(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



The Cross-Regulation Between Set1, Clr4, and Lsd1/2 in Schizosaccharomyces pombe [1]

['Haoran Liu', 'Department Of Biology', 'Wake Forest University', 'Winston-Salem', 'North Carolina', 'United States Of America', 'Bahjat Fadi Marayati', 'Department Of Biochemistry', 'Duke University School Of Medicine', 'Durham']

Date: 2024-01

Eukaryotic chromatin is organized into either silenced heterochromatin or relaxed euchromatin regions, which controls the accessibility of transcriptional machinery and thus regulates gene expression. In fission yeast, Schizosaccharomyces pombe, Set1 is the sole H3K4 methyltransferase and is mainly enriched at the promoters of actively transcribed genes. In contrast, Clr4 methyltransferase initiates H3K9 methylation, which has long been regarded as a hallmark of heterochromatic silencing. Lsd1 and Lsd2 are two highly conserved H3K4 and H3K9 demethylases. As these histone-modifying enzymes perform critical roles in maintaining histone methylation patterns and, consequently, gene expression profiles, cross-regulations among these enzymes are part of the complex regulatory networks. Thus, elucidating the mechanisms that govern their signaling and mutual regulations remains crucial. Here, we demonstrated that C-terminal truncation mutants, lsd1-ΔHMG and lsd2-ΔC, do not compromise the integrity of the Lsd1/2 complex but impair their chromatin-binding capacity at the promoter region of target genomic loci. We identified protein-protein interactions between Lsd1/2 and Raf2 or Swd2, which are the subunits of the Clr4 complex (CLRC) and Set1-associated complex (COMPASS), respectively. We showed that Clr4 and Set1 modulate the protein levels of Lsd1 and Lsd2 in opposite ways through the ubiquitin-proteasome-dependent pathway. During heat stress, the protein levels of Lsd1 and Lsd2 are upregulated in a Set1-dependent manner. The increase in protein levels is crucial for differential gene expression under stress conditions. Together, our results support a cross-regulatory model by which Set1 and Clr4 methyltransferases control the protein levels of Lsd1/2 demethylases to shape the dynamic chromatin landscape.

Histone-modifying enzymes make covalent modifications to histones. These modifications act like chemical tags that can either tighten or loosen the DNA structure, affecting whether genes are turned on or off. In fission yeast, Set1 methylates histone H3 lysine 4, which marks loosely packed DNA and is associated with gene activation; while Clr4 methylates histone H3 lysine 9, which represents a hallmark of the tightly packed DNA and is associated with gene silencing. Here, we show a regulatory relationship between these two enzymes and with two lysine-specific demethylases (Lsd1/2), which can remove the methyl tags added by Clr4 and Set1. Clr4 and Set1 have opposite effects on Lsd1 and Lsd2 protein levels. Clr4 reduces the levels of Lsd1/2, while Set1 promotes their stability. This control over the levels of Lsd1/2 is achieved through the ubiquitin-proteasome-dependent pathway. By studying these interactions, we have uncovered a novel regulatory mechanism that helps fission yeast maintain a balanced level of these active/repressive histone-modifying enzymes. Understanding these intricate regulatory networks offers important insights into gene control and enhances our comprehension of the complex interactions within cells.

Funding: KZR received a federal grant from the National Institute of General Medical Sciences (R15GM 139107-01). The funder's website: https://www.nigms.nih.gov/ . Q.S. is supported by the National Institute of General Medical Sciences (R35GM151089). The funder's website: https://www.nigms.nih.gov/ . JG was supported by the URECA Summer Fellowship from Wake Forest University ( https://ureca.wfu.edu/ ). BFM and BML obtained Fellowships from the Center for Molecular Signaling at Wake Forest University ( https://molecularsignaling.wfu.edu/ ). The funders had no role in the study design, data collection, and analysis, decision to publish, or preparation of the manuscript.

In this study, we generated viable Lsd1/2 mutants with non-functional high-mobility-group (HMG) box domains at the C-terminus, allowing for convenient manipulation in the laboratory. We identified that the chromatin binding of the Lsd1/2 complex is dependent upon the HMG domain of Lsd1/2. Utilizing a yeast two-hybrid approach, we directly detected physical interactions among Lsd1, Phf1, and Phf2. While Lsd2 was co-purified with Lsd1, Phf1, and Phf2, we did not identify direct physical interactions between Lsd2 and Lsd1 or Phf2, suggesting a divergent/non-redundant role for Lsd2. We also uncovered interactions between Lsd1/2 and Raf2 or Swd2, subunits of CLRC or COMPASS complexes, respectively. Interestingly, we observed that Set1 and Clr4 have opposite effects on Lsd1 and Lsd2 protein levels, which is mediated through the ubiquitin-proteasome-dependent pathway. Without Set1, the protein levels of Lsd1/2 are reduced, while the absence of Clr4 enhances Lsd1/2 protein levels. Under heat stress, Set1 is essential for the upregulation of Lsd1/2; these increased protein levels are critical for the differential gene expression observed during heat stress. Additionally, our findings demonstrated that CLRC exerts an antagonistic effect by controlling the protein level of Set1. Set1 protein abundance is controlled by H2B ubiquitination, which facilitates the chromatin-associated activities of Set1 during heat stress. Overall, our results unfold a cross-regulatory mechanism in which the methyltransferases Set1 and Clr4 fine-tune the protein levels of the demethylases Lsd1/2, thus shaping the dynamic landscape of chromatin and maintaining cellular homeostasis.

The protein abundance and activity of Lsd1 and Lsd2 are governed by a large and complicated regulation network [ 77 ]. However, the exact mechanism by which Lsd1/2 undergoes protein degradation remains unclear. Cul4, containing the highly conserved Cullin domain, plays an essential role in assembling the muti-subunit Cullin-RING E3 ubiquitin ligase (CRL) complexes [ 78 , 79 ]. In human cells, CUL4A and CUL4B interact with an adaptor protein DDB1(DNA damage binding protein 1) and their associated factor DDB2 to promote ubiquitination on histones [ 80 – 82 ]. In S. pombe, Cdt2 is the Ddb1-and Cul4-associated factor, and the function of the Cul4-Ddb1 Cdt2 complex has also been reported in protein degradation [ 83 , 84 ]. For example, the degradation of ribonucleotide reductase inhibitor protein, Spd1, requires the Cul4-Ddb1 Cdt2 complex, which is crucial for genome stability and cell differentiation into meiosis [ 85 , 86 ]. Additionally, Cul4-Ddb1 Cdt2 directly recognizes Epe1, an anti-silencing factor and a potential H3K9 demethylase, to promote its polyubiquitination and degradation [ 87 ]. Cul4 is also part of CLRC and functionally correlates with two other components in CLRC: the WD-40 protein, Raf1; and the β-propeller protein, Rik1 [ 41 , 88 ]. Cul4-Rik1-Raf1 structurally resembles Cul4-Ddb1 Cdt2 E3 ubiquitin ligase and exhibits ubiquitin ligase activity in vitro [ 89 , 90 ].

Histone methylation marks are reversible due to the existence of histone demethylases. LSD1/KDM1a was the first identified member of the FAD (flavin adenine dinucleotide)-dependent histone demethylase family in the mammalian system [ 67 , 68 ]. It is a highly conserved mono- and di-methylated H3K4/H3K9 demethylase, which can function as a transcriptional repressor when it removes the methyl groups from H3K4 or a gene activator when it removes the methyl groups from H3K9 in association with the androgen receptor [ 69 ]. LSD2/KDM1b, the mammalian paralog of LSD1/KDM1a, mainly targets H3K4 for demethylation [ 70 , 71 ]. In fission yeast, Lsd1 forms a complex with its paralog Lsd2, and two plant homeodomain (PHD) finger proteins, Phf1 and Phf2 [ 72 – 74 ]. This complex, known as Lsd1/2, SWIRM1/2, or SAPHIRE complex, localizes specifically at heterochromatic loci and the transcription start sites of actively transcribed genes, which indicates its dynamic role in regulating gene expression [ 75 ]. Lsd1/2 complex also maintains the boundary between euchromatin and heterochromatin at the telomeres, presumably by differentially removing methylation marks from histone H3K9 and/or H3K4 [ 76 ]. The revelation of the Lsd1/2 complex prompts us to delve into the noncatalytic functions of both Lsd1 and Lsd2 and the roles of the proteins they are linked with. Nevertheless, the precise architecture of the Lsd1/2 complex remains to be determined.

In S. pombe, the sole H3K9 methyltransferase, Clr4, initiates H3K9 methylation to silence heterochromatin [ 60 , 61 ]. Clr4 forms a protein complex named CLRC (Clr4 methyltransferase Complex), and all components are required for heterochromatin silencing [ 62 , 63 ]. The SET (Su(var)3-9, Enhancer-of-zeste and Trithorax) domain of Clr4 allows it to mediate the “writing” mechanism for heterochromatin assembly at pericentromeric repeats, sub-telomeric regions, and the mating type locus [ 60 , 63 ]. The chromodomain of Clr4 recognizes and binds to methylated H3K9 and promotes the maintenance and spreading of heterochromatin across large chromatin domains [ 63 ]. Heterochromatin is usually correlated with transcriptional or post-transcriptional gene silencing, which limits the accessibility of RNA polymerase and restricts the accumulation of heterochromatic transcripts, respectively [ 64 , 65 ]. Loss of Clr4 leads to the loss of heterochromatin and causes severe gene silencing defects [ 66 ].

In fission yeast, Schizosaccharomyces pombe (S. pombe), Set1 is the sole catalytic unit of macromolecular complex called COMPASS (Complex Associated with Set1) [ 47 , 48 ]. Set1 is responsible for mono-, di- and tri-methylation on H3K4 and is often recruited to the 5′ end near the transcriptional start site (TSS) of actively transcribed genes and associates with transcription elongation machinery [ 49 – 52 ]. Notably, the H3K4 methylation is stimulated by the upstream ubiquitination of H2B K119, which is achieved by HULC (Histone H2B Ubiquitin Ligase Complex) [ 53 , 54 ]. The ubiquitination of H2B leads to a conformational change in COMPASS, thereby promoting the catalytic activity of Set1 by inhibiting the blockage of catalytic modules [ 55 – 59 ].

Controlled protein degradation is vital for the maintenance of protein homeostasis [ 29 , 30 ]. The ubiquitin-proteasome system (UPS) is the major proteolytic system in eukaryotes that marks proteins with a 76-amino acid molecule: ubiquitin [ 31 ]. This degradation pathway is mediated by a multi-step cascade reaction encompassing E1, E2, and E3 enzymes [ 32 – 35 ]. Specifically, the E1 enzymes activate ubiquitin utilizing ATP hydrolysis and transfer it to E2 [ 36 – 38 ]. The E2 enzymes continue to pass the activated ubiquitin to the substrates through interactions with the E3 ligases [ 37 – 41 ]. Compared with E1 and E2, E3 ligases take on more responsibility to recognize the substrates and ensure the ubiquitination of the correct proteins [ 42 , 43 ]. The target proteins are shuttled to the proteasome, which catalyzes the cleavage of the proteins to amino acids and small peptides that can be recycled for new protein synthesis [ 39 , 44 , 45 ]. Previous studies have observed that about 43% of the methylation sites of proteins in budding yeast, Saccharomyces cerevisiae (S. cerevisiae), could be potential target sites for ubiquitination [ 46 ], which indicates the possibility of protein methylation competing with ubiquitination to protect proteins from degradation.

The compartmentalization of eukaryotic genomes inside the nucleus presents a complex packaging paradigm whereby DNA is wrapped around histones, forming nucleosomes, and further packaged into higher-order chromatin [ 1 – 6 ]. Chromatin is organized into either closed (silenced) heterochromatin or open (active) euchromatin domains, which control the accessibility of the transcriptional machinery and, hence, regulate gene expression [ 7 , 8 ]. Histones, including the H3, H4, H2A, and H2B subunits, form the core particle of the nucleosome [ 1 , 9 – 11 ]. The N-terminal tails of histones are post-translationally modified by various covalent modifications such as methylation, acetylation, and phosphorylation [ 12 – 14 ]. Histone-modifying enzymes that add or remove these modifications regulate the dynamic landscape of chromatin [ 14 – 20 ]. One of the most well-studied post-translational histone modifications is the methylation of multiple lysine (K) residues on histones H3 and H4 [ 21 , 22 ]. Even though the methylation of different histone lysine residues may appear chemically similar, the functional consequences of these methylations are complex and associated with distinct chromatin organization [ 23 ]. For example, methylation of histone H3 lysine 9 (H3K9me) is globally associated with heterochromatin and gene silencing, while the methylation of histone H3 lysine 4 (H3K4me) is an indication of euchromatin and active gene expression [ 24 , 25 ]. Therefore, the proper regulation and turnover of histone-modifying enzymes play essential roles in gene expression, genome stability, and cell fate determination [ 26 – 28 ]. The cross-regulation between these enzymes ensures that their activities are fine-tuned and produces a flexible regulatory circuit that achieves cellular homeostasis. Dysregulation of histone modifiers can lead to profound deviations from normal physiological conditions. Specifically, the mis-regulation of these modifiers has been implicated in the pathogenesis of chronic neurological disorders and cancer [ 23 ].

Results

The C-terminal domains are essential for Lsd1 and Lsd2 chromatin binding The complete loss of Lsd1 or Lsd2 results in severe growth defects or causes lethality in the cell, respectively [72,73,91]. Meanwhile, catalytic mutants in the amine oxidase domain of Lsd1 (from amino acids 267–775, the mutation is K603AK604A) and Lsd2 (from amino acids 516–1035, the mutation is K823AK824A) both show no apparent growth defects [91]. These results led us to wonder whether Lsd1/2 proteins have important functions beyond their amine oxidase catalytic activities. To investigate the non-catalytic functions of Lsd1 and Lsd2, we generated two C-terminal truncations named lsd1-ΔHMG (deletion of amino acids 841–1000 including the HMG domain, a High Mobility Group domain implicated in DNA binding) [92] and lsd2-ΔC (deletion of amino acids 1165–1235) (Fig 1A). Notably, deletion of the Lsd2 HMG-domain results in lethality akin to that in the complete loss of Lsd2, which further implies the unknown and important functions of the C-terminus of these two proteins. lsd2-ΔC has a shorter deletion of the C-terminus compared to the truncation of the Lsd2 HMG-domain, which may partially impair the HMG-domain of Lsd2. Both lsd1-ΔHMG and lsd2-ΔC show moderate growth and silencing defects and retain partial or full enzymatic activities (S1 Fig) [91], which demonstrates that the C-terminus, including the HMG domain, possesses functions that are independent of the amine oxidase-related catalytic activity. We tagged FTP (Flag-TEV-Protein A) at the C-terminus in wild-type or mutated Lsd1 and Lsd2 for further precipitation (IgG Sepharose beads) and detected their protein levels using peroxidase anti-peroxidase (PAP) (Fig 1A). The FTP-tagged wild-type Lsd1 and Lsd2 are fully functional, and cells carrying the wild-type FTP-tagged Lsd1 and Lsd2 have no growth defect [91]. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. The function of the C-terminal domains of Lsd1 and Lsd2 and the architecture of Lsd1/2 complex. (A) Gene schematics for full-length Lsd1 and Lsd2 with previously described C-terminus truncation mutants (lsd1-ΔHMG and lsd2-ΔC). Protein domains of Lsd1 and Lsd2 are color-coded, and numerical labels represent the respective protein lengths. Introns in Lsd2 are marked in black. Lsd1, Lsd2, and their mutant forms are all tagged with FTP (Flag-TEV-Protein A) at the C-terminus. (B) Computational analysis of the crystal structure of mammalian LSD1/KDM1a (left, PDB ID: 2Z3Y) and HMG-box protein (HMGB1) (right, PDB ID: 1NHM). Overlapping crystal structures (middle) were generated using Phyre 2.0. (C) ChIP-Seq peak enrichments of Lsd1/2-FTP and their mutants are aligned with genomic regions spanning 500 base pairs upstream of the transcriptional starting site (TSS) to 500 base pairs downstream of the transcriptional termination site (TES). Heatmaps represent the binding of Lsd1/2-FTP and their mutants to the most robustly bound genes. Each line in the heatmap represents a gene. The color scale, from blue to red, indicates increasing enrichment of Lsd1/2-FTP and their mutants at specific genome loci compared to the untagged negative control. On the top, aligned Lsd1 or Lsd2 occupancy values for all selected genes are plotted as moving averages along their relative genomic positions. (D-G) Co-immunoprecipitation followed by western blotting reveals protein-protein interactions between Lsd1 and Phf1 (D) as well as Phf2 (E). Similar interactions were observed between Lsd2 and Phf1 (F) and Phf2 (G). +/- signs indicate the presence or absence of the protein domain. IgG Sepharose beads were used for protein immunoprecipitation. WCE: whole cell extract. Peroxidase anti-peroxidase (PAP) was used in western blot to detect FTP signals. (H) ChIP-qPCR demonstrates the chromatin binding ability of Phf1/2 with or without the C-terminal domains of Lsd1/2 proteins. Phf1/2-myc was immunoprecipitated using an anti-myc antibody after preclearing Protein A with TEV cleavage. Asterisks indicate p-values ≤ 0.05 as determined by a Student’s t-test, comparing the indicated samples with WT values. Error bars represent the standard error of the mean (s.e.m.). Horizontal lines indicate significance between wild-type and mutants. (I) Cartoon illustrates the yeast two-hybrid approach. Gal4-BD is fused with the bait protein, and Gal4-AD is fused with the prey protein. When the bait protein interacts with the prey protein, Gal4-BD and AD come into close proximity, initiating reporter gene expression. (J-K) Verification of reporter gene expression and protein interactions between Lsd1, Phf1, and Phf2 using the yeast two-hybrid system. (J) Medium lacking leucine and tryptophan but containing X-α-gal (-Leu -Trp +X-α-gal). (K) Medium without adenine, leucine, and tryptophan (-Ade -Leu -Trp). (L) The cartoon demonstrates the direct interaction between Lsd1, Phf1, and Phf2 while indicating that Lsd2 does not directly interact with Lsd1 and Phf2. https://doi.org/10.1371/journal.pgen.1011107.g001 The S. pombe Lsd1 protein structure has not been resolved but is conserved to its human homolog LSD1/KDM1a (43% sequence alignment identity). We first performed a computational domain structural analysis to figure out which part of the mammalian LSD1 protein shares similarities with the HMG-box proteins. The overlapping results of the previously reported crystal structure of KDM1a (PDB ID: 2Z3Y) with a typical HMG-box protein (HMGB1)(PDB ID: 1NHM) in humans suggests that the HMG domain of HMGB1 structurally mimics that of the mammalian LSD1 Tower domain (Fig 1B). Since part of the Tower domain of LSD1 contributes to LSD1/CoREST binding to nucleosomes [93], this finding supports the hypothesis that the HMG domain may be essential for Lsd1 binding to chromatin. To confirm this hypothesis, we performed chromatin-immunoprecipitation combined with sequencing (ChIP-Seq) analysis of Lsd1 and Lsd2 in wild-type and their C-terminal domain mutations (lsd1-ΔHMG and lsd2ΔC). IgG Sepharose beads were used to pull down FTP-tagged wild-type Lsd1, Lsd2, and their C-terminal domain mutants. Previous studies have shown that Lsd1/2 proteins bind to the promoters of a few hundred genes [70,73,74], suggesting that Lsd1/2 proteins are selectively recruited to those genes. Our ChIP-Seq analysis yields a highly similar set of genomic loci where Lsd1 and Lsd2 are enriched just upstream of the transcriptional start site (TSS) (Fig 1C). This result indicates that Lsd1 and Lsd2 mostly bind to the promoter region of genes and are likely to cooperate with other transcription factors that are involved in regulating gene expression. While examining peri-centromeric and mating-type regions, we observed a notable reduction in heterochromatic silencing in lsd1 and lsd2 mutant cells [91]. When comparing wild-type Lsd1 and Lsd2 proteins, we found lower enrichments at these constitutive heterochromatic regions in contrast to their higher enrichments at promoter regions (Figs 1C and S2). However, the enrichments of lsd1 and lsd2 C-terminal mutant proteins exhibited a modest increase at specific loci associated with the loss of silencing, in accordance with the reported roles of Lsd proteins in heterochromatic silencing [91]. Notably, the localizations of Lsd1 and Lsd2 are diminished at the promoter regions in the absence of a functional C-terminus (Fig 1C). This result demonstrates that the C-terminal domains of Lsd1 and Lsd2 are involved in their chromatin binding at these regions. Lsd1/2 complex contains two zinc-finger proteins, Phf1 and Phf2, which may also participate in DNA binding. The loss of the C-terminal domains of the Lsd1/2 proteins might destabilize their association with Phf1 and Phf2, thereby affecting the ability of the complex to bind chromatin. We employed FTP-tagged Lsd1 or Lsd2 alleles and combined the tagged allele with Phf1-myc and Phf2-myc. Using co-immunoprecipitation (Co-IP) followed by western blotting, we confirmed the protein-protein interactions between Lsd1 and Phf1 (Fig 1D) and Phf2 (Fig 1E). Additionally, we established that the loss of the HMG-domain of Lsd1 does not affect the interaction between Lsd1 and Phf1 (Fig 1D) or Phf2 (Fig 1E), and hence, the loss of the HMG-domain is not sufficient to disrupt the entire complex. We also observed that the loss of the C-terminus of Lsd2 does not disrupt the interactions between Lsd2 and Phf1 (Fig 1F) or Phf2 (Fig 1G). These results suggest that the C-terminus mutants of Lsd1 and Lsd2 do not affect the structural integrity of the Lsd1/2 complex. Additionally, we tested whether the loss of the C-terminal domains of Lsd1 and Lsd2 would affect the chromatin binding of Phf1 and Phf2. We performed the ChIP assay followed by qPCR for analyzing the localization of Phf1-myc and Phf2-myc at the sah1+ promoter region, one of the most robust binding regions of Lsd1 and Lsd2 in the S. pombe genome (S3 Fig), in wild-type, lsd1-ΔHMG, and lsd2-ΔC cells. Similar to Lsd1 and Lsd2, Phf1 and Phf2 are also enriched at the promoter of sah1+ (Fig 1H). Moreover, these enrichments are diminished when Lsd1 or Lsd2 loses its C-terminal domain (Fig 1H). This result indicates that impaired C-terminal domains of Lsd1 and Lsd2 weaken the chromatin binding of Phf1 and Phf2, further supporting the notion that the HMG domains of Lsd1 and Lsd2 are essential for the chromatin binding of Lsd1/2 complex.

Lsd1 directly interacts with Phf1 and Phf2 In S. pombe, Lsd1 is copurified with Lsd2, Phf1, and Phf2 to form the Lsd1/2, SWIRM1/2, SAPHIRE complex [72,73,75], yet the exact architecture of the complex has not been revealed. To elucidate whether Lsd1 and Lsd2 have direct interactions with Phf1 or Phf2, we employed a budding yeast two-hybrid approach (Fig 1I). The bait proteins are fused with the Gal4 DNA-binding domain (BD), and the prey proteins are fused with the Gal4 activation domain (AD). When the bait protein interacts with the prey protein, the DNA-binding domain and activation domain will be close enough to initiate the reporter gene expression under the control of the Gal4 promoter in S. cerevisiae. Our results are based on two reporter genes: ADE2 and MEL1. ADE2 generates Ade2, which encodes the phosphoribosylaminoimidazole carboxylase involved in the de novo biosynthesis of purine nucleotides. The expression of ADE2 allows yeast cells to grow on minimal medium lacking adenine (-adenine). MEL1 encodes a secreted enzyme, α-galactosidase, which hydrolyzes the colorless X-alpha-Gal in the medium into a blue end product. The expression of these reporter genes allows cells to grow on -adenine medium or turn blue on a medium containing X-alpha-Gal, indicating the direct interaction between bait and prey proteins. The Lsd2-BD and Phf1-BD strains could self-activate the reporter genes without Gal4 activating domain, suggesting that Lsd2 and Phf1 alone could recruit basal transcriptional factors to initiate reporter gene transcription (S4 Fig). Despite these false positive strains, the remaining bait and prey combinations show that Lsd1-BD/Phf1-AD, Lsd1-BD/Phf2-AD, and Phf2-BD/Phf1-AD colonies turn blue on medium with X-alpha-Gal and grow robustly on medium without adenine (Figs 1I–1K and S4 and S5). These results indicate that Lsd1 directly interacts with Phf1 and Phf2, which also physically associate with each other. However, Lsd2 does not physically interact with Lsd1 or Phf2 (Figs 1L and S5).

Set1 and Clr4 oppositely regulate the genomic binding levels of Lsd1 and Lsd2 The genetic and physical interactions made us wonder whether the loss of COMPASS or CLRC would affect the genomic localization of Lsd1 or Lsd2. We performed ChIP-Seq analysis of Lsd1 and Lsd2 in wild-type, clr4Δ, and set1Δ cells (Fig 2G and 2H). We averaged the same genes and genomic regions as those shown in Fig 1C. Loss of Clr4 or Set1 does not affect the overall pattern of genomic localizations of Lsd1 and Lsd2. To our surprise, clr4Δ increases the enrichments of Lsd1/2 while set1Δ decreases the enrichments of Lsd1 at its enriched genomic loci (Figs 2G, 2H, S6 and S7). This finding suggests that clr4Δ increases, while set1Δ decreases, the genomic binding of Lsd1/2 and potentially affects their protein levels.

Clr4 and Set1 antagonistically regulate the protein levels of Lsd1/2 Next, we investigated the protein levels of Lsd1 and Lsd2 in the absence of Clr4 or Set1. We combined the Lsd1/2-FTP with clr4Δ, set1Δ, or both. Our western blot result demonstrates that without Clr4, both Lsd1-FTP and Lsd2-FTP show an increase in protein levels, compared to the wild-type cells (Fig 2I). In contrast, the protein levels of Lsd1-FTP and Lsd2-FTP show a notable decrease in the absence of Set1 (Fig 2I). Additional examination of GFP-tagged Lsd1 and Lsd2 protein abundance in cells using confocal microscopy confirms that Clr4 and Set1 deletion differentially affects Lsd1- and Lsd2-GFP levels (S8 Fig). Moreover, strain backgrounds lacking both Clr4 and Set1 show intermediate protein levels of Lsd1 and Lsd2 (Fig 2I). Consequently, we wondered whether this effect is initiated at the transcription level. Using RNA-seq and qRT-PCR, we investigated the mRNA levels of Lsd1 and Lsd2 in the absence of Clr4 or Set1 (S9 Fig). Lsd1 and Lsd2 mRNA levels do not correspond with alterations in protein levels of Lsd1 and Lsd2 without Set1 or Clr4. To sum up, it is likely that Clr4 and Set1 regulate the protein abundance of Lsd1 and Lsd2 through a post-transcriptional mechanism. Deletion of specific components of COMPASS affects the integrity of the complex and stability of Set1 [31,48,102–105]. The deletion of Spp1, Swd1, Swd2, and Swd3 significantly lowers the protein levels of Set1 [48]. To test whether the decreased Lsd1/2 proteins are due to the loss of specific members of COMPASS, we generated independent deletion mutations of each member of the complex (set1Δ, spp1Δ, swd1Δ, swd3Δ, swd2Δ, ash2Δ, shg1Δ, sdc1Δ) and combined those deletions with Lsd1-FTP and Lsd2-FTP through genetic crossing. For Lsd1, our result shows that the loss of most members of the COMPASS complex, including Set1, Spp1, Swd1, Swd3, Swd2, and Ash2 leads to a significantly decreased amount of Lsd1 protein (Fig 3A). However, the loss of Shg1 and Sdc1 seems to have little or no effect on Lsd1 protein levels. Our result indicates that COMPASS components that alter the protein levels of Set1 also lower the protein level of Lsd1 proteins [48]. A similar pattern was observed in the Lsd2 samples, although the Lsd2 protein levels also decreased with shg1Δ (Fig 3B). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. Clr4 and Set1 antagonistically regulate protein levels of Lsd1/2 proteins. (A-D) Lsd1-FTP (A & C) and Lsd2-FTP (B & D) protein abundance were analyzed by SDS-PAGE and western blot in the loss of indicated COMPASS complex members (A-B) or CLRC complex members and the cul4 mutant (cul4-1) (C-D). Asterisks denote the expected full-length size of Lsd1-FTP (A) and Lsd2-FTP (B). (E) Examination of Lsd1-FTP and Lsd2-FTP protein levels via SDS-PAGE and western blotting in the wild-type, cul4-1, and ddb1Δ backgrounds. PAP was utilized to detect Lsd1/2 protein levels. Mlo3 levels were employed as the loading control. https://doi.org/10.1371/journal.pgen.1011107.g003 Similarly, we generated independent deletions of each member of the CLRC complex (clr4Δ, raf1Δ, rik1Δ, raf2Δ, and cul4-1, which is a cul4 mutant allele previously shown to affect Cul4 functions through a reporter marker disrupting the 3′-UTR region of cul4+) [100] and combined them with Lsd1/2-FTP. The western blot results suggest that the loss of any member of CLRC complex enhances the protein amount of Lsd1 (Fig 3C) or Lsd2 (Fig 3D). Therefore, the integrity of the CLRC complex is required to restrict the protein levels of Lsd1 and Lsd2. Cul4 is the E3 ubiquitin ligase in the CLRC complex [100]. Cul4 also belongs to Cul4-Ddb1Cdt2 E3 ubiquitin ligase complex (CRL4 complex) [83,84]. Thus, we wondered whether Cul4 might also regulate Lsd1/2 protein levels through the Cul4-Ddb1Cdt2 complex. We investigated the protein levels of Lsd1 and Lsd2 without Ddb1. Loss of Ddb1 slightly enhances the Lsd1-FTP level, while having no significant effect on the Lsd2-FTP level (Fig 3E). It is likely that both CLRC and CRL4 complexes regulate Lsd1 protein levels, although CLRC may play a dominant role. Together, our results show that Clr4 restricts the protein levels of Lsd1 and Lsd2, while Set1 promotes them.

The ubiquitin-proteasome system (UPS) controls the protein amount of Lsd1 and Lsd2 In mammals, the LSD1 protein level is balanced by ubiquitination and deubiquitination [106–108]. The lysine or arginine residues of LSD1 could be the target sites for ubiquitination [109,110]. Knowing this, we asked whether Lsd1 and Lsd2 in fission yeast might be degraded by UPS, similar to mammalian LSD1. We conducted a ubiquitination assay followed by pull-down Lsd1-FTP and Lsd2-FTP combined with western blotting. Elevated ubiquitination of Lsd1 (Fig 4A) or Lsd2 (Fig 4B) was detected in the presence of a temperature-sensitive 26S proteasome subunit mutant, mts2-1, which inactivates the proteasome at 33°C [111] and thereby stabilizes Lsd1 and Lsd2 (Fig 4A and 4B). These results provide evidence that Lsd1 and Lsd2 are degraded through the UPS. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 4. Set1-dependent upregulation of Lsd1/2 proteins is crucial for regulating gene expression under heat stress. (A-B) Assessment of Lsd1-FTP (A) and Lsd2-FTP (B) ubiquitination levels in wild-type and mts2-1 strains. Cells were cultured in a rich medium (YEA) at 33°C for 8 hours. (C-D) FTP-tagged Lsd1 (C) and Lsd2 (D) protein abundance at indicated genotypes were analyzed by SDS-PAGE and western blot. The cells were cultured at the permissive temperature (30°C) and heat stress conditions (37°C) for 2 hours after the initial culture at 30°C. (E) Analysis of Lsd1-FTP ubiquitination levels in indicated strains. Cells were cultured in a rich medium (YEA) at 33°C for 8 hours. (F) Evaluation of Lsd1-FTP and Lsd2-FTP ubiquitination levels in indicated strains. Cells were cultured at 30°C in a rich medium and then subjected to a temperature shift to 37°C for 2 hours. (A, B, E, and F) IgG Sepharose beads were used to precipitate Lsd1-FTP or Lsd2-FTP, followed by western blotting using specified antibodies. PAP was employed to detect Lsd1/2 protein levels, and an anti-ubiquitin antibody was used to detect ubiquitin signals. Asterisks indicate the expected size for full-length Lsd1-FTP and Lsd2-FTP. (G) Hierarchical clustering of set1Δ, lsd1ΔHMG, and lsd2ΔC mutants at 37°C based on the similarities in their expression profiles. Gene expression was compared between mutants and wild-type under 37°C conditions using RNA-Seq data. The color scale ranges from blue to red, reflecting increasing fold expression compared to the control. (H) Pearson’s correlation coefficients were translated into color codes, illustrating similarities in gene expression pattern alterations across different genetic backgrounds. The numbers represent the degree of similarity in gene expression patterns. (I) Gene Ontology analysis of overlapping genes based on the functions of their products. The top five terms with the most assigned genes are presented. These terms encompass genes involved in metabolic regulation (89 genes), transcription (43 genes), biosynthetic processes (42 genes), translation (37 genes), and cell signaling during stress (34 genes). https://doi.org/10.1371/journal.pgen.1011107.g004

Set1 contributes to heat-induced upregulation of Lsd1/2 proteins Clr4 and its associated heterochromatin factors are recruited to facultative heterochromatin domains to modulate a genome-wide transcriptional response to suboptimal environmental conditions [112–114]. Set1 plays a major role in ribosomal gene repression during the cellular response to environmental stressors [115] and cooperates with other factors (including CENP-B, and the HDACs: Clr3 and Clr6) to mediate the transcriptional activation of a certain subset of stress-response genes [116]. We next wondered whether Lsd1/2 protein levels are modulated under stress conditions in a Set1- or Clr4-dependent manner. In S. pombe, heat stress (higher than the optimal range of 28–32°C) is extensively studied due to its highly conserved heat stress regulatory pathways, which resemble those found in higher eukaryotes [117]. We shifted the cell culture temperature from permissive 30°C to heat stress conditions (37°C) for one generation time (2 hours). Cells growing consistently at 30°C served as the control. Both Lsd1 (Fig 4C) and Lsd2 (Fig 4D) protein levels drastically increase after heat treatment, although the mRNA levels of Lsd1 and Lsd2 were not significantly altered between wild-type and set1Δ cells (S10 Fig), suggesting that enhanced Lsd1/2 proteins are required for cell survival under heat stress [61]. Protein levels of Lsd1 and Lsd2 are similar between wild-type and clr4Δ cells at 37°C, indicating that Clr4 is not responsible for Lsd1/2 upregulation in this condition (Fig 4C and 4D). However, without Set1, Lsd1/2 protein levels are no longer upregulated during heat stress (Fig 4C and 4D), which implies that Set1 is required to elevate the protein levels of Lsd1 and Lsd2 under heat stress. Indeed, set1Δ cells show severe growth defects at 37°C [47], which is consistent with previously elucidated roles of Set1 under heat stress. We next investigated whether Set1 protects Lsd1 and Lsd2 from ubiquitin-mediated protein degradation, which may be more pronounced at 37°C. Loss of Set1 does not further elevate Lsd1 protein nor ubiquitination levels in mts2-1, suggesting that the Set1-dependent degradation of Lsd1 is dependent on the UPS (Fig 4E). Under heat stress (37°C), it was consistently found that ubiquitination of Lsd1 and Lsd2 is drastically enhanced in set1Δ cells, even without mts2-1 as a background (Fig 4F). In contrast, we only observed a slight decrease in ubiquitination of Lsd1 and Lsd2 in the clr4Δ background (Fig 4F), suggesting that Clr4 likely further destabilizes Lsd1 and Lsd2 at 37°C. In brief, our data indicate that Set1 may protect Lsd1/2 proteins from degradation under heat stress, while Clr4 may participate in their degradation.

Set1-dependent upregulation of Lsd1/2 proteins is crucial for regulating gene expression under heat stress In human cells, overexpression of LSD1 is involved in the proliferation, inhibition of apoptosis, and metastasis of several types of cancers, such as gastric, breast, and prostate cancers [109, 118, 119]. The elevated levels of Lsd1/2 proteins at 37°C suggest that their functions might be essential for cell survival at stressful high temperatures. Indeed, compared to the wild-type cells, the expression of numerous genes is altered in lsd1-ΔHMG and lsd2-ΔC mutants at 37°C, and this alteration is correlated with that loss of Set1 (Fig 4G and S1 Table). Intriguingly, about 72% of downregulated genes in lsd1-ΔHMG, lsd2-ΔC, and set1Δ are antisense non-coding RNAs (S1 Table), indicating that Set1-mediated elevation of Lsd1/2 protein functions to stimulate antisense transcription under heat stress. The gene expression pattern alteration showed high similarity between lsd1-ΔHMG, lsd2-ΔC, and set1Δ during heat stress. At least 70% of genes share similar regulation patterns (between lsd2-ΔC and set1Δ) and the highest similarity can reach 90% (between lsd1-ΔHMG and lsd2-ΔC) (Fig 4H). We further analyzed and classified those differentially expressed genes according to their functional groups. Those genes participate in metabolic pathways, transcription, biosynthesis, translation, and cell signaling, which are closely related to the stress response pathways (Fig 4I and S2 Table). To conclude, Set1-dependent upregulation of Lsd1/2 proteins is critical for controlling gene expression under heat stress.

CLRC controls Set1 protein level In budding yeast, Set1 protein levels are reduced in mutants deficient in Swd1, Swd2, Swd3, or Spp1 [102,104,120]. In S. pombe, Set1 protein amount is differentially affected without COMPASS subunits. Set1 proteins were barely detectable in swd1Δ and swd3Δ and were noticeably reduced in swd2Δ and spp1Δ mutants, but were minimally affected by the loss of Ash2, Sdc1, and Shg1 [48]. Since Cul4 is an E3-ubiquitin ligase, we next investigated whether CLRC may regulate Set1 protein level by controlling Set1 degradation through the UPS, thereby modulating the protein levels of Lsd1/2. We, therefore, visualized the levels of fully functional N-terminal Flag-tagged Set1 (Flag-Set1) in wild-type or CLRC mutant cells [48]. We found that the loss of any members in CLRC promotes the protein levels of Set1, both at 30°C or 37°C (Fig 5A), indicating that the intact CLRC restricts the amount of Set1. Next, we investigated whether chromatin-bound Clr4 is necessary for regulating Set1 protein level. The binding of Clr4 to methylated H3K9 is impaired when its chromodomain of Clr4 (clr4W31G) is mutated [63]. Although no significant alterations were detected for Flag-Set1 protein levels between wild-type and clr4W31G cells at 30°C, the elevated Flag-Set1 is less pronounced in clr4W31G compared to clr4Δ at 37°C (Fig 5B), indicating that chromatin-bound Clr4 may play a role in regulating Set1 protein level, at least under heat stress. We also investigated the Set1 protein level without Ddb1. Both cul4-1 or ddb1Δ enhanced the amount of Set1 protein, indicating that both CLRC and CRL4 complexes modulate Set1 levels (Fig 5C). cul4-1 elevates Set1 significantly at 37°C compared to that at 30°C, which is due to the temperature sensitive nature of cul4-1. It is possible that Cul4 targets Set1 for degradation through the ubiquitin-proteasome system. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 5. CLRC controls Set1 protein level. (A-C) Whole-cell extracts from untagged and Flag-tagged Set1 cells were subjected to SDS-PAGE and western blot using an anti-Flag antibody. The assessed Set1 protein levels in indicated genetic backgrounds at 30°C and 37°C for 2 hours after initial culture at 30°C. Mlo3 levels served as the loading controls. https://doi.org/10.1371/journal.pgen.1011107.g005

[END]
---
[1] Url: https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1011107

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.