Chromatin Meets Condensates: Emerging Interplays Linking Nuclear Paraspeckles to Gene Activation

Abstract

Nuclear paraspeckles are membrane-less protein-protein and protein-RNA condensates that assemble co-transcriptionally upon the expression of long non-coding RNA NEAT1. While efforts in the past two decades extensively characterized the components and structures of nuclear paraspeckles, their biochemical, cellular, and physiological functions remain largely unclear. Emerging evidence has revealed that paraspeckles physically interact with active chromatin and promote its establishment, implicating the role of paraspeckles as a functional hub to maintain the active state of chromatin. Here, we discuss recent advances and key knowledge gaps in the biology of paraspeckles, including their interplay with active chromatin, disease associated mutations, and functional discrepancies between mouse and human systems.

Paraspeckle assembly: a unique requirement on NEAT1

Mammalian nuclei are highly compartmentalized and contain distinct subnuclear structures known as nuclear bodies.¹ These membrane-less organelles are macromolecular condensates formed through phase separation of proteins and RNA.¹ Emerging evidence uncovered their roles in regulating cellular processes, such as transcription, splicing, ribosome biogenesis, DNA damage response and stress response.^1,2

Among the many nuclear bodies, nuclear paraspeckles still remain functionally elusive. They are found in the interchromatin nucleoplasmic space and often located adjacent to nuclear speckles.³ Unlike transcriptional condensates formed by multivalent protein-protein and protein-RNA interactions, paraspeckle proteins assemble through well-defined protein oligomerization domains that is exclusively dependent on the expression of one specific long non-coding RNA (lncRNA), Nuclear Enriched Abundant Transcript 1 (NEAT1).^4–7 While some other nuclear bodies, such as nuclear speckles (NS) or Cajal bodies, also assemble as protein-RNA condensates, they do not rely on a single component of RNA for their assembly.^8,9 These observations highlighted unique biophysical features of nuclear paraspeckles in contrast to other nuclear condensates. There have been multiple outstanding review articles discussing the biophysical principles of paraspeckle assembly.^6,10–12 Therefore, this review mainly focuses on timely and evolving topics of nuclear paraspeckle research.

In addition to NEAT1, the core components of paraspeckles include Drosophila Behavior Human Splicing (DBHS) family proteins, NONO, SFPQ and PSPC1, as the key members of paraspeckle proteins (PSPs). The assembly of PSPs to form paraspeckles is initiated co-transcriptionally by NEAT1 expression.^13,14 The NEAT1 locus transcribes two isoforms, NEAT1_1 and NEAT1_2, which are differentiated by the alternative 3’end processing.¹⁵ Despite both isoforms are co-expressed, NEAT1_2 is the core structural component required for paraspeckle assembly and exclusively co-localized with paraspeckles while NEAT1_1 is dispensable for paraspeckle assembly and present in both paraspeckles and nucleoplasm.^15,16 DBHS proteins specifically bind to the middle domain of NEAT1_2, which is lacking in NEAT1_1, to form paraspeckles through phase separation¹⁷. Although the expression level of NEAT1_2 is a rate-limiting step for paraspeckle assembly, how NEAT1 expression is regulated in the natural course of development is unclear.^15,18 NEAT1 is absent in embryonic stem cells but becomes expressed upon lineage differentiation.¹⁹ In post-developmental tissues, NEAT1 is present in some cell populations but not universally expressed.²⁰ Intriguingly, it has been repetitively reported that in cultured cells, paraspeckles are not uniformly present and distributed.^20–22 It is unclear whether NEAT1 expression is a stochastic process in the same cell population in cultured cells or live tissues. Therefore, the dynamic nature of NEAT1 expression represents a key question towards demystifying the natural course of paraspeckle formation under variable conditions, such as cell cycles, chromatin states, environmental stimuli, and the presence of tissue-specific transcription factors (Figure 1A).^23,24 In addition, NEAT1 is also known as a stress-responsive lncRNA, whose expression is induced by various cellular stressors, including microbial infection, heat shock, and hypoxia.^25,26 These observations highlight a potential role of NEAT1 as a stress sensor in bridging cellular stimuli and paraspeckle assembly related gene activation.

Figure 1. Paraspeckle assembly and unsolved mechanistic questions.

(A Paraspeckles assemble co-transcriptionally with NEAT1_2 expression. NEAT1_2 is exclusively required for paraspeckle formation. However, how NEAT1 expression is regulated in cellular environments or physiological contexts remains unclear.

(B) DBHS proteins, NONO, SFPT, and PSPC1, are the core protein components of paraspeckles that form homo- and hetero-dimers. They further oligomerize through the extended C-terminal Coiled-coils and bind the middle domain of NEAT1_2 to assemble into paraspeckles. While all three paralogs exhibit equal affinity to dimerize and exchange rate, depletion of NONO or SFPQ abolishes paraspeckle assembly but PSPC1 appears to be dispensable and functionally promiscuous.

(C) Some evidence suggested that paraspeckle assembly alters sub-nuclear distribution of paraspeckle proteins (PSPs) until they reach an equilibrium. However, how each PSP population (DNA-bound, nucleosome-bound, or diffusive/unbound) is affected has not been rigorously tested or generalized.

(D) Paraspeckles are not evenly distributed in cultured cells or live tissues. It remains elusive whether paraspeckle formation is a stochastic or orchestrated process.

Paraspeckle assembly: the remaining questions

Paraspeckles are highly dynamic and open systems of biomolecular condensates of PSPs and RNAs. The core component DBHS proteins share 50% sequence identity and conserved domains, including two RNA-Recognition Motifs 1 and 2 (RRM1 and 2), a NonA/Paraspeckle (NOPS) domain, and an expansive Coiled-Coil tail.²⁷ DBHS proteins can form both homo- and hetero- dimers but prefer heterodimers through their N-termini and further oligomerize through their C-terminal Coiled-Coils.^28–30 The oligomerization of DBHS proteins is essential for paraspeckle formation. While DBHS oligomerization does not strictly require NEAT1 in vitro, NEAT1 is essential in living cells to guide their organization for the geometry of paraspeckles.^28,31 Microscopy and biochemical studies have demonstrated that DBHS proteins and NEAT1 form a “core-shell” structure in living cells, with DBHS proteins and NEAT1_2 middle domain in the core and the 5’ and 3’ ends of NEAT1_2 in the surface shell.²² While the three DBHS proteins exhibit similarly rapid exchange rate between the paraspeckle and diffused nucleoplasmic populations,¹⁴ NONO and SFPQ, but not PSPC1, are essential for paraspeckle formation, evident by loss-of-function studies of NONO or SFPQ in cultured cells.¹⁸ NONO-SFPQ heterodimers interact with NEAT1_2 through multivalent interactions, promoting the nucleation and stabilization of paraspeckles.¹⁷ However, while PSPC1 stably forms heterodimers with NONO or SFPQ in vitro,^29,32,33 knockdown of PSPC1 does not affect paraspeckle assembly in cultured cells (Figure 1B).^29,32,33 These studies have addressed many important questions on paraspeckle assembly. However, a key remaining question is how DBHS oligomerization is biophysically guided by NEAT1, due to a lack of biochemical kinetics studies for oligomerization in vitro and time-resolved, single-molecule resolution of paraspeckle assembly in living cells. Another intriguing unsolved question is how the stoichiometry of DBHS protein regulates paraspeckle functions. While paraspeckles are highly open and heterogeneous system assembled by DBHS homo- and hetero-dimers, how different level of DBHS expression change paraspeckle compositions and fine tune its roles toward more paralog specific functions remain an open question.

While DBHS proteins assemble into paraspeckles in the presence of NEAT1, they are not exclusively localized within paraspeckles but rather have a separate population that is diffusely distributed across the nucleoplasm.²⁴ Once NEAT1_2 is transcribed, each of the DBHS proteins is immediately recruited in situ at similar kineics.¹⁴ Thus, paraspeckle assembly is neither stochastic and thermodynamically driven nor a hierarchical, step-by-step process. Instead, it likely follows a seeding model, of which the nascent transcript of NEAT1_2 acts as a nucleation molecule to recruit DBHS proteins that randomly position themselves around the middle domain of NEAT1_2 to generate mature paraspeckles.¹⁴ Therefore, it is possible that paraspeckles recruit DBHS proteins to limit interactions with their binding partners or target loci to regulate gene expression (Figure 1C). On the other hand, paraspeckle assembly increased the local concentration of DBHS proteins, which increases the retention of A-to-I edited mRNAs and potentially other biological processes.¹⁹ It has been proposed that such RNA retention may regulate their nuclear export and translation efficiency.¹⁹ While paraspeckles assemble immediate after NEAT1_2 is transcribed, the number and size of paraspeckles are not uniformly distributed and vary from cell to cell, leaving an open question of whether this is a stochastic or orchestrated process (Figure 1D). Although loss-of-function studies implied that paraspeckles are involved in many nuclear events, including alternative splicing, transcription, and translation,^34–36 their exact biochemical functions are still poorly defined and warrant more in-depth mechanistic investigations.

Interplays with active chromatin: a “with” or “on” model?

Among the many proposed functions of nuclear paraspeckles, emerging evidence has highlighted an association between paraspeckles and chromatin. The core components of paraspeckles, including NEAT1 and the DBHS proteins, have been shown to physically interact with active chromatin sites as well as chromatin regulators, including histone modifying enzymes and chromatin remodelers, suggesting a direct participation of paraspeckles in regulating the chromatin states.^36,37

As the core RNA scaffold of paraspeckles, NEAT1 was shown to directly interact with active chromatin sites by Capture Hybridization Analysis of RNA Targets sequencing (CHART-seq), which captures genome-wide RNA-DNA interactions.³⁷ SFPQ was shown to colocalize with NEAT1 binding sites on chromatin, implicating a physical association with paraspeckles rather than diffused nucleoplasmic SFPQ.³⁷ These NEAT1/SFPQ binding sites are primarily located at transcriptional start sites (TSS) and transcriptional termination sites (TTS) of active genes, and the inhibition of transcription elongation resulted in a relocation of NEAT1 binding sites toward TSS (Figure 2A).³⁷ Biochemically, NONO and SFPQ can interact with the C-terminal domain (CTD) of RNA polymerase II.³⁸ These observations suggested that paraspeckles might participate in transcriptional elongation. Further functional interrogations are required to elucidate the exact roles of paraspeckles in the RNA Pol II dependent transcriptional cascade.

Figure 2. Mechanisms of paraspeckle-dependent gene regulation.

(A) NEAT1 and SPFQ co-localize and directly bind to active sites of chromatin, suggesting that the core structure of paraspeckles directly interacts with chromatin. They are enriched at transcriptional start site (TSS) and transcriptional termination sites (TTS) of active genes. However, the geometry and biophysical organization underlying the paraspeckle-chromatin interactions have not been defined.

(B) Emerging evidence indicated that paraspeckles are involved in gene activation by direct interactions with chromatin regulators. Top, DBHS proteins interact with ARID1B, a subunit of the SWI/SNF complex, in NEAT1-dependent manner. Such paraspeckle-ARID1B interaction regulates transcription and alternative splicing. Bottom, NONO and NEAT1 regulate the catalytic activity of H3K36 methyltransferase NSD1. NONO allosterically stimulates NSD1 through its PWWP2 domain, essential for proper H3K36me2 active chromatin formation and neural differentiation of stem cells.

(C) Some studies showed that NEAT1 binds to repressive chromatin regulators EHMT1 or EZH2 to facilitate repressive chromatin formation. However, but it is unclear NEAT1_1 or NEAT1_2 is involved and whether this is a paraspeckle-dependent mechanism.

(D) Paraspeckle assembly was shown to affect subnuclear distribution of SFPQ and dampen SFPQ binding to ADARB2 promoter, thereby reducing ADARB2 expression. However, whether this sequestration model is a generalized mechanism for gene regulation has not yet been systematically tested.

Recent studies are in line with the association of NEAT1 and DBHS proteins with gene activation. DBHS proteins were reported to interact with ARID1B in a NEAT1-dependent manner.³⁶ ARID1B is an essential component of cBAF-type SWI/SNF chromatin remodeler complex, which facilitates gene expression by increasing the accessibility of binding sites for transcription factors.³⁶ Depletion of either NEAT1 or ARID1B interrupted paraspeckle association with the SWI/SNF complex.³⁶ Moreover, depletion of NEAT1 also led to a global loss of protein-protein interactions between DBHS proteins and their interactors, including chromatin-modifying proteins, histone chaperones, and transcription factors.³⁶ At the transcriptional level, depletion of NEAT1 or ARID1B led to a global perturbation of transcription and alternative splicing events.³⁶ While these findings implied that NEAT1 is involved in transcription and splicing, the causal relationships and biochemical mechanisms remain to be defined (Figure 2B, top).

The physical interaction between nuclear paraspeckles and active chromatin is conceptually distinct from that observed for transcriptional condensates. Super-resolution Structured Illumination Microscopy (SIM) revealed that paraspeckles reside within interchromatin space adjacent to active chromatin, suggesting a “With” model of chromatin organization.²² More recently, site-specific labeling of NEAT1 combined with 3D stimulated emission depletion (STED) microscopy revealed greater details of paraspeckle organization and assembly³⁹. The 5’ and 3’ ends of NEAT1 frequently locate at opposing ends of paraspeckles, and such polarization dictates the directionality of elongation, fusion, and fission of paraspeckles, implying NEAT1 folding patterns as key determinants of different forms of paraspeckle assembly³⁹. However, how these dynamic patterns of paraspeckle assembly regulate paraspeckle functions remain a challenging but interesting question to investigate.

Biochemically, NONO and SFPQ can bind to Poll II CTD as well as transcription factors, such as SF-1 at the human CYP17 promoter.⁴⁰ A general issue that requires clarification is that many functional studies on DBHS proteins do not directly compare with a loss of NEAT1 in their systems. While DBHS proteins are well known to exert other paraspeckle-independent functions, functional studies on NEAT1 in parallel are crucial to help understand the exact functions rather than associations of paraspeckles. In contrast, transcriptional condensates, nucleated by clusters of transcription factors with defined sequence-specific DNA-binding domains or coactivators with intrinsically disordered regions, engage chromatin more directly by recruiting RNA polymerase II and coactivators to specific enhancer–promoter loops.⁷ These behaviors classify an “On” model of chromatin organization, where the condensate effectively scaffolds transcriptional machinery at target loci. These distinctions are mirrored in the role of pioneer factors, which can seed condensates at nucleosome-bound DNA to actively remodel chromatin, whereas paraspeckles do not appear to exert this remodeling function.^37,41 Since paraspeckles generally interact with active chromatin enriched at TSS without overt DNA-sequence specificity, we speculate that paraspeckles maintain an open and accessible chromatin conformation primed for non-pioneer transcription factors. Overall, while both paraspeckles and transcriptional condensates are driven by phase separation, their biophysical engagement with chromatin is very likely different: paraspeckles act as spatially adjacent organizers, whereas transcriptional condensates act as direct, functional interfaces for transcriptional regulation.

Interplays with active chromatin: emerging research priorities

More recently, DBHS protein NONO was demonstrated to bind to and allosterically stimulate a key histone H3K36 methyltransferase, NSD1, through its PWWP2 domain.⁴² H3K36me2 is a hallmark that defines the boundaries of active chromatin domains.^43,44 Depletion of NONO led to a global reduction of H3K36me2 and knockdown of NEAT1 by CRISPR interference (CRISPRi) partially phenocopied the loss of H3K36me2, implicating that paraspeckles are involved in NSD1 activation.⁴² However, NSD1 does not directly bind to SFPQ or PSPC1.⁴² A possible explanation is that SFPQ and PSPC1 are involved in paraspeckle assembly leading to an increased local concentration of NONO to stimulate NSD1. While NSD1 depletion in embryonic stem cells (ESC) led to a complete blockage of neurodevelopmental gene activation and neural differentiation, loss of NONO partially phenocopied the loss of NSD1 in ESC differentiation to neural progenitors (NPC).⁴² However, the dynamic interactions between NONO and NSD1 in the context of paraspeckles during the establishment of active chromatin remain to be tested (Figure 2B, bottom). The integration of time-resolved systems such as cell cycle synchronization in junction with chemically induced degrons on DBHS proteins or induced silencing of NEAT1 will tremendously help pinpoint the precise mechanisms of action of paraspeckle-dependent and independent functions in these biological processes.

Of note, some studies indicate that NEAT1 can also drive gene repression via its interaction with repressive chromatin regulators, such as EHMT1 and EZH2.^45,46 However, most of these studies do not specify which NEAT1 isoforms is involved, making it hard to elucidate whether paraspeckles are directly engaged in gene repression. Since NEAT1_2/paraspeckles are consistently observed in interchromatin space of euchromatin compartments, it is likely that the diffused population of NEAT1_1 independent of paraspeckles may be involved in gene repression at some loci (Figure 2C).

As paraspeckle assembly concentrates sub-nuclear localization of DBHS proteins, the rest nucleoplasmic, diffused population may reduce and therefore affect gene expression. Diffused SFPQ was shown to promote transcription of the RNA-specific adenosine deaminase B2 (ADARB2) gene.²⁴ As paraspeckles assemble, it was proposed that SFPQ becomes sequestered, leading to reduction of SFPQ bound to ADARB2 promoter and eventually repression of ADARBB2 expression.²⁴ Therefore, paraspeckles may act like a molecular hub to trap DBHS proteins and indirectly regulate DBHS-dependent genes. These observations point to an additional layer of regulation that paraspeckles may affect gene expression by modulating the ratio of aggregated versus freely diffused population of chromatin regulators. However, since NEAT1 expression and paraspeckle assembly recruit nearly 40 PSPs,¹⁵ the collectively outcome in transcriptional regulation is unlikely rendered by single PSP or single PSP-target locus. Examinations of early responses to NEAT1 expression and paraspeckle assembly using time-resolved or chemically induced degron systems are perhaps instrumental to further dissect the complex transcriptional circuits and downstream cellular processes (Figure 2D).

In the native sub-nuclear environments, paraspeckles form the core-shell structures with DBHS proteins primarily reside in the core. As NEAT1 and SFPQ colocalize on active chromatin sites, what is the biophysical organization that permits chromatin to access the core? While paraspeckles originate from NEAT1 locus, cells often present multiple paraspeckles in the interchromatin space that are not physically associated with NEAT1 locus. What are the determinants of sub-nuclear localization of paraspeckles? As NONO allosterically stimulates NSD1, how do paraspeckles participate in the establishment and inheritance of H3K36me2 active chromatin domains? These interesting observations have opened more thought-provoking questions that remain overwhelmingly unanswered. Advanced biophysical tools, such as single-particle cryo-EM, cryo-electron tomography (cryo-ET), or real-time single-molecule microscopy perhaps could help shed new lights into these questions in the future.

Genetic lesions in NEAT1 and DBHS proteins: unsettled disease mechanisms

Aside of the biological implications of NEAT1 and DBHS proteins in basic molecular and cellular processes, their causal roles in human pathogenic conditions have been increasingly recognized. Several genetic mutations in NONO and SFPQ lead to neurological disorders^47–51. Mutations of DBHS proteins were also found in a broad spectrum of cancers, of which many of them appear to be context-specific, and therefore we focus on their mutations found in congenital and developmental disorders in this review.^52,53 However, no genetic lesions in NEAT1 or PSPC1 has been reported in human congenital diseases and their roles in human development are still promiscuous (Figure 3A). Since NEAT1 is a nuclear lncRNA and does not translate into proteins, it is inherently tolerable to missense mutations. Indeed, disease-driving missense mutations found in lncRNAs are extremely rare, with exceptional cases like a RMRP mutation causing Cartilage Hair Hypoplasia (CHH)^54,55. To date, large genetic lesions, copy number variations, or promoter mutations in the NEAT1 locus have only been reported in cancer genomics studies (Table 1).⁵⁶ However, whether these genetic alterations are casual for oncogenesis remain to be investigated^57–59.

Figure 3. The potential roles of NEAT1 and DBHS proteins in human diseases.

(A) NEAT1 knockout mice develop normally except that a half of female exhibits impaired corpus luteum formation during ovary development. They also have post-developmental behavioral alterations in response to stress. In human, no congenital disease has been linked to genetic lesions in NEAT1. Whether paraspeckles play a role in human development is unclear.

(A) NONO hemizygous or NSD1 heterozygous mice exhibit post-developmental behavioral issues but no overt structural alterations in the brain during development. In human, NONO hemizygosity causes MRXS34 Syndrome and NSD1 haploinsufficiency causes Sotos Syndrome, both manifesting macrocephaly and intellectual disability. These discrepancies between mouse and human genetics indicate limitations of using mouse models to study human neurodevelopment.

(C) NONO^WT dynamically exchanges between NEAT1-bound paraspeckle and unbound diffusive populations. The patient-derived NONO^P459A mutation reduces paraspeckle localization, possibly mimicking a loss of NEAT1 to serve as a separation-of-function mutant to investigate paraspeckle-dependent functions in human neurodevelopment.

(D) SFPQ^N553H and SFPQ^L534I mutations found in ALS cause cytoplasmic aggregation in motor neurons. On the other hand, NEAT1 is also found upregulated in ALS. Whether they play paraspeckle dependent or independent roles in ALS pathogenesis has not been reconciled.

Table 1.

Genetic alterations of NEAT1 and DBHS protein in human diseases

	Genetic alteration	Altered Structural Domain(s)	Functional Annotation	Associated phenotype	References
NEAT1	Multiple missense mutations in the promoter; focally deleted	NEAT1 Promoter	Decreased NEAT1 expression	Found in breast cancer; No clearly functional data	Rheinbay et al., 2017, PMID: 28658208
NEAT1	Overexpression	N/A	Increases NEAT1 expression;	Found in Amyotrophic Lateral Sclerosis (ALS) brain tissues	Li et al., 2023, PMCID 10594666
NONO	c.90_114del, p.(Gln30Hisfs*18)	Early truncation; Retaining only the first 30 amino acids of an unstructured region	No functional NONO expression	Pathogenic; X-linked intellectual-disability (MRXS34) Syndrome with developmental delay, macrocephaly, and cardiac defects	Roessler et al., 2023, PMID: 36426740
NONO	c.217C>T, p.(Arg73*)	Early truncation in the RRM1 domain	No functional NONO expression	Pathogenic; MRXS34 Syndrome	Roessler et al., 2023, PMID: 36426740
NONO	c.224G>A, p.(Arg75His)	Missense mutation in the RRM1 domain	No functional data; possibly affecting nucleic acid binding	Found in Wilms tumor	Wegert et al., 2025, PMCID 12060375
NONO	c.1009C>T, p.(Arg337*)	Truncation in the coiled-coil domain	No functional data; possibly affecting oligomerization	Pathogenic; MRXS34 Syndrome	Roessler et al., 2023, PMID: 36426740
NONO	c.1190_1191del, p.(Asn397Lysfs*36)	Alteration in the coiled-coil domain	No functional data; possibly affecting oligomerization	Pathogenic; MRXS34 Syndrome; maternally inherited from an unaffected carrier	Roessler et al., 2023, PMID: 36426740
NONO	c.1375C>G, p.(Pro459Ala)	Missense mutation in the coiled-coil domain	Reducing paraspeckles; increasing diffused NONO in nucleoplasm in Drosophila neurons; possibly affecting oligomerization	Pathogenic; MRXS34 Syndrome; intellectual disability and developmental delay but no overt facial, neurological, or skeletal abnormalities; one patient had cariac symptoms; dysmorphic structure of compound eyes in Drosophila	Itai et al., 2023, PMCID 9849200
PSPC1	c.1568A>T, p. (Tyr523Phe)	Missense mutation in the coiled-coil domain	Reducing dimerization with NONO/SFPQ	Inducing epithelial–mesenchymal transition in hepatocellular carcinoma	Lang et al., 2019, PMCID 6914800
SFPQ	c.1657A>C, p. (Asn553His)	Missense mutation in the coiled-coil domain	Increasing binding affinity to Zinc ion; cytoplasmic aggregation of SFPQ in zebrafish primary neurons	Pathogenic; Familial amyotrophic lateral sclerosis (fALS)	Widagdo et al., 2022, PMCID 9516340
SFPQ	c.1602C>A, p. (Leu534Ile)	Missense mutation in the coiled-coil domain	Increasing binding affinity to Zinc ion; cytoplasmic aggregation of SFPQ in zebrafish primary neurons	Pathogenic; Familial amyotrophic lateral sclerosis (fALS)	Widagdo et al., 2022, PMCID 9516340

The functions of paraspeckles in development and diseases have been controversial and debatable. Mouse genetics studies demonstrated that NEAT1 is dispensable for development except that a half of female mice exhibited a defect in corpus luteum formation in a stochastic manner, implying a potential role of NEAT1 in regulating X-linked genes during ovary development.⁶⁰ While the gross brain morphology appeared normal in NEAT1-KO mice, these animals exhibited abnormal behavioral responses to stress, suggesting that paraspeckles are involved in post-developmental neural circuit formation.⁶¹ Unlike NEAT1, NONO mutations have been well documented to cause Mental Retardation, X-linked, Syndromic 34 (MRXS34) Syndrome in human (Table 1).^49,62 MRXS34 patients suffer from intellectual disabilities and exhibit a rare macrocephaly phenotype.⁴⁹ Intriguingly, NSD1 heterozygous mutations also lead to macrocephaly and intellectual disability in human Sotos Syndrome.⁶³ While NSD1^+/− or NONO^−/y mice exhibit other developmental abnormalities, they do not develop the landmark macrocephaly phenotype found in human patients, indicating a key distinction between human and mouse neurodevelopment (Figure 3B).^49,63,64 To date, there are only 65 genes associated with macrocephaly. Since NONO directly stimulates NSD1 and a loss of NONO or NEAT1 leads to global reduction of NSD1 activity and H3K36me2, it is highly possible that paraspeckles might involve in human brain development but not in mice. These observations underscore the necessity of human based systems, such as brain organoids, assembloids, or organ-on-a-chip, as future experimental models to elucidate the roles of paraspeckles in the context of development.

In addition to NONO loss-of-function mutations, a particularly intriguing case is the NONO^P459A missense mutation, which causes a mixed phenotype of cardiac overgrowth in some individuals alongside consistent intellectual disability in human patients (Table 1).⁵¹ Notably, this mutant exhibited diffused nuclear localization of NONO, reduced from paraspeckles.⁵¹ This raises the possibility that certain missense mutations mimic NEAT1 loss-of-function by dismantling paraspeckle architecture, rather than eliminating all NONO functions (Figure 3C). Such “mis-localizing” mutations could produce a clinical spectrum distinct from truncating alleles, which also remove paraspeckle-independent activities. Comparative analysis of NEAT1 knockout, NONO null, and NONO^P459A in isogenic human cell models for neural differentiation, using transcriptomics, chromatin profiling, and live-cell imaging, could directly test whether these mutations converge on a common paraspeckle-deficient state.

SFPQ stands out as an essential factor in vertebrate development.^65,66 Complete knockout of SFPQ in mice causes early embryonic lethality, with severe defects in brain and muscle development.⁶⁵ These outcomes suggest that SFPQ’s indispensable functions extend beyond paraspeckle formation, potentially encompassing transcriptional elongation, splicing regulation, and DNA repair in a paraspeckle-independent manner.⁶⁵ Decoupling these paraspeckle-dependent and -independent activities will require separation-of-function mutants and rescue experiments in null backgrounds, as well as time-resolved loss-of-function studies in early embryos to identify the first failing cellular processes. A hallmark of SFPQ missense mutations in human genetics is the familial and sporadic mutations found in human Amyotrophic Lateral Sclerosis (ALS) (Table 1).^47,48 It was reported that familial ALS mutations, such as N533H and L534I in SFPQ, increased its zinc binding in the Coiled-coil domain that led to SFPQ aggregation in cytoplasm in neurons.⁴⁷ Cytoplasmic SFPQ aggregates may reflect aberrant phase separation of its altered Coiled-coil region, a mechanism increasingly recognized for RNA-binding proteins in neurodegeneration whereby liquid-like condensates mature into more solid, fibrillar assemblies as a ‘condensatopathy’.⁶⁷ On the other hand, NEAT1 upregulation was reported in ALS patients, suggesting increased paraspeckle assembly in ALS neurons.⁶⁸ Whether SFPQ mutants have an altered nuclear function and affect paraspeckle-dependent functions in ALS pathogenesis has not been tested (Figure 3D).

PSPC1 remains the least characterized member of the DBHS family in the context of human disease. While mouse and human cell-based studies implicate PSPC1 in processes such as adipogenesis and cancer progression, its role in developmental disorders is unclear (Table 1).^52,69 This gap may be due to functional redundancy with NONO or SFPQ since PSPC1 is dispensable for paraspeckle formation. Systematic variant discovery in population-scale genome datasets, paired with functional analysis in cell types where PSPC1 is highly expressed and human-based experimental models, could clarify its contribution to human pathology.

More broadly, the field would benefit from a mechanistic characterization of DBHS functions by disease alleles or experimental mutagenesis to categorize: (1) complete loss-of-function, (2) mis-localizing mutations that disrupt paraspeckle assembly, and (3) separation-of-function variants that preserve localization but ablate specific biochemical activities. Experimentally, CRISPR saturation mutagenesis, coupled with multi-omic and imaging-based readouts for paraspeckle assembly, could rapidly assign variants to these mechanistic classes. Such systematic efforts, combined with human-specific developmental models, would move the field beyond descriptive genetics toward actionable mechanistic insight to clarify when paraspeckle disruption is the primary driver of pathology and when other DBHS functions are at play.

Concluding remarks: open questions and limitations

Since the initial discovery of paraspeckles about two decades ago³, rigorous investigations have been taken to characterize the composition and subnuclear spatial organization of this mysterious membrane-less organelle through biochemical pulldowns, proteomics, and advanced microscopy techniques^{13–15,17,22}. However, significant knowledge gaps remain. At the molecular level, biochemical pulldowns using disassociated nuclei inevitably disrupt the native environments of paraspeckles. As paraspeckles are membrane-less and highly open systems, whether the native state of protein-protein or protein-RNA interactions can survive biochemical pulldowns is questionable. Proximity labeling and crosslinking mass spectrometry approaches may further address this issue. While biochemical evidence points to a physical association between paraspeckles and active chromatin, microscopic and structural evidence are lacking. Such limitations leave critical questions of inquiry, including the geometry of paraspeckles and its influence on the topology of chromatin.

At the cellular level, how NEAT1_2 expression is transcriptionally regulated poses a critical question to understand exact roles of paraspeckles in different cell types or the cell-to-cell variation within the same cell types. Among DBHS proteins, PSPC1 is the only redundant member for paraspeckle assembly even it has the same exchange rate and binding kinetics to form heterodimers with the other two members. While PSPC1 has paraspeckle-independent functions such as interacting with and regulating TET1,⁴⁶ whether PSPC1 is truly an evolutionarily nonessential paralog for paraspeckles or has context-specific roles is still unclear. Further, while NEAT1_1 lacks the middle domain to assemble DBHS proteins and paraspeckles, it is indeed present at both paraspeckles and nucleoplasms, likely exerting both paraspeckle-dependent and independent roles.⁷⁰ Since NEAT1_1 and NEAT1_2 are transcribed from the same locus, perturbation of NEAT1_1 also affects NEAT1_2 expression, making it difficult to directly interrogate the loss-of-function studies of NEAT1_1.

At the physiological level, a lack of strong developmental phenotypes in NEAT1 knockout mice argued that paraspeckles are dispensable in the natural course of development, albeit NEAT1-KO mice exhibited certain post-developmental behavioral deviations.⁶¹ However, the discrepancies of the macrocephaly and neurological phenotypes caused by NONO mutations between mouse and human genetics highlighted a possibility that paraspeckles may regulate neurodevelopment in humans differently than in mice. Most human NONO mutations abolish both paraspeckle dependent and independent functions. Further characterization and analysis of potential separation-of-function mutants, such as NONO^P459A, in parallel with human NEAT1 loss-of-function studies would serve as a critical research strategy to dissect the paraspeckle-dependent functions of DBHS proteins and significantly advance the field. Further mechanistic studies would shed new light on the development of small molecules that selectively modulate assembly or disassembly of paraspeckles without interrupting essential functions of dimeric/monomeric DBHS proteins. Such chemical biology tools will not only serve as new research tools to dissect paraspeckle-specific functions but also provide new therapeutic interventions in paraspeckle associated human pathological conditions.

Overall, the two decades of work have gained remarkable understanding towards the biogenesis of paraspeckles but also raised far more intriguing questions. In comparison with some other macro-biomolecular condensates like stress granules, the compositions of paraspeckles are relatively well-defined. Based on current foundations, advancements in paraspeckle biology would hopefully further corroborate its role as a functional hub for the active state of chromatin and provide a broader perspective and generalize principles in the biology of nuclear and cellular condensates.