Nuclear speckles enable processing of RNA from GC-rich isochores

Michał Małszycki; Lisa Martina; İbrahim Avşar Ilık; Daniela Salgado Figueroa; Nirmalya Dasgupta; Menşura Feray Çoşar; Keun-Tae Kim; Gil Carraco; Beatrix Fauler; David Meierhofer; Thorsten Mielke; Hiroo Imai; Cantaş Alev; Ferhat Ay; Tuğçe Aktaş

doi:10.1016/j.cell.2026.01.011

. Author manuscript; available in PMC: 2026 Jun 22.

Published in final edited form as: Cell. 2026 Feb 25;189(7):2024–2039.e25. doi: 10.1016/j.cell.2026.01.011

Nuclear speckles enable processing of RNA from GC-rich isochores

Michał Małszycki ^1,^*, Lisa Martina ^1,^*, İbrahim Avşar Ilık ¹, Daniela Salgado Figueroa ^2,³, Nirmalya Dasgupta ², Menşura Feray Çoşar ^4,^†, Keun-Tae Kim ^4,^†, Gil Carraco ^4,^†, Beatrix Fauler ¹, David Meierhofer ¹, Thorsten Mielke ¹, Hiroo Imai ⁵, Cantaş Alev ⁴, Ferhat Ay ^2,³, Tuğçe Aktaş ^1,^#

PMCID: PMC13283296 NIHMSID: NIHMS2178151 PMID: 41747727

Summary

Nuclear speckles are conserved, membrane-less organelles linked to various post-transcriptional processes. Here, we examined their roles in human cells by engineered, acute removal of SON and SRRM2, two conserved speckle core components characterized by intrinsically disordered regions (IDRs). Their removal results in a significant downregulation of GC-rich genes with short introns clustered within GC-rich isochores, caused by inefficient and chaotic splicing; in contrast, expression or splicing of genes outside these isochores remains unaffected. Comparative analysis across eukaryotes, from fungi to mammals, reveals that both GC-rich isochores and speckles are found exclusively in amniotes; moreover, the IDRs of SON have undergone notable expansion in the latter. Together, these findings suggest that the expansion of IDRs in vertebrates facilitated an increase in GC content by creating a condensate essential for splicing the by-products of this process: GC-rich, levelled exon-intron architectures.

Introduction

The human genome is highly heterogeneous with respect to its base composition, local gene density, and chromatin-chromatin interactions^1–3. At the highest level of chromatin organisation lie the active (A) and inactive (B) compartments, which preferentially interact with regions of the same compartment type⁴. The A compartment is further split into two subcompartments, A1 and A2: A1 is gene-dense, has elevated GC content, and contains short GC-rich genes, whereas A2 has lower gene density and contains long genes with overall lower GC content². Why and how GC-rich short genes cluster at a small number of genomic regions, whereas GC-poor long genes are more evenly spread around the human genome is unclear.

In an unexpected convergence of two scientific fields, 3D-genome organisation and RNA processing, nuclear speckles (or speckles) are found to be physically close to the GC-rich A1 subcompartment in human cells (Figure 1A, see also ref. ^3,5). Speckles are large, irregularly shaped nuclear condensates composed of several hundred proteins that are thought to be involved in transcription, splicing, termination, RNA modifications, and nuclear export⁶. Speckles were considered to be formed by a compendium of serine-arginine (SR) proteins without an organising core^7,8, limiting direct investigation of their potential function(s). However, our earlier work revealed that speckles are formed by two large and unstructured proteins, SON and SRRM2⁹, providing a framework to directly study their function.

Recent studies have shown that genes found within the GC-rich isochores, which are physically close to speckles (Figure 1A), have a distinct gene architecture consisting of short GC-rich introns flanking GC-rich exons, also known as the “leveled” exon-intron architecture prone to intron retention as alternative splicing outcome. Genes away from speckles tend to have long GC-poor introns flanking relatively GC-poor exons, referred to as “differential” exon-intron architecture prone to exon skipping¹⁰. This separation has been described for mammalian and bird genomes, but is not common to all vertebrates and is missing from invertebrates¹¹.

We created a cell line in which SON and SRRM2 can be specifically and rapidly degraded by small-molecule degraders (Figure 1B, Methods). Depletion of both proteins is necessary and sufficient to dismantle all speckles within less than 24 hours (Figure 1C and S1A), allowing us to study the direct consequences of speckle loss Using this system, we address the following questions:

Do speckles play a role in 3D genome organisation?
Are speckles involved in transcriptional regulation?
Are speckles involved in splicing?
How did the function(s) of speckles evolve?

Here, we show that speckles restrict random chromatin movements and are essential for the processing of genes localised in GC-rich, gene-dense regions often classified as the H3 isochore, A1 subcompartment or speckle SPIN-state¹². These genes often exhibit levelled exon-intron architecture, requiring high concentrations of spliceosomal components and splicing enhancers, which is provided by speckles. Speckles likely evolved in amniotes concurrently with the emergence of the GC-rich leveled architecture and their function in the expression of GC-rich RNA is conserved between birds and mammals. Taken together, our data elucidate the link between genome structure, its evolution, and RNA processing.

Results

Nuclear speckles restrict random chromatin movements

The preferential positioning of GC-rich gene-dense chromatin around speckles (Figure 1A) implies that they may play an organisational role in the 3D structure of chromatin^13,14. To gain further insights, we used an antibody-free staining method to visualise speckles and fluorescence recovery after photobleaching (FRAP) assays to measure their mobility. First, we stained cells with NHS-488, which reveals protein-dense regions¹⁵. NHS-488 staining shows that nucleoli and speckles are the most protein-dense regions of the nucleus and occupy a large portion of the nuclear space (Figure S1A). Importantly, when we remove speckles from cells by co-depleting SON and SRRM2 (henceforth, ΔSpeckles), the only NHS-488-dense regions that remain are the nucleoli, further confirming that these two proteins are the structural scaffold of speckles⁹ (Figure S1A). We then labelled the endogenously expressed SON and SRRM2 proteins to analyse protein mobility with FRAP. While SRRM2 recovers relatively quickly, SON recovers very slowly (Figure S1B–C, Video 1), and in fact only ~58% of SON signal recovers within 5 minutes after photobleaching. These results indicate that a significant fraction of SON is immobile and, consistent with a recent report¹⁶, suggest that SON imparts viscoelastic properties to speckles. In SON-depleted (ΔSON) cells, speckles collapse into spherical bodies that resemble droplets that form via liquid-liquid phase separation (Fig. 1C and ^17,18). The integrity of these SON-less speckles depends on the presence of SRRM2, since removal of SRRM2 in addition to SON leads to the dissolution of speckles (Fig. 1C and ⁹). We will therefore refer to these bodies as “SRRM2 droplets”. Consistent with this definition, in ΔSON cells, SRRM2 droplets initially collapse into small, spherical condensates, which then merge into fewer larger condensates (Figure S1D and Video 2).

If SON is the core protein that can resist deformations, removing it could alter chromatin movements. To test this idea, we expressed the core histone protein H2B tagged with a photoactivatable GFP (PA-GFP-H2B) into cell lines where we can deplete SON alone or in combination with SRRM2, and activated only a restricted area of the nucleus. Consistent with previous work¹⁹, in control-treated cells, PA-GFP-H2B (proxy for chromatin) remains restricted to one half of the nucleus within 6-7 hours after activation. On the contrary, both in ΔSON and ΔSpeckles cells, PA-GFP-H2B signal spreads to the entire nucleus within the same time frame (Figure 1D–F and S1E, F). These results suggest that acute depletion of speckles, likely driven by the depletion of the relatively immobile core protein SON, leads to increased chromatin mobility.

Since speckles are positioned in the vicinity of the A1 subcompartment and their loss alters chromatin dynamics, we wondered whether loss of speckles would have an effect on chromatin-chromatin interactions. To capture the temporal dynamics, we performed Pore-C²⁰ at 6, 12, and 18 hours after treatment (Methods). Contact matrices show that the overall chromatin-chromatin interaction patterns are largely preserved in ΔSpeckles cells, whereas off-diagonal contacts, representing distal, inter-compartmental contacts, become less defined (Figure S1G). Saddle plots²¹ show that a reduction in compartmentalisation strength is visible already 6 hours after loss of speckles and becomes most clear after 18 hours (Figure 1H and S1H, I). Long-distance, multiway contacts showed a genome-wide increase in contacts crossing the compartmental boundaries at the expense of both A and B intracompartmental interactions (Figure S1I). Furthermore, we noticed that ΔSpeckle, but not ΔSON or ΔSRRM2 cells, exhibit a peculiar “raisin”-like nuclear morphology with a ruffled nuclear envelope and abnormal H3K9me3 distribution (Figure 1G). Transmission electron microscopy (TEM) images confirm the irregular organisation of the nucleus in ΔSpeckle cells, where the double-layered nuclear membrane becomes ruffled (Figure S1J). Since nuclear architecture and chromatin compartmentalisation are closely related to chromatin states, we investigated the localization patterns of constitutive heterochromatin using CUT&RUN²² targeting H3K9me3. In all conditions, and especially in ΔSpeckles cells, we scored a moderate but consistent decrease of H3K9me3 at loci close to speckles with an accompanying increase in loci away from speckles. This trend is also visible in radial plots, where all differential genes are plotted inside a circle such that the linear distance from the centre of the plot to the dot representing a gene is proportional to the distance of the gene from speckles in wild-type cells (Methods, Figure 1I andS1K). Taken together, these results show that speckles are important for the maintenance of nuclear architecture of human cells through the viscoelastic properties provided by SON. Next, we investigated the consequences of speckle loss on gene expression.

GC-rich genomic regions require speckles for gene expression

To measure changes in transcription following speckles loss, we pulse-labelled and sequenced nascent RNA (Methods). Despite the changes observed in chromatin morphology, nuclear architecture, and H3K9me3 distribution, we score relatively mild changes in transcription in ΔSpeckles cells (Figure 2A left, and S2B bottom). We next measured changes in steady-state gene expression by sequencing polyadenylated (p(A)+) mRNA. In contrast to the results from nascent RNA-seq data, p(A)+ RNA-seq data showed severe changes in gene expression, with a strong bias towards downregulation in ΔSpeckles cells (Figure 2B left). Depletion of either SON or SRRM2 individually, while having a smaller effect size (Figure S2A, B top), affects the expression of a very similar subset of genes (Figure S2C). These results strongly suggest that the observed effects result mostly from the disruption of speckles as an ensemble.

Figure 2. — **(A)** Nascent RNA changes mildly in ΔSpeckles cells and does not show clear correlation with speckles proximity (padj < 0.05, | l2fc | > 1). (B) Steady-state transcript levels decrease; downregulated mRNAs are speckle-proximal, (C) Genes with high GC content are more significantly downregulated in ΔSpeckles cells and reside within speckle-proximal chromatin. (D) Downregulated genes have a higher GC content than upregulated ones. (E) Downregulated genes cluster in GC-rich regions. Horizontal bars represent the number of genes downregulated on mRNA level (all other chromosomes are shown in Figure S2H). (F) Radial plot showing proximity of chromosomes 18 and 19 to speckles. (G) Exemplary cell illustrating the proximity of chromosomes 18 and 19 to speckles. Scale bar=5 μm. (H) Quantification of chromosome-speckle surface contacts in (G) shows significant reduction upon SON depletion. *** p <= 1.00e-03, **** p < 1.00e-04. (I) Downregulated mRNAs originate from loci in GC-rich H3 isochores / A1 chromatin compartment / Speckle SPIN state. (J) Depletion of SON and/or SRRM2 shows the highest enrichment of downregulated mRNAs in H3 isochore regions. TRA2A/B depletion shows the opposite trend.

Speckles are concentrated in the vicinity of GC-rich chromatin (Figure 1A and ¹²) which is enriched with short GC-rich genes^3,23. Reflecting this organisation, downregulated genes in ΔSpeckles cells are significantly more GC-rich than upregulated ones (Figure 2D and S2H bottom). In addition, radial plots show that downregulated mRNAs originate almost exclusively from GC-rich loci that are close to speckles in wild-type cells (Figure 2A right, 2B right, and 2C). GC content vs. log₂-fold-change plots of individual depletions confirm this trend (Figure S2B). Gene ontology (GO)-term enrichment analysis of differentially expressed genes is not informative for downregulated genes, but upregulated ones are enriched with the terms “RNA processing”, “ribonucleoprotein complex biogenesis”, and “mRNA processing” (Figure S2E), likely representing a compensatory response.

Plotting downregulated genes on chromosome ideograms shows concentrated downregulation zones coinciding with GC-rich segments of the genome (Figure 2E and S2H top). A striking example is chromosome 19, which shows end-to-end downregulation in ΔSpeckles cells, and is the chromosome with the highest overall GC content²⁴ and speckle-proximity (Figure 2E, and S2D,H). Radial plots place chromosome 19 very close to speckles, unlike the similarly sized but GC-poor chromosome 18 (Figure 2F). DNA-FISH experiments show that chromosome 19 is not only close to speckles but frequently wraps around them (Figure 2G), explaining the particular sensitivity of this chromosome to loss of speckles. SON depletion reduces contacts between chromosomes 19 and the SRRM2 droplets left behind, underscoring the importance of SON for the formation of functional speckles (Figure 2H).

For a more detailed analysis of GC content and gene expression in ΔSpeckles cells, we used three independent genomic segmentation strategies (Figure S2F). First, by only using the average GC content, we segmented the genome into five isochores (H3, H2, H1, L2, and L1, Methods). Genes downregulated in ΔSpeckles cells are most enriched in H3, the isochore with the highest GC content, and to a lesser extent in H2, but depleted from elsewhere (Figure 2I top). Second, we assigned each gene to one of the six subcompartments called from Hi-C data. Genes downregulated in ΔSpeckles cells are enriched in the A1 subcompartment, and depleted elsewhere (Figure 2I middle). Finally, we used SPIN states, which combine data derived from Hi-C with data from proximity labelling techniques targeting speckles, lamina, and nucleolus¹² to segment the genome into 10 states. Genes downregulated in ΔSpeckles cells are enriched in the “Speckle” state (Figure 2I bottom), which is the most GC-rich SPIN state (Figure S2G).

SPIN states and Hi-C subcompartments are complex features derived from high-throughput genomics data and can differ between cell types, whereas isochores are based solely on genomic GC content and remain constant regardless of cell identity. The enrichment of downregulated genes in ΔSpeckles cells to H3 isochore is unusual, given that GC-rich isochores are typically associated with active transcription of housekeeping genes and are thus not generally expected to show preferential downregulation following speckles perturbations. This prompted us to ask whether this enrichment is specific to speckles perturbation. We thus extended our analysis to RNA-seq data generated by the ENCORE consortium, a subset of ENCODE that focuses on RNA-binding proteins (RBPs)²⁵. In none of these experiments did we observe enrichment of downregulated genes to the H3 isochore to the extent seen in speckles perturbations (Figure 2J and S2I). However, we did observe a similar enrichment of downregulated genes to the H3 isochore in a recently published RNA-seq data of ΔSON or ΔSRRM2 cells¹⁶. Notably, even though depletion of SRRM2 alone leads to mild changes in gene expression (Figure S2A–C), these analyses show that genes that are downregulated in SRRM2 depletion are very specific to the H3 isochore (Figure 2J and S2I). These results indicate that speckles serve a unique role in supporting the expression of GC-rich genes that reside in the H3 isochore at the post-transcriptional level.

GC-rich, leveled exon-intron architecture requires intact speckles for proper splicing

Previous work and our results suggest that speckles concentrate splicing-related RBPs, most notably SR-proteins, around short GC-rich genes^26,27. To test whether this enrichment is detectable transcriptome-wide, we analysed published FLASH datasets for all SR-proteins^28,29. All SR-proteins except SRSF10, SRSF12, TRA2A, and TRA2B were enriched on speckle-proximal exons (Figure S3A), indicating that speckles locally concentrate most SR-proteins on GC-rich genes^6,30. Quantifying splicing changes using rMats³¹ identified 852 differentially regulated intron retention events in ΔSpeckles cells compared to controls (Figure 3A). 89% of these events (758/852) show increased intron retention, indicative of decreased splicing efficiency. These events originate from speckle-proximal genes (Figure 3B right), consistent with previous studies^10,32, which predict that the GC-rich leveled exon-intron architecture is prone to intron retention events.

To our surprise, we detected an order of magnitude more differentially regulated exon skipping events in ΔSpeckles cells (Figure 3A), 80% of which (4287/5359) show increased exon skipping. The majority of these events originate from speckle-proximal genes (Figure 3B left). To serve as a comparison, we co-depleted TRA2A and TRA2B, which regulate thousands of alternative splicing events in human cells³³ and showed higher enrichment on speckle-distal exons (Figure S3A). Co-depletion of TRA2A/B resulted in 4461 differentially regulated exon skipping events, with 77% showing increased exon skipping (3451/4461) (Figure 3A). Unlike exon skipping events detected in ΔSpeckles cells, TRA2A/B-regulated exons originate mostly from genes away from speckles (Figure 3C) and are bound by TRA2A/B (Figure S3B). Next, we classified exons as speckle-dependent (skipped in ΔSpeckles cells), TRA2A/B-dependent (skipped upon TRA2A/B depletion), or a matched sample of all exons (Figures 3D–G and S3B–G). Speckle-dependent units were significantly shorter (Figure 3D), exceptionally GC-rich, and displayed a leveled architecture compared to the other groups (Figure 3E,F). Furthermore, TRA2A/B-dependent exons are enriched with GAA-rich motifs, well-established targets of the TRA2 family³⁴ (Figure S3D), while ΔSpeckles-dependent exons are enriched with GC-rich motifs (Figure S3D). In summary, speckles and TRA2A/B each regulate the splicing of a distinct subset of exons that differ in their exon-intron architecture (Figure 3E and S3C).

The prevalence of exon-skipping events in ΔSpeckles cells suggests that exons of speckle-proximal genes are mostly defined via exon definition rather than intron definition¹⁰. An important aspect of exon definition is the reliance on SR-proteins to define splice sites and stabilise early spliceosomal factors. To test whether the drastic exon skipping phenotype in ΔSpeckles cells is caused by the acute reduction in the local concentration of SR proteins for speckle-proximal genes, we carried out FLASH experiments targeting the active, phosphorylated form of SR proteins (pSR³⁵). pSR binding was highly enriched over the speckle-dependent exons in wild-type cells (Figure 3G left). The same exons, but not controls, show a severe reduction of pSR binding in ΔSpeckles cells (Figure 3G). Since the total SR protein levels do not significantly change in ΔSpeckles cells (Figure S4A, B), these results suggest that the specific reduction of pSR binding to speckle-dependent exons is likely caused by a reduction in the local protein concentration over the speckle-dependent exons. Consistently, despite being significantly less abundant than SRRM2 or other SR-proteins (Figure S4C) depletion of SON alone leads to a similar number of exon skipping events and a reduction in pSR binding on speckle-dependent exons (Figure S3E, F, H).

These results indicate that speckles enable accurate processing of GC-rich, leveled-structured genes by efficiently concentrating splicing factors on their exons.

Disrupted splicing in ΔSpeckles cells impairs nuclear export of mRNAs and induces RNA decay

Exon definition is proposed to have evolved as a mechanism to cope with expanding intron sizes observed in vertebrates³⁶, which is attributed to bursts of transposable element (TE) activity³⁷. Exon skipping in such long genes where consecutive exons are separated by kilobases of RNA leads to relatively simple splicing outcomes. However, in the unusual case of speckle-proximal genes, where introns are short, but exon-intron units are recognised via exon-definition, failure to timely assemble commitment complexes can expose multiple unmatched splice sites, which could lead to complex splicing outcomes. The complexity of splicing can be quantified using Whippet³⁸, which calculates an entropy value for each splicing event. Supporting the individual examples (Figure 3H and S3G), the entropy of splicing in ΔSpeckles cells is higher compared to control-treated cells (Figure 4A). We further verified three examples of high entropy splicing events using semi-quantitative RT-PCR (Figure S4E, F). SON-depletion alone already creates highly disordered patterns of splicing as seen for ΔSpeckles cells (Figure S4E,F and S3G), reinforcing the notion of its indispensable role as the main factor driving speckles nucleation.

Figure 4. — (A) Differentially spliced genes show an increase in splicing entropy in ΔSpeckles cells. (B) SF3A2 disperses upon SON or double depletion, whereas SRRM2 depletion has a milder effect. Scale bar=5μm. (C) Specific reduction of SF3B1 binding over speckle-dependent exons in ΔSON cells. (D) 710 genes are significantly retained in the nucleus (padj < 0.05) showing depletion from the cytoplasmic fraction in ΔSON cells. (E) Downregulated and upregulated genes (p-adj < 0.05, | l2fc | > 0.5) overlap between treatments. Downregulated genes are enriched in H3 isochore and splicing perturbation, and less efficiently exported from the nucleus. (F) Splicing of speckle-independent genes is not affected by speckles perturbation, while speckle-dependent genes have many hard-to-splice, GC-rich exons, which require intact speckles nearby to ensure a high concentration of splicing factors for proper splicing.

After SR-proteins bind to nascent RNA to define exons, the first contact between the splice donor and splice acceptor site is facilitated by the SF3 complex³⁹. Our immunofluorescence images indicate that SF3A2 is completely dispersed in ΔSON and ΔSpeckles cells (Figure 4B). In contrast, SRRM1 (Figure 1C) and many other speckle-associated proteins⁹ are only dispersed in ΔSpeckles cells. In addition, SF3B1 FLASH upon SON-depletion shows loss of SF3B1 binding on speckle-dependent exons (Figure 4C), while phosphorylated SF3B1 (pSF3B1), an indicator of active splicing⁴⁰, appears largely unaffected in ΔSON, ΔSRRM2, and ΔSpeckles cells (Figure S4D), consistent with the observation that the splicing process in general is not affected. The relocation of SF3B1, together with the spread of SR-proteins upon SON-depletion (Figure S3E, F), suggests a misorganization of splicing factors specifically for genes with leveled architecture in the absence of functional speckles.

Splicing is linked to and synchronized with mRNA export⁴¹ and speckles were hypothesized to be involved in the export of nuclear RNA^42,43. Loss of speckles or SON depletion, therefore, might lead to nuclear retention of mRNAs and downregulation of gene expression. We analysed nuclear retention by comparing the relative ratio of nuclear to total RNA in control vs ΔSON cells (Figure 1G, and S1J). This analysis revealed 710 mRNAs with significantly higher nuclear RNA in ΔSON cells compared to control cells which were consistently depleted from the cytoplasm (Figure 4D). Importantly, the majority of genes that show signs of nuclear retention are downregulated in ΔSRRM2, ΔSON, and ΔSpeckles cells, and originate from the GC-rich H3 isochore (Figure 4E and S4G).

While exon skipping is not typically expected to trigger nonsense-mediated decay (NMD), the strong downregulation of affected RNAs despite exon skipping being the most prevalent splicing outcome (Figure 3A, 2B) prompted an investigation into NMD’s role. We examined three genes with complex splicing patterns in ΔSpeckles cells to determine if these splicing events could cause NMD. We found that events that can potentially trigger NMD are strongly under-represented (Figure S4E). Consistently, NMD inhibition increased the abundance and number of RNA isoforms with exon skipping events in ΔSpeckles cells (Figure S4F). These results strongly suggest that splicing deregulation followed by NMD is a major driver of the RNA downregulation in ΔSpeckles cells.

Collectively, these results show that high local concentrations of SR proteins, SF3 complex, and other factors afforded by SON-nucleated speckles are necessary to enforce orderly splicing and export of short GC-rich, leveled-architecture genes. This gene architecture is found almost exclusively proximal to speckles in humans. In contrast, genes away from speckles do not have this architecture and are not affected by the loss of speckles (Figure 4F).

Is this gene architecture a universal feature of all animals or is it restricted to certain taxa? Can organisms have this gene architecture without speckles? We address these points in the following sections.

The expanding IDRs of SON co-evolve with nuclear speckles

Our results show that the majority of genes in the human genome are spliced normally with or without speckles, but the ~1500 genes that require speckles are short, contain short GC-rich introns, and reside inside the GC-rich isochores. Active genes outside these domains have lower GC content and contain long GC-poor introns flanking exons enriched with GAA-rich motifs, which, while independent from speckles, are sensitive to TRA2A/B levels (Figures 2,3, S2, and S3). These results imply that the leveled and differential exon-intron architecture dichotomy evolved as a response to or together with the differential positioning of genes relative to speckles.

Using ExOrthist⁴⁴, we compared vertebrate orthologs of human speckle-dependent and -independent exons, and introns flanking them (Figure 5A). In fish, both classes have similar GC content. However, speckle-dependent exons show an increase in GC content among amniotes, peaking in mammals, while speckle-independent exons remain constant (Figure 5A and S5A). Similarly, introns surrounding speckle-dependent exons display low GC content in fish, but show a notable rise in GC content within amniotes, while speckle-independent introns remain GC-poor across all vertebrates (Figure 5A and S5A). Even though both speckle-dependent exons and their introns gain GC content in mammals, the increase in GC content of introns is so dramatic that they catch up with the neighboring exons, creating the leveled exon-intron architecture (Figure 5A). Additionally, the orthologs of speckle-dependent introns in fish are five times longer than those of mammals. These introns are even longer in amphibians, but undergo an apparent shortening event in amniotes reaching their lowest values in mammals (Figure 5B, S5A). The absence of GC-leveled exons in fish suggests they either lack traditional speckles or have an unrelated set of speckle-dependent genes.

Previously, we showed that the two core speckle proteins, SON and SRRM2, expanded from ~600 aa in basal animals to ~5000 aa in some vertebrates, driven by the expansion of a single exon encoding their intrinsically disordered regions (IDRs) (Figure 5C,D and S5C)^9,27. SON depletion appears to be the main factor driving the observed phenotypes in all experiments (Figure 2H, 4B and S5B), suggesting that it functions as the primary scaffold of speckles and its ability to nucleate speckles co-evolved with its expanding IDRs.

To test how different IDRs of SON influence its nucleation properties and the splicing outcomes, we generated cell lines that inducibly express siRNA-resistant SON constructs encoding partial IDRs, or full-length orthologs of SON from different species (Figure 5D). Ectopic expression of human full-length SON completely restores the typical speckles morphology (Figure 5E and S5G), as well as most of the observed splicing defects (median RescueIndex in skipped exons (SE)=1.07, retained introns (RI)=1.02, see Figure 5F, H and Methods). Invertebrate and fish SON are shorter and IDR-poor, while still encoding the highly conserved C-terminus (Figure 5D and S5C,D). Neither of the two orthologs rescued speckle morphology when expressed in human cells (Figure 5E and S5G,H). Both orthologs contain a short IDR after the conserved C-terminus, which may allow them to rescue splicing defects if SON’s function resides in its conserved domains. However, neither construct rescued splicing (fish SON median RescueIndex SE=0.15, RI=0.18, fruit-fly SON median RescueIndex SE= 0.09, RI= 0.05, Figure 5F). This suggests that speckle nucleation is linked to SON’s role in processing GC-rich, leveled exon-intron architecture.

We thus reasoned that the main factor required for speckles cohesion is encoded in the long IDRs of SON. To test this, we cloned SON^ΔC, which includes SON’s IDRs but lacks the conserved C-terminus. As a complementary construct, we cloned SON^C-term-only (Figure 5D and S5C,D). Similar to the short orthologs, SON^C-term-only rescues neither speckle morphology (Figure 5G and S5G,H) nor splicing (median RescueIndex SE=0.22, RI=0.11, Figure 5H). This construct is distributed throughout the nucleoplasm, with some enrichment in the nucleolus (Figure 5G). In contrast, SON^ΔC rescues speckles morphology (Figure 5G and S5G,H). Although its rescue efficiency is lower than full-length SON, it is able to rescue most of the splicing defects (median RescueIndex SE=0.63, RI=0.66, Figure 5H), suggesting that while the conserved C-terminus is necessary for SON to function at full capacity, the role of SON in splicing is primarily driven by the rapidly evolving IDRs. We hypothesized that an RS-rich domain may be required for the speckles localization of SON and its ability to nucleate speckles. Surprisingly, the construct containing the C-terminus together with the RS-rich stretch (SON^C-term-RSRP, Figure 5D) localizes to the nucleolus and partially recruits poly(A)+ RNA and SRRM2, merging these two nuclear compartments together (Figure 5G and S5G). This construct is able to rescue some of the splicing defects, especially exon skipping (median RescueIndex SE=0.40, RI=0.27), but the distribution is highly variable (Figure 5H). In mammals, the net charge of the unphosphorylated SON protein at physiological pH is negative, while in fish and invertebrates it is positive due to the absence of two large negatively charged regions found in mammalian SON (Figure 5D and S5C). To test whether the positive charge of the RS-rich region in humans needs to be balanced by the negatively charged region, we expressed a construct that, in addition to the C-terminus and RSRP, encodes one of the negatively charged regions (SON^{C-term-RSRP-NEG}, Figure 5D). This construct partially restores speckle morphology; however, they are more widely dispersed throughout the nucleus and partially localize to the nucleoli (Figure 5G and S5G,H). Splicing deregulation showed slight improvement compared to SON^C-term-RSRP, particularly in rescuing retained introns. However, the rescue of splicing remained less effective than with the full-length SON or SON^ΔC (median RescueIndex SE=0.54, RI=0.42, Figure 5H). The events that are less efficiently rescued by SON^C-term-RSRP, SON^{C-term-RSRP-NEG}, and to a lesser extent by SON^ΔC tend to be more GC-rich (Figure S5E), highlighting that genes with high GC content require functional speckles. Consistent with their inability to rescue splicing defects, fruit-fly SON, fish SON, and SON^C-term display spherical nuclear p(A)+ bodies, whereas full-length SON and SON^ΔC restore the elongated shape (Figure S5F–H). Interestingly, SON^C-term-RSRP and SON^{C-term-RSRP-NEG}, which rescue splicing to intermediate levels, form more elongated p(A)+ bodies than wild-type or full-length SON. This indicates that these nuclear bodies are partially functional, consistent with their ability to partially rescue splicing.

In conclusion, the IDRs of human SON, rather than its conserved domains, are both necessary and sufficient to maintain speckles cohesion and facilitate precise splicing. The complete inability of fish or fruit-fly SON to rescue speckle defects when expressed in human cells supports the idea that these organisms lack functional speckles.

The emergence of nuclear speckles coincides with the evolution of the gene-rich GC isochores in amniotes

Given these results, species with shorter SON orthologs might not possess speckles. Similarly, if the primary role of speckles is enabling proper splicing of exons flanked by short, GC-rich introns, species lacking this exon-intron architecture likely do not have or need speckles. Previous work suggests that GC content elevation, primarily observed in the GC-rich isochores of mammals and birds, drives the evolution of the leveled exon-intron architecture¹¹. Instead of directly testing if isochores exist or not in non-avian and non-mammalian species, we test a salient feature of the isochore theory: the relative enrichment or depletion of genes in GC-rich and GC-poor isochores, respectively. Plotting gene enrichment scores vs. isochore coverage shows a very specific pattern in mammals and birds: genes are enriched in GC-rich isochores, especially H3, and depleted from the GC-poor L1 and L2 (Figure 6A). In contrast, fish and invertebrate genomes do not show this pattern, suggesting they either lack isochores, or if isochores do exist, they do not affect gene distribution. As a further validation, we calculated Cramér’s V value for each species (Methods). Consistent with the enrichment-vs-coverage plots (Figure 6A), birds and mammals show the highest average values, whereas fish show the lowest (Figure S6A). Invertebrates exhibit generally low values with broad distributions, a pattern also observed in fungi. In fungi, many species contain fewer genes than expected in L1, likely reflecting repeat-induced point mutation, a process that targets repetitive elements and reduces GC content through C→T conversions⁴⁵. Furthermore, we identified the isochore contributing most to the Cramér’s V value and determined whether it drives enrichment or depletion. This analysis indicates that gene depletion in L1 is the predominant driver of gene distribution heterogeneity in fungi, whereas gene enrichment in H3 underlies the heterogeneity observed in amniotes (Figure 6B).

Figure 6. — (A) Comparison of gene density within isochores. (B) The isochore with the most substantial contribution to gene distribution heterogeneity (“driver isochore”, Methods) summarised for all genomes depicted in (A). (C) Comparisons of genome composition of all eukaryotic genomes is shown with regards to prevalence of leveled gene architecture and the clustering of exons with the leveled architecture. Each dot represents one genome. (D) UMAP representation of gene architecture with respect to intron length and GC content. (E) p(A)+ RNA FISH reveals speckles are present in turtle, chicken, mouse and human cell lines but absent in fish and fruit fly cell lines. Scale bar=5μm. Intron GC content in correlation with intron length is shown for each organism (bottom). (F) Images of p(A)+ RNA and exogenous SON in fish cells expressing SON orthologs from either human, fruit fly or fish. Only human SON is able to nucleate speckles. (G) Isochore enrichment of genes downregulated upon SON-KD in chicken cells shows enrichment in H3 isochore. (H) Scatterplots comparing gene expression changes upon SON depletion in chicken and human cells. (I) Diagram summarizing the findings on the evolution of nuclear speckles. IDRs of SON expanded in an early amniote giving rise to the ancestral speckle-like structures. This new organization, likely as a side effect, improved the recognition of short, GC-rich transcripts and facilitated the accumulation of GC content due to GC-biased gene conversion, ultimately leading to the isochore genome organization observed in all amniotes.

The positive gene enrichment scores observed in the most GC-rich isochores in amniotes are consistent with the idea that these organisms possess genes with leveled architecture. It is not clear whether the lack of positive gene enrichment scores in anamniotes and invertebrates precludes the existence of such architecture in these organisms. To address this point, we formalised the concept of GC-rich, leveled architecture and examined its presence across all eukaryotic genomes (Methods, Figure S6B). This analysis revealed that leveled architecture is predominantly found in vertebrates (Figure S6C): GC-rich leveled exons are common in mammalian genomes, as well as birds, turtles, alligators & crocodiles, and to a lesser extent in lizards & snakes, but virtually absent from fish, while amphibians showed intermediate values (Figure 6C). Interestingly, the amount of GC-rich leveled exons is higher in land vertebrates with smaller genomes (Figure S6E).

GC-rich leveled exons comprise ~2.5% of all exons in the human genome (Figure S6E), are speckle-proximal (Figure S6D) and are flanked by introns that are an order of magnitude shorter than the average intron size (134nt vs 1489nt). Relative shortness of GC-rich introns is also clearly visible as an arc extending towards the top-left corner of intron length vs GC content plots (Figure S6B, F). This trend is only visible in mammals, birds, and turtles, but absent from fish and invertebrates (Figure S6F). We used the underlying data from these plots to construct a UMAP⁴⁶ (Figure 6D). Consistent with the parametric approach (Figure 6C and S6C), this approach informed only by GC content and the length of introns, clusters birds and mammals close to each other, with turtles, alligators & crocodiles in between them, while fish, invertebrates, plants and fungi form separate clusters (Figure 6D).

If the primary function of speckles is to enable splicing of GC-rich exons surrounded by short GC-rich introns, these results strongly suggest that speckles do not exist in invertebrates and fish, which have short SON orthologs, do not seem to have GC-rich isochores enriched with genes, and have a low frequency of leveled exon-introns. On the other hand, speckles are likely present in all amniotes, with the situation in amphibians remaining ambiguous. To experimentally test the predictions of these evolutionary analyses, we used the speckled pattern of p(A)+ RNA FISH signal as a proxy to assess the presence or absence of nuclear speckles in cell lines of invertebrate (fruit fly), fish (fathead minnow, zebrafish, eel), turtle (common box turtle), bird (chicken), and mammals (human, orangutan, chimp, rabbit, rat, mouse) (Figure 6E, S6G,H).

In mouse cells⁴⁷, p(A)+ RNA signal forms condensed, irregular bodies that exclude DNA staining, similar to human speckles (Figure 6E). Birds with their long SON orthologs and the abundance of leveled GC-rich exons in their genomes (Figure 5D, S5C, 6B–E) were expected to have speckles, which is supported by the condensed, irregularly shaped p(A)+ RNA signal in chicken cells (Figure 6E). Turtles, which also possess long SON orthologs and GC-rich, leveled exons similar to those found in birds and mammals (Figure 5D, S5C, 6B–E), exhibit concentrated, irregularly shaped p(A)+ RNA signal in the nucleus that excludes DNA (Figure 6E). Finally, as predicted from the overall lack of GC-rich leveled exons and long SON orthologs (Figure 5D, S5C, 6B–E), neither fish nor invertebrate cells have concentrated p(A)+ in the nucleus, indicating a lack of condensates that could be classified as speckles (Figure 6E and S6G,H). To further verify this result, we transfected fish cells with human, fish, or fruit-fly SON. Remarkably, only human SON can nucleate condensates that resemble speckles (Figure 6F and S6I) and concentrate poly(A)+ RNA of fish origin within these structures, further confirming that SON is the primary protein responsible for the formation of speckles. As fish have very few leveled exons, human SON-induced speckles produce only limited splicing changes (Figure S6J). On the other hand, our results suggest that birds, whose last common ancestor with humans was an early amniote, possess functional speckles, which we tested by knocking down SON in chicken cells. In line with the results in human cells, depletion of SON from chicken cells resulted in downregulation of mRNAs originating in the GC-rich H3 isochore (Figure 6G, H), indicating that the speckles of birds and mammals are functionally analogous.

Discussion

The discovery of nuclear speckles as a cytological phenomenon dates back more than 100 years⁴⁸, and currently speckles are considered to be a universal feature of human interphase nuclei⁴³. In order to provide insights into the functions of speckles in human cells, we set out to deeply characterise their role in the regulation of gene expression and nuclear architecture. Taken together, our results show that SON, together with SRRM2, forms a structure with viscoelastic properties that limits random chromatin movements and concentrates factors important for defining exons and matching consecutive splice sites together. This structure is required for the orderly splicing of a relatively small subset of introns, and is not essential for splicing in general. However, failure to splice these introns has a disproportionate effect on gene expression due to the isochore structure of the human genome: these introns, which are short and GC-rich, are found almost exclusively within short, GC-rich genes which are enriched at GC-rich isochores that encompass many housekeeping genes. These genes are consistently highly expressed across diverse cell types¹³ and are localised close to speckles in the 3D space of the nucleus^12,13. Genes outside these domains tend to have long introns, are away from speckles¹², and are spliced efficiently with or without speckles. Our results suggest that the primary function of speckles is to enable the splicing of exons separated by short GC-rich introns, indicating that the current isochore structure of the human genome is contingent on the presence of speckles.

The specific exon-intron architecture that requires speckles for their splicing has been shown to lower splicing efficiency in yeast^49–51, which do not have a protein orthologous to SON. Currently, the only known mechanism of isochore formation is meiotic recombination, which can increase GC content by GC-biased gene conversion (gBGC)^52,53. A well-studied example of this process is the Fxy gene. The 3’-end of this X-linked gene has recently translocated into the pseudoautosomal region of the X chromosome in the mouse genome, where recombination rates are high. As a result, its 3’-end has a dramatically higher GC content in mice compared to other rodents. The increase in GC content, as predicted by gBGC, is observed on exons as well as introns of the gene concomitant to the loss of tens of kilobases of DNA from intronic regions⁵⁴. Large sections of DNA can be removed by meiotic recombination, but how does a gene gain DNA? Accumulating evidence suggests that bursts of transposon activity accompanying the water-to-land transition have increased the size of the genomes of vertebrates^37,55. Once inserted, TEs can only be removed via meiotic recombination. This process of DNA gain by new TE insertions has been shown to be balanced by DNA loss by large segmental deletions of > 10kb in birds and mammals⁵⁶ (Figure 6B–E). The ability to remove large segments of DNA via meiotic recombination could have provided an evolutionary advantage to organisms whose genes and genomes were inflated by transposons by recreating compact genomes but with elevated GC content. Our results strongly suggest that the IDR of SON expanded in an early amphibian-like amniote ancestor (Figure 6I). This long, IDR-rich ancestral SON could form a structure similar to human speckles, likely enhancing the efficiency of splicing by bringing splicing factors and RNA together and cushioning the vast amounts of their chromatin. This new organization may have improved the recognition of short, GC-rich transcript architectures and enabled the accumulation of GC content due to gBGC, ultimately leading to the isochore genome organization observed in amniotes.

Limitations of the Study

Our evolutionary analysis suggests that speckles emerged following the expansion of SON and SRRM2 intrinsically disordered regions in early amniote ancestors; however, the scarcity of Amphibian data and limited access to suitable cell lines as well as SON-specific antibodies reactive across multiple species hampers our ability to pinpoint the exact timing and mechanisms of their origin, or to experimentally verify if the emergence of isochores depends on the presence of speckles. It is also not trivial to fully disentangle the functions of SON from those of speckles as a whole. This gap in knowledge presents an exciting opportunity for future research employing Amphibian models, which could provide critical insights into these evolutionary changes.

Resource availability

Lead Contact

Tuğçe Aktaş, aktas@molgen.mpg.de

Materials Availability

All the materials generated in this study are available upon request from the lead contact.

STAR METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Human HCT116 (ATCC, CCL247) and derivative cell lines were cultured in HyClone McCoy’s 5A medium (Cytiva, cat. no. SH30200.01) supplemented with 10% fetal bovine serum (FBS; Thermo Fisher Scientific, cat. no. 10270106) and 1% Penicillin/Streptomycin (Thermo Fisher Scientific, cat. no. 15140148) at 37°C in the presence of 5% CO₂. Human Flp-In T-REx HEK293 cells (Thermo Fisher Scientific, catalogue no. R78007) were grown in DMEM with glutamax supplemented by Na-Pyruvate and High Glucose in the presence of 10% fetal bovine serum and penicillin/streptomycin. Unmodified cells were cultured at a final concentration of 100 μg/ml zeocin (Thermo Fisher Scientific, catalogue no. R250-01) and 15 μg/ml blasticidin (Thermo Fisher Scientific, catalogue no. A1113903).

Mouse 3T3 cells were maintained in DMEM with glutamax supplemented by Na-Pyruvate and High Glucose in the presence of 10% fetal bovine serum and penicillin/streptomycin. Cells were grown at 37°C in 5% CO₂.

Chicken DF1 cells were grown in DMEM with glutamax supplemented by Na-Pyruvate and High Glucose in the presence of 10% fetal bovine serum and penicillin/streptomycin at 39°C in 5% CO₂.

Turtle TH-1 cells were maintained in DMEM with glutamax supplemented by Na-Pyruvate and High Glucose in the presence of 10% fetal bovine serum and penicillin/streptomycin and grown at 27.5°C in 2.5% CO₂.

Fish EPC cells were cultured in EMEM (ATCC, cat. no 30-2003) supplemented with 10% fetal bovine serum and 1% Penicillin/Streptomycin at 24°C in 5% CO₂.

Fruit fly Schneider-2 (S2), S2R+ and Kc cells were grown in Schneider’s Drosophila-Medium (Thermo Fisher Scientific, cat. no 21720024) supplemented with 10% fetal bovine serum and 1% Penicillin/Streptomycin at 25°C in atmospheric CO₂ level. Drosophila OSS cells were cultured in M3 medium (S3652, Sigma-Aldrich) supplemented with 10% fetal bovine serum, 10μg/mL insulin (INS-K, Capricorn Scientific), 0.06% Glutathione (Sigma-Aldrich), 10% fly extract (Akhtar lab, MPI-IE, Freiburg) and 1% Penicillin/Streptomycin at 25°C in atmospheric CO₂ level.

Non-human primate species chimpanzee (Pan troglodytes schweinfurthii) and orangutan (Pongo abelii) fibroblast cells were obtained from Kyoto University Primate Research Institute (KUPRI). Cells were cultured on 0.2% gelatin-coated 10 cm cell culture dishes with fibroblast culture media under 37 °C, 5% CO₂. Human dermal fibroblasts, adult (HDFa) (ThermoFisher - C0135C) were cultured under the same conditions as non-human primate species and chicken cells. Rabbit, Rat and Eel; each cell line was obtained from RIKEN BRC - Japan. Rabbit (Oryctolagus cuniclus) fibroblasts (SIRC) and Rat (Rattus norvegicus) lung fibroblast cells (JTC-19) were cultured with DMEM High Glucose (ThermoFisher) containing 10 ng/μl bFGF (Nacalai Biotech), 10% FBS (Gibco), 1% MEM (Gibco), 1% Sodium Pyruvate (Nacalai Biotech), 1% Glutamax (Gibco), gentamicin (ThermoFisher) and antifungal & antibiotic solution (AA) (Nacalai Biotech) at 37 °C, 5% CO₂. Eel (Anguilla japonica) hepatocyte cells were cultured with 50:50 mix of mTESR (Stem Cell Technologies) and Leibovitz’s L-15 Medium (ThermoFisher) with 10% HEPES (Gibco), 20% FBS and 10 ng/μl bFGF medium at 26 °C, 0% CO₂.

The sex of cells is as follows: human HCT116, male; human Flp-In T-Rex HEK293, female; human HDFa, female; mouse 3T3, male; chicken DF1, sex unspecified; common box turtle TH-1, sex unspecified; fathead minnow EPC, sex unspecified; fruit fly S2, male; fruit fly Kc, female; fruit fly S2R+, male; fruit fly OSS, female; chimpanzee, female; orangutan, female; rabbit SIRC, sex unspecified; rat JTC-19, female; eel Hepa-E1, sex unspecified.

METHOD DETAILS

Generation of stable cell lines

Endogenous SON was homozygously tagged with a HaloTag^® (Promega, for sequence, see Supplementary Table 1). Construct containing HaloTag^® and a short, flexible linker surrounded by ~400bp long homology arms and a construct expressing guide RNA targeting first exon of SON, Cas9 and Puromycin resistance were transfected in 1:1 ratio to HCT116 cells using FuGene HD transfection reagent according to the transfection protocol provided by the manufacturer (Promega, cat. no. E2311). After 48 hours, the cells were subjected to selection with 1μg/m Puromycin for 48 hours to kill untransfected cells. The resulting single cell derived colonies were then left to grow for 7 days in a culturing medium without antibiotic selection. After a week, cells were labeled overnight with 100nM HaloTag^® TMR ligand (Promega, cat. no. G8252). Round colonies with a clear and uniform “specklish” fluorescence were picked manually to a separate vessel and treated as single-cell derived clones. Homozygous insertion of HaloTag^® into the SON locus was confirmed by PCR genotyping with two independent primer pairs, western blotting and Sanger sequencing of the SON 5’ region.

Endogenous Halo-SON cell line was then used for tagging endogenous SRRM2 with a construct containing SNAP-tag (NEB), FKBP^F36V,T2A and puromycin selection cassette surrounded by ~400bp homology arms targeting 3’ portion of SRRM2 locus before the STOP codon (for the sequence see Supplementary Table 1). The cells were transfected with the aforementioned construct as well as a construct expressing guide RNA and Cas9 in 1:1 ratio using FuGene HD transfection reagent. The cells were then selected with 1μg/ml Puromycin for 11 days until all unedited cells were dead. The cells were then labeled for 30 min with 3μM SNAP-Cell TMR-Star (NEB, cat.no. S9105S). Round colonies with clear and uniform “specklish” fluorescence were picked to a separate vessel and treated as single cell derived clones. Homozygous insertion of the construct was confirmed by PCR genotyping, Western Blot upon dTAG (Tocris, cat.no 6605) induced depletion and Sanger sequencing of SRRM2 3’ end.

Exogenous SON truncation constructs and SON orthologs from Fish and fruit fly were expressed in Flp-In T-REx HEK293 cells. The cells were grown in DMEM with glutamax supplemented by Na-Pyruvate and High Glucose in the presence of 10% fetal bovine serum and penicillin/streptomycin. For generation of stable cell lines, pOG44 (Thermo Fisher Scientific, catalogue no. V600520) was cotransfected with pcDNA5/FRT/TO (Thermo Fisher Scientific, catalogue no. V652020) containing the gene of interest at a 9:1 ratio. Cells were transfected with Lipofectamine 2000 (Thermo Fisher Scientific, catalogue no. 11668019) on a six-well-plate format with 1 μg of DNA (that is, 900 ng of pOG44 and 100 ng of pcDNA5/FRT/TO + GOI) according to the transfection protocol provided by the manufacturer. All transgenes were cloned with an N-terminal HaloTag^® construct separated from the SON truncation construct by a short flexible linker. The truncation constructs were amplified from the cDNA prepared from Flp-In T-REx HEK293 cells with primers that introduce 6 synonymous mutations into the siRNA target sequence (for construct sequences see Supplementary Table 1). Twenty-four hours following transfection, cells were split among three wells of a six-well plate at dilution ratios of 1:6, 2:6 and 3:6 to allow efficient selection of hygromycin B (Thermo Fisher Scientific, catalogue no. 10687010). Hygromycin selection was started 48 h following the transfection time point, with a final concentration of 150 μg ml−1, and refreshed every 3–4 days until non-transfected cells were totally dead. Induction of the transgene was performed overnight at a final concentration of 100 ng/ml doxycycline (DOX). Constructs were validated by sanger sequencing of the expression plasmids. Cell lines were validated by immunoblotting and by purification of expression constructs with Halo-Trap resin (ChromoTek, cat. no. otma) and sequencing the purified proteins with mass spectrometry.

Immunoblotting

Sample lysis and western blotting was adapted from ⁶⁵. HCT116 WT and HCT116 Halo-SON SRRM2-SNAP-FKBP cells were incubated for 18 hours in one of the 6 conditions: HCT116 WT cells with i) 1μM HaloPROTAC3 (MedChemExpress, cat #HY-111997) + 1μM dTAG13 (Tocris, cat. no 6605), or ii) DMSO, as well as HCT116 Halo-SON SRRM2-SNAP-FKBP cells with iii) DMSO, iv) DMSO + 1μM dTAG13, v) DMSO + 1μM HaloPROTAC3, or vi) 1μM dTAG13 + 1μM HaloPROTAC3. After the incubation 1 million cells was harvested, washed with DPBS, centrifuged (1 min, 1000 xg) and placed on ice. The cells were then lysed for 30 min on ice with RIPA buffer (50mM Tris-HCl, 150 mM NaCl, 0.5% Sodium Deoxycholate, 0.1% SDS, 1% Igepal and 2mM EDTA) with cOmplete protease inhibitors (Roche, cat. no. 05056489001) and PhosSTOP (Roche, cat. no. 04906837001). Next to digest the DNA and RNA benzonase wash buffer was added in 1:1 ratio (50mM TRIS-HCl, 50mM NaCl, 2mM MgCl₂, cOmplete protease inhibitors, PhosSTOP and 1:500 Benzonase HC (Millipore, cat. no. 71205)) and the sample was incubated for 20min at room temperature with rotation. After lysis the sample was centrifuged at 17000xg for 20 min at 4°C. Supernatant was mixed with 4x NuPAGE LDS Sample Buffer (Thermo Fisher Scientific, cat. no. NP0007) and 10x NuPAGE^™ Sample Reducing Agent (Thermo Fisher Scientific, cat. no. NP0004), boiled for 15min at 95°C and loaded on NuPAGE 3 to 8%, Tris-Acetate gel (Thermo Fisher Scientific, cat. no. EA0375PK2). The samples were run for 2 hours 30min at 80V in 1X tris-acetate running buffer (Thermo Fisher Scientific, LA0041). Transfer was conducted for 15 hours at 20V in Tris-glycine transfer buffer (25mM Tris, 192mM Glycine, 10% Methanol) onto 0.45 PVDF membrane (Millipore, cat. no. IPVH00010). Membrane was incubated overnight at 4°C in antibody dilution in 5% BSA (see section Antibodies), and developed with Clarity ECL (BioRad, cat. no. #1705061).

siRNA transfections and transient overexpression

Wild-type HCT116 cells were transfected with a final concentration of 5nM each (TRA2A and TRA2B siRNAs in total amount to a final siRNA concentration becomes 10nM) using silencer select siRNAs (Thermo Fisher Scientific, Catalog number: 4427037 for 1nM scale) and with RNAiMAX (Thermo Fisher Scientific Catalog number: 13778030) on 6 well plates (~200K were used per replicate). The siRNA ID for TRA2A is s26665, TRA2B is s12749. Cells were harvested on the second day of the knockdown.

Flp-In T-REx HEK293 cells (3 replicates, ~100K cells each) harbouring siRNA resistant exogenous SON truncation constructs were transfected using RNAiMAX with a final concentration of 10nM of SON silencer select siRNA (siRNA ID: s13278) to deplete the endogenous SON. 6 or 24 hours after transfection, the medium was replaced to a regular growth medium supplemented with 0.1 μg ml⁻¹ of Doxycycline (DOX) to induce the transgene expression. The cells were harvested 24 or 48 hours after transfection.

Chicken (Gallus gallus) DF1 cells (3 replicates, ~40K cells each) were transfected with 10nM siRNA pool targeting chicken SON (siPOOL against NCBI ID: 770637; siTOOLs BIOTECH), or a negative siRNA control (siPOOL Neg-C-120; siTOOLs BIOTECH) for 48 hours, using using RNAiMAX standard protocol.

To test the effect of overexpressing SON in Fish cells, the human SON construct was transiently transfected into EPC cells with Fugene transfection reagent, using a construct where EGFP is tagged with the nuclear localization sequence (NLS) as a control. Transfected cells were then FACS-sorted using either EGFP or HaloTag^® TMR high signal and lysed to be used in p(A)+ RNA sequencing.

Cytoplasmic and nuclear RNA separation

RNA fractionation was performed in ΔSON cells, which show a milder downregulation of genes compared to ΔSpeckles cells (Figure S2A, B), but do not suffer from the severe nuclear morphology disruptions observed in ΔSpeckles cells, which could disrupt the biochemical fractionation. The protocol for the separation of cytoplasmic and nuclear RNA was adapted from ⁶⁶. Briefly: Endogenous Halo-SON cells were incubated under either of two conditions: i) 300nM ent-HaloPROTAC3 (Promega, GA4110) negative control or ii) 300nM HaloPROTAC3 (Promega, GA3110) SON depletion. After 16 hours of incubation cells were washed 3 times with warm DPBS (Thermo Fisher Scientific, cat. no. 14190144) to remove mitotic cells that are enriched in PROTAC treatment.

After that the cells were detached from the plate with 0.05% trypsin, harvested with warm medium and adjusted to 750 thousand cells per replicate. Cells were then washed with warm DPBS and centrifuged at 130xg for 5min. Half of the pellet was separated and immediately lysed with RNA lysis buffer (Quick-RNA MicroPrep kit, ZYMO, cat. no R1050). The rest of the pellet was resuspended in ice-cold RLN buffer (50mM Tris.HCl pH8, 140mM NaCl, 1.5mM MgCl2, 10mM EDTA, 5mM β-mercaptoethanol, 0.5% Igepal, 1%BSA) and incubated on ice for 5 min with tube flicking every minute. This treatment results in lysis of the cytoplasm and release of the cytoplasmic RNA to the supernatant, while keeping the nuclei intact. After incubation, the tubes were centrifuged at 300xg for 5 min at 4°C, the supernatant transferred to a new tube and immediately mixed with 4 parts of RNA lysis buffer. Pelleted nuclei were washed once more with ice-cold RLN buffer, resuspended in 1mL of ice-cold S1 buffer (250mM Sucrose, 10mM MgCl2, 1%BSA) and gently flicked to detach them from the bottom of the tube. This pellet was then gently deposited with cut-tip pipette on top of 3mL of ice-cold S3 buffer (880mM Sucrose, 0.5mM MgCl2, 1%BSA), and centrifuged at 2800xg for 10 min at 4°C. Lastly, the supernatant was removed, and the pellet resuspended in the RNA lysis buffer. All the fractions were then passed through QIAshredder (Qiagen, cat. no. 79656) and the RNA was purified with the Quick-RNA MicroPrep kit (ZYMO) with DNase treatment. The quality of fractionated RNA was tested with RNA denaturing gel as well as RT-PCR test with known nuclear and cytoplasmic RNAs.

RNA-sequencing

p(A)+ RNA from total cell lysates was purified using Dynabeads mRNA DIRECT Purification Kit (Thermo Fisher Scientific, cat. no. 61012) and processed into libraries using KAPA RNA HyperPrep Kit (KK8544, Roche).

Total RNA from the whole cell lysate and fractionated cells (nuclear and cytoplasmic fractions) was processed into the libraries with the KAPA Stranded RNA-Seq kit with RiboErase (HMR) protocol (KAPA RiboErase kit KK8481 and KAPA Stranded RNA-Seq kit KK8400, Roche).

RT-PCR

HCT116 Halo-SON SRRM2-SNAP-FKBP were cultured for 16 hours in the presence of DMSO (control) or 1μM SMGi-11j (NMD inhibitor MedChemExpress, HY-124719) as well as one of i) DMSO ii) DMSO + 1μM HaloPROTAC3 (MedChemExpress, cat #HY-111997) iii) DMSO + 1μM dTAG13 (Tocris, cat. no. 6605) or iv) 1μM HaloPROTAC3 + 1μM dTAG13. After the incubation the RNA from the whole cell lysates was purified with Quick-RNA MicroPrep kit (ZYMO, cat. no R1050). p(A)+ RNA was purified using Dynabeads mRNA DIRECT Purification Kit. cDNA from p(A)+ RNA was prepared using PrimeScript^™ RT Master Mix (Takara cat. no. RR036A). 10% of the cDNA from each sample was used for a PCR reaction with DreamTaq PCR Master Mix (Thermo Scientific cat. no. K1081) according to manufacturer specifications. See Supplementary Table 1 for primer sequences.

Nascent 4sU-labeled RNA-seq

HCT116 Halo-SON;SRRM2-SNAP-FKBP cells were incubated for 18 hours in either DMSO, 1μM dTAG13, 1μM HaloPROTAC3 or 1μM dTAG13 + 1μM HaloPROTAC3. Labelling was carried out at a final concentration of 0.5mM 4sU (Merck/Sigma Aldrich cat. #T4509) for 15 min at 37°C and 5% CO₂. Cells were then placed on ice to stop transcription, scraped in cold DPBS, pelleted, snap-frozen in liquid nitrogen, and left at −80°C until library prep. Two replicates were measured for each condition. The RNA was obtained following the NucleoSpin RNA kit (Macherey-Nagel, cat. no. 74095550), according to the manufacturer’s instructions. Buffer amounts were scaled-up to account for the number of cells (600μl buffer RA1 and 6μl β-mercaptoethanol). Cells were passed through a 0.9mm 20G syringe and the optional filter. RNA was eluted in 100ul TE buffer. Fragmentation was performed with BioRuptor Pico for 20 cycles with high power 30s/30s on/off. 15ug of RNA in 300ul of TE buffer were present in each tube that underwent sonication. RNA was then purified with the kit RNA clean and conc-25 (Zymo research, cat. no. R1017), following the manufacturer’s instructions. Elution was carried out in 200μl TE buffer.

RNA was biotinylated using MTSEA biotin-XX (Biotium, cat. no. 90066-1) as described in ⁶⁷. The eluted RNA was purified from the excess free biotin linker using MaXtract high-density phase-lock-gel tubes (Qiagen, cat. no. 129056). MaXtract high-density phase-lock-gel tubes were prepared by centrifugation at 12000xg for 30sec at RT. An equal volume of ROTI^®Phenol/Chloroform/Isoamyl alcohol (ROTH, cat. no. A156.1) was added to the biotinylated RNA and the mixture was added to the phase-lock-gel tubes. Centrifugation at 12000xg for 5min at 4°C followed and the upper aqueous phase containing the RNA was transferred to a new tube. 0.1 volumes of 5M NaCl and 1.1 volumes of isopropanol were added, and the mixture was left at −20°C overnight. The following morning the mixture was spun at 20000xg for 20min at 4°C to pellet the RNA. After discarding the supernatant the pellet was washed with 85% ethanol, spun again at 20000xg for 5min at 4°C, let air-dry for a few minutes and reconstituted in water.

The RNA was then purified by μMACS streptavidin MicroBeads (Miltenyi Biotec, cat. no. 130-133-282) as described in ⁶⁷, with a few minor modifications. After the μMACS streptavidin MicroBeads were added to the RNA sample, the mixture was rotated for 20 min instead of 15 min. After the RNA was eluted from the columns, it was purified with the kit RNA clean and conc-5 (Zymo research, cat. no. R1015) and eluted in 8μl water. It was stored at −80°C until library preparation. The KAPA Stranded RNA-Seq kit KK8400, and KAPA RiboErase kit KK8481 were used (following the KAPA Stranded RNA-Seq kit with RiboErase HMR protocol). 0.04pmol of each sample library were pooled and sequenced pair-ended with Illumina NovaSeq 6000 at a depth of ca 30 M/reads per sample.

Reads were mapped using the snakePipes RNA-seq pipeline ⁶⁸. Expression analysis was performed in R using the countOverlaps function to count reads on the whole gene length (exons and introns) and the featureCounts matrix results were fed to DESeq2 for comparative purposes.

CUT&RUN

HCT116 Halo-SON;SRRM2-SNAP-FKBP cells were incubated for 18 hours in one of the 4 conditions: i) DMSO, ii) DMSO + 1μM dTAG13, iii) DMSO + 1μM HaloPROTAC3, or iv) 1μM dTAG13 + 1μM HaloPROTAC3. On the day of the experiment, cells were washed with 5ml warm media, while tapping the bottom of the dish to detach mitotic cells. Cells were then fastly rinsed with DPBS, and detached with 0.05% trypsin (Thermo Fisher Scientific, cat. no. 25300-054) at 37°C for 3min. A light fixation with 0.01% formaldehyde (Thermo Fisher Scientific, cat. no. 28906) followed. After 1min, crosslinking was stopped with the addition of glycine to a final concentration of 125mM (ROTH, cat. no. 3790.2) on ice.

The activation of concA beads, the binding of the cells to the beads, and the binding of the antibody were carried out following the EpiCypher CUTANA^™ CUT&RUN Protocol v2.1 with few minor changes. 500,000 cells per replicate were spun 5min at 800xg at RT. 1μl of the antibodies was used in each case (See “Antibodies” section). The binding and activation of pAG-MNase were carried out following the EpiCypher CUTANA^™ CUT&RUN Protocol v2.1 with few changes. The buffers have the same composition unless indicated differently. 50μl of pAG-MNase mixture was added to each tube: this mixture consists of digitonin buffer (20mM HEPES, 150mM NaCl, 0.5mM Spermidine, 0.01% digitonin, 1x Protease Inhibitor Cocktail) with a tested dilution of home-made pAG-MNase enzyme. After the addition of this mixture, the reaction was nutated at 4°C for 1h. After two washes with digitonin buffer, beads were rinsed with low-salt buffer (20mM HEPES, 0.5mM spermidine, 0.01% digitonin). Activation of the pAG-MNase was carried out in an ice bath: after the low-salt buffer was removed, the enzyme was activated by adding 100ul of ice-cold incubation buffer (3.5mM HEPES, 10mM CaCl₂, 0.01% digitonin). Samples were incubated 30min in the ice bath, and the reaction was stopped with the addition of 100ul stop buffer (170mM NaCl, 20mM EGTA, 50μg/ml RNAse A, 25μg/ml glycogen, 0.01% digitonin) incubated for 30min at 37°C. Reverse cross-linking was performed as follows: the supernatant was transferred to a new tube, 0.9μl 10% SDS and 1.2μl proteinase K 20μg/ul were added to it, and the mixture was incubated overnight at 55°C. The following day, the DNA was purified with the kit Zymo DNA clean and conc-5 (Zymo research, cat. no. D4014).

The library was prepared according to NEBNext^® Ultra^™ II DNA Library Prep Kit for Illumina^® (NEB #E7645S/L) with the following modifications.The end-repair was performed in a thermal cycler, with the heated lid set to ≥ 75°C, and with the following program: 30min 20°C, 1h 50C, hold at 4°C. The two bead clean-ups were carried out using AMPure beads (Beckman coulter, cat. no. A63881), with a ratio of 1.1x and 0.9x respectively pre- and post- PCR amplification. The annealing/extension time of the PCR amplification was reduced to 10sec to amplify only short fragments. 0.08pmol of each sample library was pooled. Libraries were sequenced pair-ended with Aviti (Element Bioscience) at a depth of ca 15 M/reads per sample.

Reads were mapped using the snakePipes ‘DNA-mapping’ pipeline ⁶⁸. Expression analysis was performed in R using the countOverlaps function to count reads on the whole gene length (exons and introns) and the featureCounts matrix results were fed to DESeq2 for comparative purposes.

Pore-C library preparation and sequencing

Protocol was adapted from the published method ²⁰ with additional RNA digestion using 10U of RNase A/T1 Mix (ThermoScientific, cat. no. EN0551) at 37°C for 20min before Pronase decrosslinking.

4.5x10⁶ endogenous Halo-SON SRRM2-SNAP-FKBP HCT116 cells were incubated with 1μM HaloPROTAC3 (MedChemExpress, cat #HY-111997) and 1μM dTAG13 (Tocris, cat. no 6605) for 0 hours (untreated) 6 hours, 12 hours or 18 hours. After the incubation the cells were fixed with 4mL 1% formaldehyde (final) in PBS (Pierce^™ 16% Formaldehyde (w/v), Methanol-free, cat. no. 28908) for 10 min at room temperature and blocked with 125mM Glycine. Pellet of the fixed cells was frozen at −80°C for at least 1 hour. Nuclei were then isolated at 4°C for 30 min with Cytoplasm lysis buffer (10mM Tris-HCl, 10mM NaCl, 0.2% Igepal, 1x cOmplete protease inhibitors (MERCK, cat. no. COEDTAF-RO). Cells were then permeabilized with 0.5% SDS at 62°C, which was subsequently inactivated with 2% Triton-X-100 (37°C, 15min). DNA was then fragmented in situ using 100U of NlaIII (NEB, cat. no. R0125S) in 1x CutSmart buffer (NEB, cat. no. B6004S) at 37°C for 16 hours. Next, the restriction enzyme was heat inactivated at 62°C for 20 min and restriction fragments ligated for 4 hours in 16°C followed by 1 hour in 25°C using 4000U of T4 ligase (NEB, cat. no. M0202L) in 1x T4 ligase buffer (NEB, cat. no. B0202S) with 1.25% Triton-X-100 and 10μg of BSA. The sample was decrosslinked for 16 hours at 63°C by the addition of SDS to 0.5% and proteinase K to 1mg/mL (Invitrogen, cat. no. 25530–049). DNA was then extracted using phenol:chloroform: isoamyl alcohol (25:24:1). RNA in purified sample was digested with 10U of RNase A/T1 Mix (ThermoScientific, cat. no. EN0551) at 37°C for 20min followed by inactivation and second round of decrosslinking for 40min at 37°C with 2mg/mL pronase in 1x pronase buffer (5mM Tris, 10mM EDTA, 0.5% SDS). After decrosslinking the DNA was extracted as previously and used for ONT native barcoding library preparation according to ONT protocol SQK-NBD114.24 (Ligation sequencing amplicons) and sequenced on a PromethION P24 device with R10.4.1 flow-cells for 3 days with flow cell recharging every 24hrs (Flow Cell Wash protocol: EXP-WSH004). The repeated sequencing runs were prepared using ONT ligation-based library preparation (SQK-LSK114, Oxford Nanopore Technologies) and sequenced as previously.

Each time point was sequenced using 10-13M reads with average read length of ~3kb. ~20% of the reads contained 3 fragments (>91% of the fragments with read quality score of MAPQ >10) and more than half had more than 3 fragments, hence, captured interactions beyond pairs of regions. Overall, this provided us ~100-150M pairwise contacts after all filters. The coverage and QC metrics were very similar among the different time points. 0hr and 18hr libraries were regenerated and sequenced deeper, then merged with the earlier experiments. In total the 0hr sample had 1,939,944 and the 18hr sample had 2,597,482 3-way cis-interactions.

Immunofluorescence

Cells were plated on gelatine coated glass coverslips and left in the incubator to attach for at least 24 hours and treated according to the experimental setup. After the treatment the cells on the coverslips were fixed for 10 min at Room Temperature with 4% methanol-free Formaldehyde in 1x PBS (Pierce^™, cat no. 28908). After fixation coverslips were washed with fresh PBS and the cells were permeabilized with 0.5% Triton-X-100 in PBS for 5 minutes at RT then blocked with 4% BSA in PBS for 30 min at RT. A 50μl drop of primary antibody (See the “Antibodies” section) dilution in 4%BSA was placed on a flat parafilm in a humidified chamber and the coverslips were gently deposited on top of the drop. The coverslips were then incubated with the primary antibody for one hour at RT. After the incubation the coverslips were washed with PBS and incubated for one hour at RT with fluorophore-conjugated secondary antibody in 4% BSA (dilution factor 1:500) and labeled with 4μg/ml Hoechst 33258. Coverslips were then mounted on glass slides in Fluoromount G mounting medium (Invitrogen, cat no. 00-4958-02), sealed and imaged with the Leica Stellaris 8 microscope.

RNA FISH

RNA FISH (RNA fluorescence in-situ hybridisation) was performed according to the Stellaris RNA FISH protocol for adherent cells (www.biosearchtech.com/support/resources/stellaris-rna-fish/stellaris-protocols). HCT116 and derivatives were plated on gelatine coated glass coverslips and left in the incubator to attach for at least 24 hours. Drosophila S2 cells were seeded on concavalin A (Sigma-Aldrich, cat. no. C5275) coated glass coverslips. Other cell types were plated on uncoated glass coverslips. Human HCT116, mouse 3T3, chicken DF1, turtle TH-1, fish EPC and fruit fly S2 cells were grown according to the provider’s specifications before fixation.

To test the dependence of nuclear mRNA and MALAT1 clusters on SON and SRRM2 in human cells Halo-SON;SRRM2-SNAP-FKBP HCT116 cells were incubated for 16 hours with either: i) DMSO ii) DMSO + 1μM HaloPROTAC3 (MedChemExpress, cat #HY-111997) iii) DMSO + 1μM dTAG13 (Tocris, cat. no. 6605) or iv) 1μM HaloPROTAC3 + 1μM dTAG13.

Flp-In T-REx HEK293 cells expressing SON constructs were transfected with 10nM SON siRNA for 24 hours, after this time 0.1μg/mL Doxycycline was added to medium to induce expression of the construct for another 24 hours.

Fish EPC cells were transfected with either human, fish or fruit fly SON cloned into the pcDNA^™5/FRT expression vector using RNAiMax, Lipofectamine 3000 or Fugene. Fugene transfection reagent showed the highest transfection efficiency and was used for the quantification of nuclear speckles formation.

After the treatment the cells were fixed with 4% methanol-free Formaldehyde in 1x PBS (Pierce^™, cat. no. 28908) and incubated with 70% ethanol at 4°C for at least 1 hour. Probes were hybridized at concentration of 125nM for 16°C at 37°C. For this paper Stellaris probes were used: mRNA positive control (Biosearch technologies, cat. no. T30-TAMRA-1) and MALAT1 with Quasar 570 (Stellaris Biosearch technologies, cat. no. SMF-2035-1) or FAM modified T30 probe (IDT, order code: /56-FAM/TTTTTTTTTTTTTTTTTTTTTTTTTTTTTT). Samples were imaged using Leica Stellaris 8 confocal microscope.

For orangutan, chimpanzee, rat, rabbit, eel and zebrafish, samples were mounted on glass slides using 2 μl/slide VECTASHIELD^® Antifade Mounting Medium (H-1000-10) and dried in room temperature for 5 minutes before imaging. For these species, images were acquired by a Nikon N-SIM S Super Resolution Microscope, using CFI SR HP Apochromatic TIRF 100XC Oil objective and ORCA-Fusion Digital CMOS Hamamatsu camera (Model: C14440-20UP). Post acquisition processing was performed using FIJI.

To quantify the morphology of speckles upon the expression of different SON constructs, the maximum projection images over the entire nuclei were generated with Fiji/ImageJ. The images were then segmented with Cellprofiler 4 (2D IdentifyPrimaryObjects function). Nuclei were identified as objects between 80 and 160 pixels (~10—18 μm) and used to mask the p(A)+ RNA-FISH image to include only nuclear RNA signal. RNA speckles were then identified as objects between 5 and 160 pixels in diameter (~0.6—18 μm). The average Eccentricity of RNA speckles per nucleus was plotted using the seaborn package. Statistical significance was calculated using a two-sided Mann-Whitney-Wilcoxon test.

Since automated software cannot definitively identify the absence of an object, the morphology of nuclear speckles in fish EPC cells was quantified manually. For this purpose, cells expressing the construct of interest were imaged using a Leica Stellaris 8 confocal microscope with a 100x oil objective. The presence of speckles, appearing as discrete condensates of concentrated SON signal, was identified by the observer. At least 12 cells were analyzed for each transfection. The experiment was replicated with two different transfection reagents, yielding similar results. For verification, quantification was also performed using Cellprofiler4 with the 2D IdentifyPrimaryObjects module, which identified speckles as objects with the area below 400 pixels. Despite a high false positive rate, the experiment produced similar, statistically significant results.

3D-DNA FISH

3D DNA FISH (DNA fluorescence in-situ hybridisation) was performed according to published protocol ⁶⁹. Briefly, Halo-SON SRRM2-SNAP-FKBP HCT116 cells were plated on glass coverslips and left in the incubator to attach for 24 hours. After that time SRRM2 was labelled overnight with 100nM SNAP-Cell 647-SiR (NEB, cat. no. S9102S). At the same time, the cells were either incubated with DMSO as a control or 1μM HaloPROTAC3 for 18hrs to deplete SON. Flp-In T-REx HEK293 cells expressing SON constructs were transfected with 10nM SON siRNA for 24 hours, after this time 100nM Doxycycline was added to medium to induce expression of the construct for another 24 hours. After the treatments the cells were washed, fixed with 4% PFA, permeabilized and denatured for 3 min at 78°C or and incubated under a sealed coverslip for 18h at 37°C with a mixture of 5μl XCP18-orange and 5μl XCP19-green (cat no.: D-0318-050-OR and D-0319-050-FI respectively, MetaSystems probes). The coverslips were washed with 2x and 0.1x SSC and mounted according to the protocol and imaged on Stellaris 8 confocal microscope using either 100x oil immersion objective, acquiring z-stacks with optimal x-y-z-resolution (voxel size: 114×114×298 nm), or 20x air objective with the resolution of 0.114×114×685nm (Flp-In T-REx HEK293 cells).

Quantification of speckle-to-chromosome contacts was done using Cellprofiler 4 ⁶² for at least 30 cells in each treatment. SRRM2-SNAP (SNAP-Cell 647-SiR) speckles were called using the watershed function (3D processing), whereas Halo-SON (JaneliaFluor646) speckles and spots of DNA FISH signal were closed using a disk with a size of 2 pixels and called as objects using ConvertImageToObjects function, to avoid splitting them into smaller objects. In the experiment involving expression of different SON constructs SON speckles were also called as DNA-FISH spots to avoid speckles were filtered to include only objects larger than 20 voxels and DNA FISH spots to include only objects between 80 and 3000 voxels. The percent of surface area of chromosome spots contacting a speckle object was calculated with MeasureObjectNeighbors function determining speckles as neighbors if they reside within 1 pixel (~114 nm) surrounding the chromosome spot. The plot was drawn using the seaborn package. The statistical significance between groups was calculated using a two-sided Mann-Whitney-Wilcoxon test.

Transmission Electron Microscopy (TEM)

For TEM analysis, Halo-SON SRRM2-SNAP-FKBP HCT116 cells were grown on Thermanox^® plastic coverslips (Nunc, Thermo Fisher). Cells were incubated for 16 hours with either: i) DMSO ii) DMSO + 1μM HaloPROTAC3 (MedChemExpress, cat #HY-111997) iii) DMSO + 1μM dTAG13 (Tocris, cat.no 6605) or iv) 1μM HaloPROTAC3 + 1μM dTAG13. Next, 5.0 % (v/v) glutaraldehyde (Grade I, Sigma) was freshly prepared in PBS and 1:1 given to the cell culture medium (final concentration 2.5%) for fixation. Cells were fixed for 1 hour at RT.

After fixation the cells were washed 3 times in PBS and contrasted in 0.5 % (v/v) osmium tetroxide (Science Services) in PBS for 1 hour at RT.

Cells were washed again 4 times in distilled water for 20 min. Then samples were incubated for 30 min in 100 mM Hepes buffer pH 7.4 containing 0.1 % (w/v) tannic acid (Electron Microscopy Sciences), washed 3 times for 10 min in distilled water, contrasted in 2 % (w/v) uranyl acetate (Fluka) for 1.5 hours at RT and washed once in distilled water followed by dehydration through a series of increasing ethanol concentrations (5 min in 30 %, 10 min in 50 % and 70%, 90%, 96%, 15 min each and finally 3 times 10 min in absolute ethanol, respectively).

Afterwards samples were incubated overnight at 4°C in 100 % SPURR (Low Viscosity Spurr Kit, Ted Pella). Spurr’s resin mixture was changed out for 2 times, 2 hours each, before embedding.

For polymerization, coverslips were transferred to resin filled Beem capsules and incubated for 2 days at 60°C. 70 nm sections were cut using a Leica UC7 ultramicrotome equipped with a 3 mm diamond knife (Diatome) and placed on 3.05 mm Formvar Carbon coated TEM copper slot grids (Science Services). Sections were post contrasted using Uranyless EM Stain and ready to use 3% Lead Citrate (both Science Services).

In order to visualize ultrastructural details, sections were imaged on a Tecnai Spirit transmission electron microscope (FEI) operated at 120 kV, which was equipped with a 4kx4k F416 CMOS camera (TVIPS).

paGFP-H2B

The H2B sequence was cloned with In-Fusion^® Snap Assembly Master Mix (Takara Bio, cat # 638944) followed by a short linker (GATCCACCGGTCGCCACC) and the paGFP (photo-activable green fluorescent protein) sequence. One clone was analysed by whole-plasmid-sequencing (Eurofins), which showed a point mutation T59A. The integrity of the chromophore activity was tested with transient transfection using Fugene HD reagents (Promega, cat. no. E2311) before generating the stable cell lines HCT116 Halo-SON;SRRM2-SNAP-FKBP;paGFP-H2B and HeLa Halo-SON;SRRM2-SNAP-FKBP;paGFP-H2B and HeLa WT paGFP-H2B. Cells were selected and kept under 1mg/ml Geneticin (Thermo Fisher Scientific, cat.no. 10131-027), the high-GFP population was enriched by FACS before the imaging experiment. HCT116 cells were imaged 3 h after addition of either 1 μM HaloPROTAC3 (MedChemExpress, cat #HY-111997) or DMSO to deplete SON or as control, respectively, and of 100nM SNAP-Cell 647-SiR (NEB, cat. no. S9102S) to visualise SRRM2. A Celldiscoverer 7 confocal microscope (Zeiss) equipped with an incubation chamber with 5% CO₂ and a heated stage at 37 °C was used. Images were acquired using a 50x objective (1.2 NA) with 1x zoom. paGFP activation was performed using the bleaching module, with scan speed=3 and 100% intensity 405nm laser. Time series were acquired using 50 cycles of 30min intervals, during which the paGFP signal was collected using a 488 nm laser with 0.2% intensity and SNAP-Cell 647-SiR using a 640nm laser with 0.2% intensity. Image processing was performed with ZEN software blue version 3.4 and 3.9 (Carl Zeiss, Germany) and Fiji ImageJ2 V2.14.0. Images underwent deconvolution, and slight smoothing (1.5 for SRRM2-SNAP-Cell 647-SiR signal, and 2.0 for paGFP-H2B).

The two HeLa cell lines were plated together and then imaged 30min after addition of 1 μM HaloPROTAC3 (MedChemExpress, cat #HY-111997) + 1μM dTAG13 (Tocris, cat.no 6605) and 100nM SNAP-Cell 647-SiR (NEB, cat. no. S9102S) to distinguish between WT and Halo-SON;SRRM2-SNAP-FKBP. A Celldiscoverer 7 microscope (Zeiss) equipped with an incubation chamber with 5% CO₂ and a heated stage at 37 °C was used. Images were acquired using a 20x objective (0.95 NA) with 2x zoom. paGFP activation was performed using the bleaching module, with scan speed=3 and 100% intensity 405nm laser. Time series were acquired every 30min intervals during which the paGFP signal was collected using the 470nm channel at 40% and SNAP-Cell 647-SiR using the 625nm channel at 30%. Image processing was performed with ZEN software blue version 3.4 and 3.9 (Carl Zeiss, Germany) and Fiji ImageJ2 V2.14.0. For quantification of changes in chromatin mobility the signal intensity was quantified in 11 and 17 cells, for HCT116 and HeLa respectively, at two timepoints (t=30min and t=6h30min/7h30min). In more detail, a line perpendicular to the activation area was drawn in the central axis of the cell in imageJ2, then the grey values obtained with the function Plot Profile were exported as csv file, and the data for each cell was loaded into python for plotting.

Fluorescence Recovery After Photobleaching (FRAP)

The cell line expressing endogenously Halo-tagged SON and SNAP-tagged SRRM2 were plated on plastic bottom IBIDI plates (IBIDI, cat no 80806) and labeled for 16 hours with 100nM Halo-TMR ligand (Promega, cat. no. G8252) and 100nM SNAP-Cell 647-SiR (NEB, cat. no. S9102S). After 16 hours the cells were washed twice with a pre-warmed medium and imaged with a Stellaris 8 confocal microscope using 86x water immersion objective and white light laser (WLL) exciting the fluorophores at two separate laser lines (552 and 647). FRAP was performed using the Stellaris FRAP module by bleaching the region of interest encompassing the entire speckle or splitting the speckle in half with 6 imaging cycles at frequency of 200 and 4 laser lines at 100% laser power (647, 532, 522, 512). The region was then imaged every 5 seconds for a total of 5 minutes with two separate laser lines (SRRM2-SNAP: 647, Halo-SON: 552).

Pixel intensity values were extracted using Fiji/Image, by selecting the minimal bleached ROI and extracting the mean intensity value of the bleached area, whole cell area and the background using ROI manager (multi measure). The resulting intensity tables were analysed with EasyFRAP-web⁶³. The normalised data were plotted with the python seaborn package (seaborn.lineplot, mean + CI).

FLASH

FLASH was carried out essentially as described in ²⁸. Halo-SON HCT116 were either treated with 1μM HaloPROTAC3 in DMSO or with DMSO for 16hrs. Cells were washed with PBS and UV-crosslinked with 0.2mJ/cm2 UV-C light on ice, nuclei were isolated and resuspended in 1xNLB + 5mM MgCl₂, with protease and phosphatase inhibitors, and sonicated with a Branson sonifier. After centrifugation to remove insoluble material, the supernatant is incubated with an agarose resin (Pierce, 26150) for 20 min in the cold-room. Supernatant was then incubated with Dynabeads Protein G beads pre-bound to p-SR antibody (10μL per IP, 1H4, Santa Cruz, sc-13509) for 90 minutes in the cold-room. The supernatant from 1H4 immunoprecipitation is used for DHX9 immunoprecipitation (2.5μL per IP, Abcam, ab26271). Beads are washed once with 1xNLB, once with NDB and treated with 0.02U/μL RNaseI (Thermo) in 100μL NDB for 3 minutes at 37°C, and immediately placed on ice to stop the reaction. Beads were then washed once 1xNLB, and once with NDB. RNA-ends were repaired with T4 PNK, after which barcoded, pre-dephosphorylated s-oligos were ligated with T4 RNA Ligase 1 for 90 minutes at 25°C. Beads are then washed once with twice with 1xNLB, and once with NDB. The RNA was released with proteinase K treatment, and purified using Oligo Clean and Concentrator columns (Zymo). Reverse transcription was carried out with SuperScript III, samples were then treated with RNaseH (NEB) in order to phosphorylate the 5’-end of the cDNA molecule. After a final round of purification with Oligo Clean and Concentrator columns (Zymo), the cDNA was circularized with CircLigaseII (Lucigen) and amplified with Q5 polymerase (NEB). PCR products were purified with SPRI beads, QCed with Bioanalyzer and subjected to high-throughput sequencing.

Proteome profiling

Around 1 million Halo-SON SRRM2-SNAP-FKBP HCT116 cells were used per replicate. Degradation of SON protein was induced with HaloPROTAC3 (Promega, GA3110), as a control inactive ligand ent-HaloPROTAC (Promega, GA4110) was used. Degradation of SRRM2 was induced with dTAG13 (Tocris, cat.no 6605). The cells were plated in triplicates and incubated for 16 hours with one of the four conditions: i) DMSO + 300nM ent-HaloPROTAC3, ii) DMSO + 300nM HaloPROTAC3, iii) 300nM ent-HaloPROTAC3 + 1μM dTAG13 or iv) 300nM HaloPROTAC3 + 1μM dTAG13. After the incubation the cells were harvested by trypsinization with 0.05% trypsin, centrifuged, washed with PBS and the pellets snap frozen in liquid nitrogen. Later the frozen cells were lysed under denaturing conditions in 200 μL of a buffer containing 3M guanidinium chloride (GdmCl), 10 mM tris(2-carboxyethyl)phosphine (TCEP), 40 mM chloroacetamide and 100 mM Tris-HCl pH 8.5. Lysates were denatured at 95°C for 10 min shaking at 1000 rpm in a thermal shaker and sonicated in a water bath for 10 min. The protein concentration of each sample was measured with a BCA protein assay kit (23252, Thermo Scientific). 500 ng protein was used per sample and diluted with a dilution buffer containing 10% acetonitrile and 25 mM Tris-HCl, pH 8.0, to reach a 1 M GdmCl concentration. Then, proteins were digested with LysC (Roche; enzyme to protein ratio 1:50, MS-grade) shaking at 800 rpm at 37°C for 3.5 hours. The digestion mixture was diluted again with the same dilution buffer to reach 0.5 M GdmCl, followed by tryptic digestion (Roche, enzyme to protein ratio 1:50, MS-grade) and incubation at 37°C overnight in a thermal shaker at 800 rpm. Peptides were acidified with formic acid to a final concentration of 2% and the digests were loaded onto Evotip Pure (Evosep) tips according to the manufacturer’s protocol. Peptide separation was carried out by nanoflow reverse phase liquid chromatography (Evosep One, Evosep) using the Aurora Elite column (15 cm x 75 μm ID, C18 1.7 μm beads, IonOpticks) with the 20 samples a day method (Whisper-100-20-SPD). The LC system was online coupled to a timsTOF SCP mass spectrometer (Bruker Daltonics) applying the data-independent acquisition (DIA) with parallel accumulation serial fragmentation (PASEF) method. MS data were processed with Dia-NN (v1.8.2 beta 22) and searched against a library of in-silico predicted human spectra. The “match between run” feature was used and the mass search range was set to m/z 400 to 1000 dalton. The resulting LFQ intensities for each protein of interest were plotted using seaborn package.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical tests, significance thresholds, sample sizes, and dispersion and precision measures are indicated in the figure legends. Specific details of the analyses performed in this study are provided in method details.

Analysis of Gene Expression

Differentially expressed genes were identified using the ‘mRNA-seq’ pipeline of snakePipes ⁶⁸(v.2.7.1). Briefly, fastq files were trimmed (cutadapt, v.4.6) to remove adaptor sequences and aligned to the human genome (hg38) with STAR ⁷⁰ (v.2.7.10). Alignments overlapping with exons (Gencode v.29) were quantified with featureCounts ⁵⁸ (v.2.0.1), which was then used to call differentially expressed genes using DESeq2 ⁵⁹ (unless otherwise noted, padj<0.05, Wald’s test). DESeq2 outputs are available as Supplementary Tables 2–10. The downregulated genes were plotted on chromosome ideograms using chromPlot R package⁶⁰.

To analyse enrichment of downregulated genes in isochores, human genome is split into five isochores based on GC content of non-overlapping 10kb bins L1 (<37%), L2 (37–42%), H1 (42–47%), H2 (47–52%), and H3 (≥52%). Each gene is assigned to an isochore based on the genomic overlap of the gene with the 10kb bins used to calculate GC content. Enrichment score is the log2 ratio of observed number of downregulated genes in an isochore to the expected number of downregulated in that isochore which is calculated by multiplying the total number of downregulated genes with the genomic coverage of a given isochore.

Retention scores for transcripts are determined by comparing the nuclear-to-total expression ratios derived from RNA-seq data. First, variance-stabilized transformation (VST) is applied using DESeq2 to normalize gene expression levels. For each gene, the nuclear-to-total ratio is then calculated in both control and SON-depleted cells. The difference in these ratios follows a normal distribution, allowing each transcript to be assigned a retention score as a z-score. Transcripts with z-scores greater than 1.96 (p < 0.05) are classified as retained in the nucleus following SON depletion.

Analysis of Splicing

Exon skipping and intron retention events were analysed using rMATS-turbo ³¹ as part of the ‘mRNA-seq’ pipeline of snakePipes. Unless otherwise indicated, events with FDR < 0.05 and |ΔPSI| > 0.2 were considered for analysis. For intron-centric, reference-free splicing analysis, we used leafcutter ⁷¹ (v.0.2.9) where introns within differentially spliced clusters (padj < 0.05) with ΔPSI > 0.1 were considered as differentially spliced introns. In order to quantify entropy of splicing, we used Whippet ³⁸, which measures splicing complexity with Shannon’s entropy by enumerating and quantifying the use of all possible splicing paths within an intron cluster (see ref.³⁸ for more in-depth information). Entropy of events with Probability >= 0.9 and |ΔPSI| > 0.1 were plotted using the ‘ecdfplot’ function of seaborn ⁷² (v.0.13.1).

Splicing rescue index (rescueIndex) was calculated as follows. First rMats-turbo was run on RNAseq results from SON Knockdown (SON KD) in Flp-In T-REx HEK293 mother cell line using Negative control siRNA treated Flp-In T-REx HEK293 mother cell line as a control (WT) and Flp-In T-REx HEK293 cell line in which endogenous SON was depleted by SON KD and replaced by the expression of rescue construct (RESCUE) compared with WT. The results were filtered to extract only the differential splicing events which are significant in SON KD vs WT experiment using FDR < 0.05 and |ΔPSI| > 0.2 and at least 20 junction counts supporting the splicing event. Then the RescueIndex was calculated using rMats IncLevelDifference values, which can be summarized by the equation:

\frac{(PSI WT - PSI SON KD) - (PSI WT - PSI RESCUE)}{(PSI WT - PSI SON KD)}

HiC Subcompartment and SPIN state analysis

The overlap of genomic regions, genes, exons of interest with HiC subcompartments ² and/or SPIN states ¹² were determined using the ‘overlap’ function of bioframe ⁷³ (v.0.6.1). If a region overlapped with multiple domains, the domain with the largest overlap is retained for analysis and plotting.

GC content analysis and UMAP

All genomes annotated by NCBI’s Eukaryotic Genome Annotation Pipeline ⁵⁷ were considered for analysis (1952 genomes in total, as of 2025.02.26). Introns and exons of all transcripts with the tag “mRNA” were extracted from .gff files using custom python scripts, only unique intron/exons were used for calculations. Bioframe’s⁷³ ‘frac_gc’ function was used to calculate GC content. To be considered a GC-rich leveled exon, we applied three criteria to all exons preceded and followed by an intron: 1) The GC content of the exon must be equal to or less then the GC content of both the downstream and the upstream introns 2) GC content of either the upstream or downstream intron must be significantly higher than the average GC content of all introns in a given species (z-score > 2) 3) The GC content of the exon must be higher than the median GC content of all exons in a given species. The UMAP ⁴⁶ was generated by extracting the underlying values of log10(intron_length) vs GC content 2D-histrograms created with matplotlib’s ‘hist2d’ function (intron length range=10-10**6, GC-concent range=0-100, 30x30 bins), which were then normalized such that each bin represents a frequency between 0-1 and all frequencies add up to 1 and with the final function call: umap.UMAP(min_dist=0.1, n_neighbors=15, random_state=42).fit(ScaledValues).

Analysis of gene distribution heterogeneity in eukaryotes

Isochores were called at a resolution of 10kb to accommodate a wide range of genome sizes and gene lengths across all eukaryotes. Analysis was restricted to protein coding genes, and each gene was assigned to an isochore based on its overlap using the ‘overlap’ function of bioframe. Expected number of genes in isochores was calculated by multiplying total coverage of a given isochore by the total number of protein coding genes in a given organism. Enrichment score is calculated as the log₂ ratio of expected number genes vs observed number of genes. Cramér’s V values, providing an effect size for gene distribution heterogeneity were calculated by dividing the chi-squared metric of each organism by 4 (1-total number of isochores) and taking the square-root of this value with the function numpy.sqrt(). The “driver isochore” was calculated from standardized residuals where the difference between observed and expected values were divided by the square-root of expected values. The isochore with the largest absolute standardized residual value is labeled as the “driver isochore”, with negative and positive values meaning depletion or enrichment of genes in that isochore respectively.

Exon-intron evolution analysis

Speckle dependent exons were defined as exons that are skipped in ΔSpeckles cells, and are located in physical proximity to nuclear speckles (Speckle SPIN state) in human cells. Orthology of the genes containing the speckle-dependent exons was traced across 28 selected species belonging to fish, amphibians, reptiles, birds and mammals using the NCBI orthology database. Only one-to-one orthologs that were found in all the tested species were used for further analysis. Orthologous exons within the speckle-dependent genes were mapped using ExOrthist ⁴⁴ with default settings for short evolutionary distance within the taxonomic class and medium evolutionary distance between classes. The orthologs of human speckle-dependent exons along with their flanking introns were extracted from genomic GTF and sequence files of each species. The length and GC content of the orthologs of speckle-dependent exons and their flanking introns were extracted from the GTF file and genomic sequence file respectively. 500 exons per species were randomly sampled from the set of exons excluding the orthologs of speckle-dependent exons for each species to serve as a speckle-independent exon set. The features of speckle-independent exons and their flanking introns were extracted in the same manner as speckle-dependent ones. For plotting the species were grouped into 6 categories: ‘fish’: [Amblyraja radiata, Rhincodon typus, Carcharodon carcharias, Solea solea, Xiphias gladius, Danio rerio, Erpetoichthys calabaricus, Latimeria chalumnae, Protopterus annectens], ‘Amphibians’: [Microcaecilia unicolor, Geotrypetes seraphini, Rhinatrema bivittatum, Bufo bufo, Xenopus tropicalis, Bombina bombina], ‘lizards’ : [Euleptes europaea, Anolis carolinensis, Zootoca vivipara], ‘turtles and alligators’ : [Chelonia mydas, Gopherus flavomarginatus, Trachemys scripta elegans, Alligator mississippiensis], ‘birds’ : [Apus apus, Parus major, Gallus gallus], ‘mammals’ : [Felis catus, Mus musculus, Homo sapiens].

Pore-C analysis

The raw pod5 reads were basecalled and demultiplexed using Dorado basecaller v0.5.3 with a model dna_r10.4.1_e8.2_400bps_sup@v4.2.0. Sequencing reads were aligned to the hg38 human genome using the published pipeline from HiPore-C ²⁰ with default parameters. Fragments with a MAPQ score < 10 were discarded. For compartment calling, dcHiC v2.1 ⁶¹ was applied for performing p and q-arm wise PCA calculations using contact maps at 100 kb resolution. In the 18 hours post-NS disruption sample, PC1 and PC2 were manually reselected for the q-arms of chromosomes 1 and 9, respectively. PC2 was then manually reoriented, and chromosome 4 was excluded from downstream analysis. Genomic regions with positive and negative principal component values were classified as compartment A and B, respectively. Saddle plot analyses were performed to visualize cis-interactions between A/B compartments of varying strength using KR-normalized 100kb bins ⁷⁴. Briefly, sorted compartment scores were equally divided into 50 classes. For each pair of classes (P, Q), we listed all pixels or locus pairs (i, j) where one locus belonged to P and the other belonged to Q. We then calculated the average of observed/expected (O/E) counts across the whole collection of such pairs {Aij | i in p, j in q}²¹. Differential saddle plots were calculated by taking the difference of the aggregated matrices across two conditions, and then the values were smoothed. To smooth differential saddle plots, the values on the edge of the matrices were unaltered. For interior points, the smoothed values are defined in terms of the originals as follows: S[i,j] = (2*M[i,j] + M[i-1,j] + M[i+1,j] + M[i,j-1] + M[i,j+1])/6. To quantify the strength of compartmentalization, we calculated the ratio between (AA+BB) / (AB+BA), which in the saddle plots visually corresponds to the ratio between the boxes spanning upper right and lower left corners, versus those that span lower right and upper left corners. This ratio was calculated for different pixel extensions (box of size 1 to 30) by taking the mean across all pixels for each extension setting. Reads from the two replicates of the 0 and 18 hours post-NS disruption samples were combined and aligned to the genome as previously described. A/B compartments were then extracted using dcHiC v2.1, with PC1 manually reselected for the p-arm of chromosome 7. For downstream analyses, PC2 was manually reoriented for the q-arm of chromosome 9, and chromosome 4 was excluded. Multi-way interactions at 100kb resolution involving three loci on the same chromosome (cis interactions) were annotated according to their affiliation to A or B compartment.

Supplementary Material

Video S1

Supplemental Video 1: Video summary of fluorescence recovery after photobleaching (FRAP), related to Figure 1E, F and Supplemental Figure 1B, C, D. Endogenous SON is labeled with Halo-TMR (green) and endogenous SRRM2 with SNAP-Cell 647-SiR (red). Video shows the recovery after FRAP in approximately half of a single Speckle. SON shows gradual recovery, mixing slowly with the bleached area, while SRRM2 recovers rapidly and uniformly. Recovery time: 5 minutes; frame rate: 5 seconds; scale bar: 5 μm.

Download video file^{(445.9KB, mp4)}

Video S2

Supplemental Video 2: Visualisation of SRRM2 droplets, related to Figure 1C and Supplemental Figure 1D. Live cell imaging following two nuclei 1 hour after SON depletion by addition of 1μM HaloPROTAC3. SRRM2 is labelled with SNAP-Cell 647-SiR (red), and the DNA with Hoechst 33258 (blue). SRRM2 droplets initially collapse into smaller, round condensates which gradually merge to form larger, fewer droplets. Frame rate: 5 min; scale bar: 5 μm.

Download video file^{(3.1MB, mp4)}

Supplementary Figure 1

Figure S1: SON and SRRM2 form a viscoelastic structure that limits random chromatin movements and enables the maintenance of constitutive heterochromatin, related to Figure 1. (A) Immunofluorescence images of WT, ΔSON, ΔSRRM2 and ΔSpeckles cells illustrate that in ΔSpeckles the only protein-dense (NHS-488 staining, green) regions that remain are the nucleoli (NOLC1, red). NS: nuclear speckles. (B,C) Nuclear speckles are formed by a liquid-like protein; SRRM2 and a more rigid counterpart SON as indicated by the fluorescence recovery after photobleaching (FRAP) experiment. Endogenous SON labelled with HaloTag^® TMR Ligand and the endogenous SRRM2 with SNAP-Cell^® 647-SiR. Half-time recovery: SRRM2: 53s; SON: 95s. (B) Single frames of representative cell illustrating the signal recovery. (C) Quantification of the signal recovery from at least 20 cells per treatment. Mobile fraction of SON is 0.58 and of SRRM2 is 0.88 with a half-life of 95s and 53s, respectively. (D) Single frames of the live-cell imaging video show that SON depletion leads to collapse of speckles into smaller SRRM2-droplets, which merge and grow over time (also see Video 2 in the supplementary materials). (E,F) Approximately half of the cell’s nucleus is activated with a blue laser (405nm) in ΔSpeckles cells, which carries a photo-activatable GFP-tagged H2B (PA-GFP-H2B) expression construct. In ΔSpeckles cells the chromatin displays higher mobility. Cells are imaged every 30 minutes in a single Z-plane to allow cell tracking. (E) Example images following single nuclei depicting the change of signal over time. The Grey dashed line outlines the nucleus. Scale bar=5 μm. (F) The normalised intensities of 17 cells along the line depicted in (F) are plotted using FiJi/ImageJ. (G) Pore-C contact matrices of a fragment of chromosome 5 show that while the overall chromatin-chromatin interaction patterns are largely preserved in ΔSpeckles cells, the inter-compartmental contacts become less defined. (H) Saddle plots showing compartmentalisation strength for each condition before and after speckles disruption using Pore-C contact maps (See Methods). (I) (left) Quantification of the compartmentalization strength (Saddleplot strength) at different timepoints after speckles disruption. (right) Quantification of long-range multiway contact in control and 18h ΔSpeckles cells show increase in cross-compartment contacts and corresponding decrease in within-compartment contacts. (J) Representative transmission electron microscopy images for control treated cells, single SON- or SRRM2-depletion and double-depletion show ruffled nuclear membrane already upon single SON-depletion, which becomes more pronounced in double-depleted cells. Scale bar=2μm (top), 1μm (middle), and 500nm (bottom). (K) Differences in H3K9me3 CUT & RUN signals show a gain of the constitutive heterochromatin mark H3K9me3 in speckle-distal chromatin (violet) and a loss in speckle-proximal chromatin (bordeaux). The differences become increasingly strong from single SRRM2 depletion, to single SON depletion, to double-depletion. p-value <0.05. Plots on the top show changes of H3K9me3 over annotated genes, whereas radial plots on the bottom show changes over 100kb genomic bins.

NIHMS2178151-supplement-Supplementary_Figure_1.pdf^{(27.2MB, pdf)}

Supplementary Figure 2

Figure S2: Nuclear speckles loss affects gene expression, targeting mainly GC-rich genes, related to Figure 2. (A) Volcano plots (top) and radial plots (bottom) of poly(A)+ RNAs upon SRRM2-, SON- and double-depletion show that SON depletion is the major driver of the dysregulation at the mRNA level seen in ΔSpeckles cells, with SRRM2 contribution apparently less pronounced but very specific to speckle-proximal genes. (B) Scatterplots of all affected genes (padj<0.05) show downregulation of the mRNAs originating in GC-rich isochores. A similar trend is not visible in differentially expressed nascent RNAs (bottom). (C) Scatterplot showing correlation between genes affected by SON and SRRM2 depletions. The overlap is additionally represented as Venn diagrams of affected genes (padj<0.05; | l2fc | > 0) upon SRRM2-, SON- and double-depletion. (D) Length and speckle-proximity of human chromosomes represented as a fraction of the chromosome annotated as “Speckle” SPIN state. Over 50% of Chromosome 19 belongs to the speckle SPIN state. (E) Gene ontology (GO)-term enrichment analysis of differentially expressed genes in ΔSpeckles cells is not informative for downregulated genes, while upregulated genes are enriched with the terms “RNA processing”, “ribonucleoprotein complex biogenesis” and “mRNA processing”. (F) Genomic segments can be classified based on HiC compartmentalisation, GC isochores and SPIN states, and they display a specific overlap between the three classification methods. (G) Genes within speckle-associated domains, as represented by SPIN states, have the highest GC content (dark red). The GC content of genes decreases as the distance from speckles increases (from Interior_Act1 - closest to speckle- to Lamina - further from speckles). (H) An overview of all human chromosomes indicates a strong overlap between downregulation of genes and their GC content. Horizontal bars represent the number of genes downregulated on mRNA level originating from the corresponding genomic bin. Chromosome ideograms are coloured according to the underlying genomic GC content (see legend). Black circles represent centromeres. (Same as in Figure 2E). GC content of the genes affected in ΔSpeckles cells is illustrated as a violin plot below the chromosome ideograms. Downregulated genes (red) have dramatically higher GC content than both unaffected genes (grey) and upregulated genes (green). (I) Depletion of SON and SRRM2 (red), shows the highest enrichment of downregulated mRNAs in H3 isochore regions, followed by H2, while these genes are depleted from H1, L1 and L2 GC-poorer isochores. TRA2A/B depletion (red) shows the opposite trend. Analysed data includes the ENCORE dataset comprising 471 RNA-seq experiments targeting 263 proteins in 2 different cell lines.

NIHMS2178151-supplement-Supplementary_Figure_2.pdf^{(17.5MB, pdf)}

Supplementary Figure 3

Figure S3: SR-protein accessibility determines splicing differences in ΔSpeckles cells, related to Figure 3. (A) SPIN-state related binding status of SR-proteins and TRA2A and TRA2B (based on the re-analysis of the data published in ²⁸) reveal the differential preference of the binding. Except for SRSF12, TRA2A and TRA2B, all the displayed RNA binding proteins show a larger fraction of binding on the exons of speckle-associated chromatin domains. SRSF12, TRA2A and TRA2B show a larger fraction of binding on the exons of Interior-Active3 chromatin domains. (B) Metagene plots show that TRA2A (right) and TRA2B (left) binding, as determined by FLASH, is highly enriched on exons that are skipped following TRA2A/B depletion. (C) Scatterplot where every dot is a gene shows that loss of speckles and co-depletion of TRA2A+TRA2B affect two different types of gene architectures. TRA2-sensitive genes are long and GC-poor, whereas speckle-dependent genes are short and GC-rich. (D) Motif enrichment analysis indicates GC-rich sequences for the exons skipped in ΔSpeckles cells, whereas GA-rich exons are skipped upon TRA2A/B depletion. (E,F) Meta-coverage analysis of phosphorylated SR proteins (pSR) on SON-dependent, TRA2A/B-dependent, and all other exons in ΔSON cells reveals that while the overall levels of pSR remain unchanged, their distribution is altered specifically on SON-dependent exons. In these exons (E), and retained introns (F), pSR shows a broadened distribution into adjacent splicing units. (G) Genome browser screenshot of an example speckle-dependent gene (NPRL2) upon SON-, SRRM2- and double-depletion shows the differences in RNA processing upon perturbation of speckles. Splicing of the same gene becomes disordered upon SON and double-depletion as illustrated by sashimi plots. Also see Figure 3H. (H) rMATs analysis of RNA-seq experiments carried in triplicates from ΔSpeckles, ΔSON, ΔSRRM2, and TRA2A-B double-depletion.

NIHMS2178151-supplement-Supplementary_Figure_3.pdf^{(2.7MB, pdf)}

Supplementary Figure 4

Figure S4: The total amount of SR proteins is not affected by speckle perturbation, the observed splicing defects are due to change in their local concentration; splicing defects and disruption of RNA export result in gene downregulation, related to Figure 4 (A) Total proteome Mass Spectrometry upon 16h of SRRM2 depletion, SON depletion, or double depletion identifies the intended targets as most significantly affected proteins in comparison to control treatment, while the protein level of the other SR-proteins, SF3A2 and SF3B1 does not change. Each dot represents normalised intensity from a single replicate, dashed blue line represents average intensity of the DMSO control treatment. (B) SR-protein levels are not significantly altered in treatment conditions to deplete SON or SRRM2 or both as quantified by the pan-phospho-SR-protein antibody 1H4. (C) iBAQ values of protein abundance in WT HCT116 cells reveal that SON depletion causes very strong phenotype, the protein is expressed at a level at least twice as low as any individual SR protein or member of the SF3 complex. (D) Immunofluorescence of SRRM2, pSF3B1 (which recognizes phosphorylated T313 of SF3b155) and SF3A2 upon single SON or double depletion after 16 hours of treatments show dispersed pSF3B1 staining, whereas single SRRM2 depletion has a milder effect. Scale bar=5μm. (E) Genome view snapshots of AUP1, ERAL1 and TPRA1 genes show changes in splicing which were detected in the RNA-seq. Colored lines in ΔSpeckles snapshots represent splicing events that either introduce a STOP codon before the last intron (induces NMD, red) or do not change the reading frame in a way that introduces a STOP codon (does not induce NMD, green). (F) Changes in splicing are validated by reverse transcription and PCR for the depicted intervals in (E). Fully spliced RNAs are expected to have lengths of 620nt (AUP1), 650nt (ERAL1), and 704nt (TPRA1). The addition of the NMD inhibitor SMG1i increases the abundance and number of spliceoforms with skipped exons. (G) Transcripts with intron retention events (center), exon skipping events (left), or which are retained in the nucleus (right) are predominantly downregulated.

NIHMS2178151-supplement-Supplementary_Figure_4.pdf^{(4.5MB, pdf)}

Supplementary Figure 5

Figure S5: Expanding IDRs of SON drive the nuclear speckle evolution, related to Figure 5 (A) Evolution of the length and GC content of introns (top panels) flanking exons that are either speckle-dependent (blue dots) or speckle-independent (grey dots). Bottom panels show the evolution of length and GC content of these exons (Same as Figure 5A,B). (B) Immunofluorescence images of SRRM2, and RNA-FISH images of MALAT1 lncRNA and p(A)+ RNA upon single SON, single SRRM2 or double depletion after 16 hours of treatment in human cells. Scale bar=5μm. (C) MUSCLE protein sequence alignment of SON orthologs in representatives of major animal clades. The protein sequence is represented as tracks of isoelectric point (pI) for each ortholog to signify the changing charge distribution among the organism (see Figure 5D for comparison). An average sized and positively charged protein in anamniotes (~1000aa, pI≈11) dramatically expanded in Frogs and amniotes especially due to the accumulation of negatively charged repetitive sequences and the expansion of short RS-rich mixed charge domains. (D) AlphaFold prediction of human SON structure superimposed on the simplified diagram (as in Figure 5D). Arrows point to the structured segments of the protein, which are mostly restricted to the conserved C-terminal region. The only other structures are short alpha helices at positions 10-25, 58-71, 993-1072. (E) Deeper analysis of splicing events rescued upon replacing endogenous SON with full-length human SON, SON^ΔC, SON^C-RSRP, and SON^C-RSRP-NEG shows that the rescue capacity of truncation construct depends on the GC content of the exon. GC-rich exons are rescued to a lesser extent or, in many cases with SON^C-RSRP and SON^C-RSRP-NEG, are spliced even less efficiently than in SON knockdown. (F) poly(A)+ RNA-FISH and DNA staining along with HaloTMR labelling in human Flp-In T-REx HEK293 cells treated with control siRNA or siRNA targeting endogenous SON. (G) Same staining as in (F) of the cell lines where the endogenous SON was replaced with one of the truncation constructs (similar to Figure 5E, G. (H) Quantification of the eccentricity of p(A)+ RNA speckles using the images from (G). Higher eccentricity corresponds to more elongated speckles, while lower eccentricity to more spherical bodies.

NIHMS2178151-supplement-Supplementary_Figure_5.pdf^{(17.3MB, pdf)}

Supplementary Figure 6

Figure S6: Leveled gene architecture emerges in land vertebrates, related to Figure 6 (A) Cramér’s V values, calculated from the chi-squared metric (Methods) summarized for each taxa shown in Figures 6A–B. (B) (Left) Upstream intron length (log10) vs GC content plot of all introns in the human genome, genes with less than two introns are ignored. (Right) Same plot as in left, but the upstream introns in GC-rich leveled exon-introns (Methods) are plotted. (C) Comparisons of genome composition of all eukaryotic genomes (from NCBI) is shown with regards to prevalence of leveled gene architecture (y-axis) and the clustering of exons with the leveled architecture (x-axis). Each dot represents one genome and different eukaryotes groups are represented by the indicated colors on the legend panel. (D) Radial plot shows that human GC-rich leveled exons, shown in (B), are found in proximity to speckles. (E) Respective percentages of leveled gene architecture (top) and genome size (bottom) are shown for invertebrates and selected taxa of vertebrates. Genome size rapidly expands upon water-to-land transition, but stabilizes in amniotes. The point of inflection in genome size corresponds with the emergence of GC-rich leveled exons. (F) Representation of intron architectures with respect to their length (x-axis) and GC content (y-axis) demonstrates the presence of a specific class of short and GC-rich introns in turtles (green sea turtle), birds (crow) and mammals (cat), but lack of these in other presented organisms. Data underlying these plots for all genomes available through NCBI was used to generate UMAP in Figure 6D. (G) p(A)+ RNA-FISH in multiple species’ fibroblasts (human, chimpanzee, orangutan, rabbit, rat and zebrafish) and hepatocytes (eel). Samples were counterstained with DAPI to label nuclear DNA. Scale bar=5 μm. (H) p(A)+ RNA-FISH (as in Figure 6E) of different cell lines originating from different tissues of fruit fly show lack of speckles in all tested cell types. Scale bar=5 μm. (I) Fish cells expressing fish, fruit fly or human SON. p(A)+ RNA FISH staining (left) and the manual quantification of the prevalence of speckle-like bodies in the cells expressing tested constructs (right). (J) RNA-seq data from fish cells expressing human SON show minimal changes in splicing. Normalized counts mapping to human SON in FACS-sorted fish cells (left) rMats analysis of splicing as in Figure 3A (middle), GC content of alternatively spliced exons-introns in hSON-expressing fish cells (right).

NIHMS2178151-supplement-Supplementary_Figure_6.pdf^{(10.6MB, pdf)}

Supplementary Table 1

Supplemental Table 1: Oligonucleotides and Plasmids Used, related to Figures 1B, 5D, 6F and Supplemental Figures 4F and 5G.

The sequences of oligos used in cloning and RT-PCR assays, the plasmid sequences used in stable cell line generation are listed.

NIHMS2178151-supplement-Supplementary_Table_1.xlsx^{(23.8KB, xlsx)}

Supplementary Table 2

Supplemental Table 2: Differential H3K9me3 analysis of CUT&RUN libraries generated from single SRRM2-depleted double tagged cell line, related to Figure 1G, I and Supplemental Figure 1K.

Differential H3K9me3 analysis of CUT&RUN libraries generated from single SRRM2-depleted double tagged cell line.

NIHMS2178151-supplement-Supplementary_Table_2.csv^{(2.9MB, csv)}

Supplementary Table 3

Supplemental Table 3: Differential H3K9me3 analysis of CUT&RUN libraries generated from single SON-depleted double tagged cell line, related to Figure 1G, I and Supplemental Figure 1K.

Differential H3K9me3 analysis of CUT&RUN libraries generated from single SON-depleted double tagged cell line.

NIHMS2178151-supplement-Supplementary_Table_3.csv^{(2.9MB, csv)}

Supplementary Table 4

Supplemental Table 4: Differential H3K9me3 Loci in SON/SRRM2 Double-Depleted Cells (CUT&RUN), related to Figure 1G, I and Supplemental Figure 1K.

Differential H3K9me3 analysis of CUT&RUN libraries generated from double SON & SRRM2-depleted double tagged cell line.

NIHMS2178151-supplement-Supplementary_Table_4.csv^{(2.9MB, csv)}

Supplementary Table 5

Supplemental Table 5: Differential Gene Expression (Nascent RNA-seq) in SRRM2-Depleted Cells, related to Figure 2A and Supplemental Figure 2B.

Differential gene expression analysis of Nascent (pulse) RNA seq libraries generated from single SRRM2-depleted double tagged cell line.

NIHMS2178151-supplement-Supplementary_Table_5.csv^{(3.4MB, csv)}

Supplementary Table 6

Supplemental Table 6: Differential Gene Expression (Nascent RNA-seq) in SON-Depleted Cells, related to Figure 2A and Supplemental Figure 2B.

Differential gene expression analysis of Nascent (pulse) RNA seq libraries generated from single SON-depleted double tagged cell line.

NIHMS2178151-supplement-Supplementary_Table_6.csv^{(3.3MB, csv)}

Supplementary Table 7

Supplemental Table 7: Differential Gene Expression (Nascent RNA-seq) in SON/SRRM2 Double-Depleted Cells, related to Figure 2A and Supplemental Figure 2B.

Differential gene expression analysis of Nascent (pulse) RNA seq libraries generated from double SON& SRRM2-depleted double tagged cell line.

NIHMS2178151-supplement-Supplementary_Table_7.csv^{(3.4MB, csv)}

Supplementary Table 8

Supplemental Table 8: Differential Gene Expression (PolyA+ RNA-seq) in SRRM2-Depleted Cells, related to Figure 2B and Supplemental Figure 2A, B, C.

Differential gene expression analysis of polyA(+) RNA seq libraries generated from double SON&SRRM2-tagged cell line upon SRRM2 depletion with dTAG13

NIHMS2178151-supplement-Supplementary_Table_8.tsv^{(4.5MB, tsv)}

Supplementary Table 9

Supplemental Table 9: Differential Gene Expression (PolyA+ RNA-seq) in SON-Depleted Cells, related to Figure 2B and Supplemental Figure 2A, B, C.

Differential gene expression analysis of polyA(+) RNA seq libraries generated from double SON&SRRM2-tagged cell line upon SON depletion with Halo-PROTAC3 degrader.

NIHMS2178151-supplement-Supplementary_Table_9.tsv^{(4.6MB, tsv)}

Supplementary Table 10

Supplemental Table 10: Differential Gene Expression (PolyA+ RNA-seq) in SON/SRRM2 Double-Depleted Cells, related to Figure 2B and Supplemental Figure 2A, B, C, E, H.

Differential gene expression analysis of polyA(+) RNA seq libraries generated from double SON&SRRM2-tagged cell line upon double depletion with Halo-PROTAC3 & dTAG13 degrader.

NIHMS2178151-supplement-Supplementary_Table_10.tsv^{(4.6MB, tsv)}

Supplementary Table 11

Supplemental Table 11: Mass Spectrometry Identified Proteins in Single and Double Depletion Cell Lines, related to Figure 3G, Figure 4B, C, Supplemental Figure 4A, C.

List of proteins from Mass Spectrometry proteomics measurement of degron cell lines with single or double treatments.

NIHMS2178151-supplement-Supplementary_Table_11.csv^{(3MB, csv)}

Supplementary Table 12

Supplemental Table 12: Gene Expression and Splicing Changes in SON/SRRM2 Single and Double Depletions, related to Figures 3, 4 and Supplemental Figures 3 and 4.

Detailed gene expression and splicing comparisons between individual depletions of SON, SRRM2 and the double-depletion.

NIHMS2178151-supplement-Supplementary_Table_12.xlsx^{(1.9MB, xlsx)}

Supplementary Table 13

Supplemental Table 13: ENCORE Data Accession Numbers for Gene Expression and H3 Isochore Analysis, related to Figure 2J and Supplemental Figure 2I.

Accession numbers for the ENCORE datasets to analyse changes in gene expression and its enrichment in the H3 isochore.

NIHMS2178151-supplement-Supplementary_Table_13.csv^{(50.3KB, csv)}

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit polyclonal anti-SRRM2	Thermo Fisher Scientific	Cat# PA5-66827
Rabbit polyclonal anti-SON	Sigma-Aldrich	Cat# HPA023535
Mouse monoclonal anti-SON	Santa Cruz	Cat# sc-398508
Mouse monoclonal anti-PRP8	Santa Cruz	Cat# sc-55533
Rabbit polyclonal anti-SRRM1	abcam	Cat# ab221061
Mouse monoclonal anti-1H4 (p-SR)	Sigma-Aldrich	Cat# MABE50
Mouse monoclonal anti-pSR	Santa Cruz	Cat# sc-13509
Rabbit polyclonal anti-SF3B1	Bethyl	Cat# A300-996A
Rabbit monoclonal anti-Phospho-SF3B1 (Thr313)	Cell Signaling	Cat# 25009
Mouse monoclonal anti-SF3A2	Santa Cruz	Cat# sc-130563
Mouse monoclonal anti-U1-70K	SySy	Cat# 203011
Mouse monoclonal anti-U2AF2 (U2AF65)	Santa Cruz	Cat# sc-53942
Rabbit monoclonal anti-RNAPII NTD	Cell Signaling	Cat# 14958
Mouse monoclonal anti-alpha-tubulin	Santa Cruz	Cat# sc-32293
Rabbit polyclonal anti-NOLC1	Sigma-Aldrich	Cat# HPA050388
Rabbit polyclonal anti-H3K9me3	abcam	Cat# ab8898
Rabbit polyclonal anti-lamin B1	abcam	Cat# ab16048

Bacterial and virus strains
Stellar^™ Competent Cells	Takara	Cat# 636763

Biological samples
10% fly extract	Asifa Akhtar lab	N/A

Chemicals, peptides, and recombinant proteins
HaloTag^® TMR ligand	Promega	Cat# G8252
SNAP-Cell TMR-Star	NEB	Cat# S9105S
dTag	Tocris	Cat# 6605
HaloPROTAC3	MedChemExpress; Promega	Cat#HY-111997; Cat# GA3110
ent-HaloPROTAC3	Promega	Cat# GA4110
Halo-Trap resin	ChromoTek	Cat# otma
bFGF	Nacalai Biotech	Cat# 19155-07
Sodium pyruvate	Nacalai Biotech	Cat# 09859-65
Gentamicin	Thermo Fisher Scientific	Cat# 15750060
Antibiotic-Antimycotic Mixed Stock Solution(100x)	Nacalai Biotech	Cat# 02892-54
mTeSR	Stem Cell Technologies	Cat# 100-1215
Leibovitz’s L-15 medium	Thermo Fisher Scientific	Cat# 11415064
M3 medium	Sigma-Aldrich	Cat# S3652
Insulin	Capricorn Scientific	Cat# INS-K
cOmplete EDTA-free protease inhibitors cocktail	Roche	Cat# 05056489001
PhosSTOP	Roche	Cat# 04906845001
Benzonase HC	Millipore	Cat# 71205
SMGi-11j	MedChemExpress	Cat# HY-124719
4-thiouridine	Sigma-Aldrich	Cat# T4509
MTSEA Biotin-XX	Biotium	Cat# BOT-90066
uMACS streptavidin MicroBeads	Miltenyi Biotec	Cat# 130-133-282
RNAse A/T1 mix	Thermo Fisher Scientific	Cat# EN0551
NIaIII	NEB	Cat# R0125S
T4 ligase	NEB	Cat# M0202L
Concavalin A	Sigma-Aldrich	Cat# C5275
SNAP-Cell 647-SiR	NEB	Cat# S9102S
Nunc Thermanox Coverslips	Thermo Fisher Scientific	Cat# 150067
Glutaraldehyde	Sigma-Aldrich	Cat# 111-30-8
Osmium tetroxide	Science Services	Cat# E19100
Tannic acid	Electron Microscopy Sciences	Cat#21700
Uranyl acetate	Fluka	Cat# 179632
TEM Grids, Formvar/Carbon Film coated, 2x1mm Slot, Cu, 3.05mm	Science Services	Cat# EFCF2010-Cu-50
Uranyless EM Stain	Science Services	Cat# DM22409
Lead Citrate	Science Services	Cat# DM22410
Plastic bottom μ-Slide 8 Well plates	ibidi	Cat# 80806
Agarose resin	Pierce	Cat# 26150
Dynabeads Protein G beads	Invitrogen	Cat# 10003D
RNaseI	Thermo Fisher Scientific	Cat# EN0601

T4 Polynucleotide Kinase	NEB	Cat# M0201
T4 RNA Ligase 1	NEB	Cat# M0204
SuperScript III	Invitrogen	Cat# 18080093
RNaseH	NEB	Cat# M0297
CircLigase II	Lucigen	Cat# 131402
Q5 polymerase	NEB	Cat# M0491
Geneticin	Thermo Fisher Scientific	Cat# 10131-027
LysC	Roche	Cat# 11420429001
Evotip Pure	Evosep	Cat# EV2011
VECTASHIELD Antifade Mounting Medium	VECTASHIELD	Cat# H-1000-10
XCP-18-orange	Metasystems	Cat# D-0318-050-OR
XCP19-green	Metasystems	Cat# D-0319-050-FI

Critical commercial assays
FuGene HD transfection assay	Promega	Cat# E2311
Lipofectamine 2000	Thermo Fisher Scientific	Cat# 11668019
RNAiMAX	Thermo Fisher Scientific	Cat# 13778030
Quick-RNA MicroPrep kit	Zymo research	Cat# R1050
Dynabeads mRNA DIRECT Purification kit	Thermo Fisher Scientific	Cat# 61012
KAPA RNA HyperPrep kit	Roche	Cat# KK8544
KAPA RiboErase kit	Roche	Cat# KK8481
KAPA Stranded RNA-Seq kit	Roche	Cat# KK8400
PrimeScript RT	Takara	Cat# RR036A
DreamTaq PCR	Thermo Fisher Scientific	Cat# K1081
NEBNext Ultra II DNA Library Prep Kit for Illumina	NEB	Cat# E7645S/L
ONT native barcoding kit	Oxford Nanopore Technologies	Cat# SQK-NBD114.24
Spurr low viscosity kit	Ted Pella	Cat# 18108

In-Fusion Snap Assembly Master Mix	Takara	Cat# 638944
Deposited data
Pore-C data	This study; NCBI Gene Expression Omnibus	GSE264354
RNA sequencing data	This study; NCBI Gene Expression Omnibus	GSE264354
FLASH data	This study; NCBI Gene Expression Omnibus	GSE264354
SON/SRRM2 KD RNA sequencing data	Publicly available dataset from Zhang, Gu, Guo et al. ¹⁶	GSE242898
TSA sequencing data	Publicly available dataset from Zhang et al. ¹³	4DNFI8HIU45G; 4DNFIBY8G6RZ
FLASH data	Publicly available dataset from Ilık, Glažar et al. ²⁸	GSE223253
RNA sequencing data	ENCORE	Accession numbers are listed in Supplemental Table 13


All genomes available	Publicly available from NCBI’s Eukaryotic Genome Annotation pipeline (Thibaud-Nissen et al.) ⁵⁷	https://www.ncbi.nlm.nih.gov/refseq/annotation_euk/all/

Experimental models: Cell lines
Human HCT116	ATCC	CCL247
Human Flp-In T-Rex HEK293	Thermo Fisher Scientific	R78007
Human HDFa	Thermo Fisher Scientific	C0135C
Mouse FlpIn 3T3	Thermo Fisher Scientific	R76107
Chicken DF-1	Mundlos lab, Max Planck Institute for Molecular Genetics	ATCC: CRL-3586
Common box turtle TH-1	Heyd lab, Freie Universität Berlin	ATCC: CCL-50
Fathead minnow EPC	Boehm lab, (from J. Swann) Max Planck Institute of Immunobiology and Epigenetics	ATCC: CRL-2872
Fruit fly S2	Akhtar lab, Max Planck Institute of Immunobiology and Epigenetics	N/A
Fruit fly Kc	Akhtar lab, Max Planck Institute of Immunobiology and Epigenetics	N/A
Fruit fly S2R+	DGRC	RRID: CVCL_Z831
Fruit fly OSS	Akhtar lab, Max Planck Institute of Immunobiology and Epigenetics	N/A
Chimpanzee fibroblasts	Kyoto Univeristy Primate Research Institute (KUPRI)	GAIN ID: 0166
Orangutan fibroblasts	Kyoto Univeristy Primate Research Institute (KUPRI)	GAIN ID: 0091
Rabbit fibroblasts	RIKEN BRC	SIRC
Rat lung fibroblasts	RIKEN BRC	JTC-19
Eel	RIKEN BRC	Hepa-E1
Zebrafish	RIKEN BRC	BRF4
Experimental models: Organisms/strains
N/A

Oligonucleotides
RT-PCR primers	This study	Supplemental Table 1
HaloTag	Promega	Supplemental Table 1
TRA2A silencer select siRNA	Thermo Fisher Scientific	Cat# s26665
TRA2B silencer select siRNA	Thermo Fisher Scientific	Cat# s12749
SON silencer select siRNA	Thermo Fisher Scientific	Cat# s13278
siRNA pool targeting chicken (Gallus gallus) SON	siTOOLs BIOTECH	siPOOL against NCBI ID 770637
Negative control siRNA	siTOOLs BIOTECH	siPOOL Neg-C-120
Stellaris probe mRNA positive control	Biosearch technologies	Cat# T30-TAMRA-1
Stellaris probe MALAT1	Biosearch technologies	Cat# SMF-2035-1
Primers to introduce synonymous mutations into the siRNA target sequence of the truncation constructs	This study	Supplemental Table 1

Recombinant DNA
Construct to tag endogenous SRRM2 containing SNAP-tag, FKBP^F36V, T2A, puromycin selection, and homology arms	This study	Supplemental Table 1
pOG44	Thermo Fisher Scientific	Cat# V600520
pcDNA5/FRT/TO	Thermo Fisher Scientific	Cat# V652020
Construct to tag endogenous SON containing HaloTag^®	This study	Supplemental Table 1
Software and algorithms
snakePipes	MPI-IE	V.2.7.1; https://snakepipes.readthedocs.io/en/stable/index.html
featureCounts	Ref ⁵⁸	v.2.0.1; https://rnnh.github.io/bioinfo-notebook/docs/featureCounts.html
DESeq2	Ref ⁵⁹	https://bioconductor.org/packages/release/bioc/html/DESeq2.html
chromPlot	Ref ⁶⁰	https://bioconductor.org/packages/release/bioc/html/chromPlot.html
rMats-turbo	Ref ³¹	https://github.com/Xinglab/rmats-turbo

ExOrthist	Manuel Irimia lab ⁴⁴	https://github.com/biocorecrg/ExOrthist

Nanopore Pore-C pipeline	Oxford Nanopore Technologies	V.0.2.0; https://nanoporetech.com/document/epi2me-workflows/wf-pore-c
dcHiC	Ferhat Ay lab ⁶¹	v.2.1; https://github.com/ay-lab/dcHiC

Cellprofiler	Broad Institute⁶²	v.4; https://cellprofiler.org/
LAS X software FRAP module	Leica	Leica Microsystems, https://www.leica-microsystems.com/products/microscope-software/p/leica-las-x-ls/
EasyFRAP-web	Ref⁶³	https://easyfrap.vmnet.upatras.gr/
ZEN software	Zeiss	Blue version 3.4 and 3.9; https://www.zeiss.com/microscopy/en/products/software/zeiss-zen.html
Dia-NN	Ref ⁶⁴	v.1.8.2. beta22; https://github.com/vdemichev/DiaNN
Other

Open in a new tab

Acknowledgements

We thank the entire Aktas lab, and Denes Hnisz for their input and comments. We thank the FACS, Sequencing and Microscopy and Cryo-electron Microscopy facilities at MPI-MG, especially René Buschow. We thank Daniel Ibrahim for DF1 cells, Florian Heyd for TH1 cells, Jeremy Swann for fish cells, Akhtar lab for OSS and Kc cells, Lukas Behrens for NHS-488, Alessa Ringel for DNA FISH protocol, and Kraushar lab for the SMG1 inhibitor. Primate cells were established by Miho Hakukawa and Yuriko Hirai (Imai lab), supported by NBRP and Cooperative Research (2025-056) of eHUB, Kyoto university. Work in the group of C.A. is supported by WPI, MEXT, KAKENHI Grant-in-aid for Scientific Research (A) from the Japan Society for the Promotion of Science (24H00569), Japan Agency for Medical Research and Development (AMED) (25bm1123047h0002) and JST FOREST Program (JPMJFR2332). D.S.F and F.A were partially supported by the NIH grant R35-GM128938 (F.A.). Research in the laboratories of T.A. and T.M. is funded by the Max-Planck Society. M.M. is paid from project no. 507866755 of DFG Schwerpunktprogramme 2202 (T.A.).

Footnotes

Declaration of Interests

I.A.I. and T.A. are inventors on a patent (EP3325621B1) regarding the s-oligo design used in FLASH experiments.

Declaration of Generative AI and AI-assisted technologies

The authors used ChatGPT/Gemini for grammar checks and readability improvements. The authors reviewed, edited the content, and take full responsibility for the final publication.

Data and Code availability

The data generated in this study is available at GEO with the accession number GSE264354.

The custom scripts for the analyses are available at https://github.com/aktas-lab/speckle_paper

References:

1.Eyre-Walker A, and Hurst LD (2001). The evolution of isochores. Nat. Rev. Genet 2, 549–555. [DOI] [PubMed] [Google Scholar]
2.Rao SSP, Huntley MH, Durand NC, Stamenova EK, Bochkov ID, Robinson JT, Sanborn AL, Machol I, Omer AD, Lander ES, et al. (2014). A 3D map of the human genome at kilobase resolution reveals principles of chromatin looping. Cell 159, 1665–1680. [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Chen Y, Zhang Y, Wang Y, Zhang L, Brinkman EK, Adam SA, Goldman R, van Steensel B, Ma J, and Belmont AS (2018). Mapping 3D genome organization relative to nuclear compartments using TSA-Seq as a cytological ruler. J. Cell Biol 217, 4025–4048. [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Lieberman-Aiden E, van Berkum NL, Williams L, Imakaev M, Ragoczy T, Telling A, Amit I, Lajoie BR, Sabo PJ, Dorschner MO, et al. (2009). Comprehensive mapping of long-range interactions reveals folding principles of the human genome. Science 326, 289–293. [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Quinodoz SA, Ollikainen N, Tabak B, Palla A, Schmidt JM, Detmar E, Lai MM, Shishkin AA, Bhat P, Takei Y, et al. (2018). Higher-order inter-chromosomal hubs shape 3D genome organization in the nucleus. Cell 174, 744–757.e24. [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Spector DL, and Lamond AI (2011). Nuclear speckles. Cold Spring Harb. Perspect. Biol 3. 10.1101/cshperspect.a000646. [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Tripathi V, Song DY, Zong X, Shevtsov SP, Hearn S, Fu X-D, Dundr M, and Prasanth KV (2012). SRSF1 regulates the assembly of pre-mRNA processing factors in nuclear speckles. Mol. Biol. Cell 23, 3694–3706. [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Dundr M, and Misteli T (2010). Biogenesis of nuclear bodies. Cold Spring Harb. Perspect. Biol 2, a000711. [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Ilik İA, Malszycki M, Lübke AK, Schade C, Meierhofer D, and Aktaş T (2020). SON and SRRM2 are essential for nuclear speckle formation. Elife 9. 10.7554/eLife.60579. [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Tammer L, Hameiri O, Keydar I, Roy VR, Ashkenazy-Titelman A, Custódio N, Sason I, Shayevitch R, Rodríguez-Vaello V, Rino J, et al. (2022). Gene architecture directs splicing outcome in separate nuclear spatial regions. Mol. Cell 82, 1021–1034.e8. [DOI] [PubMed] [Google Scholar]
11.Amit M, Donyo M, Hollander D, Goren A, Kim E, Gelfman S, Lev-Maor G, Burstein D, Schwartz S, Postolsky B, et al. (2012). Differential GC content between exons and introns establishes distinct strategies of splice-site recognition. Cell Rep. 1, 543–556. [DOI] [PubMed] [Google Scholar]
12.Wang Y, Zhang Y, Zhang R, van Schaik T, Zhang L, Sasaki T, Peric-Hupkes D, Chen Y, Gilbert DM, van Steensel B, et al. (2021). SPIN reveals genome-wide landscape of nuclear compartmentalization. Genome Biol. 22, 36. [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Zhang L, Zhang Y, Chen Y, Gholamalamdari O, Wang Y, Ma J, and Belmont AS (2020). TSA-seq reveals a largely conserved genome organization relative to nuclear speckles with small position changes tightly correlated with gene expression changes. Genome Res. 10.1101/gr.266239.120. [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Khanna N, Hu Y, and Belmont AS (2014). HSP70 Transgene Directed Motion to Nuclear Speckles Facilitates Heat Shock Activation. Curr. Biol 24, 1138–1144. [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Reza MH, Dutta S, Goyal R, Shah H, Dey G, and Sanyal K (2024). Expansion microscopy reveals characteristic ultrastructural features of pathogenic budding yeast species. J. Cell Sci 137, jcs262046. [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Zhang M, Gu Z, Guo S, Sun Y, Ma S, Yang S, Guo J, Fang C, Shu L, Ge Y, et al. (2024). SRRM2 phase separation drives assembly of nuclear speckle subcompartments. Cell Rep. 43, 113827. [DOI] [PubMed] [Google Scholar]
17.Sharma A, Takata H, Shibahara K-I, Bubulya A, and Bubulya PA (2010). Son is essential for nuclear speckle organization and cell cycle progression. Mol. Biol. Cell 21, 650–663. [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Ahn E-Y, DeKelver RC, Lo M-C, Nguyen TA, Matsuura S, Boyapati A, Pandit S, Fu X-D, and Zhang D-E (2011). SON controls cell-cycle progression by coordinated regulation of RNA splicing. Mol. Cell 42, 185–198. [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Nozaki T, Shinkai S, Ide S, Higashi K, Tamura S, Shimazoe MA, Nakagawa M, Suzuki Y, Okada Y, Sasai M, et al. (2023). Condensed but liquid-like domain organization of active chromatin regions in living human cells. Sci. Adv 9, eadf1488. [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Zhong J-Y, Niu L, Lin Z-B, Bai X, Chen Y, Luo F, Hou C, and Xiao C-L (2023). High-throughput Pore-C reveals the single-allele topology and cell type-specificity of 3D genome folding. Nat. Commun 14, 1250. [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Open2C, Abdennur N, Abraham S, Fudenberg G, Flyamer IM, Galitsyna AA, Goloborodko A, Imakaev M, Oksuz BA, Venev SV, et al. (2024). Cooltools: Enabling high-resolution Hi-C analysis in Python. PLoS Comput. Biol 20, e1012067. [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Skene PJ, and Henikoff S (2017). An efficient targeted nuclease strategy for high-resolution mapping of DNA binding sites. Elife 6. 10.7554/eLife.21856. [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Wu J, Xiao Y, Liu Y, Wen L, Jin C, Liu S, Paul S, He C, Regev O, and Fei J (2024). Dynamics of RNA localization to nuclear speckles are connected to splicing efficiency. Sci. Adv 10, eadp7727. [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Grimwood J, Gordon LA, Olsen A, Terry A, Schmutz J, Lamerdin J, Hellsten U, Goodstein D, Couronne O, Tran-Gyamfi M, et al. (2004). The DNA sequence and biology of human chromosome 19. Nature 428, 529–535. [DOI] [PubMed] [Google Scholar]
25.Van Nostrand EL, Freese P, Pratt GA, Wang X, Wei X, Xiao R, Blue SM, Chen J-Y, Cody NAL, Dominguez D, et al. (2020). A large-scale binding and functional map of human RNA-binding proteins. Nature 583, 711–719. [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Mintz PJ, and Spector DL (2000). Compartmentalization of RNA processing factors within nuclear speckles. J. Struct. Biol 129, 241–251. [DOI] [PubMed] [Google Scholar]
27.Ilık İA, and Aktaş T (2022). Nuclear speckles: dynamic hubs of gene expression regulation. FEBS J. 289, 7234–7245. [DOI] [PubMed] [Google Scholar]
28.Ilık İA, Glažar P, Tse K, Brändl B, Meierhofer D, Müller F-J, Smith ZD, and Aktaş T (2024). Autonomous transposons tune their sequences to ensure somatic suppression. Nature 626, 1116–1124. [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Ilik IA, Aktas T, Maticzka D, Backofen R, and Akhtar A (2020). FLASH: ultra-fast protocol to identify RNA–protein interactions in cells. Nucleic Acids Res. 48, e15–e15. [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Bhat P, Chow A, Emert B, Ettlin O, Quinodoz SA, Strehle M, Takei Y, Burr A, Goronzy IN, Chen AW, et al. (2024). Genome organization around nuclear speckles drives mRNA splicing efficiency. Nature 629, 1165–1173. [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Wang Y, Xie Z, Kutschera E, Adams JI, Kadash-Edmondson KE, and Xing Y (2024). rMATS-turbo: an efficient and flexible computational tool for alternative splicing analysis of large-scale RNA-seq data. Nat. Protoc 19, 1083–1104. [DOI] [PubMed] [Google Scholar]
32.Barutcu AR, Wu M, Braunschweig U, Dyakov BJA, Luo Z, Turner KM, Durbic T, Lin Z-Y, Weatheritt RJ, Maass PG, et al. (2022). Systematic mapping of nuclear domain-associated transcripts reveals speckles and lamina as hubs of functionally distinct retained introns. Mol. Cell 82, 1035–1052.e9. [DOI] [PubMed] [Google Scholar]
33.Best A, James K, Dalgliesh C, Hong E, Kheirolahi-Kouhestani M, Curk T, Xu Y, Danilenko M, Hussain R, Keavney B, et al. (2014). Human Tra2 proteins jointly control a CHEK1 splicing switch among alternative and constitutive target exons. Nat. Commun 5, 4760. [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Cléry A, Jayne S, Benderska N, Dominguez C, Stamm S, and Allain FH-T (2011). Molecular basis of purine-rich RNA recognition by the human SR-like protein Tra2-β1. Nat. Struct. Mol. Biol 18, 443–450. [DOI] [PubMed] [Google Scholar]
35.Neugebauer KM, and Roth MB (1997). Distribution of pre-mRNA splicing factors at sites of RNA polymerase II transcription. Genes Dev. 11, 1148–1159. [DOI] [PubMed] [Google Scholar]
36.Berget SM (1995). Exon recognition in vertebrate splicing. J. Biol. Chem 270, 2411–2414. [DOI] [PubMed] [Google Scholar]
37.Falcon F, Tanaka EM, and Rodriguez-Terrones D (2023). Transposon waves at the water-to-land transition. Curr. Opin. Genet. Dev 81, 102059. [DOI] [PubMed] [Google Scholar]
38.Sterne-Weiler T, Weatheritt RJ, Best AJ, Ha KCH, and Blencowe BJ (2018). Efficient and Accurate Quantitative Profiling of Alternative Splicing Patterns of Any Complexity on a Laptop. Mol. Cell 72, 187–200.e6. [DOI] [PubMed] [Google Scholar]
39.Wahl MC, Will CL, and Lührmann R (2009). The spliceosome: design principles of a dynamic RNP machine. Cell 136, 701–718. [DOI] [PubMed] [Google Scholar]
40.Girard C, Will CL, Peng J, Makarov EM, Kastner B, Lemm I, Urlaub H, Hartmuth K, and Lührmann R (2012). Post-transcriptional spliceosomes are retained in nuclear speckles until splicing completion. Nat. Commun 3, 994. [DOI] [PubMed] [Google Scholar]
41.Björk P, and Wieslander L (2017). Integration of mRNP formation and export. Cell. Mol. Life Sci 74, 2875–2897. [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Galganski L, Urbanek MO, and Krzyzosiak WJ (2017). Nuclear speckles: molecular organization, biological function and role in disease. Nucleic Acids Res. 45, 10350–10368. [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Chen Y, and Belmont AS (2019). Genome organization around nuclear speckles. Curr. Opin. Genet. Dev 55, 91–99. [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Márquez Y, Mantica F, Cozzuto L, Burguera D, Hermoso-Pulido A, Ponomarenko J, Roy SW, and Irimia M (2021). ExOrthist: a tool to infer exon orthologies at any evolutionary distance. Genome Biol. 22, 239. [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Galagan JE, and Selker EU (2004). RIP: the evolutionary cost of genome defense. Trends Genet. 20, 417–423. [DOI] [PubMed] [Google Scholar]
46.McInnes L, Healy J, Saul N, and Großberger L (2018). UMAP: Uniform Manifold Approximation and Projection. J. Open Source Softw 3, 861. [Google Scholar]
47.Takei Y, Zheng S, Yun J, Shah S, Pierson N, White J, Schindler S, Tischbirek CH, Yuan G-C, and Cai L (2021). Single-cell nuclear architecture across cell types in the mouse brain. Science 374, 586–594. [DOI] [PubMed] [Google Scholar]
48.Cajal S.R. y. (1910). El núcleo de las células piramidales del cerebro humano y de algunos mamíferos (Travaux du Laboratoire de Recherches Biologiques de l’Universite de Madrid). [Google Scholar]
49.Perchlik M, Sasse A, Mostafavi S, Fields S, and Cuperus JT (2023). Impact on splicing in Saccharomyces cerevisiae of random 50-base sequences inserted into an intron. RNA 30, 52–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
50.Gnan S, Matelot M, Weiman M, Arnaiz O, Guérin F, Sperling L, Bétermier M, Thermes C, Chen C-L, and Duharcourt S (2022). GC content, but not nucleosome positioning, directly contributes to intron splicing efficiency in Paramecium. Genome Res. 32, 699–709. [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Schirman D, Yakhini Z, Pilpel Y, and Dahan O (2021). A broad analysis of splicing regulation in yeast using a large library of synthetic introns. PLoS Genet. 17, e1009805. [DOI] [PMC free article] [PubMed] [Google Scholar]
52.Meunier J, and Duret L (2004). Recombination drives the evolution of GC-content in the human genome. Mol. Biol. Evol 21, 984–990. [DOI] [PubMed] [Google Scholar]
53.Duret L, and Galtier N (2009). Biased gene conversion and the evolution of mammalian genomic landscapes. Annu. Rev. Genomics Hum. Genet 10, 285–311. [DOI] [PubMed] [Google Scholar]
54.Galtier N, and Duret L (2007). Adaptation or biased gene conversion? Extending the null hypothesis of molecular evolution. Trends Genet. 23, 273–277. [DOI] [PubMed] [Google Scholar]
55.Schartl M, Woltering JM, Irisarri I, Du K, Kneitz S, Pippel M, Brown T, Franchini P, Li J, Li M, et al. (2024). The genomes of all lungfish inform on genome expansion and tetrapod evolution. Nature, 1–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
56.Kapusta A, Suh A, and Feschotte C (2017). Dynamics of genome size evolution in birds and mammals. Proc. Natl. Acad. Sci. U. S. A 114, E1460–E1469. [DOI] [PMC free article] [PubMed] [Google Scholar]
57.Thibaud-Nissen F, DiCuccio M, Hlavina W, Kimchi A, Kitts PA, Murphy TD, Pruitt KD, and Souvorov A (2016). P8008 The NCBI Eukaryotic Genome Annotation Pipeline. J. Anim. Sci 94, 184–184. [Google Scholar]
58.Liao Y, Smyth GK, and Shi W (2014). featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics 30, 923–930. [DOI] [PubMed] [Google Scholar]
59.Love MI, Huber W, and Anders S (2014). Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550. [DOI] [PMC free article] [PubMed] [Google Scholar]
60.Oróstica KY, and Verdugo RA (2016). chromPlot: visualization of genomic data in chromosomal context. Bioinformatics 32, 2366–2368. [DOI] [PubMed] [Google Scholar]
61.Chakraborty A, Wang JG, and Ay F (2022). dcHiC detects differential compartments across multiple Hi-C datasets. Nat. Commun 13, 6827. [DOI] [PMC free article] [PubMed] [Google Scholar]
62.Stirling DR, Swain-Bowden MJ, Lucas AM, Carpenter AE, Cimini BA, and Goodman A (2021). CellProfiler 4: improvements in speed, utility and usability. BMC Bioinformatics 22, 433. [DOI] [PMC free article] [PubMed] [Google Scholar]
63.Koulouras G, Panagopoulos A, Rapsomaniki MA, Giakoumakis NN, Taraviras S, and Lygerou Z (2018). EasyFRAP-web: a web-based tool for the analysis of fluorescence recovery after photobleaching data. Nucleic Acids Res. 46, W467–W472. [DOI] [PMC free article] [PubMed] [Google Scholar]
64.Demichev V, Messner CB, Vernardis SI, Lilley KS, and Ralser M (2020). DIA-NN: neural networks and interference correction enable deep proteome coverage in high throughput. Nat. Methods 17, 41–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
65.Roseman SA, Siegenfeld AP, Lee C, Lue NZ, Waterbury AL, and Liau BB (2023). DNA methylation insulates genic regions from CTCF loops near nuclear speckles. bioRxiv, 2023.07.26.550645. 10.1101/2023.07.26.550645. [DOI] [PMC free article] [PubMed] [Google Scholar]
66.Lubelsky Y, and Ulitsky I (2018). Sequences enriched in Alu repeats drive nuclear localization of long RNAs in human cells. Nature 555, 107–111. [DOI] [PMC free article] [PubMed] [Google Scholar]
67.Gregersen LH, Mitter R, and Svejstrup JQ (2022). Elongation factor-specific capture of RNA polymerase II complexes. Cell Reports Methods 0. 10.1016/j.crmeth.2022.100368. [DOI] [PMC free article] [PubMed] [Google Scholar]
68.Bhardwaj V, Heyne S, Sikora K, Rabbani L, Rauer M, Kilpert F, Richter AS, Ryan DP, and Manke T (2019). snakePipes: facilitating flexible, scalable and integrative epigenomic analysis. Bioinformatics 35, 4757–4759. [DOI] [PMC free article] [PubMed] [Google Scholar]
69.Szabo Q, Cavalli G, and Bantignies F (2021). Higher-order chromatin organization using 3D DNA fluorescent in situ hybridization. Methods Mol. Biol 2157, 221–237. [DOI] [PubMed] [Google Scholar]
70.Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, Batut P, Chaisson M, and Gingeras TR (2013). STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
71.Li YI, Knowles DA, Humphrey J, Barbeira AN, Dickinson SP, Im HK, and Pritchard JK (2017). Annotation-free quantification of RNA splicing using LeafCutter. Nat. Genet 50, 151–158. [DOI] [PMC free article] [PubMed] [Google Scholar]
72.Waskom M (2021). seaborn: statistical data visualization. J. Open Source Softw 6, 3021. [Google Scholar]
73.Open2C, Abdennur N, Fudenberg G, Flyamer IM, Galitsyna AA, Goloborodko A, Imakaev M, and Venev S (2024). Bioframe: operations on genomic intervals in Pandas dataframes. Bioinformatics 40. 10.1093/bioinformatics/btae088. [DOI] [PMC free article] [PubMed] [Google Scholar]
74.Knight PA, and Ruiz D (2013). A fast algorithm for matrix balancing. IMA J. Numer. Anal 33, 1029–1047. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Video S1

Download video file^{(445.9KB, mp4)}

Video S2

Download video file^{(3.1MB, mp4)}

Supplementary Figure 1

NIHMS2178151-supplement-Supplementary_Figure_1.pdf^{(27.2MB, pdf)}

Supplementary Figure 2

NIHMS2178151-supplement-Supplementary_Figure_2.pdf^{(17.5MB, pdf)}

Supplementary Figure 3

NIHMS2178151-supplement-Supplementary_Figure_3.pdf^{(2.7MB, pdf)}

Supplementary Figure 4

NIHMS2178151-supplement-Supplementary_Figure_4.pdf^{(4.5MB, pdf)}

Supplementary Figure 5

NIHMS2178151-supplement-Supplementary_Figure_5.pdf^{(17.3MB, pdf)}

Supplementary Figure 6

NIHMS2178151-supplement-Supplementary_Figure_6.pdf^{(10.6MB, pdf)}

Supplementary Table 1

Supplemental Table 1: Oligonucleotides and Plasmids Used, related to Figures 1B, 5D, 6F and Supplemental Figures 4F and 5G.

The sequences of oligos used in cloning and RT-PCR assays, the plasmid sequences used in stable cell line generation are listed.

NIHMS2178151-supplement-Supplementary_Table_1.xlsx^{(23.8KB, xlsx)}

Supplementary Table 2

Supplemental Table 2: Differential H3K9me3 analysis of CUT&RUN libraries generated from single SRRM2-depleted double tagged cell line, related to Figure 1G, I and Supplemental Figure 1K.

Differential H3K9me3 analysis of CUT&RUN libraries generated from single SRRM2-depleted double tagged cell line.

NIHMS2178151-supplement-Supplementary_Table_2.csv^{(2.9MB, csv)}

Supplementary Table 3

Supplemental Table 3: Differential H3K9me3 analysis of CUT&RUN libraries generated from single SON-depleted double tagged cell line, related to Figure 1G, I and Supplemental Figure 1K.

Differential H3K9me3 analysis of CUT&RUN libraries generated from single SON-depleted double tagged cell line.

NIHMS2178151-supplement-Supplementary_Table_3.csv^{(2.9MB, csv)}

Supplementary Table 4

Supplemental Table 4: Differential H3K9me3 Loci in SON/SRRM2 Double-Depleted Cells (CUT&RUN), related to Figure 1G, I and Supplemental Figure 1K.

Differential H3K9me3 analysis of CUT&RUN libraries generated from double SON & SRRM2-depleted double tagged cell line.

NIHMS2178151-supplement-Supplementary_Table_4.csv^{(2.9MB, csv)}

Supplementary Table 5

Supplemental Table 5: Differential Gene Expression (Nascent RNA-seq) in SRRM2-Depleted Cells, related to Figure 2A and Supplemental Figure 2B.

Differential gene expression analysis of Nascent (pulse) RNA seq libraries generated from single SRRM2-depleted double tagged cell line.

NIHMS2178151-supplement-Supplementary_Table_5.csv^{(3.4MB, csv)}

Supplementary Table 6

Supplemental Table 6: Differential Gene Expression (Nascent RNA-seq) in SON-Depleted Cells, related to Figure 2A and Supplemental Figure 2B.

Differential gene expression analysis of Nascent (pulse) RNA seq libraries generated from single SON-depleted double tagged cell line.

NIHMS2178151-supplement-Supplementary_Table_6.csv^{(3.3MB, csv)}

Supplementary Table 7

Supplemental Table 7: Differential Gene Expression (Nascent RNA-seq) in SON/SRRM2 Double-Depleted Cells, related to Figure 2A and Supplemental Figure 2B.

Differential gene expression analysis of Nascent (pulse) RNA seq libraries generated from double SON& SRRM2-depleted double tagged cell line.

NIHMS2178151-supplement-Supplementary_Table_7.csv^{(3.4MB, csv)}

Supplementary Table 8

Supplemental Table 8: Differential Gene Expression (PolyA+ RNA-seq) in SRRM2-Depleted Cells, related to Figure 2B and Supplemental Figure 2A, B, C.

Differential gene expression analysis of polyA(+) RNA seq libraries generated from double SON&SRRM2-tagged cell line upon SRRM2 depletion with dTAG13

NIHMS2178151-supplement-Supplementary_Table_8.tsv^{(4.5MB, tsv)}

Supplementary Table 9

Supplemental Table 9: Differential Gene Expression (PolyA+ RNA-seq) in SON-Depleted Cells, related to Figure 2B and Supplemental Figure 2A, B, C.

Differential gene expression analysis of polyA(+) RNA seq libraries generated from double SON&SRRM2-tagged cell line upon SON depletion with Halo-PROTAC3 degrader.

NIHMS2178151-supplement-Supplementary_Table_9.tsv^{(4.6MB, tsv)}

Supplementary Table 10

Supplemental Table 10: Differential Gene Expression (PolyA+ RNA-seq) in SON/SRRM2 Double-Depleted Cells, related to Figure 2B and Supplemental Figure 2A, B, C, E, H.

Differential gene expression analysis of polyA(+) RNA seq libraries generated from double SON&SRRM2-tagged cell line upon double depletion with Halo-PROTAC3 & dTAG13 degrader.

NIHMS2178151-supplement-Supplementary_Table_10.tsv^{(4.6MB, tsv)}

Supplementary Table 11

Supplemental Table 11: Mass Spectrometry Identified Proteins in Single and Double Depletion Cell Lines, related to Figure 3G, Figure 4B, C, Supplemental Figure 4A, C.

List of proteins from Mass Spectrometry proteomics measurement of degron cell lines with single or double treatments.

NIHMS2178151-supplement-Supplementary_Table_11.csv^{(3MB, csv)}

Supplementary Table 12

Supplemental Table 12: Gene Expression and Splicing Changes in SON/SRRM2 Single and Double Depletions, related to Figures 3, 4 and Supplemental Figures 3 and 4.

Detailed gene expression and splicing comparisons between individual depletions of SON, SRRM2 and the double-depletion.

NIHMS2178151-supplement-Supplementary_Table_12.xlsx^{(1.9MB, xlsx)}

Supplementary Table 13

Supplemental Table 13: ENCORE Data Accession Numbers for Gene Expression and H3 Isochore Analysis, related to Figure 2J and Supplemental Figure 2I.

Accession numbers for the ENCORE datasets to analyse changes in gene expression and its enrichment in the H3 isochore.

NIHMS2178151-supplement-Supplementary_Table_13.csv^{(50.3KB, csv)}

Data Availability Statement

The data generated in this study is available at GEO with the accession number GSE264354.

The custom scripts for the analyses are available at https://github.com/aktas-lab/speckle_paper

[R1] 1.Eyre-Walker A, and Hurst LD (2001). The evolution of isochores. Nat. Rev. Genet 2, 549–555. [DOI] [PubMed] [Google Scholar]

[R2] 2.Rao SSP, Huntley MH, Durand NC, Stamenova EK, Bochkov ID, Robinson JT, Sanborn AL, Machol I, Omer AD, Lander ES, et al. (2014). A 3D map of the human genome at kilobase resolution reveals principles of chromatin looping. Cell 159, 1665–1680. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R3] 3.Chen Y, Zhang Y, Wang Y, Zhang L, Brinkman EK, Adam SA, Goldman R, van Steensel B, Ma J, and Belmont AS (2018). Mapping 3D genome organization relative to nuclear compartments using TSA-Seq as a cytological ruler. J. Cell Biol 217, 4025–4048. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R4] 4.Lieberman-Aiden E, van Berkum NL, Williams L, Imakaev M, Ragoczy T, Telling A, Amit I, Lajoie BR, Sabo PJ, Dorschner MO, et al. (2009). Comprehensive mapping of long-range interactions reveals folding principles of the human genome. Science 326, 289–293. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R5] 5.Quinodoz SA, Ollikainen N, Tabak B, Palla A, Schmidt JM, Detmar E, Lai MM, Shishkin AA, Bhat P, Takei Y, et al. (2018). Higher-order inter-chromosomal hubs shape 3D genome organization in the nucleus. Cell 174, 744–757.e24. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R6] 6.Spector DL, and Lamond AI (2011). Nuclear speckles. Cold Spring Harb. Perspect. Biol 3. 10.1101/cshperspect.a000646. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R7] 7.Tripathi V, Song DY, Zong X, Shevtsov SP, Hearn S, Fu X-D, Dundr M, and Prasanth KV (2012). SRSF1 regulates the assembly of pre-mRNA processing factors in nuclear speckles. Mol. Biol. Cell 23, 3694–3706. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R8] 8.Dundr M, and Misteli T (2010). Biogenesis of nuclear bodies. Cold Spring Harb. Perspect. Biol 2, a000711. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R9] 9.Ilik İA, Malszycki M, Lübke AK, Schade C, Meierhofer D, and Aktaş T (2020). SON and SRRM2 are essential for nuclear speckle formation. Elife 9. 10.7554/eLife.60579. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R10] 10.Tammer L, Hameiri O, Keydar I, Roy VR, Ashkenazy-Titelman A, Custódio N, Sason I, Shayevitch R, Rodríguez-Vaello V, Rino J, et al. (2022). Gene architecture directs splicing outcome in separate nuclear spatial regions. Mol. Cell 82, 1021–1034.e8. [DOI] [PubMed] [Google Scholar]

[R11] 11.Amit M, Donyo M, Hollander D, Goren A, Kim E, Gelfman S, Lev-Maor G, Burstein D, Schwartz S, Postolsky B, et al. (2012). Differential GC content between exons and introns establishes distinct strategies of splice-site recognition. Cell Rep. 1, 543–556. [DOI] [PubMed] [Google Scholar]

[R12] 12.Wang Y, Zhang Y, Zhang R, van Schaik T, Zhang L, Sasaki T, Peric-Hupkes D, Chen Y, Gilbert DM, van Steensel B, et al. (2021). SPIN reveals genome-wide landscape of nuclear compartmentalization. Genome Biol. 22, 36. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R13] 13.Zhang L, Zhang Y, Chen Y, Gholamalamdari O, Wang Y, Ma J, and Belmont AS (2020). TSA-seq reveals a largely conserved genome organization relative to nuclear speckles with small position changes tightly correlated with gene expression changes. Genome Res. 10.1101/gr.266239.120. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R14] 14.Khanna N, Hu Y, and Belmont AS (2014). HSP70 Transgene Directed Motion to Nuclear Speckles Facilitates Heat Shock Activation. Curr. Biol 24, 1138–1144. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R15] 15.Reza MH, Dutta S, Goyal R, Shah H, Dey G, and Sanyal K (2024). Expansion microscopy reveals characteristic ultrastructural features of pathogenic budding yeast species. J. Cell Sci 137, jcs262046. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R16] 16.Zhang M, Gu Z, Guo S, Sun Y, Ma S, Yang S, Guo J, Fang C, Shu L, Ge Y, et al. (2024). SRRM2 phase separation drives assembly of nuclear speckle subcompartments. Cell Rep. 43, 113827. [DOI] [PubMed] [Google Scholar]

[R17] 17.Sharma A, Takata H, Shibahara K-I, Bubulya A, and Bubulya PA (2010). Son is essential for nuclear speckle organization and cell cycle progression. Mol. Biol. Cell 21, 650–663. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R18] 18.Ahn E-Y, DeKelver RC, Lo M-C, Nguyen TA, Matsuura S, Boyapati A, Pandit S, Fu X-D, and Zhang D-E (2011). SON controls cell-cycle progression by coordinated regulation of RNA splicing. Mol. Cell 42, 185–198. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R19] 19.Nozaki T, Shinkai S, Ide S, Higashi K, Tamura S, Shimazoe MA, Nakagawa M, Suzuki Y, Okada Y, Sasai M, et al. (2023). Condensed but liquid-like domain organization of active chromatin regions in living human cells. Sci. Adv 9, eadf1488. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R20] 20.Zhong J-Y, Niu L, Lin Z-B, Bai X, Chen Y, Luo F, Hou C, and Xiao C-L (2023). High-throughput Pore-C reveals the single-allele topology and cell type-specificity of 3D genome folding. Nat. Commun 14, 1250. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R21] 21.Open2C, Abdennur N, Abraham S, Fudenberg G, Flyamer IM, Galitsyna AA, Goloborodko A, Imakaev M, Oksuz BA, Venev SV, et al. (2024). Cooltools: Enabling high-resolution Hi-C analysis in Python. PLoS Comput. Biol 20, e1012067. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R22] 22.Skene PJ, and Henikoff S (2017). An efficient targeted nuclease strategy for high-resolution mapping of DNA binding sites. Elife 6. 10.7554/eLife.21856. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R23] 23.Wu J, Xiao Y, Liu Y, Wen L, Jin C, Liu S, Paul S, He C, Regev O, and Fei J (2024). Dynamics of RNA localization to nuclear speckles are connected to splicing efficiency. Sci. Adv 10, eadp7727. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R24] 24.Grimwood J, Gordon LA, Olsen A, Terry A, Schmutz J, Lamerdin J, Hellsten U, Goodstein D, Couronne O, Tran-Gyamfi M, et al. (2004). The DNA sequence and biology of human chromosome 19. Nature 428, 529–535. [DOI] [PubMed] [Google Scholar]

[R25] 25.Van Nostrand EL, Freese P, Pratt GA, Wang X, Wei X, Xiao R, Blue SM, Chen J-Y, Cody NAL, Dominguez D, et al. (2020). A large-scale binding and functional map of human RNA-binding proteins. Nature 583, 711–719. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R26] 26.Mintz PJ, and Spector DL (2000). Compartmentalization of RNA processing factors within nuclear speckles. J. Struct. Biol 129, 241–251. [DOI] [PubMed] [Google Scholar]

[R27] 27.Ilık İA, and Aktaş T (2022). Nuclear speckles: dynamic hubs of gene expression regulation. FEBS J. 289, 7234–7245. [DOI] [PubMed] [Google Scholar]

[R28] 28.Ilık İA, Glažar P, Tse K, Brändl B, Meierhofer D, Müller F-J, Smith ZD, and Aktaş T (2024). Autonomous transposons tune their sequences to ensure somatic suppression. Nature 626, 1116–1124. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R29] 29.Ilik IA, Aktas T, Maticzka D, Backofen R, and Akhtar A (2020). FLASH: ultra-fast protocol to identify RNA–protein interactions in cells. Nucleic Acids Res. 48, e15–e15. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R30] 30.Bhat P, Chow A, Emert B, Ettlin O, Quinodoz SA, Strehle M, Takei Y, Burr A, Goronzy IN, Chen AW, et al. (2024). Genome organization around nuclear speckles drives mRNA splicing efficiency. Nature 629, 1165–1173. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R31] 31.Wang Y, Xie Z, Kutschera E, Adams JI, Kadash-Edmondson KE, and Xing Y (2024). rMATS-turbo: an efficient and flexible computational tool for alternative splicing analysis of large-scale RNA-seq data. Nat. Protoc 19, 1083–1104. [DOI] [PubMed] [Google Scholar]

[R32] 32.Barutcu AR, Wu M, Braunschweig U, Dyakov BJA, Luo Z, Turner KM, Durbic T, Lin Z-Y, Weatheritt RJ, Maass PG, et al. (2022). Systematic mapping of nuclear domain-associated transcripts reveals speckles and lamina as hubs of functionally distinct retained introns. Mol. Cell 82, 1035–1052.e9. [DOI] [PubMed] [Google Scholar]

[R33] 33.Best A, James K, Dalgliesh C, Hong E, Kheirolahi-Kouhestani M, Curk T, Xu Y, Danilenko M, Hussain R, Keavney B, et al. (2014). Human Tra2 proteins jointly control a CHEK1 splicing switch among alternative and constitutive target exons. Nat. Commun 5, 4760. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R34] 34.Cléry A, Jayne S, Benderska N, Dominguez C, Stamm S, and Allain FH-T (2011). Molecular basis of purine-rich RNA recognition by the human SR-like protein Tra2-β1. Nat. Struct. Mol. Biol 18, 443–450. [DOI] [PubMed] [Google Scholar]

[R35] 35.Neugebauer KM, and Roth MB (1997). Distribution of pre-mRNA splicing factors at sites of RNA polymerase II transcription. Genes Dev. 11, 1148–1159. [DOI] [PubMed] [Google Scholar]

[R36] 36.Berget SM (1995). Exon recognition in vertebrate splicing. J. Biol. Chem 270, 2411–2414. [DOI] [PubMed] [Google Scholar]

[R37] 37.Falcon F, Tanaka EM, and Rodriguez-Terrones D (2023). Transposon waves at the water-to-land transition. Curr. Opin. Genet. Dev 81, 102059. [DOI] [PubMed] [Google Scholar]

[R38] 38.Sterne-Weiler T, Weatheritt RJ, Best AJ, Ha KCH, and Blencowe BJ (2018). Efficient and Accurate Quantitative Profiling of Alternative Splicing Patterns of Any Complexity on a Laptop. Mol. Cell 72, 187–200.e6. [DOI] [PubMed] [Google Scholar]

[R39] 39.Wahl MC, Will CL, and Lührmann R (2009). The spliceosome: design principles of a dynamic RNP machine. Cell 136, 701–718. [DOI] [PubMed] [Google Scholar]

[R40] 40.Girard C, Will CL, Peng J, Makarov EM, Kastner B, Lemm I, Urlaub H, Hartmuth K, and Lührmann R (2012). Post-transcriptional spliceosomes are retained in nuclear speckles until splicing completion. Nat. Commun 3, 994. [DOI] [PubMed] [Google Scholar]

[R41] 41.Björk P, and Wieslander L (2017). Integration of mRNP formation and export. Cell. Mol. Life Sci 74, 2875–2897. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R42] 42.Galganski L, Urbanek MO, and Krzyzosiak WJ (2017). Nuclear speckles: molecular organization, biological function and role in disease. Nucleic Acids Res. 45, 10350–10368. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R43] 43.Chen Y, and Belmont AS (2019). Genome organization around nuclear speckles. Curr. Opin. Genet. Dev 55, 91–99. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R44] 44.Márquez Y, Mantica F, Cozzuto L, Burguera D, Hermoso-Pulido A, Ponomarenko J, Roy SW, and Irimia M (2021). ExOrthist: a tool to infer exon orthologies at any evolutionary distance. Genome Biol. 22, 239. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R45] 45.Galagan JE, and Selker EU (2004). RIP: the evolutionary cost of genome defense. Trends Genet. 20, 417–423. [DOI] [PubMed] [Google Scholar]

[R46] 46.McInnes L, Healy J, Saul N, and Großberger L (2018). UMAP: Uniform Manifold Approximation and Projection. J. Open Source Softw 3, 861. [Google Scholar]

[R47] 47.Takei Y, Zheng S, Yun J, Shah S, Pierson N, White J, Schindler S, Tischbirek CH, Yuan G-C, and Cai L (2021). Single-cell nuclear architecture across cell types in the mouse brain. Science 374, 586–594. [DOI] [PubMed] [Google Scholar]

[R48] 48.Cajal S.R. y. (1910). El núcleo de las células piramidales del cerebro humano y de algunos mamíferos (Travaux du Laboratoire de Recherches Biologiques de l’Universite de Madrid). [Google Scholar]

[R49] 49.Perchlik M, Sasse A, Mostafavi S, Fields S, and Cuperus JT (2023). Impact on splicing in Saccharomyces cerevisiae of random 50-base sequences inserted into an intron. RNA 30, 52–67. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R50] 50.Gnan S, Matelot M, Weiman M, Arnaiz O, Guérin F, Sperling L, Bétermier M, Thermes C, Chen C-L, and Duharcourt S (2022). GC content, but not nucleosome positioning, directly contributes to intron splicing efficiency in Paramecium. Genome Res. 32, 699–709. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R51] 51.Schirman D, Yakhini Z, Pilpel Y, and Dahan O (2021). A broad analysis of splicing regulation in yeast using a large library of synthetic introns. PLoS Genet. 17, e1009805. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R52] 52.Meunier J, and Duret L (2004). Recombination drives the evolution of GC-content in the human genome. Mol. Biol. Evol 21, 984–990. [DOI] [PubMed] [Google Scholar]

[R53] 53.Duret L, and Galtier N (2009). Biased gene conversion and the evolution of mammalian genomic landscapes. Annu. Rev. Genomics Hum. Genet 10, 285–311. [DOI] [PubMed] [Google Scholar]

[R54] 54.Galtier N, and Duret L (2007). Adaptation or biased gene conversion? Extending the null hypothesis of molecular evolution. Trends Genet. 23, 273–277. [DOI] [PubMed] [Google Scholar]

[R55] 55.Schartl M, Woltering JM, Irisarri I, Du K, Kneitz S, Pippel M, Brown T, Franchini P, Li J, Li M, et al. (2024). The genomes of all lungfish inform on genome expansion and tetrapod evolution. Nature, 1–8. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R56] 56.Kapusta A, Suh A, and Feschotte C (2017). Dynamics of genome size evolution in birds and mammals. Proc. Natl. Acad. Sci. U. S. A 114, E1460–E1469. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R57] 57.Thibaud-Nissen F, DiCuccio M, Hlavina W, Kimchi A, Kitts PA, Murphy TD, Pruitt KD, and Souvorov A (2016). P8008 The NCBI Eukaryotic Genome Annotation Pipeline. J. Anim. Sci 94, 184–184. [Google Scholar]

[R58] 58.Liao Y, Smyth GK, and Shi W (2014). featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics 30, 923–930. [DOI] [PubMed] [Google Scholar]

[R59] 59.Love MI, Huber W, and Anders S (2014). Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R60] 60.Oróstica KY, and Verdugo RA (2016). chromPlot: visualization of genomic data in chromosomal context. Bioinformatics 32, 2366–2368. [DOI] [PubMed] [Google Scholar]

[R61] 61.Chakraborty A, Wang JG, and Ay F (2022). dcHiC detects differential compartments across multiple Hi-C datasets. Nat. Commun 13, 6827. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R62] 62.Stirling DR, Swain-Bowden MJ, Lucas AM, Carpenter AE, Cimini BA, and Goodman A (2021). CellProfiler 4: improvements in speed, utility and usability. BMC Bioinformatics 22, 433. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R63] 63.Koulouras G, Panagopoulos A, Rapsomaniki MA, Giakoumakis NN, Taraviras S, and Lygerou Z (2018). EasyFRAP-web: a web-based tool for the analysis of fluorescence recovery after photobleaching data. Nucleic Acids Res. 46, W467–W472. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R64] 64.Demichev V, Messner CB, Vernardis SI, Lilley KS, and Ralser M (2020). DIA-NN: neural networks and interference correction enable deep proteome coverage in high throughput. Nat. Methods 17, 41–44. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R65] 65.Roseman SA, Siegenfeld AP, Lee C, Lue NZ, Waterbury AL, and Liau BB (2023). DNA methylation insulates genic regions from CTCF loops near nuclear speckles. bioRxiv, 2023.07.26.550645. 10.1101/2023.07.26.550645. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R66] 66.Lubelsky Y, and Ulitsky I (2018). Sequences enriched in Alu repeats drive nuclear localization of long RNAs in human cells. Nature 555, 107–111. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R67] 67.Gregersen LH, Mitter R, and Svejstrup JQ (2022). Elongation factor-specific capture of RNA polymerase II complexes. Cell Reports Methods 0. 10.1016/j.crmeth.2022.100368. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R68] 68.Bhardwaj V, Heyne S, Sikora K, Rabbani L, Rauer M, Kilpert F, Richter AS, Ryan DP, and Manke T (2019). snakePipes: facilitating flexible, scalable and integrative epigenomic analysis. Bioinformatics 35, 4757–4759. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R69] 69.Szabo Q, Cavalli G, and Bantignies F (2021). Higher-order chromatin organization using 3D DNA fluorescent in situ hybridization. Methods Mol. Biol 2157, 221–237. [DOI] [PubMed] [Google Scholar]

[R70] 70.Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, Batut P, Chaisson M, and Gingeras TR (2013). STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R71] 71.Li YI, Knowles DA, Humphrey J, Barbeira AN, Dickinson SP, Im HK, and Pritchard JK (2017). Annotation-free quantification of RNA splicing using LeafCutter. Nat. Genet 50, 151–158. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R72] 72.Waskom M (2021). seaborn: statistical data visualization. J. Open Source Softw 6, 3021. [Google Scholar]

[R73] 73.Open2C, Abdennur N, Fudenberg G, Flyamer IM, Galitsyna AA, Goloborodko A, Imakaev M, and Venev S (2024). Bioframe: operations on genomic intervals in Pandas dataframes. Bioinformatics 40. 10.1093/bioinformatics/btae088. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R74] 74.Knight PA, and Ruiz D (2013). A fast algorithm for matrix balancing. IMA J. Numer. Anal 33, 1029–1047. [Google Scholar]

PERMALINK

Nuclear speckles enable processing of RNA from GC-rich isochores

Michał Małszycki

Lisa Martina

İbrahim Avşar Ilık

Daniela Salgado Figueroa

Nirmalya Dasgupta

Menşura Feray Çoşar

Keun-Tae Kim

Gil Carraco

Beatrix Fauler

David Meierhofer

Thorsten Mielke

Hiroo Imai

Cantaş Alev

Ferhat Ay

Tuğçe Aktaş

Summary

Introduction

Figure 1. SON and SRRM2 depletion disrupts speckles, affecting chromatin mobility and nuclear architecture.

Results

Nuclear speckles restrict random chromatin movements

GC-rich genomic regions require speckles for gene expression

Figure 2. Speckles disruption affects GC-rich genomic regions.

GC-rich, leveled exon-intron architecture requires intact speckles for proper splicing

Figure 3. Speckles disruption leads to differential splicing due to the dispersal of SR-proteins.

Disrupted splicing in ΔSpeckles cells impairs nuclear export of mRNAs and induces RNA decay

Figure 4. Speckles disruption results in high entropy splicing, which causes RNA degradation via NMD.

The expanding IDRs of SON co-evolve with nuclear speckles

Figure 5. Expanding IDRs of SON correlate with nuclear speckles evolution.

The emergence of nuclear speckles coincides with the evolution of the gene-rich GC isochores in amniotes

Figure 6. Leveled gene architecture and speckles co-emerge in land vertebrates.

Discussion

Limitations of the Study

Resource availability

Lead Contact

Materials Availability

STAR METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

METHOD DETAILS

Generation of stable cell lines

Immunoblotting

siRNA transfections and transient overexpression

Cytoplasmic and nuclear RNA separation

RNA-sequencing

RT-PCR

Nascent 4sU-labeled RNA-seq

CUT&RUN

Pore-C library preparation and sequencing

Immunofluorescence

RNA FISH

3D-DNA FISH

Transmission Electron Microscopy (TEM)

paGFP-H2B

Fluorescence Recovery After Photobleaching (FRAP)

FLASH

Proteome profiling

QUANTIFICATION AND STATISTICAL ANALYSIS

Analysis of Gene Expression

Analysis of Splicing

HiC Subcompartment and SPIN state analysis

GC content analysis and UMAP

Analysis of gene distribution heterogeneity in eukaryotes

Exon-intron evolution analysis

Pore-C analysis

Supplementary Material

Acknowledgements

Footnotes

Data and Code availability

References:

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases