Chromatin Compaction by Small RNAs and the Nuclear RNAi Machinery in C. elegans

DNA is organized and compacted into higher-order structures in order to fit within nuclei and to facilitate gene regulation. Mechanisms by which higher order chromatin structures are established and maintained are poorly understood. In C. elegans, nuclear-localized small RNAs engage the nuclear RNAi machinery to regulate gene expression and direct the post-translational modification of histone proteins. Here we confirm a recent report suggesting that nuclear small RNAs initiate or maintain chromatin compaction states in C. elegans germ cells. Additionally, we show that experimentally provided small RNAs are sufficient to direct chromatin compaction and that this compaction requires the small RNA-binding Argonaute NRDE-3, the pre-mRNA associated factor NRDE-2, and the HP1-like protein HPL-2. Our results show that small RNAs, acting via the nuclear RNAi machinery and an HP1-like protein, are capable of driving chromatin compaction in C. elegans.

together in a feed-forward loop to promote heterochromatin formation in fission yeast. Argonaute proteins have been linked to H3K9 methylation and heterochromatin formation in many other eukaryotes including, plants, insects, and mammals 18,26,27 . In some cases, a different class of small RNA, termed piRNA, can also direct heterochromatin formation 28 . In summary, an axis of nuclear small RNAs, H3K9 methytransferases, and HP1-like proteins direct heterochromatin formation and chromatin compaction in many eukaryotes.
siRNAs can also direct H3K9me3 in C. elegans 29 . To do so, cytoplasmic siRNAs engage AGO proteins (HRDE-1 in the germline; NRDE-3 in the soma), which escort siRNAs into nuclei where they bind nascent transcripts (pre-mRNA) and recruit the downstream nuclear RNAi effectors (NRDE-1/2/4) to genomic sites of RNAi [29][30][31][32] . Once recruited to genes, the NRDEs inhibit RNAP II elongation and direct the deposition of H3K9me3 via a currently unknown histone methyltransferase(s) 29 . C. elegans express an abundant class of endogenous (endo) siRNAs, which engage the nuclear RNAi machinery to regulate gene expression and chromatin states (e.g. H3K9me3) during the normal course of growth and development [29][30][31][32][33] . C. elegans possess two HP1-like proteins HPL-1 and HPL-2 34,35 . HPL-2 is required for RNAi to direct long-term (transgenerational) gene silencing, which is a process known to depend upon the other components of the C. elegans nuclear RNAi machinery 32,36,37 . Thus, HPL-2 may be a downstream component of the C. elegans nuclear RNAi machinery. Surprisingly, the relationship between HPL-2 and H3K9me3 in C. elegans is unclear; HPL-2 colocalizes with H3K9me3 on chromatin, however, HPL-2 is still largely able to localize to chromatin (albeit at attenuated levels) in mutant animals that lack detectable levels of H3K9me3 38 . Thus, other chromatin marks, in addition to H3K9me3, may contribute to HPL-2 localization in C. elegans. In plants, microchidia (MORC) GHKL ATPases are thought to act downstream of siRNAs and H3K9me3 to compact and silence chromatin surrounding transposable elements 39 . The C. elegans genome encodes a single MORC-like protein (MORC-1) 39 . A recent study showed that MORC-1 is a downstream component of the C. elegans nuclear RNAi machinery 40 . Interestingly, in C. elegans lacking MORC-1, germ cell chromatin becomes disorganized and decompacted 40 . The data suggest that endo siRNAs, acting via the nuclear RNAi machinery, may contribute to chromatin compaction in the C. elegans germline.
Here we show that two additional components of the C. elegans nuclear RNAi machinery are required for normal chromatin compaction and chromatin organization in the germline, supporting the model that endo siRNAs regulate chromatin compaction in nematodes. In addition, we show that siRNAs are sufficient to direct chromatin compaction in the soma and that this process requires a nuclear RNAi Ago (NRDE-3) as well as the HP1-like factor HPL-2. Our results support a model in which nuclear-localized small regulatory RNAs are important mediators of chromatin organization and compaction in C. elegans. Results endogenous siRNAs may compact germline chromatin. Small regulatory RNAs, such as siRNAs or piRNAs, have been linked to heterochromatin formation in many eukaryotes, suggesting that small RNAs may act as specificity factors, acting via Watson-Crick base pairing, for directing heterochromatin formation and chromatin compaction in many eukaryotes 27 . In C. elegans, the GHKL ATPase MORC-1, a component of the nuclear RNAi machinery, promotes chromatin organization and chromatin compaction in adult C. elegans germ cells, suggesting that siRNAs, acting via the nuclear RNAi machinery and MORC-1, also regulate chromatin compaction in C. elegans 40 . To further test this idea, we asked if two other components of the C. elegans nuclear RNAi machinery: the germline expressed nuclear RNAi AGO HRDE-1, and the nuclear RNAi factor NRDE-2, were, like MORC-1, needed for chromatin organization and compaction in adult C. elegans germ cells. To do so, we used fluorescence microscopy to visualize a GFP::H2B chromatin marker in wild-type, hrde-1(−), or nrde-2(−) animals. Note, it takes 2-3 generations of growth at elevated temperatures (25 °C) for chromatin organization defects to be observed in morc-1(−) animals 40 . The reason for this is not known, but may be linked to the role of nuclear RNAi in transgenerational epigenetic gene regulation in the C. elegans germline 32 . GFP::H2B was monitored over the course of three generations in wild-type, hrde-1(−), or nrde-2(−) animals grown at 25 °C (Fig. 1A,B and Fig. S1). In the first generation of growth at 25 °C, germ cell nuclei of wild-type, hrde-1 (−), and nrde-2(−) animals appeared similar ( Fig. 1A and Fig. S1). After three generations of growth at 25 °C, however, GFP::H2B marked chromatin in hrde-1 (−) or nrde-2(−) germ cells became disorganized and GFP::H2B signals appeared enlarged (Fig. 1A, Fig. S1, and Movie 1). We quantified the diameter of GFP::H2B fluorescence in randomly chosen nuclei and confirmed that GFP::H2B occupied more space in hrde-1(−) or nrde-2(−) animals than in wild-type animals after three generations of growth at 25 °C (Fig. 1B). Increases in the amount of space occupied by chromatin in hrde-1(−) or nrde-2(−) animals could be due to decompaction of chromatin or, conceivably, an increase in DNA content. To distinguish between these possibilities, we quantified DAPI fluorescence (see materials and methods) in wild type, nrde-2(−), and hrde-1(−) animals after three generations of growth at 25 °C. The analysis indicated that wild-type, nrde-2(−), and hrde-1(−) nuclei stained with similar amounts of DAPI, suggesting that the increased in space occupied by chromatin in hrde-1(−) or nrde-2(−) nuclei is due to chromatin decompaction and not increased DNA content (Fig. S2). The data suggest that germline chromatin becomes disorganized and decompacted in animals lacking the nuclear RNAi factors NRDE-2 and HRDE-1. We used whole chromosome DNA fluorescent in situ hybridization (FISH) to test this idea further. We subjected wild-type, nrde-2(−), and hrde-1(−) animals, grown at 25 °C for three generations, to whole chromosome DNA FISH targeting chromosomes I, II, and III (https://doi.org/10.7554/eLife.42823). As expected, chromosome I, II, and III DNA FISH stained three distinct regions of subnuclear space, which were concentrated near the nuclear periphery, a pattern consistent with the expected localization of chromosomes in adult pachytene germ cell nuclei (Fig. 1C) 41 . In nrde-2(−) and hrde-1(−) animals, chromosomes I, II, and III DNA FISH signals appeared disorganized and appeared to occupy more space than in wild-type animals (Fig. 1C). We used the 3D Manager software in ImageJ to quantify the amount of space occupied by DNA FISH signals in these animals and confirmed that each chromosome occupied more space in nrde-2(−) and hrde-1(−) animals than in wild-type animals (Fig. 1D) 42 . We www.nature.com/scientificreports www.nature.com/scientificreports/ conclude that, similar to what has been seen for MORC-1, the nuclear RNAi AGO HRDE-1 and the downstream nuclear RNAi effector NRDE-2 promote chromatin compaction in germ cells of adult C. elegans.

RNAi induces chromatin compaction in C. elegans.
Our data are consistent with the idea that endogenously expressed siRNAs promote chromatin organization and compaction in C. elegans. It is also possible that chromatin decompaction in hrde-1(−), nrde-2(−), or morc-1(−) animals could be an indirect consequence of the gene misregulation that occurs in animals lacking a functioning nuclear RNAi system 32,43 . To ask if small RNAs directly mediate chromatin compaction in C. elegans, we asked if experimental RNAi were sufficient to drive chromatin compaction. We targeted a repetitive multi-copy sur-5::gfp transgene, which expresses GFP in all somatic cells of C. elegans, with gfp RNAi and used gfp DNA FISH to visualize the space occupied by gfp DNA before and after gfp RNAi ( Fig. 2A and Fig. S4) 44,45 . The sur-5::gfp transgene was chosen due to its repetitive/large size which might be expected to make volume quantifications feasible. Additionally, we chose to image DNA FISH signals in intestinal cells as these cells are polyploid (32 N), large, and easy to identify 46 . After gfp RNAi, the subnuclear space occupied by sur-5::gfp DNA appeared to decrease ( Fig. 2A). Quantification confirmed that gfp RNAi caused gfp DNA FISH signals to occupy ~2 fold less space than in animals not exposed to gfp RNAi ( Fig. 2B and Fig. S3). We conclude that RNAi can induce chromatin compaction in C. elegans.
Nuclear RNAi couples siRNAs to chromatin compaction. Nuclear RNAi in somatic tissues requires the somatically expressed AGO NRDE-3 and the conserved pre-mRNA binding protein NRDE-2 29,30 . To explore how RNAi mediates chromatin compaction, we asked if nrde-2(−) or nrde-3(−) animals were able to compact sur-5::gfp chromatin in response to gfp RNAi. We conducted gfp DNA FISH on nrde-2(−);sur-5::gfp and nrde-3(−);sur-5::gfp animals that were treated +/− with gfp RNAi and found that both NRDE-2 and NRDE-3 were required for RNAi-directed chromatin compaction (Fig. 3A,B and Fig. S4A,B). Given that NRDE-3 is an AGO, the data support the idea that small RNAs are needed to mediate sur-5::gfp chromatin compaction 30 . Given that NRDE-2 is a nuclear RNAi factor, the data support the idea that small RNAs engage the nuclear RNAi machinery to mediate chromatin compaction 29 . In the absence of gfp RNAi the space occupied by the sur-5::gfp transgene www.nature.com/scientificreports www.nature.com/scientificreports/ was greater in nrde-2(−) and nrde-3(−) than in wild-type animals, suggesting that small RNA and nuclear RNAi-based sur-5::gfp compaction occurs at some level even in the absence of exogenous sources of gfp dsRNA (Fig. 3B). This latter observation is consistent with previous reports demonstrating that repetitive transgenes are often subjected to RNAi-based gene silencing in C. elegans 47 . How might siRNAs direct chromatin compaction? In many eukaryotes including C. elegans, RNAi directs H3K9me3 at genomic loci exhibiting sequence homology to RNAi triggers 27 . In addition, HP1-like proteins are recruited to H3K9me3 containing chromatin and homo-dimerize (or undergo phase separation) to compact chromatin by linking together distant sites of chromatin 13,16,48 . Finally, the HP1-like protein HPL-2 has been functionally linked to nuclear RNAi in C. elegans 37 . For all these reasons, we asked if HPL-2 might be required for RNAi-mediated chromatin compaction in C. elegans. hpl-2(−) animals failed to compact sur-5::gfp chromatin in response to gfp RNAi (Fig. 3B and Fig. S4C). The , and hpl-2(tm1489) animals, which all express the sur-5p::gfp transgene, and which were grown in the presence or absence of gfp RNAi. Data is normalized to the WT (no RNAi) control for each condition to account for differences in hybridization efficiency across experiments. p-values were calculated using a student's two tailed t-test and were calculated with respect to wild type samples feeding on gfp RNAi. ***p-value < 0.005.
www.nature.com/scientificreports www.nature.com/scientificreports/ data are consistent with the idea that HP1-like proteins contribute to chromatin compaction directed by nuclear siRNAs in C. elegans.

Discussion
Here we present evidence supporting the idea that endogenous siRNAs and the nuclear RNAi machinery can initiate/maintain chromatin architecture and compaction in C. elegans germ cells. We also show that experimentally introduced siRNAs are sufficient to compact chromatin in the soma and that this chromatin compaction requires a nuclear AGO as well as a downstream component of the nuclear RNAi machinery and the HP1-like protein HPL-2. The data support a model whereby nuclear small RNAs are important regulators of chromatin organization and chromatin compaction in C. elegans.
Our genetic analyses identified three factors (NRDE-2, NRDE-3, and HPL-2) that are required for RNAi-directed chromatin compaction in intestinal cells. These results have led us to propose the model summarized in Fig. 4. The AGOs NRDE-3 (soma) and HRDE-1 (germline) bind siRNAs and interacts with homologous nascent RNAs to recruit the NRDE nuclear RNAi factors to chromatin. The downstream nuclear RNAi factors (e.g. NRDE-2) recruit a currently unknown histone H3K9 methyltransferase(s) to chromatin to deposit H3K9me3. Finally, H3K9me3 recruits the HP1-like protein HPL-2, which dimerizes to compact chromatin (Fig. 4). Important tests of this model will include (1) asking if siRNAs are sufficient to direct chromatin compaction in germ cells, (2) asking if HPL-2 is physically recruited to sites of nuclear RNAi, (3) asking if HPL-2 dimerization is necessary for RNAi-based chromatin compaction, and (4) asking if H3K9me3 (directed by nuclear RNAi) is needed for recruitment of HPL-2 to chromatin. Regarding this latter question, three SET domain proteins (SET-25, MET-2, and SET-32/HRDE-3) have been linked to H3K9 methylation in C. elegans and are redundantly required for RNAi directed H3K9me3 [49][50][51] . Potential roles for H3K9me3 in HPL-2 recruitment and/ or chromatin compaction could be addressed by asking if animals lacking all three methytransferases are defective for small RNA-directed chromatin compaction. Interestingly, in C.  www.nature.com/scientificreports www.nature.com/scientificreports/ elegans, localization of HPL-2 to chromatin is only partially dependent upon H3K9me3, suggesting that C. elegans may possess H3K9me3-independent systems for recruiting HPL-2 to chromatin 38 . Indeed, in vitro studies show that C. elegans HPL-2 binds H3K27me3 (as well as H3K9me3) and in vivo studies show that RNAi directs the deposition of H3K27me3 on chromatin (as well as H3K9me3) in C. elegans 52,53 . Thus, H3K27me3 may act in parallel with, or independently of, H3K9me3 during HPL-2 driven chromatin compaction. Studies asking if components of the C. elegans Polymcomb 2 complex are needed for RNAi-directed chromatin compaction could test this model. Finally, the GHKL ATPase MORC-1 has been linked to chromatin compaction in plants and in C. elegans germ cells 39,40 . These observations have led to a model in which MORC-1 compacts chromatin at the direction of small RNAs. The chromatin compaction system described here will allow the role of MORC-1 in somatic chromatin compaction to be assessed.
Many eukaryotes express a diverse array of small regulatory RNAs. These RNAs are relatively mobile within cells and can interact with a high degree of specificity with other cellular nucleic acids. Such traits make small regulatory RNAs excellent candidates for precise yet versatile vectors of chromatin compaction during reproduction and development. What biological function(s) might small RNA-directed chromatin compaction play during reproduction and development? The simplest answer might be that small RNA-directed chromatin compaction regulates gene expression programs by regulating accessibility of chromatin to the transcriptional machinery. Consistent with this idea, the nuclear RNAi factors that we have linked to chromatin compaction in this work, are known to regulate gene expression in C. elegans 29,30,32,43 . Asking if nuclear RNAi-regulated genes are bound by HPL-2 (and if this binding requires nuclear RNAi factors) would be a good test of this idea. Interestingly, many of the loci regulated by nuclear RNAi are not protein-coding genes; rather, they are cryptic loci or pseudogenes 32,43 . Thus, gene regulation may not be the sole biological output for small RNA-mediated chromatin compaction. It is possible that small RNA-directed chromatin compaction could also be used to help generate higher-order chromatin structures important for chromosome segregation during mitosis or meiosis. Consistent with this latter idea, mutations in several C. elegans RNAi factors (including HRDE-1 and NRDE-2) have been linked to chromosomal nondisjunction during meiosis [54][55][56][57][58][59][60][61][62][63] . Centromeres are compacted chromosomal structures needed for chromosome segregation during mitosis and meiosis. C. elegans chromosomes are holocentric; centromere function is dispersed across hundreds of point centromeres along each chromosome 64 . Like centromeres in other eukaryotes, C. elegans holocentromeres are flanked by chromatin containing H3K9me3 and HP1 (termed pericentromeres) and this flanking heterochromatin is important for centromere function 38,65,66 . Thus, it is possible that small RNA-directed HPL-2 recruitment and, therefore, chromatin compaction, may contribute to holocentromere function in C. elegans, an idea that could be tested by asking if the localization of HPL-2 to pericentromeres depends upon nuclear small RNAs.

TTTTCACTGGAGTTGTCCCA AAAGGGCAGATTGTGTGGAC
The FISH Tag DNA Kit from Life Science Technologies (cat. number F32948) was used and the protocol suggested by the manufacturer was followed. Exceptions to the manufacturer's protocol were the amount of starting material (2 ug used) and resuspension volume (20 ul). Probes were stored at −20 C in the dark and used within 2 weeks of synthesis. gfp DNA FIsH. Frozen worm pellets were resuspended in cold 95% ethanol and vortexed for 30 seconds.
Samples were rocked for 10 minutes at room temperature. Samples were spun down (3k rpm for 30 seconds) and supernatant discarded. Samples were washed twice in 1X PBST. 1 ml of 4% paraformaldehyde solution was added and samples were rocked at room temperature for 5 minutes. Samples were then washed twice with 1X PBST before resuspension in 2XSSC for 5 minutes at room temperature. Samples were spun down and resuspended in a 50% formamide 2XSSC solution at room temperature for 5 minutes, 95 °C for 3 minutes, and 60 °C for 20 minutes. Samples were spun and resuspended in 60 ul of hybridization mixture (10% dextran sulfate, 2XSSC, 50% formamide, 2 ul of gfp FISH probe and 2 ul of RNAse A (sigma 20 mg/ml)). Hybridization reactions were incubated at 95 °C for 5 minutes before overnight incubation at 37 °C in a hybridization oven. The next day, samples were washed with 50% formamide 2XSSCT (rotating at 37 °C) for 30 minutes. Wash buffer was removed and samples were resuspended in mounting medium (vectashield with DAPI). Samples were mounted on microscope slides and sealed with nail polish.
www.nature.com/scientificreports www.nature.com/scientificreports/ oligopaint DNA FIsH. For Oligopaint staining, the same DNA FISH protocol was used with the following exceptions. In the hybridization mixture, 100 pmol of oligo was used for each chromosome. The following day, samples were washed with 50% formamide 2XSSCT at 37 C for 30 minutes. Wash buffer was removed, and samples were resuspended in in 60 ul of hybridization mixture (10% dextran sulfate, 2XSSC, 50% formamide, and 100 pmol of bridge oligo for each chromosome). Samples were incubated at 37 C for 45 minutes. Samples were washed with 50% formamide 2XSSCT at 37 C for 30 minutes. Wash buffer was removed, and samples were resuspended in 60 ul of hybridization mixture (10% dextran sulfate, 2XSSC, 50% formamide, and 100 pmol of detection oligo (labeled with Alexa 488, cy3, or Alexa647) for each chromosome). Samples were incubated at 37 C for 45 minutes. Samples were washed with 50% formamide 2XSSCT at 37 C for 30 minutes. Wash buffer was removed, and samples were resuspended in 50 ul of slowfade gold with DAPI.
DApI staining alone. Frozen worm pellets were resuspended in 1 ml of cold methanol (−20 C). Samples were rocked for 15 minutes at room temperature. Samples were washed twice with 1XPBST, and resuspended in 75 ul of vectashield with DAPI.
Microscopy. gfp DNA FISH images were captured by standard fluorescent microscopy using a widefield Zeiss Axio Observer.Z1 microscope using a Plan-Apochromat 63X/1.40 Oil DIC M27 objective and an ORCA-Flash 4.0 CMOS Camera. The Zeiss Apotome 2.0 was used for structured illumination microscopy using 3 phase images for DNA FISH imaging. For each animal imaged, the exposure time was set to fill 35% of the camera's dynamic range to account for differences in hybridization efficiency between animals. For intestinal nuclei, all images were taken at the anterior end of the worm to ensure we were imaging the same set of nuclei across all images. All image processing was done using the Zen imaging software from Zeiss. For in vivo imaging of GFP::H2B, animals were immobilized in M9 with 0.1% Sodium Azide. Animals were imaged immediately with the Zeiss Axio system described above. Confocal imaging ( Fig. 1C  Quantification of space occupied by GFP::H2B. To measure volume of H2B::GFP signals in Fig. 1B, nuclear diameters were measured using the Zeiss Zen software on randomly selected nuclei located in a similar region of the germline (close to gonad turn, pachytene). Nuclear volumes were calculated assuming spherical shape.
Quantification of space occupied by whole chromosome DNA FISH signals. The 3D manager plugin for ImageJ was used to quantify the surface area of each chromosome. For each condition, nuclei were randomly selected from a similar region of the germline (close to gonad turn, pachytene). After individual nucleus files were obtained the image files were split to obtain individual channel files that represent each chromosome. Thresholding was used to eliminate background noise using the default algorithm and using the "auto" function on every image. Each chromosome "object" was then added to the 3D ROI Manager. Surface area was calculated using the quantify 3D function.

Quantification of space occupied by DNA FISH signals. The Tools for Analysis of Nuclear Genome
Organization (TANGO) software was used to quantify the surface area and/or volume of space occupied by DNA FISH signals (gfp DNA FISH and whole chromosome DNA FISH 42 . For gfp DNA FISH analysis in intestinal nuclei, the standard nucleus segmentation processing chain was used with the following modifications: Nucleus edge detector, minimum nucleus size = 15,000. Size and edge filter, minimum nucleus size = 15,000. For object (gfp DNA FISH signal) segmentation the nucleoli processing chain was used with the following modification: Size and edge filter, minimum volume = 3.

DAPI quantifications.
As a readout of DNA quantity we quantified DAPI fluorescence using ImageJ.
Regions of interest (ROI), were generated for five germ cell nuclei per worm, and z-slices were selected to encompass the entire nucleus. The multi-measure tool in ImageJ was used to measure total fluorescence within the ROI for each slice. Three ROIs were drawn outside the worm to gain an average background level. Background levels were subtracted from each nucleus by multiplying the area of the ROI for a given nucleus with the average background value for the image.

Data Availability
All data generated or analysed during this study are included in this published article (and its Supplementary  Information files).