Proteasome-dependent truncation of the negative heterochromatin regulator Epe1 mediates antifungal resistance

Epe1 histone demethylase restricts H3K9-methylation-dependent heterochromatin, preventing it from spreading over, and silencing, gene-containing regions in fission yeast. External stress induces an adaptive response allowing heterochromatin island formation that confers resistance on surviving wild-type lineages. Here we investigate the mechanism by which Epe1 is regulated in response to stress. Exposure to caffeine or antifungals results in Epe1 ubiquitylation and proteasome-dependent removal of the N-terminal 150 residues from Epe1, generating truncated tEpe1 which accumulates in the cytoplasm. Constitutive tEpe1 expression increases H3K9 methylation over several chromosomal regions, reducing expression of underlying genes and enhancing resistance. Reciprocally, constitutive non-cleavable Epe1 expression decreases resistance. tEpe1-mediated resistance requires a functional JmjC demethylase domain. Moreover, caffeine-induced Epe1-to-tEpe1 cleavage is dependent on an intact cell-integrity MAP kinase stress signalling pathway, mutations in which alter resistance. Thus, environmental changes elicit a mechanism that curtails the function of this key epigenetic modifier, allowing heterochromatin to reprogram gene expression, thereby bestowing resistance to some cells within a population. H3K9me-heterochromatin components are conserved in human and crop plant fungal pathogens for which a limited number of antifungals exist. Our findings reveal how transient heterochromatin-dependent antifungal resistant epimutations develop and thus inform on how they might be countered.

heterochromatin to reprogram gene expression, thereby bestowing resistance to some cells within a population. H3K9me-heterochromatin components are conserved in human and crop plant fungal pathogens for which a limited number of antifungals exist. Our findings reveal how transient heterochromatin-dependent antifungal resistant epimutations develop and thus inform on how they might be countered.
The overuse in agriculture of antifungal agents, related to those used in treating human fungal infections, has caused progressive increases in resistance in soil-borne fungi 1 . Consequently, clinical treatment of patients exhibiting aspergillosis (Aspergillus), candidiasis (Candida) or cryptococcosis (Cryptococcus) is challenging due to the limited number of effective antifungal drugs and the increasing prevalence of resistance [2][3][4] . A dilemma exists as widespread use of antifungals in agriculture is required to combat major plant pathogens such as Magnaporthe oryzae (rice; blast fungus) and Zymoseptoria tritici (wheat; leaf blotch) in order to enhance crop yields. Optimised crop production is required to meet the nutritional needs of the human population that is estimated to increase by approximately 2 billion in the next 30 years 5 . It is therefore important to understand the mechanisms that allow fungi to adapt and develop resistance to external insults such as antifungals in order to design more prudent interventions in clinical and agricultural settings.
Throughout their evolutionary history fungi have been exposed to challenging environments 6 . Consequently, fungi have developed robust, intricate systems to sense and adapt to 'new' external insults such as antifungal compounds. Altered environments or manipulations are known to allow transcriptional memory and epigenetic inheritance in single-celled organisms such as fungi [7][8][9][10][11][12][13] , and also in multicellular organisms [14][15][16][17] so that resulting DNA and chromatin modifications can persist through cell division, long after the original stimulus has dissipated. Such 'epigenetic memory' encourages heterogeneity with variable persistence in otherwise genetically identical cell populations, providing a bet-hedging strategy that ensures adaptation and survival of a proportion of individual cells upon exposure to new environmental challenges. Thus, transient but metastable epigenetic states can confer a selective advantage to particular cell lineages within a clonal population, while allowing return to the initial normal state once external pressures are relaxed 13 .
In the absence of Epe1 synthetic heterochromatin induced at ectopic chromosomal locations can be transmitted through multiple cell divisions by a read-write mechanism following the release of the heterochromatin-nucleating activity 7,8 . Recently we demonstrated that heterochromatin redistributes in an acute response to caffeine stress 13 . Caffeine treatment resulted in Epe1 down-regulation, suggesting that this allows heterochromatin assembly at beneficial loci. However, the mechanism of Epe1 regulation in response to caffeine stress remained unexplained.
Here we determine how Epe1 levels are down-regulated in response to external stresses caffeine and antifungal agents. Unexpectedly, we find that the N-terminal region of Epe1 is removed in response to insults, resulting in diminished Epe1 heterochromatin nuclear foci, cytoplasmic accumulation and redistribution of heterochromatin to other locations, including those that confer resistance through resulting gene repression. Epe1 cleavage is mediated by E3 ubiquitin ligases and the proteasome in a process resembling regulated ubiquitin/ proteasome-dependent processing (RUP) 33,34 . Constitutive tEpe1 expression results in more H3K9me-heterochromatin, increased associated gene repression and elevated resistance. Epe1 processing is dependent on signalling through the cell integrity MAP kinase stress pathway which therefore modulates heterochromatin-mediated caffeine and antifungal resistance. Epe1-related histone demethylase proteins are broadly conserved, suggesting that similar mechanisms contribute to adaption to external insults and antifungal resistance in pathogenic fungi.

Epe1 exhibits altered mobility upon exposure to stress
Exposure of fission yeast expressing N-terminally 3xFLAG-tagged Epe1 to caffeine results in reduced levels of this key negative regulator of heterochromatin 13 . However, caffeine treatment of cells expressing C-terminally 13xMyc-tagged Epe1 (Epe1-Myc) revealed a faster-migrating form (hereafter tEpe1) and reduced levels of full-length Epe1-Myc (Fig.1a,b). This effect was independent of the tag as a faster-migrating form of C-terminally GFP-tagged Epe1 (Epe1-GFP) was also detected upon caffeine exposure (Extended Data Fig.1a). tEpe1 also appeared upon osmotic (1M KCl) or prolonged oxidative (1mM H2O2) stress (Fig.1b). By comparison, neither amount nor mobility of C-terminally TAP-tagged Lid2 JmjC H3K4 demethylase 35 were altered by caffeine exposure (Fig.1c). Thus, the mobility of a proportion of Epe1-Myc increases in response to several external stresses.
(Extended Data Fig.1b). Thus, stress-induced tEpe1 is unlikely to result from a downstream transcriptional start site (TSS). Indeed, no change in epe1 + TSS usage in stress was previously detected 36 .
Mass spectrometry analysis of affinity-selected Epe1-GFP indicated that several residues gained phosphorylation following caffeine treatment (Extended Data Fig.1c). However, the mobility of neither normally migrating nor tEpe1 was altered by lambda phosphatase treatment, suggesting that phosphorylation does not cause the stress-mediated increase in Epe1 mobility (Extended Data Fig.1d).
To investigate if exposure to caffeine results in post-translational cleavage of Epe1, a strain expressing Epe1 tagged at both the N-terminus (3xMyc) and C-terminus (GFP) from the endogenous locus was constructed (Myc-Epe1-GFP). Anti-GFP westerns following caffeine treatment detected tEpe1-GFP but only the normal, slower-migrating form was detected by anti-Myc western, albeit it at a reduced level (Fig.2a). This observation suggests that in response to stress a region containing the N-terminal 3xMyc tag is cleaved away from Myc-Epe1-GFP to leave both full-length (FL-) Epe1-GFP and truncated tEpe1-GFP detectable.

Epe1 has a portable signal for stress-mediated processing
To investigate the requirements for stress-induced Epe1 cleavage strains expressing a series of N-terminal truncations of Epe1-GFP were generated (Fig.2b). Epe1ΔN50-GFP, Epe1ΔN100-GFP, Epe1ΔN150-GFP, and Epe1ΔN200-GFP truncated proteins were expressed at similar levels to FL-Epe1-GFP. Following caffeine treatment, cells expressing Epe1-GFP, Epe1ΔN50-GFP, or Epe1ΔN100-GFP exhibited faster-migrating forms. However, no caffeine-induced change in Epe1ΔN150-GFP or Epe1ΔN200-GFP mobility was observed (Fig.2b), suggesting that a region required for caffeine-induced cleavage resides between Epe1 residues 100 and 150.
To determine if the N-terminal region is sufficient to mediate caffeine-induced cleavage the first 150 residues of Epe1 were fused to GFP (Epe11-150-GFP; Fig.2e). Treatment with caffeine released an entity from Epe11-150-GFP that co-migrated with recombinant GFP (27kDa, Fig.2f). We conclude that the first 150 residues of Epe1 are necessary and sufficient to drive cleavage of a heterologous protein in response to external stress.

Caffeine-induced Epe1 cleavage is proteasome dependent
To gain insight into the mechanism of Epe1 stress-mediated cleavage we affinity selected Epe1-GFP from cells grown with or without 14mM caffeine and applied proteomics to identify associated proteins. Notably, caffeine treatment resulted in increased association of 23 of 35 known proteasome subunits 38 , representing components of both the 20S core (9) and 19S regulatory (14) particles, with Epe1-GFP (Fig.3a). Caffeine-induced cleavage of 3xMyc-Epe1-GFP should release an N-terminal product of approximately 180 residues with a predicted mass of ~20kDa (including 3xMyc). However, no such product was detected, suggesting that the N-terminal fragment is rapidly degraded by the proteasome following its initial cleavage (Extended Data Fig.3a). To test proteasome involvement in stressmediated post-translational processing of Epe1 we examined Epe1-13xMyc processing in cells harbouring the mts2-1 mutation in the Rpt2/ATPase subunit of the 19S proteasome regulatory particle 39 . tEpe1 was generated in wild-type but not mts2-1 cells grown at the permissive temperature, indicating that proteasome function contributes to caffeinedependent Epe1 cleavage (Fig.3b). A shorter Mst2 histone acetyltransferase (HAT) isoform is made in response to stress by a distinct mechanism involving use of a downstream TSS 13,36 . Production of N-terminally truncated Mst2-13xMyc in caffeine was unaffected by the mts2-1 mutant: indicating that proteasome-mediated processing is Epe1 specific and that mts2-1 does not prevent other changes in response to caffeine stress (Extended Data Fig.3b).
The proteasome inhibitor bortezomib primarily inhibits the Pts1/β5 peptidase of the 20S core particle 40 . Addition of bortezomib prevented processing of Epe1-Myc to tEpe1 in the presence of caffeine, and also inhibited caffeine-induced release of GFP from Epe1 1-150 -GFP (Fig.3c,d). In contrast, bortezomib did not inhibit caffeine-induced Mst2/HAT truncation (Extended Data Fig.3c), indicating that stress-induced protein truncation is not generally inhibited by defective proteasome function.
Epe1 is a nuclear protein concentrated in 2-4 heterochromatic foci representing clustered centromeres, telomeres and the silent mating type locus 28,41 . Treatment of wild-type cells with caffeine resulted in loss of all major nuclear Epe1-GFP foci but not the distinct centromere-specific histone CENP-A Cnp1 signal (Extended Data Fig.3d,e). Nuclear Epe1-GFP accumulated in proteasome compromised mts2-1 cells (permissive temperature), consistent with the known Cul4-Ddb1 Cdt2 -mediated ubiquitin-dependent removal of Epe1 from within major blocks of heterochromatin 42 . Although nuclear Epe1 levels were somewhat decreased by addition of caffeine to mts2-1 cells, substantial amounts remained.
Thus, proteasome function, and presumably proteasome-dependent Epe1 processing, is required to deplete nuclear Epe1 in response to caffeine stress.
Following caffeine treatment increased ubiquitin associated with affinity-purified Epe1-Myc and Epe1-Myc was enriched on ubiquitin-binding beads (Fig.3e,f). Thus, ubiquitylation of Epe1 is associated with Epe1-to-tEpe1 caffeine-induced proteasome-dependent processing. The Cul4-Ddb1 Cdt2 E3 ubiquitin ligase targets Epe1 for degradation within major heterochromatin domains 42 , however, Ddb1 Cdt2 loss had little impact on Epe1-to-tEpe1 processing (Extended Data Fig.3f) suggesting involvement of other E3 ligases. Our proteomics analysis of affinity-selected Epe1-GFP detected enrichment of the Hul5 and Pub3 E3 ubiquitin ligases in the presence or absence of caffeine, respectively (Fig.3a). Cells lacking Hul5 exhibited decreased caffeine-induced Epe1-to-tEpe1 processing, whereas deletion of pub3 + had little impact alone or in combination with hul5Δ. Thus Hul5, but not Pub3, E3 ubiquitin ligase contributes to proteasome-mediated Epe1 processing in response to caffeine (Fig.3g). Consistent with decreased Epe1-to-tEpe1 processing in response to external stress, both hul5Δ and hul5Δ pub3Δ cells, but not pub3Δ or ddb1Δ cells, exhibited a lower frequency of caffeine and fluconazole resistant colonies compared to wild-type cells (Fig.3h). We conclude that Hul5 E3 ubiquitin ligase contributes to the removal of the N-terminal region from Epe1 to generate tEpe1 by promoting its ubiquitin-dependent proteasome-mediated processing, however this does not rule out the possible involvement of other E3 ligases in Epe1-to-tEpe1 processing.

N-terminally truncated Epe1 accumulates in the cytoplasm
To further investigate the cellular location of processed Epe1-GFP we utilized constitutively expressed N-terminal truncations. Epe1ΔN100-GFP gave a similar pattern and nuclear signal intensity to FL-Epe1-GFP. Removal of an additional 50 (Epe1ΔN150-GFP) or 100 (Epe1ΔN200-GFP) residues resulted in complete loss of nuclear GFP foci, reduced nuclear signal intensity, with a corresponding increase in cytoplasmic signal intensity (Fig.4a,b). These analyses indicate that the N-terminal 150 residues of Epe1 are required to maintain normal nuclear Epe1 levels and its concentration at heterochromatin domains, and that their removal increases the cytoplasmic location of tEpe1.
Lysates from cells expressing Epe1-Myc grown with or without caffeine were fractionated into soluble cytosolic supernatants and insoluble chromatin-containing pellets (Fig.4c,d). As expected, most α-tubulin was released into the supernatant whereas chromatin (histone H3) was retained in the pellet. Without caffeine most Epe1 stayed in the insoluble pellet, as expected for this chromatin-associated protein. Upon caffeine-induced processing, most tEpe1 was present in the soluble fraction whereas unprocessed Epe1-Myc remained in the pellet. This finding indicates that caffeine-induced removal of Epe1's N-terminus results in its loss from chromatin. In agreement with this observation, constitutively truncated Epe1ΔN150-GFP was detected mainly in the soluble fraction and was unaffected by caffeine treatment. Moreover, comparison of full-length Epe1-GFP and Epe1ΔN150-GFPassociated proteomes revealed that they occupy distinct cellular compartments (Extended Data Fig.4a,b). Epe1-GFP associated with several other chromatin-associated nuclear proteins, including histones, the chromodomain protein Swi6, and the bromodomain proteins Bdc1 and Bdf2. In contrast, constitutively truncated Epe1ΔN150-GFP (equivalent to the caffeine-induced tEpe1-GFP) was associated with cytoplasmic proteins, consistent with a more cytosolic location.
To minimize the deleterious effects of caffeine-induced stress on cell morphology and cytology, the location and cytoplasmic/nuclear signal intensity of Epe1-GFP was assessed in the presence and absence of a lower caffeine concentration (7mM). Under these less stressful conditions less Epe1-to-tEpe1 processing occurs than at higher caffeine concentrations (Fig.1e), and most Epe1-GFP remained in the nucleus, nevertheless, cytoplasmic Epe1-GFP signals were elevated (Fig.4e,f). Moreover, employment of the significantly brighter Halo-tag in place of GFP confirmed FL-Epe1-Halo to be primarily nuclear whereas Epe1ΔN150-Halo was distributed throughout the nucleus and cytoplasm (Fig.4g). Together these data indicate that stress-mediated Epe1-to-tEpe1 processing, for which Epe1ΔN150-GFP and Epe1ΔN150-Halo provide constitutively expressed proxies, results in loss of tEpe1 from chromatin and its increased presence in the cytoplasm.

Epe1 N-terminal removal increases heterochromatin formation
Exposure of fission yeast to caffeine reduces Epe1 association with constitutive heterochromatin regions, alters the distribution of H3K9me-dependent heterochromatin and instigates caffeine resistance through the repression of underlying genes 13 . We next investigated if caffeine-induced N-terminal Epe1 truncation results in changes to its chromatin distribution and that of heterochromatin. Silencing assays at the euchromatinheterochromatin border on the left side of cen1 revealed that, as in epe1Δ cells, repression of the nearby inserted euchromatic ura4 + marker gene occurred in cells expressing Epe1ΔN150 (growth on counter-selective FOA plates; Extended Data Fig.5a). Quantitative ChIP showed that Epe1-GFP was enriched at centromere repeats (dg), a cen1 adjacent boundary region (IRC1-L) and a region near a telomere (tlh2), where Epe1 normally associates 22,42 (Fig.5a). In contrast, removal of Epe1 N-terminal 150 residues (Epe1ΔN150-GFP) resulted in loss of association with these regions (Fig 5a). As expected, epe1Δ cells exhibit higher H3K9me2 levels at known facultative heterochromatin islands 22 (isl3, isl6; Fig.5b). However, cells expressing only Epe1ΔN150-GFP exhibited greater H3K9me2 levels at several islands (isl1, isl3, isl6, isl9). Importantly, cells expressing non-cleavable Epe1ΔN101-110-GFP exhibited no change in its association with major heterochromatin regions compared to Epe1-GFP before or after caffeine treatment (dg, IRC, tel2R; Fig.5a,c) nor in H3K9me2 levels at dg centromere repeats or any island examined (Fig.5b). Thus removal of the Epe1 N-terminal region mimics caffeine treatment 13 and Epe1 loss 22 in increasing H3K9 methylation-dependent heterochromatin at various chromosomal locations.
ChIP-seq analysis confirmed that H3K9me2 levels increase at sub-telomeric and island regions in epe1Δ null and Epe1ΔN150-GFP expressing cells (Fig.5d,e). Consistent with ChIP-qPCR analysis (Fig. 5b), ChIP-seq also shows that at some locations heterochromatin spreads further in Epe1ΔN150-GFP cells relative to epe1Δ cells, indicating that Epe1ΔN150-GFP somehow promotes increased heterochromatin formation (tel1L, isl9/mei4; Fig.5d,e). This increased spreading is particularly prominent at sub-telomeric regions and RNA-seq shows that many telomere proximal genes display even greater repression in Epe1ΔN150-GFP cells compared to epe1Δ cells (Extended Data Figure 5b,c). Therefore, since Epe1ΔN150-GFP cells are not equivalent to epe1Δ cells, these data reveal that Epe1ΔN150-GFP, a surrogate for tEpe1, retains functions that can regulate heterochromatin distribution.

tEpe1 enhances resistance by forming more heterochromatin
To further explore the influence of Epe1 processing on resistance of fission yeast to caffeine and antifungals we measured the frequency of caffeine resistance amongst viable cells expressing truncated or mutant forms of Epe1. Like epe1Δ, cells expressing Epe1ΔN150-GFP displayed an increased frequency of caffeine-resistant colonies relative to cells expressing wild-type Epe1-GFP or Epe1ΔN100-GFP (Fig.6a). In contrast, cells expressing non-cleavable Epe1Δ101-110-GFP, which in response to caffeine remains enriched at major heterochromatic regions (dg, IRC-1L, tel2R; Fig.5c), exhibited greater sensitivity to caffeine than wild-type cells (Fig.6b). Removal of Clr4 H3K9 methyltransferase from epe1Δ, Epe1ΔN100-GFP or Epe1ΔN150-GFP cells reduced the frequency of caffeine-resistant colonies, demonstrating that heterochromatin is required to confer resistance 13 (Fig.6a). Thus, Epe1 N-terminal truncation to residue 150 results in its increased cytoplasmic location and phenocopies the H3K9me heterochromatin-dependent resistance phenotype of epe1Δ cells. Reciprocally, non-cleavable Epe1Δ101-110-GFP remains chromatin associated in the presence of caffeine, thus the reprogramming required to elicit resistance cannot take place, rendering these cells super-sensitive.
ChIP-qPCR performed on caffeine-resistant isolates from Epe1ΔN150-GFP cells revealed significantly higher H3K9me2 levels over the isl14/ncRNA394 locus in eight of ten isolates analyzed (Fig.6c). Previously we showed that caffeine resistance directly results from heterochromatin-associated gene repression over isl14/ncRNA394 13 . We therefore conclude that cells expressing constitutively truncated Epe1ΔN150-GFP exhibit increased caffeine resistance due to similar heterochromatin-mediated epimutations.
Caffeine resistant epimutants were previously shown to also exhibit resistance to clinical and agricultural crop-plant antifungals 13 . As with caffeine and salt stress (Fig.1b), treatment with a clinical antifungal (fluconazole, FLC), or crop-plant fungicides enilconazole (ENL) or tebuconazole (TEB), induced Epe1-to-tEpe1 processing (Fig.6d). Moreover, N-terminally truncated Epe1ΔN150-GFP, but not Epe1ΔN100-GFP, increased the frequency of colonies resistant to these antifungals to levels similar to epe1Δ cells (Fig.6e,f). Removal of Clr4 from epe1Δ, Epe1ΔN100-GFP or Epe1ΔN150-GFP cells reduced the frequency of fluconazole-resistant colonies, indicating that this increase in anti-fungal resistance also requires H3K9me-dependent heterochromatin (Fig.6g).
The phenotype of cells expressing constitutively truncated Epe1ΔN150-GFP appears similar to cells completely lacking Epe1 (epe1Δ) in that heterochromatin accumulates at particular chromosomal locations and both exhibit a Clr4-dependent increase in caffeine and antifungal resistance. However, insult-induced tEpe1 and Epe1ΔN150 retain an intact JmjC demethylase domain and accumulate in the cytoplasm (Fig.4). Moreover, cells expressing Epe1ΔN150-GFP exhibit elevated H3K9me2 levels at certain locations (Fig.5b,d). To determine if the JmjC domain of Epe1ΔN150 contributes to their resistance we constructed cells expressing a predicted catalytically dead version of FL-Epe1-GFP and Epe1ΔN150-GFP bearing H297A and K314A substitutions to disrupt iron/Fe(II) and 2-oxoglutarate binding, respectively (Epe1-cd-GFP, Epe1ΔN150-cd-GFP). Both Epe1-cd-GFP and Epe1ΔN150-cd-GFP were expressed from the epe1 locus at similar levels to their catalytically active counterparts (Extended Data Fig.6a). However, cells expressing Epe1-cd-GFP (naturally induced tEpe1) or Epe1ΔN150-cd-GFP (constitutive tEpe1) displayed reduced frequencies of caffeine and fluconazole resistant colonies relative to Epe1-GFP and Epe1ΔN150-GFP cells (Extended Data Fig.6b). Since tEpe1/Epe1ΔN150 are more cytoplasmically localized, their intact JmjC demethylase domains may promote heterochromatin-mediated resistance via other, perhaps cytoplasmic, substrates. Detailed future investigation will be required to identify potential additional Epe1 substrates and how they might contribute to resistance phenotypes.

A MAP kinase pathway regulates stress-induced Epe1 cleavage
TOR and MAP kinase-dependent signalling pathways sense and mediate adaptation of fission yeast to the external environment. TOR serine/threonine kinase dependent pathways are major regulators of cellular metabolism, growth and cell cycle in response to nutrient availability and other stresses 43,44 . TORC2 signalling acts through the single downstream Gad8 effector kinase which is chromatin-associated and known to influence heterochromatin integrity 45,46 (Extended Data Fig.7a). However, gad8Δ cells exhibited relatively normal Epe1-to-tEpe1 processing but a reduced frequency of caffeine resistant colonies, suggesting that TORC2-mediated signalling may counteract the effects Epe1 processing has on heterochromatin (Extended Data Fig.7b,d). Indeed, TORC2/Gad8 is known to antagonise Epe1 function by promoting heterochromatin formation 45 .
Two fission yeast MAP kinase signalling pathways elicit responses to extracellular changes, resulting in altered gene expression enabling cell adaptation and survival [47][48][49] . The general stress-activated (SAP, Fig.7a) and the cell integrity (CIP, Fig.7b) pathways are interconnected,and signal through Sty1 and Pmk1 MAP kinases, respectively, to activate the key downstream transcription factor Atf1. Loss of Pek1/MAPKK or Pmk1/MAPK, but not Sty1/MAPK, dramatically reduced tEpe1-GFP generation in the presence of caffeine, retaining FL-Epe1-GFP levels similar to wild-type (Fig.7c,d). Cells lacking Atf1 also retained high levels of FL-Epe1, although some cleavage to tEpe1 was apparent. In contrast, cells lacking the Pmp1 phosphatase, which inhibits Pmk1, displayed even lower levels of FL-Epe1 than wild-type cells because more caffeine-induced tEpe1 was evident (Fig.7d). These results indicate that the CIP pathway is mainly responsible for regulating Epe1 processing in response to external caffeine. Epe1-to-tEpe1 processing in response to fluconazole also required intact CIP/MAPK signalling but in this case Atf1 was required for tEpe1 generation (Extended Data Fig.7c); cells lacking Atf1 displayed an equivalent frequency of caffeine-resistant colonies but higher frequency of fluconazoleresistant colonies, compared to wild-type cells (Extended Data Fig.7d,e). These observations suggest that caffeine and fluconazole activate Epe1 cleavage through overlapping routes.
Consistent with Pmk1 and Pmp1 promoting and hindering stress-induced Epe1 cleavage respectively, Epe1-GFP association with major heterochromatic regions in response to caffeine decreased similarly in wild-type and pmp1Δ cells (more cleavage), but not pek1Δ cells (less cleavage) (cen dg, IRC1-L; Fig.7f). Moreover, as in epe1Δ cells, pmp1Δ cells exhibited an increased frequency of caffeine and fluconazole resistant colonies compared to wild-type (Fig.7e,h), consistent with increased Epe1-to-tEpe1 processing ( Fig.7d; Extended Data Fig.7c). Reciprocally, loss of the upstream Pek1/MAPKK resulted in reduced frequencies of caffeine-and fluconazole-resistant colonies. Double mutant pek1Δ epe1-ΔN150 cells exhibited a similar frequency of caffeine and fluconazole resistant colonies as epe1-ΔN150 cells. Thus, Pek1 function is bypassed by constitutive Epe1 truncation, confirming that Pek1 acts to promote resistance through Epe1-to-tEpe1 processing (Fig.7g). We conclude that the CIP/MAPK pathway is required to sense and signal the presence of external stresses that can, in part, be overcome by triggering proteasome-mediated Epe1 processing, its loss from chromatin and thus increased heterochromatin-mediated gene repression which confers resistance 13 (Fig.8).

Discussion
We have uncovered an unusual mechanism by which external stresses alter the location of the pivotal heterochromatin regulator Epe1 by inducing its proteasome-mediated processing to a shorter form lacking the N-terminal 150 residues. As previously demonstrated 13 , epimutations arise due to increased heterochromatin formation over genes at various locations, allowing the survival of cells where reduced expression of underlying genes imparts beneficial resistance phenotypes in the face of external challenges. Such genes include hba1 + , encoding a factor involved in nuclear export of the Pap1 stress-induced transcription factor, and cup1 + , encoding a mitochondrial LYR domain protein.
Proteasome-mediated ubiquitin-dependent protein degradation is well known. However, it is less appreciated that some proteins are subject to proteasome-mediated processing in which only part of a target protein is degraded in a process termed regulated ubiquitin proteasome-dependent processing (RUP) 33,[50][51][52][53][54][55][56][57][58] . Mono-ubiquitylation of target proteins can be sufficient to stimulate such processing events 59,60 which frequently terminate at nearby tightly folded structural domains 37,56 . Thus, elevated Epe1-Myc associated ubiquitylation and enrichment of 23 proteasome subunits with Epe1-GFP upon caffeine treatment (Fig.3a, Extended Data Fig.3d) suggest that the translocating activities of the 19S regulatory particle may stall at the JmjC domain. Subsequent cleavage and removal of the N-terminal 150 residues by peptidases within the 20S proteasome core would produce truncated tEpe1 (Fig.8). Proteasome-dependent processing is known to result in ill-defined cleavage sites in substrate proteins due to bidirectional proteasome processivity following initiating cleavage events in the proteasome's catalytic chamber 37 . Thus, our inability to detect a precise cleavage site is consistent with Epe1 cleavage being mediated by the proteasome. Such proteasome-mediated Epe1 processing may be relevant to the observation that 19S AAA-ATPase mutants, that do not affect ubiquitin-dependent proteolysis, exhibit an altered heterochromatin distribution 61 .
Epe1-to-tEpe1 processing requires an intact cell integrity signalling pathway (CIP/MAPK) to communicate the presence of external insults (Fig.7). Both Pmk1/MAPK and Pek1/ MAPKK are needed for stress-induced Epe1 truncation whereas the counteracting MAPK phosphatase Pmp1 acts to restrain tEpe1 generation. We conclude that activation of the CIP/MAPK pathway by external insults results in regulated N-terminal truncation of Epe1 by proteasome-mediated cleavage (Fig.8). CIP/MAPK may directly phosphorylate Epe1 triggering its ubiquitylation. Alternatively, it might act equivalently to MAPK/Slt2 Mpk1 , its Removal of residues 1-150 from Epe1 reduces its chromatin association and nuclear levels, increasing its presence in the cytoplasm, thereby permitting H3K9-methylation, and thus heterochromatin, to increase at known facultative heterochromatin islands and other locations (Fig. 5). Constitutive tEpe1 (Epe1ΔN150) expression also increases the frequency of caffeine and antifungal resistance relative to wild-type cells. The majority of Epe1ΔN150 caffeine-resistant isolates examined exhibited high H3K9me2 levels at isl14/ncRNA394 (Fig. 6) where reduced expression of the nearby cup1 + encoded mitochondrial protein confers caffeine and anti-fungal resistance 13 . Mitochondrial dysfunction is known to confer antifungal resistance in fission yeast: resistance may result from increased intracellular oxidative stress and associated Pap1 nuclear import, a key stress-regulated transcription factor which elevates efflux pump expression [63][64][65] . Moreover, higher mitochondrial membrane potential in S. cerevisiae correlates with greater anti-fungal resistance; possibly due to efflux pump upregulation 66 . Thus, pathways that respond to mitochondrial-generated stress appear to confer resistance. Interestingly, the JmjC demethylase domain of constitutively truncated Epe1ΔN150 contributes to resistance since mutants expected to disrupt its catalytic activity display reduced resistance (Extended Data Fig.6b). Given the increased cytoplasmic presence of Epe1ΔN150 (Fig.4a,d), it is possible that processed tEpe1 acts through unknown cytoplasmic substrates that perhaps alter metabolism and the levels of metabolites that contribute to heterochromatin regulation via chromatin modifiers, thereby influencing resistance 67 . Although the frequency of caffeine and fluconazole resistance was lower in cells expressing Epe1ΔN150-cd relative to Epe1ΔN150, it is possible that cytoplasmic tEpe1 is mostly non-functional and that its accumulation is mainly incidental resulting from inefficient proteasomal degradation. The identification of potential cytoplasmic tEpe1 substrates that contribute to resistance will be required to confirm a role for tEpe1 in mediating resistance phenotypes.
Constitutive Epe1 truncation causes H3K9me/heterochromatin to redistribute without any selection imposed by external insults. H3K9me2 increases over genes at various locations including heterochromatin islands and especially telomeres, where gene expression is substantially decreased (Fig.5b,d,e; Extended Data Fig.5). We therefore propose that in wild-type cells continual exposure to insults triggers Epe1-to-tEpe1 proteasome-mediated processing and selects for the survival of cells where the resulting heterochromatin-imposed gene repression confers resistance through epimutations such as those at isl14/ncRNA394 that reduce cup1 + expression (Fig.6c) 13 . Insults that trigger Epe1-to-tEpe1 processing include caffeine, salt, hydrogen peroxide and several clinical and agricultural azole-based antifungal compounds ( Fig.1b; Fig.6d). Hence fission yeast can transiently adapt to, and resist, external insults by altering the properties and location of Epe1.
The results presented identify a route by which wild-type fungal cells use chromatin-based epigenetic reprogramming to generate epimutations that survive environmental challenges such as those posed by antifungals. Unlike genetic mutants, such resistant epimutants remain wild-type and can thus return to the normal baseline wild-type state upon removal of the triggering insult. However, long-term exposure to that insult will impose selection amongst surviving epimutants for irreversible genetic mutations which fix the resistant phenotype but may also reduce overall fitness. Thus, insult-triggered proteasome-mediated Epe1 processing provides a bet-hedging strategy by which wild-type cells can develop transient resistance with only temporary fitness impacts. Such epigenetic regulation may explain in part the prevalence and variability of resistance in pathogenic plant and animal fungi and may also contribute to the dearth of available effective antifungal treatments.
Azole-based antifungal overuse has resulted in increasing incidence of resistance in both clinical and agricultural settings. H3K9-methylation-dependent heterochromatin is conserved in human pathogenic fungi such as Cryptococcus 68 and Aspergillus 69 , which threaten the health of immunocompromised individuals, and in major crop pathogens such as Zymoseptoria tritici 70 and Magnaporthe oryzae 71 which regularly and significantly reduce global cereal grain yields. Our findings suggest that the development of fungalspecific inhibitors of processes that regulate or mediate heterochromatin formation, and do not affect related host activities, could increase the sensitivity of pathogenic fungi to antifungal compounds. Such compounds would be expected to improve the prognosis of affected patients and reduce crop destruction.

Yeast strains and manipulations
S. pombe strains used in this study are described in Supplementary

Protein extraction and western analysis
Cells were treated with 14mM caffeine, 1M KCl, 1mM H2O2 for 16 hours, with fresh H2O2 added every 2 hours to compensate for its decomposition. Protein samples were prepared essentially as previously detailed 3 . Briefly, a 10 ml culture of log phase cells was harvested, resuspended in 1 ml of cold water and transferred to an eppendorf tube. Cells were pelleted and resuspended in 775μl water. 150μl 1.85M NaOH and β-mercaptoethanol 7.5% were added and the samples incubated on ice for 15 minutes. 150μl 55% TCA was added and incubated for a further 10 minutes on ice. Samples were centrifuged at 16,800xg for 10 minutes and pellets were resuspended in HU-buffer (8M urea, 5% SDS, 200mM Tris pH 6·8, 1mM EDTA, with bromo-phenol blue, 1·5 DTT).
Samples were run on a 4-12% NuPAGE Bis-Tris gel (ThermoFisher) and blotted using the MiniBlot Module (Life Technologies) 20V for 1 hour. Western blotting detection was performed using anti-myc 9B11 (Cell Signalling) 1:1000, anti GFP (Roche) 1:1000 and anti-Mouse IgG (whole molecule)-Peroxidase antibody (Sigma) 1:10000. Gels were visualised using the ChemiDoc imaging system (BioRad) and analysed with ImageJ. Where quantified, levels of full-length (FL) Epe1 and truncated tEpe1 were normalised to no treatment and adjusted relative to the α-tubulin loading control. Resulting numbers are an average of 3 biological replicates.

MultiDsk2 Pull Down assay
MultiDsk2 pull-down assays were performed as previously described with slight modifications 4 . Ground cell power was resuspended in D-buffer (150 mM Tris-Acetate pH 7.4, 100 mM potassium acetate, 1 mM EDTA, 0.1% Triton X-100, 10% glycerol, 20 μM MG132 and 50 μM PR-619 Protease Inhibitor (Merck Life Sciences)) and incubated 30' on ice. After centrifugation, 5 mg of protein from each sample was incubated for two hours with purified Multidsk2 recombinant protein bound to glutathione agarose magnetic beads (Sigma G0924). After incubation beads were washed four times with D-buffer and resuspended in NuPage LDS sample buffer (ThermoFisher) and subjected to western analysis as described above.

Serial dilution and colony count assays
Equal amounts of cells were grown in YES without agents then serially diluted five-fold and then spotted onto appropriate media. Cells were grown at 30-32°C for 7 days and then imaged. Images were captured with Epson Scan 2 Perfection V600 (6.4.99). For resistance assays agents were added at the following concentrations; 16

Immunoprecipitation and mass spectrometry analysis
IP and mass spectrometry were performed as previously described 5 . Cell were grown in 4X-YES and 2x10 10 cells were used per immunoprecipitation experiment. After harvesting, cells were washed twice with water and frozen in liquid nitrogen. Frozen cell pellets were ground using Retsch MM400 mill. Ground cells were resuspended in lysis buffer ( Following digestion, samples were diluted with equal volume of 0.1% TFA and spun onto StageTips 6 . Peptides were eluted in 40μl of 80% acetonitrile in 0.1% TFA and concentrated down to 1μl by vacuum centrifugation (Concentrator 5301, Eppendorf, UK). The peptide sample was then prepared for LC-MS/MS analysis by diluting it to 5μl by 0.1% TFA.

LC-MS analyses were performed on an Orbitrap Fusion™ Lumos™ Tribrid™ Mass
Spectrometer (Thermo Fisher Scientific, UK) both coupled on-line, to an Ultimate 3000 HPLC (Dionex, Thermo Fisher Scientific, UK). Peptides were separated on a 50 cm (2 μm particle size) EASY-Spray column (Thermo Scientific, UK), which was assembled on an EASY-Spray source (Thermo Scientific, UK) and operated constantly at 50°C. Mobile phase A consisted of 0.1% formic acid in LC-MS grade water and mobile phase B consisted of 80% acetonitrile and 0.1% formic acid. Peptides were loaded onto the column at a flow rate of 0.3 μl min -1 and eluted at a flow rate of 0.25 μl min -1 according to the following gradient: 2 to 40% mobile phase B in 150 min and then to 95% in 11 min. Mobile phase B was retained at 95% for 5 min and returned back to 2% a minute after until the end of the run (190 min).

Europe PMC Funders Author Manuscripts
Europe PMC Funders Author Manuscripts Survey scans were recorded at 120,000 resolution (scan range 350-1500 m/z) with an ion target of 4.0e5, and injection time of 50ms. MS2 was performed in the ion trap at a rapid scan mode, with ion target of 2.0E4 and HCD fragmentation 7 with normalized collision energy of 27. The isolation window in the quadrupole was 1.4 Thomson. Only ions with charge between 2 and 7 were selected for MS2. Dynamic exclusion was set at 60s.
The MaxQuant software platform 8 version 1.6.1.0 was used to process the raw files and search was conducted against Schizosaccharomyces pombe complete/reference proteome set of PomBase database (released in July 2017), using the Andromeda search engine 9 for the first search, peptide tolerance was set to 20 ppm while for the main search peptide tolerance was set to 4.5 pm. Isotope mass tolerance was 2 ppm and maximum charge to 7. Digestion mode was set to specific with trypsin allowing maximum of two missed cleavages. Carbamidomethylation of cysteine was set as fixed modification. Oxidation of methionine, phosphorylation of serine, threonine and tyrosine and ubiquitination of lysine were set as variable modifications. Additionally, methylation of lysine and arginine, di-and tri-methylation of lysine were also set as variable modifications. Label-free quantitation analysis was performed by employing the MaxLFQ algorithm as described 10 . Absolute protein quantification was performed as described previously 11 . Peptide and protein identifications were filtered to 1% FDR.
For imaging cells with Halo tagged proteins, 250 to 500nM of HaloTag ligand JF646 dye was added to 400μl of log phase cells and incubated at 32°C for 3 hours. Cells were washed with PMG and plated on a PMG agarose pad, dried before addition of a coverslip. Images were taken with an inverted microscope (Nikon TiE) equipped with an EMCCD Camera (iXion Ultra 897, Andor), a SpectraX Line engine (Lumencor) and a 100X Nikon objective (NA 1.45,oil immersion). For HaloTag ligand JF646, a red LED and a CY5 filter cube were used. Images were taken with an EM Gain of 300 and exposure times between 20 and 200ms per frame. For all colour channels, a series of 11 z slices (step size 0.2 μm) were taken to cover the entire depth of the cells. Live cells were kept at 32°C during imaging. All image analysis was performed in FIJI as follows: a region of interest was manually selected (nuclear/cytoplasmic spot). Background intensity for each cell was subtracted from the mean intensity measurement. Significance of the difference between samples was evaluated using Student's t-test.  qRT-PCR histograms represent 3 technical replicates; error bars correspond to standard error of the mean.

Fractionation
Cells were grown in 4xYES and harvested by centrifugation at 3500×g, washed twice with water and flash frozen in liquid nitrogen. Frozen cell pellets were ground using Retsch MM400 mill. Ground cells were resuspended in lysis buffer (10mM Tris pH7.4, 5mM CaCl2, 5mM MgCl2, 50mM NaCl, 0.1% IGEPAL-CA630 and supplemented with protease inhibitor (P8215, Sigma) and 2 mM PMSF After thawing on ice for 30 minutes, sample were vortexed 15 times for 30", with one minute on ice between each wash. After centrifugation at 5,000xg, supernatant was collected as soluble fraction, pellet was washed twice with same lysis buffer and then collected as insoluble fraction.

ChIP
ChIP experiments were performed as previously described 13

ChIP-seq, RNA-seq and Bioinformatic analyses
Samples were prepared as previously described 15 . ChIP-seq and RNA-seq samples were compared using DESeq2 statistics of read counts per feature, accessed by HTSeq-count and genomic 1kb bins for ChIP-seq or by Salmon-quant and annotated genes for RNA-seq analyses. Features with adjusted p-values <0.1 were selected as significantly changed. Features with log2 (Fold Change) values >1 (RNA-seq) or >2 (ChIP-seq) are presented in figures.

Statistics and Reproducibility
No sample size calculation was performed, the samples sizes used are considered standard in the field and were selected to ensure robust and statistically significant comparisons. Unless otherwise stated the experiments presented were all repeated as three independent biological replicates. There were no data exclusions and findings were reliably reproducible. Samples were not randomized and no blinding was employed.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material. a. Schematic of cells expressing tagged Epe1 from endogenous promoter exposed to various stresses followed by western. b. C-terminally tagged Epe1-13xMyc or α-tubulin western from cells untreated (-)/treated (+) with 14mM caffeine, 1M KCl or 1mM H2O2 for 16h. Western analysis was performed at least twice. c. Lid2-TAP or a-tubulin western from cells untreated(-)/treated(+) with 14mM caffeine. Western analysis was performed at least twice.