Similar to the formation of message RNA (mRNA), the formation of pri-miRNAs follows the same processes including transcription, capping, 3′ polyadenylation, and splicing. MIR genes are transcribed by DNA-dependent RNA polymerase II (Pol II). Thus, some factors associated with Pol II may also affect MIR transcription (Figure 2A). The mediator complex plays a critical role in recruiting Pol II to the promoters of MIR genes, such as MEDIATOR 20A (MED20A), MEDIATOR 17 (MED17), and MEDIATOR 18 (MED18). In these mutants, the transcription level of pri-miRNAs is suppressed (Chadick and Asturias, 2005; Kim et al., 2011). Tho2/Hpr2 Phenotype 1 (THP1) and Yeast Sac3 Homolog A (SAC3A), two core subunits of the Transcription Coupled Export 2 (TREX-2) complex, interact and colocalize with Pol II to promote MIR gene transcription (Zhang et al., 2020). The C-terminal domain (CTD) of Pol II can be phosphorylated by Cyclin-Dependent Kinase Ds (CDKDs) and Cyclin-Dependent Kinase (CDKF;1), and the phosphorylation of CTD facilitates MIR transcription, 5′-end capping, cotranscriptional RNA processing, and 3′-end polyadenylation (Hajheidari et al., 2012). Histone modification and chromatin remodeling are two important epigenetic modifications for gene activation and/or inhibition, and they are important for the transcription of MIR genes. For example, General Control Non-repressed protein 5 (GCN5), a histone acetyltransferase, targets a number of MIR genes, and it is responsible for the acetylation of H3K14 of these loci (Kim et al., 2009) and others, including SWINGER (SWN), CURLY LEAF (CLF), and PICKLE (PKL), which regulate MIR transcription as well. Different from the above, these three epigenetic factors affect only a few MIRs, for example, MIR156A/C, and accelerate the transition of juvenile to adult phase (Xu et al., 2016). Chromatin Remodeling factor 2 (CHR2), an ATPase subunit of the large Switch/Sucrose Non-Fermentable (SWI/SNF) chromatin-remodeling complex, positively regulates transcription of MIR loci by its chromatin remodeling activity (Wang Z. et al., 2018). Some transcription factors, such as Cell Division Cycle 5 (CDC5) and Cycling DOF Factor 2 (CDF2), also impair MIR transcription. CDC5, an MYB-related protein, positively regulates the occupancy of Pol II at MIR promoters and its activities (Zhang et al., 2013). CDF2, a member of DNA binding with One Finger (DOF) gene family, regulates MIR transcription positively and negatively by recruiting different transcriptional machinery to different MIR promoters (Sun et al., 2015). Apart from transcription factors, other regulators play an important role in MIR transcription, including Negative On TATA less2 (NOT2), ELONGATOR PROTEIN 2/5 (ELP2/5), Suppressor of npr1-1 Constitutive 1 (SNC1), Topless Related 1 (TPR1), SMALL1 (SMA1), Short Valve 1 (STV1), Increased Level of Polyploidy1-1D (ILP1), NTC-Related protein 1 (NTR1), and HASTY (HST). NOT2a/2b, a pair of NOT2_3_5 domain-containing proteins, promotes the transcription of MIR genes via interacting with Pol II (Wang et al., 2013). ELP2/5, which are two elongator proteins, and loss functions of ELP2/5 reduce Pol II occupancy at miRNA loci and pri-miRNA transcription (Fang et al., 2015a). SNC1, a disease-resistant gene, and its transcriptional corepressor TPR1 repress the transcription of MIR genes (Cai et al., 2018). Recent research suggests that STV1 impresses the transcription of MIR by affecting the binding of Pol II to the promoter of MIR genes and also SMA1 (Li et al., 2017c,2018). ILP1/NTR1, which are two conserved disassembly factors of the Intron-Lariat Spliceosome complex, positively regulate miRNA biogenesis by facilitating the transcriptional elongation of MIR genes (Wang et al., 2019a). HST, which is the ortholog of EXPORTIN5, is considered as an exportin of plant miRNAs from the nucleus to the cytoplasm for the last two decades. However, at present, Cambiagno et al. (2021) have suggested that HST plays a role in promoting MIR gene transcription.

After transcription, the pri-miRNAs with a typical stem-loop structure are produced in the nucleus. Then, pri-miRNAs follow two constitutive cuts to produce pre-miRNAs and to release the miRNA/miRNA* duplex. This process is called processing, which occurs in two distinct subnuclear bodies, namely, Dicing-bodies (D-bodies), or SmD3/SmB bodies (Fang and Spector, 2007; Fujioka et al., 2007). DCL1, HYL1, and SE are the three core components of a microprocessor. DCL1, an RNAse III endoribonuclease, is responsible for the cleavage of pri-/pre-miRNAs. The knockout of Arabidopsis DCL1 is embryo lethal (Errampalli et al., 1991). HYL1 and SE are the two cofactors of DCL1, which interact with each other in D-bodies to ensure the accuracy and efficiency of processing (Fang and Spector, 2007; Dong et al., 2008; Liu et al., 2012; Zhu et al., 2013). Increasing studies have been conducted on DCL1, HYL1, and SE because of its importance in processing (Figure 2B).

DCL1 was first identified in pri-miRNAs and pre-miRNAs processing (Kurihara and Watanabe, 2004). Several factors that could regulate the transcription of DCL1, including GCN5, Stabilized 1 (STA1), and XAP5 Circadian Timekeeper (XCT), were identified (Figure 2B); the first one is a negative factor and the last two are positive ones (Kim et al., 2009; Ben Chaabane et al., 2013; Fang et al., 2015b). Otherwise, the abundance of DCL1 is fine-tuned by two feedback regulating levels. On the one hand, the biogenesis of miR162 is dependent on DCL1, but the increase of miR162 suppresses the abundance of DCL1mRNA by target cleavage. On the other hand, the generation of miR838 depends on the processing of DCL1 pre-mRNA, which leads to a decrease in productive mRNA processing of DCL1 (Xie et al., 2003; Rajagopalan et al., 2006). Except for the above, a recent research suggested that SMA1 not only affects MIR transcription but also plays an important role in the correct splicing of DCL1 (Li et al., 2018).

HYL1, an important chaperone of DCL1, encodes a double-stranded RNA-binding protein and plays a critical role in processing accuracy and efficiency. In hyl1 mutant, more pri-miRNAs but less mature miRNAs accumulate (Vazquez et al., 2004). HYL1 protein comprises three representative domains, including two double-stranded RNA-binding domains (dsRBD1 and dsRBD2) in its N-terminal and one protein–protein interaction domain in its C-terminal. Yang et al. (2010, 2014) suggested that HYL1 binds to the double-stranded region of pri-miRNAs in dimer and the homodimer functions as a molecular anchor for correct DCL1 cleavage. In the past few years, several groups approached forward genetic screen of hyl1 mutant for its suppressor. Several dominant DCL1 alleles were identified, and all of them can rescue the phenotypes of hyl1 mutant defects in varying degrees (Tagami et al., 2009; Liu et al., 2012; Gao et al., 2020). Notably, apart from DCL1, Gao et al. (2020) identified new suppressors such as SOP1 (Suppressor of pas2 1), which is a novel factor of hyl1. Further study showed that HYL1 antagonizes nuclear exosome to protect pri-miRNAs from degradation. In recent years, considerable studies have been conducted on the function of HYL1, particularly in D-body formation, nuclear-cytoplasm translocation, and post translational modification (Figure 2B). MOS4-associated Complex 7 (MAC7), a component of the MAC complex, promotes the localization of HYL1 to D-bodies. In mac7 mutant, the number of D-bodies is decreasing, which may decrease the miRNA level (Jia et al., 2017). In addition, considerable research has shown that HYL1 can localize not only in the nucleus but also in the cytoplasm. Karyopherin Enabling the Transport of the Cytoplasmic HYL1 (KETCH 1), a member of the importin β-family, transports HYL1 from the cytoplasm to the nucleus to take part in miRNA biogenesis (Zhang et al., 2017a). In dark conditions, HYL1 suffers from degradation in the cytoplasm, whereas in light conditions, Constitutive Photomorphogenic 1 (COP1), which is a RING-finger E3 ligase, translocates into the cytoplasm from the nucleus and protects the degradation of HYL1 by suppressing the activity of an unknown protease (Cho et al., 2014). Phosphorylation plays an important role in regulating the activity and stability of HYL1 (Figure 2B). The activity and stability of the phosphorylation of HYL1 affect the accurate processing of miRNA. Phosphorylated HYL1 is non-functional. C-terminal domain Phosphatase-Like 1 (CPL1) shares a domain with yeast and human Fcp1 phosphatases (Koiwa et al., 2002) and dephosphorylates HYL1, thereby leading to accurate processing and strand selection (Manavella et al., 2012). Similar to CPL1, a PP4/Suppressor of MEK 1 (SMEK1) complex can also dephosphorylate and stabilize HYL1 (Su et al., 2017). Some kinases, such as Mitogen-Activated Protein Kinase (MAPK3) and SNF1-related Protein Kinase 2 (SnRK2), can phosphorylate HYL1 and waken the position of HYL1 during miRNA processing (Raghuram et al., 2015; Yan et al., 2017). The regulator of CBF gene expression 3 (RCF3), a KH-domain protein, interacting with phosphatases CPL1 and CPL2, mediates the dephosphorylation of HYL1 and then increases its activity in specific tissues (Karlsson et al., 2015).

SE, another cofactor of DCL1, encodes a C₂H₂ zinc finger protein. se mutant exhibits some developmental defects similar to hyl1 mutant. Further analysis suggested that SE plays a critical role in pri-miRNA processing (Yang L. et al., 2006). After years of studies, research on SE has made great progress (Figure 2B). The transcription level of SE is regulated by miR863-3P in a feedback loop manner (Niu et al., 2016). Apart from transcriptional regulation, post translational modification is important for SE. SnRK2 can phosphorylate HYL1 and SE in vitro, but the real function in vivo is largely unknown (Yan et al., 2017). Several SE-interacting proteins have been identified for the past few years. CHR2, a typical chromatin remodeling factor, which can interact with SE to remodel pri-miRNAs’ conformation and impair its processing (Wang Z. et al., 2018). MAC5, a component of MOS4-associated complex (MAC), interacts with SE and protects pri-miRNAs from nuclease degradation (Li et al., 2020). Recent study has found another SE-interacting protein, namely, Serrate-Associated Protein 1 (SEAP1), which positively promotes miRNA biogenesis by modulating pri-miRNA splicing, processing, and/or stability (Li M. et al., 2021). SE also interacts with the RNA helicase HEN2, RNA-binding protein RBM7, and one of the two ZCCHC8A/ZCCHC8B, which are members of the Nuclear Exosome Targeting (NEXT) complex. SE and NEXT complex promote the degradation of pre-miRNA, but unlike SE, the NEXT complex is not necessary for miRNA processing (Bajczyk et al., 2020). SE is not only involved in miRNA biogenesis, but also in regulating the expression of mRNA. First, SE promotes pre-mRNA splicing by interacting with Cap Binding Complex (CBC) (Laubinger et al., 2008). Second, SE plays an unexpected role in promoting the expression of intronless genes via promoting the association of Pol II with direct chromatin binding (Speth et al., 2018). Third, SE is also involved in the expression of TE by promoting H3K27me1 mediated by Trithorax Related Protein 5 and 6 (ATXR5/6) and by suppressing RNA silencing via RNA-dependent RNA polymerase 6 (RDR6) (Ma et al., 2018). Notably, SE also participates in phase separation. SE-mediated phase separation is necessary for the formation of D-bodies, in which accurate and efficient miRNA processing occurs (Xie et al., 2021). A recent study also reported that several RNA helicase RH6/8/12 interact with SE and promote the phase separation and the formation of D-bodies (Li Q. et al., 2021).

A number of factors, which regulate miRNA 3′ end modification such as methylation and uridylation, have been identified (Figure 2C). First, HEN1 catalyzed miRNA/miRNA* and siRNA/siRNA* duplex methylation; although the binding of its two dsRNA-binding domains to the RNA substrate is significant, the contribution is different (Baranauskė et al., 2015). After the action of HEN1, the methyl group is added at 2′ OH of 3′ terminal nucleotide of miRNA (Yu et al., 2005; Yang Z. et al., 2006). To date, HEN1-mediated methylation is considered as the core mechanism of miRNA stability regulation. In 2017, a group from Chinese Taipei suggested that light activated the expression of HEN1. Moreover, the activation depends on a number of photoreceptors and transcription factors HY5. Further analysis found that HY5 is also targeted by miR157d, whose abundance is regulated by HEN1 (Tsai et al., 2014). This project focuses on the transcriptional and post transcriptional regulation of HEN1. Apart from methylation and nuclease media degradation of methylated miRNAs (see blow, SDNs), miRNA also suffered from uridylation. In Arabidopsis, HEN1 Suppressor 1 (HESO1) is the first gene identified for miRNA uridylation (Li et al., 2005; Zhao et al., 2012). HESO1 catalyzes the addition of uracils to the 3′-end of miRNAs in the absence of HEN1. Therefore, HESO1 catalyzes unmethylated miRNAs. During the overexpression of HESO1 in hen1 mutant, the developmental defect was more serious than hen1 mutant. Also, the miRNAs level is less, indicating that HESO1 promotes miRNA degradation (Ren et al., 2012b). The second one is UTP: RNA uridylyltransferase 1 (URT1), a functional paralog of HESO1, which functionally reduant and cooperates with HESO1 (Tu et al., 2015; Wang et al., 2015). HESO1 and URT1 colocalize and interact with AGO1 and uridylate AGO1-bound miRNAs (Ren et al., 2014b; Tu et al., 2015; Wang et al., 2015). Although HESO1 and URT1 share many similarities in uridylation, there are also some differences. For miRNA, HESO1 has been proven to be the major enzyme responsible for uridylation of unmethylated miRNAs and siRNAs, while URT1 mainly uridylates miRNAs sequentially and cooperatively with HESO1 (Tu et al., 2015; Wang et al., 2015). For mRNA, URT1 becomes the main TUTase to cover 70%–80% of mRNAs uridylation; however, HESO1 targets mostly the mRNAs with short tails (Sement et al., 2013). HESO1 and URT1 also exhibit distinct substrate preferences, especially the 3′ end nucleotide of its substrate sRNAs, in detail, HESO1 has the strongest preference for U, while URT1 for A (Tu et al., 2015; Wang et al., 2015).

RNA degradation occurs either from its 5′ or 3′ terminal guide by exoribonucleases or cleavage by endoribonuclease theoretically. Different from mRNA, miRNA primarily includes 3′ to 5′ degradation, according to methylation mediated by HEN1. Small RNA-degrading nucleases (SDNs), a family of DEDDh 3′ to 5′ exonucleases, showed nuclease activity by removing 3′ end methylated nucleotides directly and promoted the degradation of some methylated miRNAs (Ramachandran and Chen, 2008; Yu et al., 2017; Figure 2C). SDN1 is responsible for the degradation of single-stranded small RNAs, which is 17–24 nt long. In addition, it is sensitive to methylated miRNAs but not to the miRNA/miRNA* duplex and long single-stranded RNAs (Ramachandran and Chen, 2008; Yu et al., 2017; Chen et al., 2018). Atrimmer 2 (ATRM2), a DEDDy type exoribonucleases, is involved in the degradation of unmethylated miRNA/miRNA* duplexes during RISC assembly. The loss function of ATRM2 in hen1 mutant partially rescues the developmental defects and the abundance of numerous miRNAs but decreases the expression of the corresponding miRNA targets (Wang X. et al., 2018; Figure 2C).

AGO protein not only serves as an effector, but also plays critical roles in maintaining the stability of miRNA (Figure 2C). As previously mentioned, several degradative factors, including HESO1, URT1, SDN1, and ATRM2, interact with AGO1 and play an important role in AGO1-bound miRNAs or unmethylated miRNA/miRNA* duplexes. Therefore, AGO1 protein can recruit these regulative factors for the degradation of AGO1-associated miRNAs or unmethylated miRNA/miRNA* duplexes (Ramachandran and Chen, 2008; Chen et al., 2018; Wang X. et al., 2018).

The Arabidopsis genome encodes 10 AGO proteins, and it can be grouped into three clades: AGO1/5/10, AGO2/3/7, and AGO4/6/8/9 (Vaucheret, 2008). Different AGO proteins had different functions on their bound miRNAs. AGO1 binds most miRNAs, and the 5′ end of these miRNAs tends to be uridine (Mi et al., 2008). In ago1 mutant, the abundance of miRNAs at a low level suggests that AGO1 can also stabilize miRNA (Vaucheret et al., 2004). AGO7 and AGO10 are primarily associated with miR390 and miR165/166, respectively. In ago7 mutant, the expression level of miR390 is high, but the cause remains unknown (Li et al., 2017a). Over-accumulation of miR165/166 in ago10 mutant is suppressed when AGO10 is overexpressed (Yu et al., 2017). In Oryza sativas, low abundant of miRNAs and phased secondary sRNAs in AGO18 loss function mutants suggested that AGO18 can also stabilize sRNAs including miRNAs (Das et al., 2020). In addition AGO18 can sequester miR168 and prevent it from repressing its target AGO1 essential for antiviral RNAi (Wu et al., 2015). Likely, AGO18 can sequesters miR528 apart from AGO1 to prevent the formation of an effective RISC, as a result, the function of miR528 targeted L-ascorbate oxidase (AO) is then released, following the initiation of ROS-mediated resistance against Rice stripe Tenuivirus (RSV) infection (Wu et al., 2017). In brief, AGO proteins exhibit different functions in miRNA stability control. On the one hand, it protects miRNAs from degradation. On the other hand, it can recruit some degradative factors facilitating the degradation of AGO-bound miRNAs.

miRNAs are associated with AGO proteins to direct post-transcriptional target gene repression by the manner of miRNA-target complementation. miRNA-target complementation can also trigger destabilization of miRNAs, for example, target mimicry, which is initially discovered in plants. Phosphate Starvation1 (IPS1), a non-coding RNA, whose sequence provides a complementary motif of miR399, due to its loosely complementary around cleavage site, as a result, miR399 is sequestered from its endogenous targets by IPS1 binding. Consequently, miR399 is sequestered from its endogenous targets (Franco-Zorrilla et al., 2007). Based on the abovementioned principles, some researchers designed target mimicry technologies artificially, such as MIM and Short Tandem Target Mimic (STTM), which can be used to reduce endogenous miRNA function (Todesco et al., 2010; Yan et al., 2012). In 2018, these two research groups found that the loss function of an F-box gene Hawaiian Skirt (HWS) leads to the recovery of MIM and STTM-induced developmental and molecular defects (Lang et al., 2018; Mei et al., 2019). In hws mutant, miRNA and its mimicry targets coexisted with AGO1 stably, showing that HWS may touch off unoptimizable RISCs degradation (Mei et al., 2019; Figure 2C). RISC interacts with Clearing 3′–5′ exoribonucleases 1/2 (RICE 1/2), which are two members of DnaQ-like exonucleases. Malfunction of RICE1 and RICE2 resulted in reduced miRNA levels. RICE1 and RICE2 interact with AGO1 and AGO10, thereby degrading uridylated 5′ products of miRNA cleavage products to maintain the normal function of RISC (Ren et al., 2014a; Zhang et al., 2017b; Zuber et al., 2018; Figure 2C). Therefore, the stability of RISC-associated miRNAs is impaired by the excessive accumulation of cleaved products.