sign in sign up
<<
Home , ChIP sequencing, Chromosome regions

Normal conformation depends on subtelomeric facultative heterochromatin in Neurospora crassa

Andrew D. Klockoa, Tereza Ormsbya, Jonathan M. Galazkab,1, Neena A. Leggetta, Miki Uesakac, Shinji Hondac, Michael Freitagb, and Eric U. Selkera,2

aInstitute of Molecular , University of Oregon, Eugene, OR 97403; bDepartment of Biochemistry and Biophysics, Oregon State University, Corvallis, OR 97331; and cDivision of Chromosome Biology, Faculty of Medical Sciences, University of Fukui, Fukui 910-1193, Japan

Contributed by Eric U. Selker, October 21, 2016 (sent for review September 19, 2016; reviewed by Victor Corces and Shiv I. S. Grewal) High-throughput chromosome conformation capture (Hi-C) analy- chromosome regions that predict 3D arrangements (2–4, 6, 9). TADs ses revealed that the 3D structure of the Neurospora crassa ge- within active or repressed genomic zones are typically correlated with nome is dominated by intra- and interchromosomal links between corresponding marks (e.g., or H3K27me2/3, regions of heterochromatin, especially constitutive heterochromatin. respectively) and may constrain (2, 3). To un- Elimination of trimethylation of lysine 9 on H3 () or derstand the basis for the segregation of the genome into functionally its binding partner Heterochromatin 1 (HP1)—both prominent distinct domains, it is desirable to test mutants defective in factors features of constitutive heterochromatin—have little effect on the Hi- thought to be key. Unfortunately, at least in higher cells, such mu- C pattern. It remained possible that di- or trimethylation of lysine 27 tants are usually not viable, and chromatin factors, including meth- on histone H3 (H3K27me2/3), which becomes localized in regions of yltransferases (MTases), are frequently redundant. However, simpler constitutive heterochromatin when H3K9me3 or HP1 are lost, plays a eukaryotes such as fission yeast and N. crassa provide opportunities critical role in the 3D structure of the genome. We found that for such experiments. H3K27me2/3, catalyzed by the Polycomb Repressive Complex 2 N. crassa In , facultative heterochromatin marked by H3K27me2/3 GENETICS (PRC2) member SET-7 (SET domain protein-7), does indeed play a is principally subtelomeric, collectively covering ∼7% of the ge- prominent role in the Hi-C pattern of WT, but that its presence in nome (871 genes completely covered by H3K27me2/3; 215 border regions normally occupied by H3K9me3 is not responsible for main- genes) (10–12). Di- and trimethylated K27 are frequently detected tenance of the genome architecture when H3K9me3 is lost. The Hi-C together by common , but both forms have been detected pattern of a mutant defective in the PRC2 member N. crassa p55 (NPF), by MS and ChIP experiments (12). The relative level of these forms which is predominantly required for subtelomeric H3K27me2/3, set-7 may vary somewhat in different regions; for simplicity, we will wasequivalenttothatofthe deletion strain, suggesting that “ ” subtelomeric facultative heterochromatin is paramount for normal primarily refer to the two forms collectively as H3K27me2/3. chromosome conformation. Both PRC2 mutants showed decreased H3K27me2/3 requires at least three subunits of Polycomb Re- heterochromatin–heterochromatin contacts and increased euchro- pressive Complex 2 (PRC2), including EED, Suz12, and SET matin–heterochromatin contacts. Cytological observations sug- gested elimination of H3K27me2/3 leads to partial displacement Significance of clusters from the nuclear periphery. Transcriptional pro- filing of Δdim-5, Δset-7, Δset-7; Δdim-5,andΔnpf strains detailed Two forms of heterochromatin, constitutive and facultative, cause anticipated changes in gene expression but did not support the idea gene silencing in eukaryotes. In Neurospora crassa, H3K27me2/3- that global changes in genome architecture, per se, led to altered marked facultative heterochromatin reversibly represses scores of . specialized genes, whereas H3K9me3-marked constitutive hetero- chromatin permanently silences repetitive DNA. Interactions be- Hi-C | facultative heterochromatin | H3K27me2/3 | PRC2 | tween heterochromatin provide a structural framework for the Neurospora crassa genome, and this is thought to be functionally important. Histone marks underlying constitutive and facultative heterochromatin are ecently developed methods to globally assess the 2D and 3D nonessential in N. crassa, permitting tests of their roles in genome Rorganization of have shed light upon the im- organization and gene expression. Although linkages between portance of chromatin for genome structure. Chromosome confor- regions of constitutive heterochromatin are the most prominent mation capture coupled with high-throughput sequencing (Hi-C) feature of the 3D structure of the genome, loss of the facultative assesses the pairwise contact probability (i.e., “strength”) between all mark has a much greater effect on genome architecture than does genomic loci (1). The genome organization of several eukaryotes has loss of key features of constitutive heterochromatin, i.e., H3K9me3 been examined by Hi-C, including humans, mice, plants, fruit flies, and Heterochromatin Protein 1. yeasts, and Neurospora crassa (2–7). These eukaryotic genomes are separated into relatively active and inactive zones; the quiescent Author contributions: A.D.K., J.M.G., S.H., M.F., and E.U.S. designed research; A.D.K., T.O., “ ” and J.M.G. performed research; M.U. and S.H. contributed new reagents/analytic tools; heterochromatin includes two distinct subtypes characterized by A.D.K., T.O., J.M.G., N.A.L., S.H., M.F., and E.U.S. analyzed data; and A.D.K., T.O., J.M.G., S.H., specific covalent modifications of . “Constitutive” hetero- M.F., and E.U.S. wrote the paper. chromatin, found, for example, within or near gene-sparse centro- Reviewers: V.C., Emory University; and S.I.S.G., National Institutes of Health. mere regions, is commonly marked by trimethylation of lysine 9 on The authors declare no conflict of interest. “ histone H3 (H3K9me3) and cytosine methylation, whereas faculta- Data deposition: The data reported in this paper have been deposited in the Gene Ex- tive” heterochromatin is generally marked by di- or trimethylation of pression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE82222). lysine 27 on histone H3 (H3K27me2/3) (8). Each Hi-C experiment 1Present address: Space Biosciences Division, NASA Ames Research Center, Moffett Field, has revealed strong intra- and interchromosomal associations be- CA 94035. tween constitutive and facultative heterochromatic domains (1, 3). In 2To whom correspondence should be addressed. Email: [email protected]. addition, some genomes show topologically associated domains This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. (TADs) or globules, defined as interactions or contacts between 1073/pnas.1615546113/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1615546113 PNAS Early Edition | 1of6 Downloaded by guest on September 25, 2021 domain protein-7 (SET-7), which is the histone MTase; a fourth pro- the Δset-7 strain showed greater cross- interactions (Fig. tein, N. crassa p55 (NPF), is required for subtelomeric H3K27me2/3 S2C). Direct comparison of the Δset-7 observed vs. expected heat (11, 13). Although set-7 deletion (Δset-7) strains lack H3K27me2/3, map vs. that of WT showed striking changes in intra- and inter- which derepresses <100 genes, they show no obvious growth defects, chromosomal interactions. The H3K27me2/3-marked consistent with evidence that PRC2 represses ancillary genes (11). In (e.g., Fig. 1 B–D, squares) and centromeric flanks (e.g., Fig. 1 B and contrast, Δnpf strains show stunted growth, presumably because NPF C, black arrowheads) showed decreased contacts. Conversely, the is an important member of histone deacetylase (HDAC) and nucle- mutant broadly showed increased centromere–euchromatin (e.g., Fig. osome remodeling complexes (11, 13, 14). 1 B and C, elongated ovals) and intracentromere (e.g., Fig. 1C,violet In contrast, H3K9me3 enriched in N. crassa constitutive het- arrowhead) contacts. The overall ratio of inter- to intrachromosomal erochromatin depends on all members of a five-protein complex, contactsincreasedintheΔset-7 strain (Fig. S2D). Similar, albeit DCDC [DIM-5 (Defective in Methylation-5), DIM-7, DIM-9, somewhat larger, changes were observed in the genome organization dim-8 CUL4 DDB-1 Complex]; DIM-5 is the single N. crassa ho- of a dim-3 strain (7) (Fig. S2E). Comparison of the Δset-7 and Δdim-5 molog of the H3K9me3 MTase Su(var)3–9 (15, 16). Hetero- (7) datasets (Fig. 1D) shows differences much like those seen in the chromatin Protein 1 (HP1) (17), encoded by the hpo gene, binds Δset-7–WT comparison, but additional centromeric flank contacts H3K9me3 and recruits other to constitutive hetero- (Fig. 1D, black arrow) and decreased interactions between centro- chromatin, including the DNA MTase DIM-2 (18, 19) and an meres and smaller heterochromatic regions (Fig. 1D,oval)were HDAC complex, HCHC (20). H3K9me3 is required for telo- apparent, consistent with the subtle changes observed in the simple meric silencing and largely does not overlap with Δdim-5 strain (7). The constitutive heterochromatin-defined globules (10–12, 21). In strains devoid of H3K9me3 (e.g., Δdim-5) or HP1 (“triangles” of interactions along the x-axis) in WT appear more (Δhpo), H3K27me2/3 moves to former constitutive heterochro- disordered in the Δset-7 strain (compare the globules in Fig. 1 E and matic regions, i.e., , subtelomeres, and islands of F from a sample region). To assess chromosome contact changes in dispersed transposon relicts, whereas most facultative hetero- detail, we plotted the strongest observed vs. expected intra- and in- chromatic regions lose H3K27me2/3 (12, 21). No H3K27me2/3 terchromosomal interactions in WT and Δset-7 strains with Circos redistribution occurs in dim-2 strains, and no ectopic H3K9me3 plots. Subtelomeric contacts and interactions between H3K9me3- localization occurs in PRC2 mutants (12, 21). and H3K27me2/3-marked domains in WT were depleted in Δset-7 Recent Hi-C studies with fission yeast and N. crassa revealed that (Fig. 2 and Fig. S2 F and G). Thus, loss of H3K27me2/3 impacts the loss of H3K9me3 or HP1 minimally altered heterochromatin in- normal N. crassa genome organization. teractions (6, 7). However, an N. crassa strain with a neomorphic importin α (dim-3) allele that substantially enlarges nuclei had re- Ectopic H3K27me2/3 Does Not Substitute for Lost H3K9me3 in a Δdim-5 duced heterochromatin interactions, potentially as a result of re- Background. Considering that most H3K27me2/3 localizes to con- duced –nuclear membrane (NM) association (7). Here stitutive heterochromatic regions after elimination of H3K9me3 we used Hi-C to test the possible role of H3K27me2/3 in genome or HP1 (Fig. 3A and Fig. S3 A and B, dark purple tracks) (12, 21), organization and to test the possibility that the ectopic H3K27me2/3 we asked if this ectopic H3K27me2/3 compensates for the absence in Δdim-5 and Δhpo strains may serve a function normally per- of H3K9me3 with respect to a role in genome structure. Two formed by H3K9me3. We discovered that loss of H3K27me2/3, or biological Hi-C replicates of a Δset-7; Δdim-5 strain gave re- both H3K27me2/3 and H3K9me3, caused an altered genome ar- producible datasets (Fig. S1 E and F). The merged Δset-7; Δdim-5 chitecture consistent with untethered . Selective loss of dataset provided high-resolution information similar to our other subtelomeric H3K27me2/3 in a Δnpf strain produced Hi-C results datasets (40-kb resolution overall and 10-kb resolution for strong similar to those of strains lacking all H3K27me2/3. Thus, sub- contacts; Fig. S1 C and D). The raw Hi-C heat map revealed telomeric facultative heterochromatin is important for maintaining typical interactions between subtelomeres as well as intercentro- the overall genome conformation and seems to play a role in sub- meric interactions on a genome-wide basis (Fig. S3C)thatare telomere–NM association in N. crassa. emphasized upon normalization (Fig. S3D). The Δset-7; Δdim-5 double mutant and the Δset-7 single mutant also gave comparable Results patterns in a Pearson correlation analysis (Fig. S3E). Relative to SET-7 Is Required for Normal Chromosomal Conformation. To assess WT, the Δset-7; Δdim-5 strain gained centromere–euchromatin if SET-7 or H3K27me2/3 plays a role in chromosome confor- contacts and lost subtelomeric contacts (Fig. 3B and Fig. S3F;com- mation, we performed Hi-C on a Δset-7 strain (statistics in Table pare Fig. 3A and Fig. S3 A and B vs. Fig. 2 and Fig. S2 F and G), S1). The genome architecture observed in two biological repli- similar to the Δset-7 dataset. In fact, the Hi-C results for the Δset-7; cates was reproducible (Fig. S1 A and B), allowing us to merge Δdim-5 strain were remarkably similar to those from the single Δset-7 replicates into a single Δset-7 dataset to analyze the entire ge- strain (Fig. 3C), but there were subtle differences in the Δset-7; Δdim-5 nome at 40-kb resolution (≥99% of bins have ≥1,000 mapped dataset. For example, regions flanking constitutive heterochromatin reads) and strong contacts at 10-kb resolution (Fig. S1C). As showed decondensation (Fig. 3C)similartothesingleΔdim-5 mu- expected, the Δset-7 dataset showed a strong inverse relationship tant (7). Of all of the strains examined, intra- and intersubtelomeric between contact probability and genomic distance, similar to the contacts, as well as centromeric flank contacts, were most impacted WT dataset (Fig. S1D), with local contacts appearing as an in- in the Δset-7; Δdim-5 strain (Fig. 3C and Fig. S4). As constitutive tense diagonal (Fig. S2A). In addition, interactions not explained heterochromatin interactions still dominate the Δset-7; Δdim-5 ge- by linear proximity were visible, such as intercentromere and nome structure, similar to Δhpo and Δdim-5 strains (7), ectopic intertelomere contacts (Fig. S2A). To highlight these contacts, H3K27me2/3 at centromeres must not maintain constitutive het- we calculated the median contact frequency at each genomic erochromatic contacts in mutants defective in DCDC or HP1. distance (i.e., the “expected” map) and applied this to our raw N. crassa (i.e., “observed”) data, producing a log2-transformed observed Subtelomeric H3K27 Methylation Is Important for Chromosome vs. expected heat map (7). Similar to the case in WT, Δdim-5, Conformation. We closely examined H3K27me2/3 ChIP-sequencing and Δhpo datasets (7), the Δset-7 genomic heat map showed (ChIP-seq) data to glean possible insight regarding the chromosome prominent interactions among telomeres and between centro- organization changes observed in Δset-7 strains. Constitutive het- meric flanks, and less than expected interactions between cen- erochromatin mutants retain normal levels of H3K27me2/3 at tromeres and euchromatic arms (Fig. 1A and Fig. S2B). A Pearson immediately adjacent to the telomeres, i.e., at correlation analysis revealed that euchromatin interactions are subtelomeres (Fig. S5A) despite the widespread loss of this mark mainly within each chromosome arm in the WT strain; interestingly, from facultative heterochromatin and gain of it in domains of

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1615546113 Klocko et al. Downloaded by guest on September 25, 2021 A Δset-7 LGVII B LGI LGII LGIII LGIV LGV LGVI LGVII 0

0 Distance (Mb) 40 0 2.0

) E

2 7

) 2 2.0 WT LGI

Bins LGVII L arm

0 Hi-C signal (log LGII 0.3 Mb 1.1 Mb Fold change of Δ set-7 relative to WT Fold change of expected (log 100

-2.0 H3K27me2/3 H3K9me3 %GC genes LGIII 04Distance (Mb) C D 0 0 Bins LGIV Bins Bins F 6

) LGV 2 Δset-7

LGVII L arm

0 Hi-C signal (log LGVI 100 100 0.3 Mb 1.1 Mb -2.0 LGVII 04Distance (Mb) 04Distance (Mb) 1000 -2.0 2.0 -2.0 2.0 GENETICS Δset-7 Fold change of Δset-7 relative to Δdim-5 (log ) Fold change of relative to WT (log2) 2 H3K27me2/3 H3K9me3 %GC genes

Fig. 1. Facultative heterochromatin influences genome structure. (A) Heat map of observed vs. expected contact frequency ratio (log2) for a representative chromosome (LGVII) in a Δset-7 strain at 40-kb resolution. Chromosome schematic above and to the right; filled green circles and black rectangle represent telomeres and the centromere, respectively. Centromeric flank (arrowhead) and intertelomere (square) interactions and centromere–euchromatin repulsion

(ovals) are marked, as is the region expanded in E and F (triangle). (B) Heat map of differences in contact frequency (as log2 fold change) of Δset-7 relative to WT (7) for the whole genome. Schematics of the seven chromosomes (LGI–LGVII) as in A; thin blue lines mark empty bins. Examples of decreased centromeric flank (arrowheads) and intertelomere (squares) contacts and increased centromere–euchromatin contacts (ovals) are indicated. (C) Heat map of differences in contact

frequency (log2)ofΔset-7 relative to WT for LGVII. WT ChIP-seq H3K9me3 (green) and H3K27me2/3 (purple) tracks are shown, and genomic feature changes are marked as in B; the increase in internal centromeric contacts (violet arrowhead) is indicated. (D) Heat map of differences in contact frequency of Δset-7 relative to Δdim-5,asinC; the decreased H3K9me3–H3K27me2/3-marked region contact (green arrowhead) is shown. (E and F) Local interaction raw heat maps of ∼0.9 Mb of the left arm of LGVII at 10-kb resolution of (E)WT(7)or(F) Δset-7 strains. ChIP-seq enrichment, %guanine:cytosine (%GC) basepairs, and gene tracks are below.

constitutive heterochromatin (12, 21). Subtelomeres are the only subtelomeres cluster less. That the genome structure of Δdim-5 and regions where H3K9me3, H3K27me2/3, and DNA methylation Δhpo strains is nearly identical to that of WT whereas those from overlap in N. crassa (Fig. S5A). To explore if loss of H3K27me2/3 Δset-7, Δset-7; Δdim-5,orΔnpf strains show considerable disorder from subtelomeres may be responsible for the altered genome or- implies that the presence of H3K27me2/3 at chromosome ends is ganization observed in the Δset-7 and Δset-7; Δdim-5 strains, we important for normal genome organization. Curiously, H3K27me2/3 performed Hi-C on a Δnpf strain, which selectively loses telomere- deposition is normal in dim-3 strains, despite the similarity of proximal but retains centromere-proximal H3K27me3 peaks (Fig. dim-3 and Δset-7 Hi-C patterns (Fig. S5B and S2D). 4A; Δnpf H3K27me3 enrichment begins ∼319 kb from telomeres on average). The combined dataset from merging two replicates displays H3K27me2/3 Loss Results in Telomere Mislocalization. The apparent the whole genome at 50-kb resolution and strong contacts at 10-kb changes in genome structure in strains that lose subtelomeric resolution (Fig. S1 C and D and Fig. S1 G and H). The Δnpf raw H3K27me2/3 might be explained by telomere mislocalization. To examine telomere position relative to the centromeres and the Hi-C dataset showed typical long-range heterochromatin interac- NM, we labeled these elements with different fluorescent tions, which were highlighted upon normalization, and the Pearson markers by constructing WT and Δset-7 strains with GFP-labeled correlation analysis showed that, again, chromosome arms were A–C telomere repeat-binding protein (TRF-1-GFP), IR fluorescent mostly segregated (Fig. S6 ). Comparison of the normalized protein-labeled centromeric histone H3 (CenH3-iRFP), and a Δnpf and WT datasets showed that intrasubtelomeric and centro- – blue fluorescent protein-labeled trans-NM protein (ISH-1-BFP). meric flank contacts were decreased whereas centromere euchro- We measured nucleus diameter, NM–telomere distance, NM– B D matic contacts were increased (Fig. 4 and Fig. S6 ), as in the centromere distance, and closest telomere–centromere distance, Δset-7 Δset-7 Δdim-5 Δnpf and ; strains. Direct comparison of the and assigned telomeric foci to one of three zones of equal area Δset-7 and datasets highlighted this similarity; the only differences (zones 1–3; Fig. 5A) (22). In WT, the 14 telomeres typically evident in a Δnpf strain are reduced compaction of the centro- clustered into two to four NM-associated zone 1 foci and all mere core (Fig. 4C, black arrowhead) and slightly reduced inter- centromeres clustered into a single zone 1 focus (Fig. 5B and Fig. actions within euchromatin (Fig. 4C, oval). As with the Δset-7 and S7 A–C). In the Δset-7 strain, most telomeres were still NM- Δset-7; Δdim-5 strains (Figs. 2 and 3 and Figs. S2 F and G and S3 associated (Fig. 5C, image 1, and Fig. S7B), but zone 3 telomeric A and B), the strongest intra- and interchromosomal contacts in foci increased (Fig. 5C, images 2–5, and Fig. S7B) and the Δnpf were depleted relative to those in WT (Fig. 4D and Fig. S6 E number of telomeric foci per nucleus decreased (Fig. S7A). and F), including intrasubtelomeric contacts, which may indicate Overall, the fraction of nuclei containing at least one zone 3 telomeric

Klocko et al. PNAS Early Edition | 3of6 Downloaded by guest on September 25, 2021 0 Mb 3D contacts and affect transcription. We therefore checked if A 4 Mb WT changes in genome structure and gene expression were correlated. LGVII H3K27me2/3 Our previous survey of Δset-7 transcriptional changes identified 0.5 CenH3 H3K9me3 only a small number of SET-7 up-regulated genes (17%) marked by H3K27me2/3 (11), possibly because of low detection of rare tran- 3.5 scripts affecting fold-change calculations (Fig. S8 A and B). We extended this study with replicate experiments of WT, Δset-7, Δnpf,

Δdim-5,andΔset-7; Δdim-5 strains and identified transcripts more 1 Mb 1 than fourfold (log2 ≥ 2.0)changedfromWTinthesemutants (Dataset S1). The replicate datasets revealed that 83% of genes up- Δset-7 A–

3 Mb 3 regulated in strains were H3K27me2/3-marked (Fig. S8 C). Differentially expressed genes were plotted; most of the 76 genes up-regulated in Δset-7 strains are found in H3K27me3- 1.5 marked subtelomeres (Fig. 2 and Fig. S2 E and F,redlines),

Circos Hi-C 2.5 whereas the 75 down-regulated genes were scattered across chro- links (log )

2 E F 2 Mb 2 >3.0 mosomes (Fig. 2 and Fig. S2 and , blue lines). We examined if B 0 Mb 2.75 to 3.0 strong Hi-C intrachromosomal interactions were correlated with Δset-7 4 Mb 2.5 to 2.75 changed gene expression. Approximately 46% of 10-kb bins marked LGVII 2.25 to 2.5 by H3K27me2/3 included up-regulated genes. In WT, each such bin 0.5 had an average of 4.9 strong contacts, whereas the bins showed an average of 2.0 contacts in the Δset-7 strain (Fig. S8D). Although all

3.5 H3K27me2/3-marked regions lost 3D contacts, no specific corre- lation between changes in gene expression and changes in contacts

was observed (Fig. S8D). This suggests that the global change in 3D 1 Mb 1

3 Mb 3 A 0 Mb Δ mRNA levels Δ mRNA levels 4 Mb from WT (log ) from WT (log )

2 2 Δset-7;Δdim-5 3.5 to 4.0+ 1.5 -3.5 to -4.0+ 3.0 to 3.5 -3.0 to -3.5 LGVII 0.5

2.5 to 3.0 -2.5 to -3.0 2.0 to 2.5 2.5 -2.0 to -2.5

C Mb 2 3.5 1729 4472 1800 WT 4500 WT 1171 Δset-7 Δset-7 1200 3000 2466 2341 625 609 1520 Mb 1 600 295 331 1500 988 981 242 149 454 303

# Intra contacts 0 # Inter contacts 0

K9 to K27 to K9 to K9 to K9 to K27 to K9 to K9 to K9 K27 K27 K27only K9 K27 K27 K27only Mb 3

Fig. 2. H3K27me3 loss reduces strong Hi-C contacts and alters gene expres-

sion. (A and B) Circos plots of the strongest (observed vs. expected, log > 2.25; H3K9me3 2 1.5 CenH3 color scales at right) intrachromosomal LGVII contacts at 10-kb resolution of Δ WT H3K27me2/3 (A)WT(7)or(B) set-7 strains. WT H3K9me3 (green), CenH3 (orange), and Δdim-5 H3K27me2/3 Δdim-5 changed genes

H3K27me2/3 (purple) ChIP-seq tracks are displayed around the outside. Ex- 2.5

Δset-7;Δdim-5 changed genes pression data of Δset-7 (log2 ≥ 2.0 or ≤−2.0, color scales below; red indicates Mb 2 increased expression, blue indicates decreased expression; bars do not reflect 0 0 gene size) are shown in the central (white background) track. (C)Quantifica- B C tion of intra- (Left) or inter- (Right) chromosomal contacts between H3K9me3- marked regions, H3K27me2/3-marked regions, or contacts originating from a H3K9me3-marked region and ending in an H3K9me3- or H3K27me2/3-marked Bins Bins region using program K9-K27_Quant.py, or a region marked only by H3K27me2/3 using program K9-K27only_Quant.py.

focus doubled in Δset-7 nuclei (Fig. 5D). Similarly, the distance be- tween the chromocenter and the closest telomeric bundle was shorter 100 100 in Δset-7 nuclei (Fig. 5E); even colocalization of centromeric and 0 Distance (Mb) 4 0 Distance (Mb) 4 C 0.2- 0.2 0.2- 0.2 telomeric foci was observed (Fig. 5 , panels 6 and 7), even though the Fold change of Δset-7; Δdim-5 Fold change of Δset-7; Δdim-5 – Δset-7 centromere NM distance and NM diameter remained unchanged relative to WT (log2) relative to (log2) (Fig. S7 C and D). Interestingly, Δset-7 but not WT nuclei occa- sionally showed two centromeric foci (12 of 249 in Δset-7 vs. 0 of 201 Fig. 3. Reduced heterochromatic contacts and altered gene expression in C Δset-7; Δdim-5.(A) Plot of the strongest LGVII intrachromosomal contacts and in WT; Fig. 5 , panels 7 and 8). These findings suggest that telomeres Δ Δ lose their NM association upon H3K27me2/3 loss, which may in- expression changes in set-7; dim-5 relative to WT, as in Fig. 2, except that H3K27me2/3 ChIP-seq data from Δdim-5 (dark purple) and WT strains (light crease genome disorder and produce the abnormal Hi-C pattern. purple) and expression data of a Δdim-5 strain (as in Fig. 2) in the yellow background track are shown. (B and C) Heat map of differences in contact Differential Gene Expression upon H3K27me2/3 Loss. Movement of frequency (log2)ofΔset-7; Δdim-5 relative to WT (B)orΔset-7 (C), as in Fig. 1C, the telomeres from the predominantly silent nuclear periphery except that the LGVII schematic is shown as in Fig. 1A. Decreased centromeric (23, 24) and associated genome disorder might give rise to aberrant and interspersed heterochromatic contacts are shown (arrowheads).

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1615546113 Klocko et al. Downloaded by guest on September 25, 2021 A LGVII H3K27me3 of prominent heterochromatic features, namely H3K9me3 or HP1, 0 1.0 2.0 3.0 4.0 Mb 4.1 4.2 Mb has little effect on Hi-C patterns (7). We found that disruption of 15 30 WT facultative heterochromatin by elimination of all H3K27me2/3 in 0 0 Δset-7 Δset-7 Δdim-5 Δnpf 15 30 and ; strains reduces intra- or interchromosomal 0 0 interactions among heterochromatic regions. NPF is found in com- plexes in addition to PRC2, so, even though the same Hi-C pattern B C Δnpf 0 0 was obtained with the strain, we cannot be sure the effect is through this particular complex. Interestingly, the Δset-7, Δset-7; Δdim-5,andΔnpf strains not only affect interactions between H3K27me2/3-marked domains but also interactions between

Bins Bins H3K9me3 regions and interactions between H3K27me2/3 and H3K9me3 regions. All three strains also showed increased cen- tromere–euchromatin interactions, indicating that subtelomeric H3K27me2/3 plays a role in preventing such interactions in WT despite lack of H3K27me2/3 in WT centromere regions. Deletion of set-7 alters genome structure and gene expression without noticeably 80 80 compromising vegetative growth, raising questions regarding the 0 Distance (Mb) 4 0 Distance (Mb) 4 -2.0 2.0 -2.0 2.0 functional importance of these changes. Although no global corre- Δnpf Δnpf Δset-7 Fold change of relative to WT (log2) Fold change of relative to (log2) 0 Mb lation between changes in genome organization and gene expression were detected, globule-like domains appeared less ordered in the D 4 Mb Δnpf Δset-7 strains. The apparent disorganization of globules, which could 0.5 – LGVII result indirectly from reduced telomere NM association, might cause subtle changes in gene regulation that were not detected in our

3.5 analyses. We conclude that subtelomeric facultative heterochromatin is required to maintain the normal genome structure, but normal

genome structure is not functionally paramount in N. crassa grown 1 Mb 1 under standard laboratory conditions. Curiously, in Fusarium species GENETICS that have been examined, deletion of SET-7 homologs causes severe growthdefects(25)orislethal(26). 3 Mb 3 Our cytological observation of mislocalized telomeres in the Δset-7 strain suggests that H3K27me2/3 plays a role in anchoring

H3K27me2/3 1.5 α CenH3 telomeres to the NM. Interestingly, a mutant form of importin , H3K9me3 DIM-3 (27), has an altered genome organization resembling that 2.5 of Δset-7 strains and shows decreased telomere–NM association

2 Mb 2 (7). These may be mechanistically distinct, however. dim-3 Fig. 4. Δnpf and Δset-7 strains show comparable Hi-C patterns. (A)(Left)LGVII Nuclear diameter is increased in relative to WT, which H3K27me3 ChIP-seq tracks of WT (purple) or Δnpf (gray) strains. (Right)Expansion may hinder dim-3 telomere anchoring; the distribution of of subtelomeric ∼150 kb (red bar). (B and C) Heat map of differences in contact H3K27me2/3 in dim-3 is normal. Conversely, nuclear diameter is Δ Δ frequency (log2)of npf relative to WT (B)or set-7 (C) across LGVII, as in Fig. 1C, unchanged in Δset-7 strains despite H3K27me2/3 loss. Together, at 50-kb resolution. Subtle decreases in centromeric bundling (black arrowhead) and euchromatic contacts (oval) are indicated. (D)PlotofthestrongestLGVII intrachromosomal contacts and expression changes in Δnpf relative to WT, as in Fig. 2; note the Δnpf strain gained many moderately strong contacts. 1234 A Zone 1 Zone 2 B WT Δset-7 ISH-1 genome structure in the strain did not itself cause the Zone 3 C 1234 derepression. CenH3 Most of the 145 derepressed genes in the Δset-7; Δdim-5 strain TRF-1 were also subtelomeric or near former constitutive heterochromatic 5678 A A B D * Δ set-7 regions (Fig. 3 and Fig. S3 and , red lines); the 97 repressed 80 78 WT genes were often near centromeric flanks that gained H3K27me2/3 Δset-7 or strong Hi-C interactions lost in the Δset-7; Δdim-5 strain (Fig. 3A 57 E 1.5 60 * and Fig. S3 A and B, blue lines). The double mutant had considerably 43 m) fewer expression changes than the single Δdim-5 mutant (Fig. 3A and 40 μ 1.0 A B Fig. S3 and ), consistent with some ectopic H3K27me2/3 in the 22 Δdim-5 Δnpf % Nuclei strain causing multiple abnormalities (12, 21). The 20 0.5

strain showed no obvious correlation between its altered Hi-C profile Distance ( 0 and its gene expression changes (656 up- and 687 down-regulated Z1 &/or Zone 3 0 D D E Telomere Centromere - genes; Fig. 4 and Fig. S6 and ), presumably because NPF has Z2 Foci Focus Only WT Δset-7 roles in complexes besides PRC2 (13, 14). No pattern was found between gene function and expression changes, as the majority of Fig. 5. H3K27me3 loss mislocalizes telomeres. (A) Schematic of ISH-1-BFP changed genes had an unknown function, per Gene Ontology analysis (blue, NM), CenH3-iRFP (red, centromere), and TRF-1-GFP (green, telomeres) E markers with dotted lines indicating three equal area zones. (B and C) Sin- (Fig. S7 ), consistent with the idea that H3K27me2/3 marks Δ N. crassa-specific genes (11). gle-section images of WT and set-7 nuclei expressing the three marker proteins. (D) Percentage of WT or Δset-7 nuclei with zone 1 or 2 telomeric Discussion foci (Left) or at least one zone 3 focus (Right). Asterisks denote significant changes (P < 0.001, χ2 test; all zone changes P < 0.001). (E) Centromere– Paradoxically, constitutive heterochromatin contacts dominate closest telomere distance box plot; the WT (dark gray) and Δset-7 (light gray) N. crassa chromosome interactions detected by Hi-C, but elimination difference is significant (P < 0.001, ANOVA).

Klocko et al. PNAS Early Edition | 5of6 Downloaded by guest on September 25, 2021 our observations suggest that normal telomere–NM contacts, which dominated by interactions among regions of heterochromatin. We depend on subtelomeric H3K27me2/3 and appropriate nuclear size, also demonstrated that the overall architecture of the genome is are involved in normal genome organization; conceivably, loss of robust, remaining largely intact even when the key features of con- subtelomere tethering following H3K27 demethylation of faculta- stitutive and facultative heterochromatin are eliminated. Additional tive heterochromatin facilitates rapid gene activation in response to studies are needed to further explore the functional significance of environmental stimuli. How H3K27me2/3 could promote NM– the 3D organization of eukaryotic genomes. telomere interactions is unknown. N. crassa lacks orthologs of the H3K27me3-binding proteins found in other systems, such as Poly- Materials and Methods comb/CBX of PRC1 in animals (28) and Like Heterochromatin Strains are listed in Table S1. Duplicate Hi-C (7) and ChIP-seq (12) experiments Protein-1 in plants (29). One possibility is that PRC2 itself is directly were performed as described previously except that the final ChIP-seq PCR had involved in chromosome conformation, as the PRC2 component eight cycles. Duplicate RNA-seq experiments were performed as described in SI EED has been reported to bind H3K27me3 in other systems (30). Materials and Methods (31–39). Deconvolution microscopy was performed as Although it was possible that ectopic H3K27me2/3 localization described previously (7). All Hi-C, ChIP-seq, RNA-seq data, and python scripts partly substituted for lost H3K9me3 or HP1 binding at former described here were deposited to the National Center for Biotechnology In- constitutive heterochromatin regions to maintain the genome formation (NCBI) Gene Expression Omnibus (GEO) database (accession no. structure in Δdim-5 or Δhpo mutants (7), our observation that the GSE82222). WT Hi-C (accession no. GSE71024) (7), WT H3K9me3 ChIP-seq, WT H3K27me2/3 ChIP-seq, Δdim-5 H3K27me2/3 ChIP-seq, Δhpo H3K27me2/3 ChIP- Hi-C pattern of a Δset-7; Δdim-5 strain was nearly the same as that Δset-7 Δnpf seq (accession no. GSE68897) (12), and WT Bisulfite-seq (accession nos. GSE61173 of and strains suggests that this is not the case. All and GSE81129) (27, 40) data were previously reported. strains showed prominent interactions of constitutive hetero- chromatin, even though they were less pronounced than in the ACKNOWLEDGMENTS. We thank Vince Bicocca (University of Oregon) for the Δdim-5 or Δhpo mutants (7). These findings, together with the use of unpublished ChIP-seq data, Diana Libuda (University of Oregon) for the use observation that Δset-7; Δdim-5 strains grow better than single of the deconvolution microscope, and Ayumi Yokoyama (University of Fukui) for dim-5 deletion strains (12, 21), argue against the hypothesis that technical support. This study was supported by NIH Grants GM035690 (to E.U.S.), GM093061 (to E.U.S.), and GM097637 (to M.F.); a Competitive Funds in Program to ectopic H3K27me2/3 functionally replaces H3K9me3. Disseminate Tenure Tracking System, Ministry of Education, Culture, Sports, Science In conclusion, our studies have revealed an unexpected role of and Technology (MEXT), Japan, grant (to S.H.); NIH Postdoctoral Fellowship H3K27me2/3 in the 3D organization of the genome, which is GM097821 (to A.D.K.); and a NASA postdoctoral fellowship (to J.M.G.).

1. Lieberman-Aiden E, et al. (2009) Comprehensive mapping of long-range interactions 21. Basenko EY, et al. (2015) Genome-wide redistribution of H3K27me3 is linked to reveals folding principles of the . Science 326(5950):289–293. genotoxic stress and defective growth. Proc Natl Acad Sci USA 112(46):E6339–E6348. 2. Hou C, Li L, Qin ZS, Corces VG (2012) Gene density, transcription, and insulators 22. Meister P, Gehlen LR, Varela E, Kalck V, Gasser SM (2010) Visualizing yeast chromo- contribute to the partition of the Drosophila genome into physical domains. Mol somes and nuclear architecture. Methods Enzymol 470:535–567. 48(3):471–484. 23. Towbin BD, Gonzalez-Sandoval A, Gasser SM (2013) Mechanisms of heterochromatin 3. Rao SS, et al. (2014) A 3D map of the human genome at kilobase resolution reveals subnuclear localization. Trends Biochem Sci 38(7):356–363. principles of chromatin looping. Cell 159(7):1665–1680. 24. Mekhail K, Moazed D (2010) The nuclear envelope in genome organization, expres- 4. Sexton T, et al. (2012) Three-dimensional folding and functional organization prin- sion and stability. Nat Rev Mol Cell Biol 11(5):317–328. ciples of the Drosophila genome. Cell 148(3):458–472. 25. Connolly LR, Smith KM, Freitag M (2013) The Fusarium graminearum histone H3 K27 5. Feng S, et al. (2014) Genome-wide Hi-C analyses in wild-type and mutants reveal high- methyltransferase KMT6 regulates development and expression of secondary me- resolution chromatin interactions in Arabidopsis. Mol Cell 55(5):694–707. tabolite gene clusters. PLoS Genet 9(10):e1003916. 6. Mizuguchi T, et al. (2014) Cohesin-dependent globules and heterochromatin shape 26. Studt L, et al. (2016) Knock-down of the methyltransferase Kmt6 relieves H3K27me3 3D genome architecture in S. pombe. Nature 516(7531):432–435. and results in induction of cryptic and otherwise silent secondary metabolite gene 7. Galazka JM, et al. (2016) Neurospora chromosomes are organized by blocks of im- clusters in Fusarium fujikuroi. Environ Microbiol, 10.1111/1462-2920.13427. portin alpha-dependent heterochromatin that are largely independent of H3K9me3. 27. Klocko AD, et al. (2015) Neurospora importin α is required for normal heterochro- Genome Res 26(8):1069–1080. matic formation and DNA methylation. PLoS Genet 11(3):e1005083. 8. Beisel C, Paro R (2011) Silencing chromatin: Comparing modes and mechanisms. Nat 28. Cao R, et al. (2002) Role of histone H3 lysine 27 methylation in Polycomb-group si- Rev Genet 12(2):123–135. lencing. Science 298(5595):1039–1043. 9. Nora EP, et al. (2012) Spatial partitioning of the regulatory landscape of the X- 29. Zhang X, et al. (2007) The Arabidopsis LHP1 protein colocalizes with histone H3 Lys27 inactivation centre. Nature 485(7398):381–385. trimethylation. Nat Struct Mol Biol 14(9):869–871. 10. Smith KM, et al. (2008) The fungus Neurospora crassa displays telomeric silencing 30. Jiao L, Liu X (2015) Structural basis of histone H3K27 trimethylation by an active mediated by multiple sirtuins and by methylation of histone H3 lysine 9. polycomb repressive complex 2. Science 350(6258):aac4383. Chromatin 1(1):5. 31. Ebbole DJ, Sachs MS (1990) A rapid and simple method of isolation of Neurospora 11. Jamieson K, Rountree MR, Lewis ZA, Stajich JE, Selker EU (2013) Regional control of his- crassa homokaryons using microconidia. Fungal Genet Newslett 37:17–18. tone H3 lysine 27 methylation in Neurospora. Proc Natl Acad Sci USA 110(15):6027–6032. 32. Robinson JT, et al. (2011) Integrative genomics viewer. Nat Biotechnol 29(1):24–26. 12. Jamieson K, et al. (2015) Loss of HP1 causes depletion of H3K27me3 from facultative 33. Xu S, Grullon S, Ge K, Peng W (2014) Spatial clustering for identification of ChIP- heterochromatin and gain of H3K27me2 at constitutive heterochromatin. Genome enriched regions (SICER) to map regions of histone methylation patterns in embryonic Res 26(1):97–107. stem cells. Methods Mol Biol 1150:97–111. 13. Song J-J, Garlick JD, Kingston RE (2008) Structural basis of histone H4 recognition by 34. Quinlan AR (2014) BEDTools: The Swiss-Army tool for genome feature analysis. Curr p55. Genes Dev 22(10):1313–1318. Protoc Bioinformatics 47:11.12.1–34. 14. Tie F, Furuyama T, Prasad-Sinha J, Jane E, Harte PJ (2001) The Drosophila Polycomb 35. Catchen J, Hohenlohe PA, Bassham S, Amores A, Cresko WA (2013) Stacks: An analysis Group proteins ESC and E(Z) are present in a complex containing the histone-binding tool set for population genomics. Mol Ecol 22(11):3124–3140. protein p55 and the histone deacetylase RPD3. Development 128(2):275–286. 36. Trapnell C, et al. (2010) Transcript assembly and quantification by RNA-Seq reveals 15. Tamaru H, Selker EU (2001) A histone H3 methyltransferase controls DNA methylation unannotated transcripts and isoform switching during cell differentiation. Nat in Neurospora crassa. Nature 414(6861):277–283. Biotechnol 28(5):511–515. 16. Lewis ZA, et al. (2010) DNA methylation and normal chromosome behavior in Neu- 37. Anders S, Pyl PT, Huber W (2015) HTSeq–a Python framework to work with high- rospora depend on five components of a histone methyltransferase complex, DCDC. throughput sequencing data. Bioinformatics 31(2):166–169. PLoS Genet 6(11):e1001196. 38. Anders S, Huber W (2010) Differential expression analysis for sequence count data. 17. Freitag M, Hickey PC, Khlafallah TK, Read ND, Selker EU (2004) HP1 is essential for Genome Biol 11(10):R106. DNA methylation in Neurospora. Mol Cell 13(3):427–434. 39. Thomas PD, et al. (2003) PANTHER: A library of protein families and subfamilies in- 18. Nielsen PR, et al. (2002) Structure of the HP1 chromodomain bound to histone H3 dexed by function. Genome Res 13(9):2129–2141. methylated at lysine 9. Nature 416(6876):103–107. 40. Honda S, et al. (2016) Dual chromatin recognition by the histone deacetylase complex 19. Honda S, Selker EU (2008) Direct interaction between DNA methyltransferase DIM-2 HCHC is required for proper DNA methylation in Neurospora crassa. Proc Natl Acad and HP1 is required for DNA methylation in Neurospora crassa. Mol Cell Biol 28(19): Sci 113(41):E6135–E6144. 6044–6055. 41. Hulsen T, de Vlieg J, Alkema W (2008) BioVenn - a Web application for the com- 20. Honda S, et al. (2012) Heterochromatin protein 1 forms distinct complexes to direct parison and visualization of biological lists using area-proportional Venn diagrams. histone deacetylation and DNA methylation. Nat Struct Mol Biol 19(5):471–477, S1. BMC Genomics 9:488.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1615546113 Klocko et al. Downloaded by guest on September 25, 2021