Monoallelically-expressed Noncoding RNAs form nucleolar territories on NOR-containing chromosomes and regulate rRNA expression

Out of the several hundred copies of rRNA genes that are arranged in the nucleolar organizing regions (NOR) of the five human acrocentric chromosomes, ∼50% remain transcriptionally inactive. NOR-associated sequences and epigenetic modifications contribute to differential expression of rRNAs. However, the mechanism(s), controlling the dosage of active versus inactive rRNA genes in mammals is yet to be determined. We have discovered a family of ncRNAs, SNULs (Single NUcleolus Localized RNA), which form constrained sub-nucleolar territories on individual NORs and influences rRNA expression. Individual members of the SNULs monoallelically associate with specific NOR-containing chromosome. SNULs share sequence similarity to pre-rRNA and localize in the sub-nucleolar compartment with pre-rRNA. Finally, SNULs control rRNA expression by influencing pre-rRNA sorting to the DFC compartment and pre-rRNA processing. Our study discovered a novel class of ncRNAs that by forming constrained nucleolar territories on individual NORs contribute to rRNA expression.

processing. Our study discovered a novel class of ncRNAs that by forming constrained 48 nucleolar territories on individual NORs contribute to rRNA expression. 49

INTRODUCTION 51
The nucleolus is the most well-characterized non-membranous nuclear domain, where 52 ribosome biogenesis and maturation takes place and is formed around the nucleolus 53 organizer regions (NORs) 1 . NORs are comprised of rRNA gene tandem arrays, and in 54 human cells, they are located on the short arms (p-arm) of the five acrocentric 55 chromosomes (Chrs. 13,14,15,21 & 22) 2 . Human cells contain >400 copies of rRNA 56 (18S/28S/5.8S) genes, yet only ~50% of the copies are transcriptionally active 3 . The 57 expression of rRNA genes is tightly controlled during physiological processes, such as 58 cellular development by epigenetic mechanisms 2, 4-6 . However, the mechanism that 59 precisely maintains the dosage of active versus inactive rRNA genes within a cell is yet 60 to be determined. 61 The nucleolus harbors a diverse set of small and long noncoding RNAs (ncRNAs), 62 which play crucial roles in organizing the nucleolar genome as well as regulating rRNA 63 (ASOs) targeting individual SNUL-1 CS candidate did not reduce total SNUL-1 levels 280 (data not shown). The repeat sequence within SNULs was highly conserved among all 281 the SNUL-1 CSs and was also shared by the SNUL-2 transcript. By using an ASO 282 targeting this region (ASO-SNUL) we efficiently depleted both SNUL-1 and SNUL-2 283 ( Figure 5a). Interestingly, SNUL-depleted cells showed enhanced 5-FU incorporation in 284 the nucleolus (Figure 5b-c), and also showed increased levels of nascent 285 quantified by single molecule RNA-Fluorescent hybridization (smRNA-FISH) using a 286 probe set (5'ETS-2) that preferentially detects nascent 47S pre-rRNA . 287 These results imply that SNUL depletion either enhanced the expression of nucleolus-288 enriched rRNA genes and/or reduced the co-transcriptional pre-rRNA processing. SNUL 289 depletion did not alter the overall distribution of the nucleolus-localized proteins   12,34 . However, we observed that SNUL-depleted cells showed increased 291 number of FC/DFC compartments/nucleolus, which could be a consequence of enhanced 292 pre-rRNA levels in these cells (Figures 5d,. 293 The nascent 47S pre-rRNA is co-transcriptionally sorted from its transcription site at 294 the DFC/FC boundary to DFC 21 . The DFC-localized FBL binds to the 5' end upstream of 295 the first cleavage site (01 site) (Figure S1m & 5g) of the 47S pre-rRNA co-296 transcriptionally and facilitates pre-rRNA sorting for efficient RNA processing and DFC 297 assembly 21 . Due to this, the 5' end of 47S pre-rRNA is localized in the DFC region, as 298 shown by SR-SIM of smRNA-FISH using the 5' external transcribed spacer (5'ETS)-1 299 probe set targeting the first 414 nts of 47S pre-rRNA (Figure 5d) 21 . On the other hand, 300 the region within the 47S pre-rRNA 5'ETS located downstream of the 01 cleavage site 301 (detected using 5'ETS-2 & 3 probes [Figures S1m & 5g;) 302 associated with the rRNA transcription sites (FC or FC/DFC junction)   21 . 303 Interestingly, in the SNUL-depleted cells, FBL-interacting 5'ETS-1 region within the 47S 304 pre-rRNA failed to sort to DFC, and instead preferentially accumulated in the FC ( Figure  305 5d-e). Depletion of pre-rRNA processing factors, including FBL, rRNA sorting at DFC, resulting in the accumulation of pre-rRNA in the FC region 21 . Our 307 results suggest the possibility that SNULs could influence pre-rRNA biogenesis by 308 modulating FBL-mediated pre-rRNA sorting. In support of this, our SR-SIM imaging 309 data revealed enriched association of SNUL-1 in the FBL-localized DFC. We therefore 310 evaluated whether SNULs influence the interaction between FBL and pre-rRNA. 311 Towards this, we performed FBL RNA-immunoprecipitation followed by quantitative 312 RT-PCR to quantify the interaction between FIB and nascent pre-rRNA in control and 313 SNUL-depleted cells. Strikingly, SNUL-depleted cells showed reduced association 314 between FBL and pre-rRNA (Figure 5h), indicating that DFC-enriched SNULs could 315 enhance the FBL interaction with pre-rRNA. 316 The sequence upstream of the 01-cleavage site within the 47S pre-rRNA (detected by 317 5'ETS-1 probe), is co-transcriptionally cleaved after it is sorted to DFC by FBL. Defects 318 in the pre-rRNA sorting to DFC were shown to affect pre-rRNA processing 21 . SNUL-1-319 depleted cells showed defects in the initial cleavage at the 5'end of the 47S pre-rRNA, as 320 observed by the reduced levels of 30S+1 intermediate and +1-01 cleaved product ( Figure  321 5i & S5f) by Northern blot analyses. The +1-01 is the unstable product processed from 322 the 5'end of 47S pre-rRNA due to the cleavage at the 01 site. All these results indicate 323 potential involvement of SNULs in pre-rRNA, sorting and/or co-transcriptional 324 processing. 325

DISCUSSION 327
We have discovered SNUL, a novel family of ncRNAs, which display non-random 328 association to specific NOR-containing chromosomes within the nucleolus. Our data 329 suggest that SNUL-1 is a member of a family of RNAs sharing similar sequence features. 330 The most striking feature of the SNUL-1 sequence is its resemblance to the 21S pre-331 rRNA intermediate. Recent genomic mapping of acrocentric chromosome arms revealed 332 that most of the sequences in the NOR-containing p-arms are shared among all the 5 333 chromosomes 15,18,35 . However, these studies have also identified inter-chromosomal 334 sequence variations 18 . Our observations showing the association of individual members 335 of SNULs to specific alleles of one of the acrocentric chromosomes support the idea that 336 the acrocentric arms encode chromosome-and allele-specific transcripts. 337 The underlying mechanism(s) controlling differential expression rRNA gene copies 338 in mammals is yet to be determined. Nucleolar dominance (NuD) is a developmentally 339 regulated process that is speculated to act as a dosage-control system to adjust the number 340 of actively transcribed rRNA genes according to the cellular need 5, 36 . However, NuD is 341 primarily observed in the 'interspecies hybrids' of plants, invertebrates, amphibians and 342 mammals 37-41 . NuD is reported in certain nonhybrid or 'pure species' of plants and fruit 343 flies, but has not yet been observed in mammals, primarily due to lack of information 344 about the allele-specific rRNA sequence variations [42][43][44]  We observe that SNUL-1 forms a spatially constrained RNA territory that associates 359 next to the NOR of Chr. 15, but is devoid of pre-rRNA. Furthermore, SNUL-depleted 360 cells show elevated levels of pre-rRNA, along with defects in pre-rRNA sorting and 361 processing. One-way SNUL-1 could modulate rRNA expression is via modulating the 362 levels of bioprocessing machinery that control rRNA biogenesis and processing. For 363 example, high sequence similarity between SNUL-1 and pre-rRNAs helps SNUL-1 to 364 compete for and/or recruit factors that regulate rRNA biogenesis in a spatially 365 constrained area within the nucleolus. A recent study, by visualizing the distribution of 366 tagged pre-rRNAs from an NOR-containing chromosome, reported that similar to 367 SNULs, pre-rRNAs transcribed from individual NORs form constrained territories that 368 are tethered to the NOR-containing chromosomal regions 53 . It is possible that SNUL-1, 369 by forming a distinct RNA territory on the NOR of the Chr. 15 allele, influences the 370 expression of rRNA genes from that NOR in an allele-specific manner. Such organization of SNULs and rRNA territories in a constrained area within the nucleolus would help to 372 control the expression of a subset of rRNA genes without affecting the rRNA territories 373 on other acrocentric chromosomes. 374 Our observation of compartmentalized distribution of individual members of SNUL 375 RNA within specific sub-nucleolar regions challenges the current view that all the 376 nucleoli within a single nucleus are composed of identical domains. Future work will 377 entail determining the mechanism(s) underlying the constrained formation of ncRNA 378 territories and allele-specific spreading and regulation on autosomal regions. 379 380

Limitations of the present study 381
Presently, very little is known about the sequences in the short-arms of NOR-containing 382 chromosomes, the region that harbors novel ncRNA genes such as SNULs. A recent 383 study, by utilizing long-read sequencing in a haploid cell line revealed that p-arms of 384 NOR-containing chromosomes are enriched with repeat sequences 35 . Higher levels of 385 sequence similarity observed between SNUL-1 candidates and pre-rRNA made it 386 impossible for us to precisely map the genomic coordinates of SNUL-1 genes from the 387 available long-read sequencing data set. Complete genome assembly of p-arms from 388 SNUL-1-expressing diploid cells would be essential to map SNUL-1 genes in the genome 389 and also to identify the regulatory elements controlling monoallelic expression of SNULs. 390 Genomic annotation of the full-length SNUL-1 genes is also crucial for designing 391 strategies to specifically alter the expression of individual SNUL genes, without targeting 392 other SNUL-like genes, furthering mechanistic understanding of SNUL functions. Even 393 with these technical limitations, the current study is highly impactful because our observations of the association of autosome arms by SNULs supports a paradigm-shifting 395 model that ncRNA-coating of chromosomes and their roles in gene repression are not 396 restricted only to sex chromosomes. In addition, our study will serve as a starting point 397 towards the understanding of how differential rDNA expression is achieved during 398 physiological processes. Altogether, this study will form the basis for an entirely new 399 avenue of investigations, which would help to understand the role of ncRNAs on 400 were washed with PBS for 3 times and were recovered in fresh growth medium for 30 463 min or 60 min. For epigenetic mark inhibition, cells were treated with 80nM TSA and 464 500nM 5-Aza-dC for 6 days. 465

RNA-Fluorescence in situ hybridization (FISH) 466
For all of the FISH and Immunofluorescence staining done with adherent cells, cells were 467 seeded on #1.5 coverslips at least two days before experiments. For GM12878 and 468 isolated HeLa nucleoli, suspension was smeared onto the Poly-L-lysine-coated (Sigma-469 Aldrich) coverslips prior to fixation. 470 For RNA-FISH using probes prepared by nick translation, cells were fixed by 4% PFA 471 for 15 min at room temperature (rt) and permeabilized with 0.5% Triton X-100 for 5 min 472 on ice. Alternatively, cells were pre-extracted by 0.5% Triton X-100 in CSK buffer for 5 473 min on ice and then fixed by 4% PFA for 10 min. Probes were made using Nick 474 Promega) and purified by G-50 column (GE Healthcare). For RNA-FISH using ribo-496 probes, cells on coverslips were fixed by 4% PFA for 10 min at rt, and then treated with 497 0.25% acetic anhydride in 0.1 M triethanolamide (pH 8.0) for 10 min. Coverslips were 498 washed in 1XSSC for 5min, treated with 0.2N HCl for 10 min, and pre-hybridized in 499 50% formamide, 5XSSC for at least 6 h at rt. Dig-labeled RNA probes were added to the 500 hybridization buffer (50% formamide, 5XSSC, 1X Denhardt's solution, 0.1% Tween20, 501 0.1% [w/v] CHAPS, 100 μg/ml Heparin, 5 mM EDTA, and 50 μg/ml Yeast tRNA) at a 502 final concentration of 2 μg/ml. Hybridization was carried out in a humidified chamber in 503 the dark overnight at 50 ℃. The coverslips were then washed with 0.2XSSC for 1 h at 55 504 ℃, blocked in 4% BSA, PBS for 30 min at 37 ℃, and incubated with anti-Dig-FITC or -505 Rhodamine (1:200) (Roche) in 1% BSA, PBS for 1 h at 37 ℃. The coverslips were 506 washed twice with washing buffer (0.1% Tween20, 2XSSC) and refixed with 4% PFA 507 for 15 min at rt. 508 For RNase A treatment, pre-extracted cells were incubated with 1 mg/ml RNase A in 509 CSK buffer for 30 min at 37 ℃. Cells were then fixed by 4% PFA for 15 min at rt and 510 processed to RNA-FISH. For DNase I treatment, fixed and permeabilized cells were incubated with 200 U/ml DNase I (Sigma) in DNase I buffer prepared with PBS for 2 h at 512 37 ℃, followed by incubation in Stop solution for 10 min at room temperature. RNA-513 FISH was then performed as described above. 514

RNA-DNA FISH 515
For DNA-FISH using chromosome paint probes (Chrs. 13, 15, 22) (MetaSystems), after 516 fixation and permeabilization, coverslips were incubated in 20% glycerol overnight and 517 then went through freeze-thaw by liquid nitrogen for at least 6 cycles. Coverslips were 518 then treated with 0.1N HCl for 5 min and prehybridized in 50% formamide, 2XSSC for 519 30 min at rt. Probe mix was made by adding the RNA-FISH probe into the chromosome 520 paint probe. Probes were applied to the coverslips and denatured with the coverslips at 521 75-80 ℃ on a heating block. Hybridization was carried out in a humidified chamber in 522 the dark for 48 h at 37 ℃. 523 For FISH using DNA-FISH probes made by nick translation, cells were pre-extracted 524 and fixed. Salmon sperm DNA and Human Cot-I DNA were added to the hybridization 525 buffer. Denaturation and hybridization were performed as described above. 526

DNA-FISH on Metaphase Spread 527
Cells were grown to ~70% confluence and treated with KaryoMax Colcemid solution 528 (Gibco) at a final concentration of 0.1 μg/ml in growth medium for 3 h. Mitotic cells 529 were then shaken off and pelleted by centrifuge. Cells were then gently resuspended in 75 530 mM KCl and incubated at 37 ℃ for 30-40 min. Cells were then fixed by freshly prepared 531 fixative (methanol: acetic acid 3:1 [v/v]) and dropped onto pre-cleaned microscope slides 532 from height. After air-drying, slides were stored at -20 ℃ for a least overnight before the 533 DNA-FISH. For the DNA-FISH on metaphase chromosomes, slides were rehydrated with 534 PBS and then treated with 50 μg/ml Pepsin in 0.01N HCl at 37 ℃ for 9 min. Slide were 535 then rinsed by PBS and 0.85% NaCl sequentially and dehydrated by a series of Ethanol at 536 different concentration (70%, 90%, and 100%). Air-dried slides were then subjected to 537 hybridization as described above. 538

Northern Blot 564
For the pre-rRNA Northern, 2 μg of total RNA extracted from WI-38 cells treated with 565 Ctr-ASO or ASO-SNUL was separated on 1% denature agarose gel prepared with 566 NorthernMax Denaturing Gel Buffer (Ambion) and run in NorthernMax Running Buffer 567 (Ambion). RNA was then transferred to Amersham Hybond-N+ blot (GE Healthcare) by 568 capillary transfer in 10 x SSC and crosslinked to the blot by UV (254 nm, 120mJ/cm2). 569 The DNA probes were labeled with [α-32P] dCTP by Prime-It II Random Primer 570 Labeling Kit (Stratagene) as per manufacturer's instructions. Hybridization was carried 571 out using ULTRAhyb Hybridization Buffer (Ambion) containing 1 X 10 6 cpm/ml of 572 denatured radiolabeled probes overnight at 42 ℃. Blots were then washed and developed 573 using phosphor-imager. 574 Native RNA Immunoprecipitation 575 WI38 cells were washed twice with cold PBS and collected by centrifuge (1,000g, 10 576 mins at 4 ℃). Cells were then lysed in 2ml RIP buffer (50 mM  Aprotinin, and 2mM VRC(NEB)) for three times and RIP buffer twice, followed by RNA 587 isolation and RT-qPCR. 588 589 590

Imaging Analyses 610
For colocalization analyses, 3D SIM stacks were imported into Fiji/ImageJ. The To generate an accurate version of SNUL-1 transcripts, we generated a consensus 679 sequence from the long-read alignments, using samtools and bcftools 59 . To evaluate the 680 specificity of the assembled transcripts, we performed a similarity comparison between 681 the generated consensus sequence against rRNA and PacBio Iso-Seq CS clones. We 682 observed that the sequences identified/generated from our analysis were more analogous 683 to the Iso-Seq clones over rRNA. 684 In order to verify the error rate of PacBio sequencing technology, for each isoform in 685 the high-quality PacBio database, we ran BLAST against the human transcript database 686 (GRCh38.p13 assembly). The maximum number of target sequences and the maximum 687 number of high-scoring segment pairs were set to 20 and 1 respectively, and the rest of 688 the arguments were set to default in the BLAST runs. LAGAN-v2.0 60 was then used to 689 perform pairwise global alignment between each isoform and its corresponding top 20 690 best matches, found by BLAST. A dissimilarity score was assigned to each matched 691 candidate by taking the ratio of mismatching sites to all the sites where both isoform and 692 the matched candidate did not contain gaps. The matched candidate with the least 693 dissimilarity score was taken as the best match to the isoform. The mean of the 694 dissimilarity scores, associated with isoforms having GC content greater than 60% 695 (matching the GC content of ITS1), being ~0.5% verifies the PacBio error rate (<1%).
To determine whether the 5 isoforms capable of detecting SNUL-1 are different 697 transcripts, and their difference is not due to sequencing error, we propose the following 698 three hypotheses to be tested: 699 H 0 : There is one known gene whose transcripts are I 1 ,…,I 5 . 700 H 1 : There is one unannotated gene, i.e., with no transcripts present in human transcript 701 database, whose transcripts are I 1 ,…,I 5 . 702 H 2 : There are multiple unannotated/annotated genes whose transcripts are I 1 ,…,I 5 . 703 If H 0 is true, then there exists a known transcript such that the dissimilarity score 704 between isoform i (I i ) and the transcript's dissimilarity score should follow the empirical 705 distribution of the dissimilarity scores in Figure S1l. For each isoform, the empirical 706 probability of observing a dissimilarity score greater than or equal to its associated 707 dissimilarity score was computed (Table S2). The product of the empirical p-values being 708 in the order of 10 -10 suggests that H 0 does not hold. 709 If H 1 is true, there should be one unannotated transcript whose dissimilarity score with 710 each read is about 0.5% (the empirical mean of the dissimilarity scores). Therefore, the 711 pairwise dissimilarity scores for the isoforms should be about 1%. We computed the real 712 pairwise dissimilarity by doing global pairwise alignment for each pair of isoforms using 713 LAGAN (Table S3) . The pairwise dissimilarities being 4% or above suggest that H 1 is 714 not true. Approximating the empirical distribution of the dissimilarity scores with an 715 exponential probability density function with mean 0.5%, if H 1 holds then the pairwise 716 dissimilarity should follow erlang distribution with shape and scale parameters being 2 717 and 200 respectively, as the sum of two independent exponential random variables with 718 the same rate parameter has erlang distribution. The probability of observing a pairwise 719 dissimilarity score greater than or equal to each real pairwise dissimilarity score under 720 erlang distribution was computed. The product of these probabilities being in the order of 721 10 -39 rejects the H 1 hypothesis which leaves us with accepting H 2 hypothesis. 722 For the analyses of alignment between SNUL-1 candidates and pre-rRNA, the candidate 723 sequences were aligned to the canonical 21S sequence and to each other using the LAST 724 algorithm as described previously 61 . 725

Data analyses and statistics 726
The data used in this study are performed at least biological triplicates. For each isoform the empirical p-value is the empirical probability of observing a 769 dissimilarity score greater than or equal to its dissimilarity score. The product of the 5 p-770 values being in the order of 10 rejects the hypothesis of all isoforms being the 771 transcripts of the same known gene.                           CAAATAACCCCCCCATCATACCCTGCCCC ATGTCTTCCTCTACTCTCTCCCTCATGCTTTCTCTCTCTCTCTCTCTCTCTGTCTCTCTCTCTCTCTCTCT :: :::: : :: : :: :::: :::::::: ::::::: ::::::::: ::::: ::::::: ATTA ATTTCTTGAGC-ACGCCCTTCCTCCC----CCTCTCTCTGTCTCTCTGTCTGTCTCTGTCTCTGTCTCTCT  Dissimilarity Score Count of isoforms (log Extended Data Fig. 1| SNUL-1 forms RNA clouds in human cell lines. a, RNA-FISH of SNUL-1 (green) in different human cell lines. For all images, nucleoli are visualized by rRNA (red). Scale bars, 5µm. b, RNA-FISH of SNUL-1 after nuclease treatments in WI-38 cells. Scale bars, 5µm. c, Schematic showing the truncated probes designed to determine the minimum region required for SNUL-1 hybridization. d, RNA-FISH performed with the strand-specific ribo-probes listed in c. Scale bars, 5µm. e, RNA-FISH to detect SNUL-1 (green) and SNUL-2 (red) clouds in WI-38 cells. Nucleoli are visualized by rRNA (blue). Scale bars, 5µm. f, Local alignment between SNUL-1 Probe 4 and SNUL-2 probe. Note the imperfect [CT] repeat in SNUL-2 probe and the poor alignment between the two probes beyond the [CT]-rich region. g, Schematic showing the workflow of the unbiased strategies to determine the full-length SNUL-1 sequence. h, RNA-FISH using probes designed from the CSs (green) and SNUL-1 (red) Probes 1 in WI-38 cells. Scale bars, 5µm. i, RNA-FISH with CS probes and SNUL-1 Probe 1 in WI-38 cells. Scale bars, 5µm. j. Signal profiles of the lines marked in i. Note that the signals shown by different probes are not completely colocalized. k, Pairwise sequence comparisons between 21S and CS8644, 21S and CS2615, and 21S and CS9572. Red lines indicate forward aligned regions, blue lines indicate reverse aligned regions, and yellow boxes indicate unaligned regions. l, Histogram of the dissimilarity score between each isoform in the high-quality PacBio database and its best match in the human transcript database. m, Schematic showing the positions of the rRNA and ITS1 probes. n, Representative SIM image showing the relative distribution of SNUL-1 (green) and pre-rRNA hybridized by ITS1 probe (red) within a single nucleolus. Scale bars, 1µm. o, RNA-FISH of SNUL-1 (green) and pre-rRNA hybridized by ITS1 probe (red) in WI-38 cells transfected by Ctr-ASO or ASO-SNUL. Scale bars, 5µm. DNA is counterstained with DAPI.   Fig. 3| SNUL-1 is associated with the NOR of one Chr. 15 allele. a, RNA-FISH showing the distribution of SNUL-1 in WI-38 cells during mitosis. Arrows point at the prominent SNUL-1 cloud in early G1 daughter nuclei. Arrow heads point at the relatively weak SNUL-1 cloud in early G1 phase of daughter nuclei. Scale bars, 5 µm. b, RNA-FISH of SNUL-1 and rRNA in early G1 daughter nuclei. Arrows point at the prominent SNUL-1 clouds. Arrow heads point at the relatively weak SNUL-1 clouds. Scale bars, 5 µm. c, DNA-RNA-FISH to detect SNUL-1 RNA and 15CEN in early G1 hTRET-RPE1 daughter nuclei. Scale bars, 5 µm. d, DNA-RNA-FISH to detect SNUL-1 RNA and Chr.15 in the nucleus. The two alleles of Chr.15 are marked by either probe painting the q-arms of the chromosome (IMR-90 cells), or 15CEN (hTRET-RPE1 and MCH065 cells). e, DNA-RNA-FISH to detect SNUL-2 RNA and Chr. 13 marked by the probe painting the q-arm of the chromosome in WI-38 nucleus. Scale bars, 5 µm. f, Quantification of the association rates between SNUL-2 and Chr13. Data are presented as Mean ± SD from biological triplicates. > 100 cells were counted for each of the biological repeats. g, DNA-FISH showing the rDNA contents on Chr.15 in hTRET-RPE1 metaphase chromosomes. The two alleles of Chr.15 are marked by 15Sat III and 15CEN. rDNA arrays are detected by a probe within the IGS region. h, Relative integrated density of the rDNA contents on the two Chr.15 rDNA arrays is calculated by dividing the measurement of the larger rDNA signal by the smaller rDNA cloud. Data are presented as Median and interquartile range. n = 15. i, Immuno-RNA & DNA-FISH showing SNUL-1 (green), Chr. 15 alleles (red) and UBF (blue) in early G1 phase WI-38 daughter nuclei. Scale bars, 5 µm (main images) and 2 µm (insets). j, SNUL-1 localization and 5-FU incorporation in WI-38 telophase/early G1 daughter nuclei. Scale bars, 5 µm. DNA is counterstained with DAPI.   the relative integrated density of the 15Sat III signal in WI-38 nuclei. The relative integrated density is calculated by dividing the measurement of the larger DNA-FISH signal by that of the smaller DNA-FISH signal. Data are presented as Median and interquartile range. n = 60. c, Representative RNA-FISH images showing the association of SNUL-1 cloud with 15Sat III or 15CEN in MCH065 nuclei. d, Quantification of the association rates between SNUL-1 and the smaller 15Sat III in MCH065 nuclei. Data are presented as Mean ± SD from biological triplicates. > 50 cells were counted for each of the biological repeats. e, Representative RNA-FISH images showing the localization SNUL-1 along with the SNRPN and SPA2 transcription site on the paternal allele of Chr. 15 in WI-38 nucleus. Dotted lines mark the boundary of the nucleus. f. Quantification of the association rate between SNUL-1 and the transcription sites of SNRPN and SPA2 in WI-38 cells. Data are presented as Mean ± SD from biological triplicates. > 35 cells were counted for each of the biological repeats. g, Representative RNA-FISH images showing the distribution of SNUL-1 (green) and SNUL-2 (red) in control and Aza-dC (500nM) and TSA (80nM) treated WI-38 nuclei. h, Quantification of the percentage of cells showing one or two SNUL-1clouds in control and Aza+TSA-treated WI-38 cells. Data are presented as Mean ± SD from biological triplicates. > 50 cells were counted for each of the biological repeats. Student's unpaired two-tailed t-tests were performed. *p < 0.05. i, RNA-FISH to detect SNUL-1 clouds in control and Aza+TSA-treated WI-38 nuclei. Nucleoli are visualized by rRNA (red). j, DNA-RNA-FISH of SNUL-1 RNA and 15CEN in control and Aza+TSA-treated WI-38 nuclei. k, DNA-RNA-FISH to detect SNUL-1 RNA and 15Sat III in control and Aza+TSA-treated in WI-38 nuclei. All scale bars, 5µm. DNA is counterstained by DAPI. Extended Data Fig. 4| The SNUL-1 cloud displays mitotically-inherited random monoallelic association. a, DNA-RNA-FISH showing the localization SNUL-1 RNA cloud and 15Sat III in hTERT-RPE1 cell nucleus. b, Plot showing the relative integrated density of the 15Sat III signals. Relative integrated density is calculated by dividing the score of the larger 15Sat III signal by that of the smaller 15Sat III signal. Data are presented as Median and interquartile range. n = 149. c, DNA-FISH to detect 15Sat III and 15CEN in Ctr and SNUL-depleted WI-38 nuclei. d, Plot showing the relative integrated density of the 15Sat III or 5CEN signals in control and SNUL-depleted WI-38 cells. Relative integrated density is calculated by dividing the measurement of the larger DNA-FISH signal by that of the smaller DNA-FISH signal. Data are presented as Median and interquartile range. n = 30. Mann-Whitney tests are performed. *p < 0.05; ns, not significant. e, RNA-FISH showing the distribution of SNUL-1 and rRNA in MCH2-10 nuclei. Arrows point at the SNUL-1 cloud. Arrowheads mark two non-nucleolar foci of unknown origin hybridized by the SNUL-1 probe only in MCH2-10 nuclei. f, DNA-RNA-FISH of SNUL-1 RNA and 15CEN or 15Sat III in MCH2-10 IPSC nuclei. Please note that the SNUL-1 probe-hybridized non-nucleolar foci is observed only after RNA-FISH and not after RNA-DNA-FISH treatments. g, Representative RNA-FISH image showing the localization of SNUL-1 cloud and SNRPN transcription site in MCH2-10 IPSC nucleus. Arrows point at the SNUL-1 cloud. Arrowheads mark two non-nucleolar foci of unknown origin hybridized by the SNUL-1 probe only in MCH2-10 nuclei. h, Quantification of the association rates between SNUL-1 and SNRPN in MCH2-10 IPSCs. Data are presented as Mean ± SEM from biological triplicates. > 50 cells were counted for each of the biological repeats. i, Representative DNA-RNA-FISH image to detect SNRPN RNA and 15Sat III in MCH2-10 nucleus. j, Quantification of the association rates between SNRPN RNA signal and the smaller and larger 15Sat III in MCH2-10 cells. > 100 cells were counted. All scale bars, 5µm. DNA is counterstained with DAPI.