Biomolecular condensates in plant RNA silencing: insights into formation, function, and stress responses

Introduction

Cells are highly organized and compartmentalized to facilitate the numerous chemical reactions they host. In addition to membrane-bounded organelles, cells also possess membraneless compartments/organelles, also known as biomolecular condensates, that significantly contribute to cellular organization and function (Banani et al. 2017; Boeynaems et al. 2018; Zhao and Zhang 2020). Biomolecular condensates are dynamic and liquid-like in nature, harboring a concentrated mix of diverse biomolecules, including proteins, nucleic acids (DNAs and RNAs), and others. These structures are capable of undergoing fusion and separation (Brangwynne et al. 2009; Hyman et al. 2014; Gomes and Shorter 2019; Ismail et al. 2021). Differing from membrane-bound organelles, biomolecular condensates enable rapid exchange of components with their surroundings. Moreover, they possess the ability to assemble and disassemble in response to stimuli, thereby influencing cellular responses to environmental stresses (Alberti and Hyman 2021; Xu et al. 2021). Liquid-liquid phase separation (LLPS) has been considered as one of the critical driving forces behind biomolecular condensate formation. LLPS refers to the spontaneous separation of homogenous mixture into 2 or more distinct liquid phases with kinetic and thermodynamic alteration (Hyman et al. 2014; Brangwynne et al. 2015; Alberti and Dormann 2019). In certain instances, these liquid-like condensates can even transit into solid-like or gel-like structures over time (Banani et al. 2017; Shin and Brangwynne 2017). LLPS is facilitated by multivalent interactions among different components, often involving proteins with intrinsically disordered regions (IDRs), prion-like domains (PrLDs), and/or DNA/RNA binding domain, as well as RNA molecules (Alberti et al. 2019; Gomes and Shorter 2019; Martin and Holehouse 2020; Borcherds et al. 2021). Local concentrations of constituents, modification of protein/RNA components, and ATPase energy-dependent regulation collectively influence the LLPS capability and the homeostasis of biomolecular condensates (Gomes and Shorter 2019; Ismail et al. 2021). In addition, other forces also contribute to the assembly of biomolecular condensates. One such force is percolation, where biomolecules with high multivalency come together through processes like oligomerization or interactions between proteins or nucleic acids, thus forming condensates (Tibble and Gross 2023).

Biomolecular condensates play critical roles in various biological processes within eukaryotes (Banani et al. 2017; Emenecker et al. 2021; Xu et al. 2021). RNA molecules are not only major components of numerous biomolecular condensates but also act as essential regulators impacting their dynamics and functions (Rhine et al. 2020). These condensed structures, often termed RNA granules, play crucial roles across the entire spectrum of RNA's life cycle, including transcription, splicing, processing, modification, translation, and quality control of RNA molecules (Gomes and Shorter 2019; Lin and Fang 2021). In plants, biomolecular condensates have emerged as key players in regulating various aspects of growth and development, such as phytohormone- and photosynthesis-mediated growth and development (Wunder et al. 2018; Atkinson et al. 2020; He et al. 2020; Ouyang et al. 2020; Song et al. 2021; Cairo et al. 2022; Chen et al. 2022; Jing et al. 2022; Zhang et al. 2022; Kuan et al. 2023), water homeostasis (Dorone et al. 2021), osmotic stress sensing (Wang et al. 2022), response to heat or cold stress (Jung et al. 2020; Tong et al. 2022; Zhu et al. 2022), pathogen resistance (Huang et al. 2021; Wang and Gu 2022), and RNA silencing (Li et al. 2021b; Xie et al. 2021; Han et al. 2023). Notably, a significant proportion of these plant biomolecular condensates are closely associated with RNA, highlighting their crucial roles in RNA-related processes.

The RNA silencing pathway, a highly RNA-involved process, operates in coordination with various cellular condensates. In plants, a range of endogenous small RNA (sRNA) classes exists, including microRNAs (miRNAs), trans-acting small interfering RNAs (ta-siRNAs, also known as phased siRNAs or phasiRNAs), heterochromatic small interfering RNAs (hc-siRNAs), and natural antisense siRNAs (nat-siRNAs) (Huang et al. 2016; Song et al. 2019). These sRNAs are crucial components of the RNA silencing mechanism, and their biogenesis and action are closely linked to RNA-associated processes that potentially benefit from subcellular compartmentation. As an example, dicing bodies (D-bodies) serve as sites for miRNA biogenesis, pivotal in the processing of miRNAs by the dicing complex (Fang and Spector 2007; Xie et al. 2021). Processing bodies (P bodies), well characterized as locales for mRNA decay and translation inhibition (Maldonado-Bonilla 2014), have recently been observed to correlate with miRNA action and turnover (Solis-Miranda et al. 2023). Moreover, cytoplasmic siRNA bodies have been shown to integrate key components for siRNA biogenesis (Jouannet et al. 2012; Tan et al. 2023). As for hc-siRNA biogenesis, the presence of Dicer-like endonuclease 3 (DCL3) and RNA-dependent RNA polymerase 2 (RDR2) within Cajal body-like structures suggests potential connections between hc-siRNA production and these compartments (Li et al. 2006; Pontes et al. 2006).

In this review, we aim to provide a comprehensive overview of the pivotal role played by biomolecular condensates in RNA silencing pathways. We intend to summarize the current literature on the formation mechanisms and the functions of these biomolecular condensates in the biogenesis, loading, and action of various types of sRNAs. Additionally, we will explore the regulatory mechanisms that govern these biomolecular condensates in plant responses to different stress, with a special emphasis on viral infections. By synthesizing the existing research, we hope to shed light on the intricate interplay existing between biomolecular condensates and RNA silencing, offering insights into the significance of these condensates in plant defense against viral pathogens.

D-body and miRNA biogenesis

P body: site for action and turnover of miRNAs?

P bodies, along with other biomolecular condensate, play important roles in the action and turnover of miRNAs. After processing, miRNA/miRNA* duplexes generated within D-bodies are carried out of D-bodies by HYL1 and loaded into Argonaute 1 (AGO1) (Xie et al. 2021). This loading process is facilitated by the interaction between HYL1 and AGO1 with a nuclear-localized protein called CARP9, which contains IDRs (Fig. 1) (Tomassi et al. 2020).

Biomolecular condensate has been found to be involved in the action of miRNAs and siRNAs. In human cells, AGO can form miRNA-induced silencing complex droplets with TNRC6B in the cytoplasm, which accelerates miRNA target deadenylation, leading to the downregulation of target gene expression (Sheu-Gruttadauria and MacRae 2018). Interestingly, both fly and human AGO can undergo lipid-mediated phase separation on the endoplasmic reticulum (ER) membrane and facilitates nascent-peptide ubiquitination, which thus couples posttranscriptional gene silencing with protein quality control (Gao et al. 2022). In plants, miRNA-loaded AGO1 inhibits the translation of miRNA targets on the ER (Li et al. 2013). The integral ER membrane protein AMP1 interacts with AGO1 and inhibits the association of mRNA target transcripts with membrane-bound polysomes (Fig. 1) (Li et al. 2013). Transiently expressed AGO1 has been observed to accumulate in cytoplasmic granules that colocalize with the ER (Li et al. 2013). It is worth noting that plant AGO1 also contains an IDR at the N terminal (Fig. 2). It would be interesting to determine whether plant AGO also undergo phase separation on the ER.

P bodies are conserved biomolecular condensates that play important roles in RNA decay, RNA storage, and translation repression (Jang et al. 2020; Kearly et al. 2022). They are believed to be involved in sRNA-mediated gene silencing. The formation of P bodies is facilitated by proteins such as RH6/8/12 (Chantarachot et al. 2020). Regarding miRNA-mediated gene silencing, the exonuclease XRN4 has been implicated in the degradation of both 5′ and 3′ cleavage products generated by AGO1 slicing of target mRNAs (Fig. 1) (Souret et al. 2004; Ren et al. 2014). XRN4 interacts with AGO2 and forms cytoplasmic foci that co-localize with P bodies (Weber et al. 2008; Liu et al. 2020). Additionally, P body–localized proteins, such as VARICOSE (Brodersen et al. 2008), a component of the decapping complex, and SUO (Yang et al. 2012), a GW-repeat protein, have been identified as regulators of miRNA-mediated translational inhibition in Arabidopsis. However, it remains to be determined whether plant AGOs carry out target mRNA cleavage and translational inhibition within P bodies. Considering that plant AGO1 inhibits the translation of miRNA targets on the ER (Li et al. 2013) and human AGO1 undergoes phase separation on the ER (Gao et al. 2022), investigating the relationship between AGO granules on the ER and P bodies stands as an intriguing avenue for further exploration. In humans, the Importin β protein Imp8 has been shown to form a complex with AGO within P bodies, facilitating the delivery of AGO complexes to target mRNAs (Weinmann et al. 2009). In Arabidopsis, 2 importin β family proteins, EMA1 (Wang et al. 2011) and TRN1 (Cui et al. 2016), also have been implicated in the regulation of miRNA loading into AGO1 (Fig. 1). Moreover, TRN1 but not EMA1 has been shown to interact with AGO1 in both the nucleus and cytoplasm (Cui et al. 2016).

The turnover of miRNAs plays a crucial role in their homeostasis, and the integration of proteins and RNAs may involve biomolecular condensates. In Arabidopsis, unmethylated miRNA duplexes can be degraded by the exoribonuclease ATRIMMER 2 (ATRM2) in a 3′ to 5′ direction (Fig. 1) (Wang et al. 2018a). ATRM2 colocalizes and interacts with AGO1 in stress granules, biomolecular condensates often associated with P bodies, upon heat stress. For methylated miRNAs, proteins SDN1-3 are responsible for their truncation, both when free and bound to AGO1 (Ramachandran and Chen 2008; Yu et al. 2017; Chen et al. 2018). Following truncation, HESO1 and URT1 cooperate to uridylate the substrates of SDN1-3, thereby promoting their degradation (Fig. 1) (Ren et al. 2012; Zhao et al. 2012; Tu et al. 2015; Wang et al. 2015). HESO1 and URT1 colocalize and interact with AGO1 in cytoplasmic speckles in Nicotiana benthamiana (Ren et al. 2014; Wang et al. 2015). Considering that AGO1 is also found in P bodies and stress granules (Wang et al. 2018a; Solis-Miranda et al. 2023), these cytoplasmic speckles may represent P bodies and/or stress granules. Furthermore, AtXRN4, an exonuclease, contributes to the turnover of some miRNA*s, the passenger strand of the miRNA duplexes. Transiently expressed AtXRN4 interacts with AGO2 and forms granules within P bodies in N. benthamiana (Fig. 1) (Liu et al. 2020). A recent review has comprehensively summarized the presence of the previously mentioned components within P bodies. These components include various AGOs (such as AGO1, AGO2, AGO5, and AGO9), XRN4, and RH6/8/12 (Solis-Miranda et al. 2023). This compilation of evidence suggests a potential connection between P bodies and the regulation of miRNA turnover.

P bodies play significant roles in responses to abiotic and biotic stresses (Jang et al. 2019). Viruses, on the other hand, have developed various mechanisms to target specific components of P bodies and modulate their accumulation, thereby influencing their functions (Table 3) (Makinen et al. 2017). For example, infection of Tobacco mosaic virus in N. benthamiana leads to transcriptional up-regulation of genes associated with RNA decay, as well as an increase in the size of P bodies and siRNA bodies (Conti et al. 2017). Similarly, plants recovered from Tobacco rattle virus infection also display an increased number of P bodies (Ma et al. 2015). Although the exact impact of viruses on P body accumulation in other cases remains ambiguous, it is evident that the functions of P body are affected. For example, the HC-Pro protein of Potato Virus A has been found to relocate components of P bodies and stress granules to form of Potato Virus A-induced granules. The number of Potato Virus A-induced granules decreases when HC-Pro is impaired (Hafren et al. 2015). The BV1 protein of Cabbage leaf curl Virus activates the decapping activity of DCP2 and siRNA production and promotes plant susceptibility to viral infection (Ye et al. 2015). The VPg protein of Plum pox virus and Turnip Mosaic Virus interacts with RH6/8/12, core components that drive the formation of P bodies and D-bodies, and promotes virus proliferation (Huang et al. 2010; Chantarachot et al. 2020; Li et al. 2021b). Cauliflower mosaic virus relocates P body components DCP5, VCS, and LSM1 into viral factories, facilitating the translation of viral RNA (Hoffmann et al. 2022). Additionally, the decapping activator LSm1-7 complexes directly bind to the genomic RNA of Brome mosaic virus within P bodies, exerting regulatory effects that activate viral translation or promote viral replication (Galão et al. 2010).

siRNA body and ta-siRNA biogenesis

AGO7, SGS3, RDR6, and DCL4, the key components involved in ta-siRNA biogenesis, have been found to localize within a plant-specific cytoplasmic biomolecular condensate called the siRNA body, suggesting that this biomolecular condensate serves as a center for ta-siRNA biogenesis (Fig. 3) (Kumakura et al. 2009; Jouannet et al. 2012; Kim et al. 2021; Han et al. 2023; Tan et al. 2023). Unlike P bodies, these granules are primarily referred to as SGS3/RDR6 bodies due to the interaction and co-localization of SGS3 and RDR6 within them (Kumakura et al. 2009). Notably, SGS3, which contains a PrLD domain, is believed to play a critical role in the formation of siRNA bodies by LLPS (Kim et al. 2021). While full-length SGS3 can form cytoplasmic foci, SGS3 lacking the PrLD domain is evenly distributed in the nucleus (Kim et al. 2021). Further investigations have demonstrated that SGS3 interacts with RDR6 and drives the formation of siRNA bodies, which is necessary and sufficient for siRNA actions (Han et al. 2023; Tan et al. 2023). Moreover, the phosphorylation statue of SGS3 regulates the phase separation of SGS3 and the formation of SGS3/RDR6 bodies (Han et al. 2023). Furthermore, the phase separation of SGS3 contributes to plant antiviral immunity, and the phosphorylation status of SGS3 affect this immune response (Han et al. 2023). Both SGS3 and RDR6 possess RNA binding activity (Elkashef and Ding 2009; Kunej et al. 2021). Notably, RDR6 prefers to bind aberrant mRNAs without poly(A+) tails (Baeg et al. 2017), suggesting that RNA molecules may also regulate siRNA body formation. Indeed, RNAs targeted by SGS3 have been found to promote the formation of siRNA bodies (Tan et al. 2023).

siRNA bodies play a pivotal role in plant resistance to viruses and are sensitive to viral infections and various stresses. The formation and dynamics of siRNA body can be influenced by multiple viruses (Table 2) (Makinen et al. 2017). For example, the p2 protein of Rice stripe virus interacts with rice SGS3 and reduce the functions of ta-siRNAs (Du et al. 2011). The βC1 protein of Tomato yellow leaf curl China virus upregulates the expression of calmodulin-like protein (NbCaM), an endogenous RNA silencing suppressor, and repress RDR6 expression (Li et al. 2014). Further studies have shown that NbCaM also negatively regulates siRNA body formation by interacting with SGS3 and guiding SGS3 into the autophagy pathway for degradation, thereby altering plant immunity (Li et al. 2017). The TGBp1 protein of Plantago asiatica mosaic virus interacts with SGS3 and RDR6 (Okano et al. 2014), altering the subcellular localization of SGS3 and the synthesis of dsRNA in the ta-siRNA pathway. Moreover, the virus-induced small peptide VISP1 interacts with both SGS3 and autophagy-related protein 8, promoting siRNA body degradation through the autophagy pathway and facilitating viral infection (Tong et al. 2021). The TAV protein of Cauliflower mosaic virus impairs the production of secondary but not primary siRNAs (Shivaprasad et al. 2008). In contrast, although the VPg protein of Potato virus A also interacts with SGS3 within siRNA body, the silencing of SGS3 actually reduces the accumulation of viral RNA (Rajamaki et al. 2014).

Interestingly, proteins associated with RNA quality control, which are predominantly concentrated within P bodies, also play a role in suppressing posttranscriptional gene silencing (Christie et al. 2011; Liu and Chen 2016; Kim 2023). For instance, factors related to both 5′-3′ and 3′-5′ cytoplasmic RNA decay pathways also repress endogenous and transgene posttranscriptional gene silencing (Zhang et al. 2015). Moreover, the key decapping enzyme DCP2 prevents transgene silencing in Arabidopsis (Thran et al. 2012). Notably, mutations in RNA Quality Control factors such as xrn4, dcp2, and vcs activate RDR6-dependent siRNAs targeting endogenous transcripts (Martinez de Alba et al. 2015). This observation raises an intriguing perspective on the interplay between the RNA silencing and RNA decay pathways, suggesting that when the quality control pathway is overwhelmed or compromised, decay substrates may spill into the RDR6/DCL2/DCL4 pathway, leading to sRNA biogenesis and subsequent RNA silencing processes (Christie et al. 2011; Liu and Chen 2016; Kim 2023).

Cajal body and heterochromatin siRNA production

Cajal bodies are a type of nuclear biomolecular condensate involved in the processing and maturation of nuclear small non-coding RNAs, such as small nuclear RNA and small nucleolar RNA in yeast, animals, and plants (Machyna et al. 2013; Taliansky et al. 2023). These specialized organelles play regulatory roles in various biological processes, including transcription, splicing, ribosome biogenesis/rRNA maturation, and telomere maintenance (Love et al. 2017). In humans, more than 50 proteins have been identified to localize in Cajal bodies, while in plants, approximately 13 proteins have been observed to localize in Cajal bodies (Machyna et al. 2013; Love et al. 2017). While the involvement of Cajal bodies in hc-siRNA production in plants remains a subject of debate, several studies suggest their association with this process. RDR2, DCL3, AGO4, NRPD1b/NRPE1 of Pol IV/V, and 24-nt hc-siRNAs tend to be enriched in Cajal bodies (Li et al. 2006) or nucleolar dots that are likely to be Cajal bodies or related entities (Pontes et al. 2006) (Fig. 4). However, it should be noted that the DNA methyltransferase DRM2, which is essential for hc-siRNA function, is not localized in Cajal bodies. This suggests that Cajal bodies likely serve as the sites for hc-siRNA processing and storage rather than action (Li et al. 2006).

Coilin, an evolutionarily conserved protein, contains a typical IDR domain and plays a crucial role in the formation of Cajal bodies in both animals and plants (Collier et al. 2006; Navascues et al. 2008; Liu et al. 2009; Makarov et al. 2013; Courchaine et al. 2022). Additionally, NRPD1b/NRPE1, subunit protein of RNA polymerase IV and V that is involved in hc-siRNA production, also possesses a PrLD domain (Lei et al. 2021). Interestingly, AGO4 has been observed to localize in nuclear compartments other than Cajal bodies, suggesting the involvement of additional biomolecular condensates in hc-siRNA action (Li et al. 2008).

Cajal bodies, like other biomolecular condensates, display dynamic behavior influenced by various factors such as cell type, developmental stage, and life cycle in animals (Platani et al. 2000; Walker et al. 2009). In plants, Cajal bodies have been found to participate in stress response, particularly in plant-virus interactions (Table 2). Numerous studies have demonstrated interactions between viral proteins and components of Cajal bodies, shedding light on the viral manipulation of these organelles (Kim et al. 2007a; Rajamaki and Valkonen 2009; Semashko et al. 2012; Ruiz-Ruiz et al. 2013; Zheng et al. 2018; Shaw et al. 2019; Wang et al. 2020). For example, the 16-K protein of Tobacco rattle virus (TRV) interacts with Coilin, resulting in the delocalization of Coilin from Cajal body and the activation of salicylic acid-mediated resistance (Shaw et al. 2019). Similarly, the ORF3 protein of Groundnut rosette virus associates with fibrillarin (Kim et al. 2007a), another Cajal body component, leading to the reorganization of Cajal bodies and facilitating the systemic movement of the virus (Kim et al. 2007b). The VPg protein of Potato virus A interacts with fibrillarin in both nucleolar and Cajal bodies, promoting virus accumulation (Rajamaki and Valkonen 2009). Additionally, the V2 protein of Tomato yellow leaf curl virus interacts with AGO4 within Cajal bodies, suppressing its antiviral methylation activity (Wang et al. 2020). Furthermore, apart from virus-induced stress, Cajal bodies also participate in response to various abiotic stresses, such as hypoxia, salt stress, osmotic stress, UV irradiation, and heat shock. These roles were extensively discussed in a recently published review by Taliansky et al. (2023), providing comprehensive insights into the multifunctional roles of Cajal bodies in cellular biology.

Conclusion and future perspective

In conclusion, D-bodies, P bodies, siRNA bodies, and Cajal bodies have emerged as essential biomolecular condensates with demonstrated or predicted roles in regulating RNA silencing processes. The functional significance of biomolecular condensates has been extensively discussed in mammalian cells (Banani et al. 2017; Shin and Brangwynne 2017), providing valuable insights into their roles in RNA silencing and various other plant biological processes. These condensates can be categorized into 3 distinct functional roles: reaction crucible, sequestration, and organization hub (Shin and Brangwynne 2017). Biomolecular condensates can enhance cellular reactions by concentrating specific molecules within the condensates, thereby increasing reaction efficiency. This concept is illustrated by the enhancement of miRNA processing efficiency within D-bodies (Xie et al. 2021). Furthermore, biomolecular condensates may function as sequestration factors, aggregating specific components for future utilization. P bodies and stress granules, cytoplasmic condensates linked to translation complex storage, provide examples of this role, particularly during stress responses. These functions have been extensively reviewed by Solis-Miranda and colleagues (Solis-Miranda et al. 2023). Moreover, biomolecular condensates can contribute to spatial organization within cells. Cajal bodies, for example, are specifically associated with various RNA-related cellular processes, including small nuclear RNA and small nucleolar RNA biogenesis (Taliansky et al. 2023). However, whether and how the condensations truly facilitate RNA silencing processes remains to be fully elucidated.

The dynamic nature of biomolecular condensates enables them to rapidly respond to cellular stresses. Their ability to form and dissolve, along with their capacity to concentrate and organize specific components, contribute to the precise modulation of sRNA production and activities. These mechanisms enable cells to finely tune their stress responses and ensure the appropriate regulation of gene expression in diverse biological contexts. Future research is anticipated to unravel the intricate regulatory mechanisms that govern the formation and dynamic behavior of RNA silencing-related biomolecular condensates, and their impact on RNA silencing and stress responses. Investigations will aim to comprehend the mechanisms underlying their formation, dynamic regulation, factors influencing assembly and disassembly, as well as the interplay between different biomolecular condensates and RNA silencing pathways.

Author contributions

X.Z. conceptualized the review outline. Q.L. and Y. L. curated information, prepared the figures, and wrote the original draft. X.Z. reviewed and edited the manuscript.

Funding

This work is supported by the National Key R&D Program of China (2022YFD1400800), National Natural Science Foundation of China (NSFC 32090010 and 32325042), Program of CAS (ZDBS-LY-SM027), and Hainan Yazhou Bay Seed Lab (B21HJ0104).

Dive Curated Terms

The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:

• SGS3 Gramene: AT5G23570

• SGS3 Araport: AT5G23570

• AGO4 Gramene: AT2G27040

• AGO4 Araport: AT2G27040

• AGO2 Gramene: AT1G31280

• AGO2 Araport: AT1G31280

• AGO7 Gramene: AT1G69440

• AGO7 Araport: AT1G69440

• MTA Gramene: AT4G10760

• MTA Araport: AT4G10760

• TGH Gramene: AT5G23080

• TGH Araport: AT5G23080

• TRN1 Gramene: AT2G16950

• TRN1 Araport: AT2G16950

• DCP2 Gramene: AT5G13570

• DCP2 Araport: AT5G13570

• DRM2 Gramene: AT5G14620

• DRM2 Araport: AT5G14620

• VCS Gramene: AT3G13300

• VCS Araport: AT3G13300

• HYL1 Gramene: AT1G09700

• HYL1 Araport: AT1G09700

• DCL1 Gramene: AT1G01040

• DCL1 Araport: AT1G01040

• DCL3 Gramene: AT3G43920

• DCL3 Araport: AT3G43920

• DCL4 Gramene: AT5G20320

• DCL4 Araport: AT5G20320

• PRL1 Gramene: AT4G15900

• PRL1 Araport: AT4G15900

• MAC7 Gramene: AT2G38770

• MAC7 Araport: AT2G38770

• DBR1 Gramene: At4g31770

• DBR1 Araport: At4g31770

• PP4R3A Gramene: AT3G06670

• PP4R3A Araport: AT3G06670