expressed at endogenous levels) and Swi6 is compensated by a gain of an interaction between a heterochromatin enhancer, Clr3 and Swi6 (Figure 6A). Therefore, we disagree that our manuscript does not establish a non-enzymatic function associated with Epe1 when the full-length protein has been expressed at endogenous levels.

Our tethering experiments were designed based on findings in Figure 6A to overcome the need for Swi6 to be involved in Clr3 recruitment to sites of heterochromatin formation. We were surprised that the tethering experiments conformed to our expectations where Clr3 recruitment appears to be rate-limiting for epigenetic inheritance. Intriguingly, our ectopic system recapitulates a DNA sequence dependent maintenance mechanism where Atf1/Pcr1 (two DNA binding proteins) are thought to recruit Clr3 to the fission yeast mating type locus (Wang et al., 2017). Therefore, the Clr3 tethering approach recapitulates what potentially occurs in fission yeast cells at natural loci.

Our Discussion section is agnostic about the possibility of Epe1 having an additional enzymatic function. We certainly have not ruled out this possibility. We have stated that the expression of the Epe1 C-terminus only partially suppresses heterochromatin assembly. It is eminently possible that our reconstitution conditions failed to capture histone demethylation. Our Discussion section highlights some potential strategies to consider while attempting to reconstitute its putative catalytic functions. We have added the following lines to the Discussion:

“Nevertheless, given the similarity of the JmjC domain of Epe1 to active demethylases, we favor a model where different proteins that interact with Epe1 could act as positive or negative allosteric regulators. For example, Bdf2, a BRD4 homolog that localizes to heterochromatin boundaries along with Epe1, represents a possible candidate that activates its latent enzymatic functions given that H3K9 methylation levels are negligible at the heterochromatin-euchromatin interface (Wang et al., 2013).”

Same of Clr3 tethering – I find it equally plausible that Clr3 and Epe1 have antagonistic catalytic activities, and tethering Clr3 tips the balance in favour of heterochromatin. I don't find the claim for a noncatalytic function of Epe1 to be strongly supported.

It is important to note that whether Clr3 acts directly or indirectly to oppose Epe1 activity does not negate the central claims of our manuscript. Nevertheless, we deeply appreciate and recognize the need to be cautious with regards to our conclusions. Hence, we have amended the Results section of the manuscript relating to the Clr3 tethering experiments to explicitly state:

“We hypothesized that directly tethering Clr3 at an ectopic site would eliminate its dependency on Swi6 to localize at sites of heterochromatin formation. In this genetic context, we rationalized that Epe1 would be unlikely to counteract Clr3 recruitment. One caveat of this experimental strategy is that it does not rule out the possibility that Epe1 and Clr3 have antagonistic catalytic activities but simply seeks to test whether Clr3 recruitment itself might be rate-limiting at sites of heterochromatin formation.”

Second, the authors propose that binding of H3K9me to Epe1 relieves Epe1 autoinhibition and enables Epe1 to interact with Swi6. This is supported by the observations that (1) a domain of Epe1 interacts more strongly than the entire protein with Swi6 in vitro, that (2) the N- and C-terminal halves of Epe1 interact in vitro, that (3) the interaction between Epe1 and Swi6 is stimulated by H3K9me in vitro, that (4) a mutation that reduces Swi6 affinity for H3K9me does not eliminate this effect, and that (5) several mutations disrupt Epe1 localization to heterochromatin and binding to Swi6. However, the first two lines of evidence do not demonstrate autoinhibition, they just show in vitro effects that are consistent with autoinhibition and may not be relevant in vivo. The evidence that H3K9me stimulates the interaction with Swi6 via binding to Epe1 is weak and indirect. The authors do not demonstrate H3K9me binding to Epe1 and do not identify the domain responsible for this. The authors also do not know how the cofactor binding mutations disrupt Epe1 interaction with Swi6. The conclusion that binding of H3K9me to Epe1 relieves Epe1 autoinhibition and enables Epe1 to interact with Swi6 remains a hypothesis.

We have now included three independent lines of experimental evidence to support our claims:

1) We tested the direct interaction between Epe1 and an H3K9me3 peptide and H3K9 methylated histones. Our binding assays clearly detect a direct interaction between Epe1 and an H3K9me3 peptide as opposed to an unmodified H3K9me0 peptide. Furthermore, Epe1 selectively interacts with H3K9 methylated histones as opposed to H3K4 methylated histones. These data have now been added to the revised version of the manuscript (Figure 2—figure supplement 1H and I).

2) To identify a potential domain within Epe1 that enforces an H3K9 methylation dependent mode of interaction with Swi6, we generated a C-terminal truncation mutant (Epe1-ΔC). Using in vitroand in vivomeasurements: We demonstrate that Epe1-ΔC exhibits a significantly higher degree of binding to Swi6 (Figure 5B-C). Furthermore, Epe1-ΔC fails to exhibit any H3K9me3 peptide dependent stimulation (Figure 5D). Epe1-ΔC binds to Swi6 in cells in the presence or absence of the H3K9 methyltransferase, clr4D (Figure 5E). We conclude that the C-terminus of Epe1 enforces an H3K9 methylation dependent mode of interaction between Epe1 and Swi6.

These data have now been added as a new figure to the revised version of our manuscript (new Figure 5B-E).

3) To test whether H3K9 methylation reverses an auto-inhibitory conformation of Epe1, we measured the interaction between the N- and C-terminal Epe1 fragments in the presence or absence of an H3K9me3 peptide. The addition of an H3K9me3 peptide disrupts the direct interaction between the N- and C-terminal Epe1 fragments in trans. This data has now been added to the revised version of the manuscript (new Figure 5A).

I am particularly concerned by the weak evidence because I find the overall model proposed by the authors counterintuitive (or at least nonintuitive). A simple model would be that the interaction between H3K9me and Swi6 increases the local concentration of Swi6 at heterochromatin (facilitated by the protein's phase transition properties), which allows Epe1 to bind. But the authors add a major complication – interaction between H3K9me and Epe1 is also required.

We initially considered and spent considerable time and effort exploring a model where Swi6 phase separation facilitates Epe1 binding. However, new and existing biochemical lines of evidence that we have presented in this manuscript forced us to re-evaluate these early assumptions. In particular, the most striking findings that contradict this model are:

1) A Swi6 chromodomain mutant (Swi6 W104A) that fails to bind to H3K9 methylated histones nevertheless undergoes a substantial increase in its interaction with Epe1 in the presence of an H3K9me3 peptide (Figure 4—figure supplement 1D).

2) An Epe1 C-terminal truncation mutant, Epe1-ΔC, does not undergo any stimulation in its interaction with wild-type Swi6 in the presence of an H3K9me3 peptide (Figure 5D).

3) A trans interaction between the Epe1 N- and C- terminus can be disrupted by an H3K9me3 peptide bolstering our case for a model predicated on Epe1 being responsive to the presence of H3K9 methylation (Figure 5A).

Based on these observations, we envision a model where H3K9 methylation promotes the cooperative assembly of a heterochromatin inhibitory complex comprised of Epe1 and Swi6. The model where the phase separation properties of Swi6 drive Epe1 recruitment is not fully supported by our data.

In this model, Epe1 must interact with H3K9me and Swi6, but it is also competing with Swi6 for H3K9me binding. Is this competition important?

The question of competing interactions between Epe1 and Swi6 for H3K9 methylated histone tails is undoubtedly an interesting and contentious point to consider. This is never an issue in our experiments where our in vitroassays are supplemented with an excess of histone H3K9me3 peptide. In cells, it is equally plausible that the multivalency of heterochromatin leads to non-competitive complex assembly wherein Epe1 and Swi6 bind to adjacent or distal nucleosomes. Given the dense networks of H3K9 methylation at sites of constitutive heterochromatin, we favor a model where the two proteins need not have to compete for the same binding site. We have added a line in our Discussion section which highlights how Swi6 and Epe1 could potentially compete for a shared binding site:

“The ability of Epe1 to bind to H3K9 methylation raises the possibility that it competes with Swi6 for a shared binding site. However, heterochromatin consists of dense networks of H3K9 methylation. It is possible that Epe1 while interacting with Swi6 could undergo stimulation via an interaction with an adjacent or a distal nucleosome eliminating a need for competition.”

Can tethering of Swi6 to chromatin recruit Epe1 in the absence of H3K9me? I feel that the authors' model needs substantially more support than is currently provided.

We interpreted this request by the reviewer as having to show further evidence for the non-enzymatic properties of Epe1 by developing new phenotypic assays. We were inspired by this question to design the experiment detailed in this response.

Two hybrid experiments where Epe1 is fused to an activation domain and Swi6 is fused to a DNA binding domain detect a direct interaction between the two proteins (Sorida et al., 2019; Braun et al., 2011). The two hybrid assay also detects a loss of interaction between Epe1 JmjC mutants and Swi6 (Sorida et al., 2019). Therefore, tethering Swi6 to chromatin can recruit Epe1 to activate reporter gene expression in budding yeast cells which lack H3K9 methylation. However, these findings are not at all in conflict with our data. We clearly demonstrate the existence of a weak interaction between Epe1 and Swi6 and then go on to show that this interaction is enhanced in the presence of H3K9 methylation (Figure 2 and Figure 5). Since our reconstitution experiments are performed in the absence of any additional cellular factors, we believe that our measurements a unique feature of Epe1 that cannot be detected using genetic assays.

Furthermore, we designed a completely novel ectopic silencing approach in fission yeast to provide additional evidence for a non-catalytic function associated with the Epe1 C-terminus. Since Epe1 interacts with Swi6 and excludes other effector molecules, we hypothesized that Swi6 mediated heterochromatin initiation would be highly sensitive to Epe1 activity (enzymatic or non-enzymatic). Tethering Swi6 at an ectopic site does not lead to heterochromatin establishment in fission yeast. Deleting epe1D was sufficient to trigger heterochromatin assembly and mediate epigenetic inheritance where silencing is maintained in the absence of the initiator. Next, we expressed the C-terminus of Epe1 (434-948) which lacks the JmjC domain in cells where a TetR-Swi6 CSD fusion has been tethered. The C-terminal fragment expression now completely blocks heterochromatin formation at the ectopic site. These results clearly support a model where the non-enzymatic function of Epe1 relies on its Swi6 interaction. Our in vitroand in vivodata seek to demonstrate how modified chromatin domains provide a context where this interaction is significantly strengthened.

Reviewer #2:

In fission yeast, Epe1 is a putative histone demethylase that plays important roles in regulating the heterochromatin landscape. However, the mechanism by which Epe1 functions is not clear.

In this manuscript, the authors found that point mutations that abolish co-factor binding in the JmjC domain can disrupt Epe1's binding with Swi6 and also reduces Epe1 localization at heterochromatin. in vitro binding assays confirmed that one Epe1 mutant (H297A) has significantly reduced binding with Swi6. Truncation analysis mapped the Swi6 interaction interface to Epe1 433-630, which is outside of the JmjC domain. Moreover, Epe1 433-630 interacts with Swi6 stronger than full length Epe1, which led the authors to hypothesize that full-length Epe1 might adopt an auto-inhibitive conformation to prevent its binding with Swi6. Indeed, the authors found that Epe1's N-terminal region can bind to its C-terminal region. Authors further demonstrated that H3K9me3 is required for Swi6-Epe1 interaction in vivo and that H3K9me3 peptide enhanced Swi6-Epe1 interaction in vitro, but not Epe1-H29&A. These results suggest a mechanism for Epe1 to only interact with Swi6 at heterochromatin regions. Lastly, the authors showed that Epe1 blocks Swi6's interaction with Clr3, an HDAC important for heterochromatin assembly.

Overall, the data quality is high and the proposed model is very interesting. The findings will be of high interest to the heterochromatin field. I would recommend its publication after addressing several major questions.

1) The model suggests that the JmjC domain of Epe1 binds to H3K9me3 and that the H297A mutation affects this binding. Since the authors have recombinant proteins at hand, they should test these interactions.

We tested the direct interaction between Epe1 and an H3K9me3 peptide and H3K9 methylated histones. Our binding assays clearly detect a direct interaction between Epe1 and an H3K9me3 peptide as opposed to an unmodified H3K9me0 peptide. Furthermore, Epe1 selectively interacts with H3K9 methylated histones as opposed to H3K4 methylated histones. These data have now been added to the revised version of the manuscript (Figure 2—figure supplement 1H and I).

2) The model also suggests that in the presence of H3K9me3, interaction between Epe1's N- and C-terminal regions will be reduced. This should also be tested.

To test whether H3K9 methylation releases an auto-inhibitory conformation of Epe1, we measured the interaction between the N- and C-terminal Epe1 fragments in the presence or absence of an H3K9me3 peptide. The addition of an H3K9me3 peptide disrupts the direct interaction between the N- and C-terminal Epe1 fragments in trans. This data has now been added to the revised version of the manuscript (Figure 5A).

3) In Figure 4, there appears to be very weak binding of WT Epe1 with Swi6 without H3K9me3 peptide. This is very different from results shown in Figures 2 and 3. What's the reason for this discrepancy?

We apologize to the reviewer for the lack of clarity in our presentation. This difference is due to the use of chemiluminescent based western blot assays to measure binding efficiencies. We have explicitly stated that these assays are semi-quantitative in the manuscript text although we do obtain trends based on adding different Epe1 protein concentrations. Although samples in the same blot can be compared, we cannot compare samples loaded on different gels.

In Figure 2E where we compare the interaction between wild-type Epe1 and Epe1 H297A we use high exposure times to reveal a difference in binding. To demonstrate a stimulation in the interaction between Epe1 and Swi6 in the presence of an H3K9me3 peptide, we used a low exposure time so that the assay is sensitive to reveal a substantial change in binding upon peptide addition. In this assay as well, the difference between wild-type Epe1 and Epe1 H297A binding to Swi6 is still visible but less prominent (Figure 4E).

We have now included an additional statement in our Materials and methods section regarding how we choose exposure times to capture differences in binding interactions. The Materials and methods section describing how we setup our binding assays has been amended as follows:

“The exposure times for the interaction assays were chosen and differ in each experiment in order to capture differences in the interaction between Epe1 and Swi6 depending on the assay conditions. Assays performed on different blots cannot be compared but samples loaded on the same blot can be readily compared to each other”.

4) Clr3∆ or Sir2∆ has different effects on H3K9me at centromeres, mating type region, and telomeres. If the hypothesis is correct, it is expected that Epe1 localization will change differently at different locations. ChIP analysis of Epe1 will complement Figure 4B and strengthen the hypothesis.

We are very grateful to the reviewer for this suggestion. We deleted clr3∆ in strains expressing Epe1-3XFLAG. We cross-linked cells with formaldehyde and performed ChIP experiments to map Epe1 localization at the pericentromeric repeats, mating-type locus and the telomeres. The dg repeats exhibit a 2-3 fold increase in Epe1 localization in clr3Δcells (Shimada et al.). This increase is consistent with a model where Clr3 and Epe1 trade places at sites of heterochromatin formation. Epe1 localization also exhibits a two-fold increase at the mating type locus in strains where the cenH nucleation site is intact. Deleting clr3Δ causes a complete loss of Epe1 localization at the telomeres consistent with a critical role for the SHREC complex in telomere silencing (Sugiyama et al., 2007).

Reviewer #3:

The putative histone demethylase Epe1 in fission yeast has been shown to localize heterochromatin via interaction with a HP1 homologue Swi6 and antagonize heterochromatin formation probably through its JmjC-domain dependent demethylation activity, though the enzymatic activity has not been detected in vitro. In this manuscript, authors made several interesting observations. (1) JmjC-domain mutants of Epe1 loose interaction with Swi6 and heterochromatin localization. (2) Epe1 interacts with Swi6 via its C-terminal half that does not include JmjC domain. (3) Epe1 N-terminal half including JmjC-domain interacts with C-terminal half in trans. (5) H3K9me peptide enhance interaction between Swi6 and Epe1. (4) Expression of C-terminal half Epe1 antagonize maintenance of heterochromatic silencing with non-enzymatic activity. (6) Mutation of JmjC domain strengthen the interaction between Swi6 and Clr3, a Swi6-binding silencing factor. (7) Artificial recruitment of Clr3 supports maintenance of heterochromatin in the presence of wild type Epe1. Taken together authors claimed that Epe1-Swi6 binding is auto-inhibited by cis-interaction between N-terminal half and C-terminal half of Epe1. The auto-inhibition can be released by H3K9 methylation and Epe1 compete with Clr3 for Swi6-binding resulting in removal of heterochromatin.

We are very grateful to Dr. Murakami for his encouraging and positive comments regarding our manuscript. Furthermore, we were not aware of a publication from his lab (Sorida et al., 2019) relating to the non-enzymatic functions of Epe1 at the time of submission. We have corrected this oversight on our part. We have also highlighted areas of overlap and how both our studies collectively add important mechanistic insight relating to the non-enzymatic functions of Epe1 (albeit using different phenotypic assays). We have attempted to address all of the major revisions outlined in Dr. Murakami’s scholarly critique.

Author's model is novel and interesting in terms of regulatory mechanism of heterochromatin by Epe1. However, their model is not fully supported by their experimental data. There is no direct evidence showing Epe1 N-terminal half and C-terminal half interacts in cis. Their result can be also explained by trans-interaction. Moreover, they did not present direct evidence showing this interaction is inhibitory to Swi6 binding. This could be tested by adding N-terminal half in the in vitro binding assay of Epe1^434-948 and Swi6 (Figure 3A). In addition, If H3K9me peptide release the auto inhibition through Epe1 as author claimed, addition of H3K9me peptide in the in vitro binding assay would disrupt interaction between N- and C-terminal half of Epe1.

We thank Dr. Murakami for his positive feedback about our manuscript. We have added additional experiments that lends strong support to our model. To test whether H3K9 methylation releases an auto-inhibitory conformation of Epe1, we measured the interaction between the N- and C-terminal Epe1 fragments in the presence or absence of an H3K9me3 peptide. First, we purified Epe1 (1-434) from fission yeast cells using 3XFLAG epitope tag. Despite protein expression being low in fission yeast cells, we obtained sufficient quantities of the N-terminal fragment in order to setup in vitrobinding assays. Next, we added a recombinant Epe1 (434-948) protein fragment purified from E. coli. We performed experiments in the presence and absence of an H3K9me3 peptide. The addition of an H3K9me3 peptide disrupts a direct interaction between the N- and C-terminal Epe1 fragments in trans. This data has now been added to the revised version of the manuscript (new Figure 5A). This data strongly supports our hypothesis where an interaction between the Epe1 N- and C-terminal halves is dependent on H3K9 methylation.

Furthermore, I think one of the main point of their model is that competition between Epe1 and Clr3 for Swi6, by which Epe1 inhibits stable maintenance of artificially formed heterochromatin. However, the evidence for this mechanism is not enough. At least, they should check localization of Clr3 at artificial heterochromatin by ChIP analysis. If their model is correct, amount of Clr3 at artificial heterochromatin (-Tet) would increase in epe1∆ or epe1-H297A cells and decreased in the presence of Epe1 or Epe1^434-948.

Our conclusions in this part of the manuscript builds on previous studies revealing an interplay between Epe1 and Clr3 in localizing at sites of heterochromatin formation (Shimada et al., 2009). Inspired by this study, we tested whether Clr3 could interact with Swi6 and how this interaction is altered in an Epe1 JmjC mutant background (Figure 6A).

To test whether Clr3 binding is critical to preserve epigenetic memory, we fused a 3X-V5 epitope tag to the Clr3 C-terminus and performed ChIP experiments in cells expressing WT Epe1 or an Epe1 H297A mutant. The V5 epitope tag was chosen since it is orthogonal to other FLAG tagged constructs that are expressed in this strain background. We performed ChIP experiments using formaldehyde or after adding reactive cross-linkers in addition to formaldehyde. Under both conditions, we were unable to detect Clr3 enrichment at the ectopic site relative to controls where Clr4 is absent. Furthermore, we note that this problem was not unique to the ectopic locus since we also could not detect significant Clr3 enrichment at the dg repeats compared to clr4Δcontrol strains. We compared our studies to those performed by other groups including Dr.Murakami’s lab (Shimada et al., 2010). One important difference is the use a 3X FLAG epitope tag versus a 3X V5 epitope tag.

However, we would like to emphasize that our ectopic system is explicitly dependent on Clr3 for heterochromatin establishment. The deletion of clr3Δleads to a complete loss of epigenetic silencing where cells turn white in the absence of tetracycline (Ragunathan et al., 2015). The deletion of Epe1 (epe1D clr3Δ)cannot rescue this loss of silencing. Therefore any circumstance where we see maintenance mandates the presence of Clr3. We fully expect that our ChIP experiments will conform to the proposed model where Epe1 and Clr3 trade places while interacting with Swi6 to preserve epigenetic memory. Additional experiments requested by reviewer 2 (new Figure 6C-E) also support this hypothesis.

Recent paper (Sorida et al., 2019), showed that N-terminal domain that does not include jmjC domain has transcription activation activity and deletion of this domain caused severe loss of Swi6 binding and heterochromatin localization of Epe1. In addition, they showed H297A mutation also caused reduction of Swi6 binding. Probably, this paper published after the submission of this manuscript, but authors should mention and discuss about this paper in the revised manuscript.

We apologize to Dr. Murakami for this oversight. We were not aware of the publication of this manuscript at the time of our submission. In combination with Bao et al., 2019; we believe that the publication of our manuscript will be a timely addition to demonstrate how a broad spectrum of non-enzymatic functions associated with Epe1 affect the establishment and maintenance of epigenetic silencing in fission yeast cells.

[Editors’ note: what follows is the authors’ response to the second round of review.]

Essential revisions:

1) The most important outstanding concern is about the interaction between H3K9me3 and Epe1. This interaction is central to the authors' hypothesis yet is tested only with the entire Epe1 protein and reported in two supplementary panels. One concern is that the model presented by the authors – that Swi6 and Epe1 may not compete for H3K9me tails – depends on the binding affinities of these proteins for H3K9me. There are significantly fewer H3K9me tails (~2000 nucleosomes) in fission yeast than Swi6 molecules (~20,000, Sadaie, 2008), which would argue that only molecules at the affinity of Swi6 for H3K9me2/3, or tighter, could access tails. Co-occupancy of Swi6 and Epe1 of adjacent tails therefore is possible only in a circumstance where the affinity of Epe1 is at least as high as that of Swi6 (which is not terribly high, measured at ~2-15 µM depending on the group), given the substantial Swi6 excess. The easiest way to show that H3K9me binding by Epe1 directs its assembly on chromatin is to mutate the region or residues required for this recognition and test its impact.

Given the many available truncation constructs, the authors should be able to identify the domain that interacts with H3K9me and determine the affinity of the interaction. As previously suggested, the JmjC domain is a good candidate for H3K9me binding. These data could be used to generate a mutant that abolishes binding, and the mutant phenotypes should be tested in vivo. These experiments would be very important to support the authors' conclusions.

We thank the reviewers for their scholarly comments and for highlighting a major point of interest in our manuscript. We have generated new supporting data to highlight the unique H3K9me specific binding properties associated with Epe1. We have also elevated the importance of these findings by relocating the peptide binding and histone binding assay to the main figure panel (new Figure 5).

The previous version of our manuscript established that the full-length Epe1 protein preferentially interacts with an H3K9me3 peptide (Figure 5A) and specifically binds to H3K9me3 histones (as opposed to an H3K4me3 histone, Figure 5B). Now, we have used a C-terminal truncation mutant of Epe1 (Epe1-ΔC) to test whether the H3K9me3 binding activity resides within the N-terminus of the protein. Indeed, the Epe1-ΔC protein (deletion of the Epe1 C-terminus, 600-948 amino acids) exhibits a preferential interaction with an H3K9me3 peptide (new Figure 5—figure supplement 1A) and specifically binds to H3K9me3 histones (new Figure 5—figure supplement 1B). Since the N-terminal fragment of Epe1 includes the putative catalytic JmjC domain, we are fully in agreement with reviewers that this domain might be directly responsible for H3K9me binding.

To purify The JmjC domain of Epe1 and design mutants to test H3K9 methylation binding, we dedicated considerable effort to test different solubility and affinity tags in E. coli for recombinant protein expression. However, despite our best efforts we were unable to purify sufficient amounts of recombinant Epe1 protein required for rigorous affinity measurements using either ITC or fluorescence polarization. Therefore, we are unable to compare and report the absolute H3K9 methylation binding affinities associated with Epe1 and Swi6.

However, we fully recognize the importance of this question raised by the reviewers and endeavored to provide further support for our proposed model. Swi6 is preserved in an auto-inhibited state through interactions between adjacent chromodomains (Canzio et al., 2013). A Swi6 Loop-X mutation (Swi6 R93A K94A) releases the chromodomain from an auto-inhibited conformation enabling increased engagement with an H3K9me3 peptide. Despite the lack of auto-inhibition, the interaction between Epe1 and a Swi6 Loop-X mutant is also stimulated by the presence of an H3K9me3 peptide (new Figure 4—figure supplement 1E). Therefore, Epe1 remains responsive to the presence of an H3K9 methylated peptide even in an experimental context where Swi6 is constitutively released from auto-inhibition.

The central premise of our model is that H3K9 methylation acts via Epe1 to stimulate complex formation with Swi6. Our multi-pronged genetic and biochemical analyses provide strong evidence in favor of this model. We believe that the lack of affinity measurements does not detract from the major conclusions in our revised submission.

2) The presented data clearly indicate that an H3K9me peptide can disrupt the interaction between the N-terminal and C-terminal domains of Epe1. However, the authors have not tested if the addition of the N-terminal half inhibits the interaction between the C-terminus and Swi6. This would provide support for the authors' autoinhibition model.

We thank the reviewers for suggesting a key experiment that would lend additional support for our model of H3K9 methylation dependent regulation of Epe1 activity. Our previous studies demonstrated that the N-terminal fragment of Epe1 (1-434 amino acids) interacts with a C-terminal fragment (434-948 amino acids) in trans. We also observed that the addition of an H3K9me3 peptide but not an H3K9me0 peptide disrupts a trans interaction between the two protein fragments (Figure 5C). We purified limited quantities of the Epe1-N (1-434) fragment using an M2 FLAG antibody. The protein was retained on beads and immediately used for subsequent binding assays. We added recombinant MBP-Epe1-C (434-948) in the presence of 1.5-2 fold molar excess of Swi6 (or control experiments where Swi6 was left out). Our in vitro binding experiments demonstrate that Swi6 could also efficiently disrupt complex formation between the Epe1 N-terminus and C-terminus in trans (new Figure 3B).

3) The presented experiments strongly suggest, but do not prove, an autoinhibitory mechanism, as there are no experiments that directly address Epe1 conformation. Therefore, claims regarding such a mechanism should be softened. The title could be changed to "Interaction with histone H3 methylated at lysine 9 regulates the non-enzymatic functions of the putative histone demethylase Epe1" or something similar, and strong claims like "An auto-inhibitory conformation regulates the non-enzymatic properties of Epe1 through its interaction with Swi6" in the Abstract should be amended or removed.

We appreciate the reviewers’ insights and have amended our manuscript by discussing our results as H3K9 methylation dependent changes in protein-protein interactions between Epe1 and Swi6. Consistent with the suggestions provided, we have revised the title and the Abstract of our revised submission.

Author details

1. Gulzhan Raiymbek

   Department of Biological Chemistry, University of Michigan, Ann Arbor, United States

   Contribution

   Conceptualization, Investigation, Methodology

   Competing interests

   No competing interests declared

2. Sojin An

   Department of Biological Chemistry, University of Michigan, Ann Arbor, United States

   Contribution

   Resources, Investigation, Methodology

   Competing interests

   No competing interests declared

3. Nidhi Khurana

   Department of Biological Chemistry, University of Michigan, Ann Arbor, United States

   Contribution

   Formal analysis, Validation, Investigation

   Competing interests

   No competing interests declared

4. Saarang Gopinath

   Department of Biological Chemistry, University of Michigan, Ann Arbor, United States

   Contribution

   Investigation

   Competing interests

   No competing interests declared

5. Ajay Larkin

   Department of Biological Chemistry, University of Michigan, Ann Arbor, United States

   Contribution

   Formal analysis, Investigation

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-5967-9535

6. Saikat Biswas

   Department of Biological Chemistry, University of Michigan, Ann Arbor, United States

   Contribution

   Investigation

   Competing interests

   No competing interests declared

7. Raymond C Trievel

   1. Department of Biological Chemistry, University of Michigan, Ann Arbor, United States
   2. Department of Biophysics, University of Michigan, Ann Arbor, United States

   Contribution

   Resources

   Competing interests

   No competing interests declared

8. Uhn-soo Cho

   1. Department of Biological Chemistry, University of Michigan, Ann Arbor, United States
   2. Department of Biophysics, University of Michigan, Ann Arbor, United States

   Contribution

   Resources, Investigation, Methodology

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-6992-2455

9. Kaushik Ragunathan

   Department of Biological Chemistry, University of Michigan, Ann Arbor, United States

   Contribution

   Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition, Methodology, Project administration

   For correspondence

   ragunath@umich.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-4776-8589

• 3,581

  Page views

• 539

  Downloads

• 12

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.