Exploring the interplay between PARP1 and circRNA biogenesis and function

Abstract

PARP1 (poly-ADP-ribose polymerase 1) is a multidomain protein with a flexible and self-folding structure that allows it to interact with a wide range of biomolecules, including nucleic acids and target proteins. PARP1 interacts with its target molecules either covalently via PARylation or non-covalently through its PAR moieties induced by auto-PARylation. These diverse interactions allow PARP1 to participate in complex regulatory circuits and cellular functions. Although the most studied PARP1-mediated functions are associated with DNA repair and cellular stress response, subsequent discoveries have revealed additional biological functions. Based on these findings, PARP1 is now recognized as a major modulator of gene expression. Several discoveries show that this multifunctional protein has been intimately connected to several steps of mRNA biogenesis, from transcription initiation to mRNA splicing, polyadenylation, export, and translation of mRNA to proteins. Nevertheless, our understanding of PARP1’s involvement in the biogenesis of both coding and non-coding RNA, notably circular RNA (circRNA), remains restricted. In this review, we outline the possible roles of PARP1 in circRNA biogenesis. A full examination of the regulatory roles of PARP1 in nuclear processes with an emphasis on circRNA may reveal new avenues to control dysregulation implicated in the pathogenesis of several diseases such as neurodegenerative disorders and cancers.

Graphical Abstract

Possible ways by which PARP1 and its PARylation activity might regulate circRNA biogenesis and function. PARP1 and PARylation could regulate the various cis and trans mechanisms implicated in circRNA synthesis (pink arc). PARP1 and circRNA may cooperate in fine-tuning their activities and functions (orange arc).

1. INTRODUCTION

ADP (ribosyl)ation is a post-translational modification of biomolecules catalyzed by the ADP-ribosyl transferase family of enzymes, commonly called PARPs. PARP enzymes transpose ADP-ribose units from donor nicotinamide adenine dinucleotide (NAD+) molecules to a range of acceptor proteins, including proteins involved in DNA repair, transcription, histones, and chromatin modulators (Hassa & Hottiger, 2008; Zhu et al., 2023). The covalent attachment of ADP-ribose subunits varies from monomers (MARylation) to polymers (PARylation) (Alemasova & Lavrik, 2019). The addition of PAR or MAR is dynamic and has been implicated in several biological processes. The removal of these PAR chains is catalyzed by the enzyme poly(ADP-ribose) glycohydrolase (PARG). Thus, the dynamic interplay between PARP and PARG enzymes regulates the transient nature of PAR chains within the cell.

The PARP family in human organisms consists of 18 members, each of which is encoded by a distinct gene, yet they all share homology in a conserved catalytic domain (Ame et al., 2004). PARP1 (poly (ADP)-ribose polymerase 1) is the most studied member of this family. Apart from histones, PARP1 is the most abundant, accounting for approximately 90% of the PARylation activity induced by DNA damage, while the remainder is predominantly PARP2 (Zhang et al., 2022). Both PARP1 and PARP2 are not only involved in chromatin binding at sites of DNA lesions (Bilkis et al., 2023), but also in the life cycle of RNA, including RNA splicing by interacting with components of the spliceosome machinery (Isabelle et al., 2010; Jurica et al., 2002). However, as of now only PARP1 has been shown to regulate the elongation kinetics of RNA polymerase II (Matveeva et al., 2022), a critical regulator of circRNAs biogenesis (Zhang et al., 2016).

To fine-tune a variety of physiological processes—most notably DNA damage repair, chromatin dynamics, transcriptional regulation, and cell death signaling—PARP1 regulates a multitude of target proteins, including itself, by PARylation (Ciccarone et al., 2017; Gupte, 2017; Ying & Padanilam, 2016). Although the function of PARP1 in DNA repair has been extensively explored, more recent research has focused on the involvement of PARP1 in the regulation of gene expression. In living cells, gene expression is a dynamic process where RNA is constantly being transcribed and translated. PARP1 seems to be involved in every step of RNA processing—from transcription initiation, elongation, and splicing to RNA processing and degradation (Gibson et al., 2016; Kotova et al., 2009; Matveeva et al., 2016; Matveeva et al., 2022; Matveeva, Mathbout, et al., 2019; Tulin & Spradling, 2003). Therefore, a comprehensive understanding of how PARP1 regulates these steps will yield new insight into its molecular mechanisms of action and provide a foundation for the development of PARP1-targeted therapeutic agents.

Gene expression begins with the synthesis of RNA from the DNA template by the enzyme RNA polymerase (RNAP). Co-transcriptional splicing of nascent RNA occurs when pre-mRNA is still tethered to the chromatin during transcription. We recently showed that PARP1 interacts with a component of the elongation complex (Matveeva et al., 2022) to modulate alternative splicing decisions (Matveeva et al., 2016). While most studies focus on the canonical splicing, another form of splicing, back-splicing, produces circular RNAs (circRNAs). While back-spliced events are formed by the canonical spliceosome machinery (Ashwal-Fluss et al., 2014; Starke et al., 2015; Wang & Wang, 2015), their production is considerably less favorable compared to their linear counterparts. The inefficiency of back-splicing led to the belief that cis- and trans-acting factors within flanking intronic regions are modulators of this process, adding another cell- and context-dependent regulatory layer to circRNA expression (Conn et al., 2015; Huang & Zhu, 2021; Ivanov et al., 2015; Kristensen et al., 2019). While trans-acting factors can include RNA-binding proteins (RBPs), spliceosome factors, RNA helicases, and cleavage factors, cis-acting elements include reverse complementary (repetitive and non-repetitive) sequences within the transcript itself. The spatial presence and timely synthesis of these regulatory factors may influence the likelihood of back-splicing events to occur (Conn et al., 2015; Huang & Zhu, 2021; Ivanov et al., 2015; Kristensen et al., 2019).

CircRNAs were initially considered to be non-functional. Recent studies, however, show that they are associated with diverse biological functions and their expression is often dysregulated in human diseases. In some cases, circRNAs can sponge microRNAs and RNA-binding proteins for gene regulation (A. Huang et al., 2020; X. Zhang et al., 2019). In other cases, circRNAs are even translated into functional proteins (Chen et al., 2023; Legnini et al., 2017; Pamudurti et al., 2017). In addition, some circRNAs directly interact with the genomic DNA of their host gene, resulting in altered gene expression (Conn et al., 2017). How and why circRNAs are produced is now a hot topic in research. CircRNAs are relatively less abundant but more stable than linear mRNAs due to their closed loop structure, which makes them more resistant to degradation by exonucleases (Chuang et al., 2023; Conn et al., 2017). They are also cell- and tissue-type specific and have been found in circulating blood (Conn et al., 2017; Wen & Gu, 2022). Due to these characteristics, they may serve as potential biomarkers for various diseases (Arnaiz et al., 2019; Cao et al., 2022; Wen et al., 2020). Our current knowledge of factors regulating circRNA biogenesis is very much in its infancy. In this review, we aim to elucidate the multiple ways by which PARP1 might impact the biogenesis of circRNAs and uncover any potential interplay between the newly discovered circRNA functions and PAPR1.

2. STRUCTURAL DOMAINS OF PARP1

The 18 human PARPs all share a highly conserved catalytic domain containing an ADP-ribosyl transferase (ART) fold, making the discovery of specific inhibitors to the catalytic pocket differentiating PARP members particularly challenging (Rose et al., 2020; Rudolph et al., 2022). Nevertheless, PARP members are distinguished by a variety of additional domains that determine the specificity of individual member’s cellular and biochemical functions. The most studied of these proteins is PARP1, responsible for most of the PARylation activity in human cells (Beck et al., 2014; Huber et al., 2004). The Drosophila genome encodes only one PARP gene and a homolog of Tankyrase, making it a simpler model for PARP biology studies. PARP1 has three functional domains (Kouyama et al., 2019) (Figure 1): (1) The N-terminal DNA-binding domain has three zinc-fingers (Zn1, Zn2, and Zn3), a nuclear localization signal (NLS) which comprises cleavage sites recognized by caspase-3 and caspase-7. (2) A central BRCT (BRCA1 C-terminus) fold-domain that bears the major sites of auto-modification and contains a considerable number of glutamate and aspartate residues. (3) The C-terminal catalytic region, which is the most conserved domain across the PARP family. This domain is comprised of a tryptophan-, glycine-, and arginine-rich (WGR) domain and the PARP signature sequence necessary for the catalysis of PAR synthesis (Krishnakumar et al., 2008).

Figure 1:

PARP1 is composed of multiple structural domains: 1) DNA-binding domain contains three zinc fingers, a nuclear localization signal (NLS) between Zn2 and Zn3, and cleavage site recognized by a caspase-3 and caspase-7; 2) A central fold domain containing the auto-modification sites and BRCA1 C-terminus motif; and 3) A C-terminal catalytic domain containing a tryptophan-, glycine-, and arginine-rich (WGR) domain and the PARP signature sequence.

These domains of PARP1 provide its functional diversity. The Zn1, Zn3, and WGR domains are essential for activation of PARP1 at sites of DNA damage. Upon association with a DNA break, PARP1 folds into an elegant structure that promotes extensive interdomain contact to facilitate the catalytic activity of PARP1, linking its catalytic domain (CAT) to the interface of DNA damage (Langelier et al., 2012). The first two zinc fingers of PARP1 (Zn1 and Zn2) are critical for DNA binding yet serve separate purposes in the recognition of DNA breaks and enzyme activation. While both are required to stimulate the activation of PARP1 in response to DNA single-strand breaks, only Zn1 is required for activation of PARP1 in the presence of double-stranded DNA breaks (Langelier et al., 2012). The third zinc-finger (Zn3) of PARP1, though structurally and functionally distinct from Zn1 and Zn2, is required for DNA-dependent PARP1 catalytic activity and chromatin compaction (Langelier et al., 2011). Between Zn1 and Zn2 is the nuclear localization sequence (NLS) critical for PARP1’s translocation to the nucleus. Within the NLS is a caspase-3 and caspase-7 cleavage site that is important for the PARP1-mediated apoptosis and parthanatos pathways (Liu et al., 2022; Maluchenko et al., 2021). The BRCT domain was recently discovered to bind DNA, contributing to the ‘monkey-bar’ mechanism, to mediate intra-strand DNA transfer without inducing the enzymatic activity of PARP1 (Rudolph et al., 2021). The catalytic domain (CAT) is composed of two subdomains, the helical domain (HD), which consists of six α-helices with connecting linkers of variable sizes (Langelier & Pascal, 2013), and the conserved ADP-ribosyl transferase domain (ART) (Otto et al., 2005). The active site of the CAT domain is responsible for initiation of ADP-ribose attachment to target proteins, elongation of the polymer, and introduction of branches (Reviewed in Alemasova and Lavrik (2019)).

Though PARP1 is a well-known DNA/chromatin binding protein, we recently showed that it also binds RNA (Matveeva et al., 2016; Matveeva, Al-Tinawi, et al., 2019; Matveeva, Mathbout, et al., 2019; Melikishvili et al., 2017). We further showed that PARP1 binds mRNAs, microRNAs (miRNA), small nucleolar RNAs (snoRNA), and long intergenic non-coding RNAs (lincRNAs). These findings were corroborated by Kim et al. (2019), who showed that PARP1 binds snoRNAs in the absence of DNA repair, stimulating PARP1’s catalytic activity in cancer cells. A recent study further demonstrated that the binding of PARP1 to snoRNAs both in vitro and in vivo stimulated its catalytic activity (D. Huang, D. S. Kim, et al., 2020). Not only is its activity stimulated by RNA binding but PARP1 was shown to reversibly in the mono-ADP-ribosylation RNAs (Munnur et al., 2019). Despite these findings, stimulation of PARP1’s activity by RNA is still a subject of debate (Nakamoto et al., 2019). We showed that PARP1 binds several types of RNAs, but due to the novelty of circRNA detection and the limitations of bioinformatic tools for their discovery, we did not probe for PARP1-circRNA interactome. However, a recent study showed that circTFDP2 binds directly to the DNA-binding domain of PARP1, and affecting PARP1’s activity (Lifeng Ding et al., 2023).

It is still not clear which regions of PARP1 are critical for its RNA binding. In competition assays, we showed that though PARP1 preferentially binds DNA, deletion of Zn1 and Zn2 DNA binding domains shifted PARP1 from a DNA binder to an RNA binder (Melikishvili et al., 2017). A recent finding however, showed that circTFDP2 binds PARP1 at the DNA binding domain (Lifeng Ding et al., 2023). These results suggest that there could be multiple sites on PARP1 important for PARP1-RNA and PARP1-circRNA. Indeed, deletion mutants confirmed that multiple regions of PARP1 are needed for its stable binding to RNA (Melikishvili et al., 2017). Collectively, these studies identify PARP1 as an RNA binding protein, yet a detailed mechanistic insight into the functions of PARP1 in RNA regulation is still unclear. Future studies will be critical to understanding the contributions of PARP1’s different domains in RNA binding and its functional implications. Just as with DNA, it is still not clear what recruits PARP1 to a particular location for gene regulation. Several studies have implicated sequence context. For instance, at transcriptional start sites of actively transcribing genes, PARP1 accumulates at GGAAAGG-rich sequences (Lodhi et al., 2014). On the other hand, Vidakovic et al. (2009) showed that PARP1 binds AGGCC sequences at the promoter region of PARP1 gene. We and others have shown that PARP1 associates with GC-rich nucleosomes (Matveeva et al., 2016) and G-quadruplexes (Soldatenkov et al., 2008). Interestingly, this sequence-mediated binding extended to RNA sequence. We showed that PARP1 preferred GC-rich sequences in RNA (Melikishvili et al., 2017). Additionally, J. Ke et al. (2021), using RNA-protein immunoprecipitation followed by sequencing, discovered that the most PARP1-bound sequences in HK-2 cells were GGUAAG and UUUGGG, while in MCF7 cells PARP1 binds mostly CCCCCCC and UGAA regions, suggesting that PARP1-RNA binding is cell-dependent. Thus, it is possible that PARP1 does not recognize a particular sequence, but a structure formed by these sequences. This idea is in line with previous findings of PARP1 binding to four-way junctions in DNA repair as well as to the entry/exit of the nucleosome (Clark et al., 2012; Lonskaya et al., 2005). Besides, PARP1 not only binds to RNA, potentially affecting its function, but it is also involved in the mechanism of RNA life cycle.

3. PARP1, RNAPII, AND REGULATION OF TRANSCRIPTION IN CIRCRNA BIOGENESIS

Chromatin structure plays a crucial role in the movement of RNAPII along the gene body, thus controlling transcription initiation, elongation, and termination. Since circRNAs are formed during the process of transcription, therefore chromatin structure will impact circRNA biogenesis. Numerous studies have shown that circRNA biogenesis can be mediated by an interplay between the speed of RNAPII and gene architecture (Li et al., 2018; Muniz et al., 2021), both of which are modulated by PARP1 (Figure 2). In this section, we discuss how PARP1 may influence chromatin structure to mediate RNAPII movement. Next, we will discuss how PARP1, by binding to chromatin structure, might recruit gene regulatory factors at the initiation, elongation, splicing, and polyadenylation stages, and how this might influence the production of circRNA. In Section 4, we will discuss in greater detail how PARP1 coordinates the maintenance of gene architecture to mediate circRNA synthesis.

Figure 2:

According to the kinetic model, the rate of RNAPII elongation influences transcriptional outcome, and as a result, the ratio of circRNA to linear RNA transcripts. However, this is not the only hypothesized model that can fine-tune circRNA formation, as gene architecture, which depends on other factors such as the nature of DNA sequence and the availability of RNA binding proteins, can play a crucial role in transcript circularization.

3.1. PARP1’s regulation of chromatin structure in transcription

Chromatin undergoes dynamic modifications to allow RNAPII and transcription factors to access the condensed genomic DNA for most nucleic acid-related processes. Factors that modulate the structure of chromatin to mediate transcription initiation and RNAPII elongation impact co-transcriptional splicing (de la Mata et al., 2003; X. Li et al., 2017; Z. Liu et al., 2021; Zhao et al., 2020). Since circRNAs are another form of alternative splicing, it is therefore assumed that these same factors would regulate its biogenesis. We and others have shown that PARP1 is one of such factors that regulate chromatin structure with consequences in transcription initiation (Páhi et al., 2020), RNAPII elongation (Leutert et al., 2016), and co-transcriptional splicing (Matveeva et al., 2016; Matveeva, Al-Tinawi, et al., 2019). Therefore, we discuss PARP1’s role in chromatin structural dynamics and how it might impact circRNA biogenesis.

Eukaryotic DNA is found as chromatin, with the nucleosome as the basic repeating unit. Each nucleosome consists of 146bp of DNA wrapped around two copies each of histone H2A, H2B, H3, and H4, reducing the accessibility of the wrapped DNA to regulatory factors. Chromatin is separated into the less dense euchromatin and highly dense heterochromatin. Euchromatin is histone H1-depleted, more accessible to regulatory factors, with active transcription. Heterochromatin, on the other hand, is less accessible and histone H1-rich with repressed transcription (Fan et al., 2003). PARP1 competes with histone H1, and when deposited, alleviates the transcriptionally repressed chromatin state to a more transcriptionally active chromatin (Kim et al., 2004; Krishnakumar & Kraus, 2010; Pascal & Ellenberger, 2015; Posavec Marjanovic et al., 2017). PARP1 competition with H1 drives differentiation of progenitor stem cells into neurons (Azad et al., 2018). For differentiation to occur, doublecortin (Dcx), a marker for de novo generation of neurons, needs to be expressed. Before differentiation, the proximal enhancer of Dcx is repressed by histone H1-rich chromatin fiber, keeping Dcx expression in a repressed state. Once differentiation is induced, pre-B cell leukemia homeobox 1 protein (PBX1) associates with myeloid ecotropic viral integration site (MEIS) and recruits PARP1 to the Dcx enhancer (Hau et al., 2017). PARP1 PARylates itself and histone H1, ejecting histone H1 from chromatin to activate gene expression and ultimately differentiation into neurons (Azad et al., 2018; Hau et al., 2017). Although this is gene-specific, it has been demonstrated that H1 and PARP1 compete on a genome-wide scale (Figure 3a), affecting gene expression patterns (Kaiser et al., 2020; Krishnakumar et al., 2008). In addition to histone H1, PARP1 also PARylates core histones, resulting in nucleosome destabilization, increased DNA accessibility (Martinez-Zamudio & Ha, 2012), and increased transcription (Kotova et al., 2022). While these studies investigated the functional role of histone PARylation on gene expression, others have implicated the direct role of PARP1 and histones on chromatin structural dynamics. For instance, phosphorylation of the C-terminal domain of Drosophila histone H2Av led to the exposure of leucine 23 and valine 61 on histone H4, resulting in the binding of PARP1 to chromatin (Thomas et al., 2014) and in long-term stimulation of PARP1’s enzymatic activity (Kotova et al., 2011). The increased activity modulates changes in histone posttranslational modification (PTM) code to control gene regulation (D. Huang, C. V. Camacho, et al., 2020; Thomas et al., 2019). In addition, PARP1’s activation has been found to facilitate the exchange of the repressive histone variant H2A.Z by canonical H2A at the c-fos promoter to activate its expression (O’Donnell et al., 2013). Another study has shown that PARP1 cooperates with the histone variant macroH2A1.1 to enhance the acetylation of histone H2B at lysine residues 12 and 120, leading to activation of macroH2A1 target genes (Chen et al., 2014).

Figure 3:

PARP1 can alter nucleosome compaction and accessibility to DNA to organize chromatin structure with implication in gene regulation. a) PARP1 not only competes with H1 for DNA binding, but also directly PARylates H1, leading to its release from chromatin. b) Through its PARylation activity, PARP1 can interfere with KDM5B binding ability to chromatin, preventing the demethylation of histone H3K4me3 and sustaining an active gene expression. c) CTCF induces PARP1 to auto-modify itself and to PARylate CTCF. Since DNMT1 has a higher affinity for PAR polymers, it preferably binds to PARP1 in a non-covalent manner, reducing its affinity for DNA binding and methylation activity. d) PARylation stimulates TET1’s enzymatic activity. When activated, TET1 controls transcription by promoting local chromatin decompaction through the hypomethylation of active gene sites.

While a complete picture of how PARP1 is directed to specific genomic locations to assist in transcriptional regulation remains unclear, genome-wide profiling of PARP1 occupancy has revealed PARP1 chromatin signatures. These studies show that PARP1 occupies gene regions that carry specific histone modifications and/or chromatin regulators (Lodhi et al., 2014; Nalabothula et al., 2015) to impact gene regulation. For instance, PARP1 associates with and PARylates the H3K4me3 demethylase, lysine demethylase 5B enzyme (KDM5B), resulting in its inhibition to bind chromatin (Krishnakumar & Kraus, 2010). The inability of KDM5B to bind prevents H3K4me3 demethylation, thus maintaining an active gene expression state (Figure 3b). On the other hand, at sites of DNA lesions, PARP1 auto-PARylation is essential for the recruitment and function of KDM5A to drive transcriptional activation of genes crucial for repair activities (Kumbhar et al., 2021), suggesting that PARP1’s effect on chromatin structure is context dependent.

While we have discussed PARP1’s effect at the 1D chromatin structure, PARP1 could also mediate 3D chromatin structural dynamics via regulation of factors that govern the 3D genome structure. CTCF is called the ‘master weaver of the genome’ in creating a 3D chromatin conformation critical for gene regulation (Dehingia et al., 2022; Kim et al., 2015; Zhang et al., 2021). PARP1 was shown to PARylate CTCF and stabilize CTCF’s binding to its target sites for active transcription (Lupey-Green et al., 2018; Pavlaki et al., 2018; Zhao et al., 2015). On the other hand, CTCF also regulates PARP1 activity, causing auto-PARylation of PARP1 (Guastafierro et al., 2008) for gene activation. Thus, the interplay of PARP1 and CTCF impacts the 3D chromatin structure. Additionally, PARP1 PARylates DNA-methyltransferase 1 (DNMT1) (Zampieri et al., 2012), inhibiting its binding to DNA, resulting in hypomethylation (Figure 3c) (Reale et al., 2005). DNA methylation is removed by the ten eleven translocation (TET) enzymes, which catalyze the conversion of 5-methylcytosine (5mC) into 5-hydroxymethylcytosine (5hmC), to activate gene transcription (W. Liu et al., 2021; Tahiliani et al., 2009). PARylation of TET1 increases its enzymatic activity with consequences in decreased methylation levels (Ciccarone et al., 2015) (Figure 3d). We showed genome-wide, an inverse correlation of PARP1 occupancy and DNA methylation (Nalabothula et al., 2015). Since DNA methylation plays a crucial role in circRNA biogenesis (Li et al., 2018; Xu et al., 2021), regulation of the DNA state via PARP1 therefore will be important in this process (also reviewed in Zong et al. (2022) and Eleazer and Fondufe-Mittendorf (2020)).

3.2. PARP1’s regulation of RNAPII movement

Chromatin plays a role in transcription initiation, co-transcriptional splicing, and export of the RNA. Since circRNAs are formed as part of RNA transcription and processing (Li et al., 2018), it is logical to assume that chromatin state mediated by PARP1 would regulate the movement of RNAPII from pause-release at the initiation state to pause-release at the elongation state to final release at the export state. Though not mutually exclusive from its chromatin regulation, PARP1 might recruit specific regulatory factors at each stage, possibly PARylating them to modify their transcriptional activity which will be critical in circRNA biogenesis. We discuss these below.

3.2.1. First stage: Transcription initiation

CircRNAs are frequently expressed from highly transcribed genes, implying that transcription initiation may be a critical step for their synthesis (Salzman et al., 2013). It has been determined through years of research that PARP1 is a crucial modulator of transcription initiation. PARP1 accumulates at the transcription start site (TSS) of genes, where it PARylates histones and itself, resulting in the opening of chromatin and allowing for the passage of RNAPII and its associated elongation factors (Martinez-Zamudio & Ha, 2012; Muthurajan et al., 2014; Nalabothula et al., 2015; Schiewer & Knudsen, 2014). Immediately after RNAPII starts elongation, it pauses at the promoter proximal nucleosome (25–35 nt) to ensure transcriptional fidelity and the proper recruitment of RNA processing components (Adelman & Lis, 2012; Akhtar et al., 2019; Henriques et al., 2013). This pausing is mediated by DRB sensitivity-inducing factor (DSIF) and negative elongation factor (NELF) (Vos et al., 2018). To escape this checkpoint and enter a productive elongation, the positive elongation factor (pTEFb) is recruited to phosphorylate DSIF, causing NELF to dissociate from RNAPII. PARP1 is critical in this process (Ji & Tulin, 2009; Matveeva et al., 2016; Matveeva, Al-Tinawi, et al., 2019; Petesch & Lis, 2012), as PARylation of NELF-E results in CDK9/pTEFb-mediated NELF-E phosphorylation, thus increasing the efficiency of pTEFb-mediated RNAPII pause-release into productive elongation (Gibson et al., 2016). In a feedback mechanism, phosphorylation of NELF-E promotes PARP1 auto-PARylation. As PARylated PARP1 and NELF-E have low affinity for RNA, they are subsequently released, allowing RNAPII to continue with elongation (Gibson et al., 2016; Vispe et al., 2000) (Figure 4).

Figure 4:

PARP1 can directly influence the rate of RNAPII at the promoter and within the gene body for gene regulation.

3.2.2. Elongation stage: PARP1 modulates the kinetics of RNAPII elongation

Once elongation resumes along the gene body, specific epigenetic mechanisms also control transcriptional fidelity, whereby RNAPII pauses at intron–exon and exon–intron boundaries. Aside from controlling the promoter proximal pausing, we have recently shown that PARP1 directly binds to subunit 3 of the integrator complex (INTS3) to regulate the processivity of RNAPII along the gene body, possibly by aiding in the recruitment and assembly of elongation factors (Matveeva et al., 2022). Such mechanisms aid in co-transcriptional splicing. We recently showed that PARP1-chromatin state regulates pausing at these boundaries to enact co-transcriptional splicing decisions (Eleazer et al., 2023; Gibson et al., 2016; Matveeva, Al-Tinawi, et al., 2019). Recent studies have shown that circRNA expression is mediated not only post-transcriptionally (Zhang et al., 2016), but also co-transcriptionally and that splicing is the primary regulatory step in transcript circularization (Ashwal-Fluss et al., 2014). It therefore seems that circRNA biogenesis might be not only mediated by PARP1 at the transcriptional start site, but also during co-transcriptional splicing. The speed of RNAPII is a key player in controlling not only transcription initiation but also co-transcriptional splicing. In this regard, RNAPII also controls the type and levels of transcripts (spliced variants such as mature mRNAs and circRNAs) emanating from a particular gene. This correlates with the suggested kinetic model of circRNA biogenesis (Figure 2). In line with this idea, depletion of core spliceosome components using RNAi or attenuation of transcription termination by depletion or inhibition of polyadenylation machinery resulted in increased circRNA-to-linear ratio of several transcripts (Liang et al., 2017). Furthermore, we have recently demonstrated that the knockdown of PARP1 resulted in differential RNAPII pausing events in exons compared to introns. In detail, when exonic pausing is accompanied by an increase in intronic pausing, back-splicing events are more likely to occur, increasing the production of circRNAs and thereby the circRNA-to-linear mRNA ratio. However, when exonic pausing is coupled with decreased intronic pausing, linear RNA is favored (Eleazer et al., 2023). These results suggest that PARP1 is dynamically regulating RNAPII elongation in exons and introns, mediating the transcriptional output from a particular gene. Therefore, factors that regulate the rate of RNAPII elongation should impact circRNA biogenesis. We discuss below possible steps in RNAPII kinetics modulated by PARP1 and how these might impact the types of transcripts formed (circRNAs vs. linear mRNAs).

During early transcription, RNAPII transiently pauses at the promoter to ensure the proper recruitment of RNA processing components as we discussed above. The rate of nascent RNA processing coordinates with transcription elongation by recruiting processing factors to the transcription elongation complex (Bentley, 2014; Godoy Herz & Kornblihtt, 2019). This process therefore impacts the transcriptional outcome (mature linear mRNAs and circRNAs) from a particular gene (Starke et al., 2015; Wang & Wang, 2015). Recent studies show that RNAPII elongation kinetics positively correlate with back-splicing events (Ragan et al., 2019; Zhang et al., 2016) (also reviewed in Muniz et al. (2021)). Since PARP1 is involved in RNAPII pause-release at the proximal promoter region (Gibson et al., 2016) and along the gene body (Matveeva, Al-Tinawi, et al., 2019), it comes as no surprise that PARP1 might be involved in the production of circRNAs.

3.2.3. Termination stage: PARP1’s regulation of polyadenylation

Finally, as RNAPII progresses through to the 3’ end of the gene, polyadenylation of the transcript by polyadenylation polymerase (PAP) is critical for its maturation and export. PARP1 is known to PARylate PAP, hindering polyadenylation during thermal stress (Di Giammartino et al., 2013). Interestingly, given that circRNAs are distinct from other RNAs in that they are devoid of the 5’ cap and poly(A) tail and have a covalent closed loop structure, disruption of the polyadenylation process favors the production of circRNAs (Holdt et al., 2018; Liang et al., 2017).

Nascent RNA transcripts generated by RNAPII undergo maturation through endolytic cleavage of the 3′-untranslated region (UTR) and polyadenylation through the activity of PAP. This process is highly regulated and is essential for the stability, translational efficiency, and nuclear export of the mature mRNA. Alternative polyadenylation coupled with alternative splicing is known to influence the incorporation of terminal exons, thus increasing transcript and protein diversity. Surprisingly, recent studies showed that depletion of the polyadenylation machinery increased the level of circRNAs produced compared to the level of mRNAs (Liang et al., 2017). These studies suggest that changes in the regulation of terminal cleavage and polyadenylation are important in determining the splicing outputs and circRNAs generation. We therefore speculate that factors modulating cleavage and polyadenylation can influence the ratio of circRNAs-to-linear mRNA production. PARP1 is one of such factors that was shown to regulate the assembly and function of the polyadenylation complex (Shi et al., 2009). PARP1 PARylates PAP and represses its polyadenylation activity (Di Giammartino et al., 2013; Kim et al., 2020). Thus, in so doing, PARP1 might enact a change in the levels of circRNAs-to-mRNA production, which can have a fundamental impact on the cell. However, more studies are needed to investigate PARP1’s co-regulatory role within the cleavage/polyadenylation complex in the context of circRNA biogenesis.

While regulation of RNAPII kinetics is critical for the synthesis of circRNA, this could act in concert with another key regulator, such as the architecture of the gene. Next, we address how PARP1 might work together with gene architecture to enable circRNA production.

4. Interplay of PARP1 binding, gene architecture and CIRCRNA biogenesis

Features of the gene architecture, such as the composition of its pre-mature RNA sequence (exon and intron lengths, RBP motifs, inverted repeats, and GC-content) as well as its secondary structure (G-quadruplex), along with epigenetic marks, influence back-splicing decisions (Caba et al., 2021; Razpotnik et al., 2021; Wang et al., 2020; Zhang et al., 2014). By modulating the features of gene architecture, PARP1 can control the expression of circRNAs. Our recent study shows that in the presence of PARP1, the circRNAs generated have long and symmetrical flanking introns, while those in PARP1 knockdown were asymmetrical with longer upstream introns and shorter downstream introns, with high RNAPII pausing at the downstream introns. Thus, we hypothesized that PARP1 influences the differential movement of RNAPII along the gene body resulting in different circRNAs being made. It is possible that the length of these sequences increases the likelihood for sequence biases that might be critical for circRNA biogenesis, as these sequences include RNAP/splicing motifs, repetitive elements, etc. We showed that in the presence of PARP1, satellite repeats were slightly enriched in the downstream introns, while LINEs were more enriched in upstream introns. However, in the absence of PARP1, these features were reversed; namely satellite repeats became more enriched in upstream introns, while LINEs were enriched in the downstream introns. In general, PARP1 depletion resulted in circRNAs emanating from host genes depleted of rolling circle transposons and simple repeats but more enriched with satellite repeats and LINE1 with a bias of enrichment in downstream introns. These results indicate that PARP1 might be acting with different gene features to govern circRNA expression.

Other cis-regulatory features that have been implicated in circRNA biogenesis include GC content and G-quadruplexes (Gruhl et al., 2021; Huang et al., 2017), both features that have been implicated in PARP1 regulation of alternative splicing (Georgakopoulos-Soares et al., 2022; Zhang et al., 2011). We showed that circRNAs formed in the presence or absence of PARP1 showed no difference in GC content. However, G-quadruplexes were depleted in host genes of circRNA producing genes and this depletion was exaggerated in PARP1 knockdown conditions. Lastly, differential RNAP binding motifs were seen in the presence and absence of PARP1, suggesting the possibility of PARP1 recruiting different RBPs to mediate circRNA biogenesis (Eleazer et al., 2023). In general, PARP1 seems to act together with cis-regulatory elements of the genome to mediate the production of different types of circRNAs. Next, we discuss how PARP1 can influence the synthesis of various forms of circRNA.

4.1. Possible roles of PARP1 in the proposed models of circular RNA formation

The recent advancements in next-generation sequencing technology have led to the discovery of various subtypes of circRNA. The four main subtypes are exonic circRNAs (ecircRNAs), derived primarily from single or multiple exons; circular intronic RNAs (ciRNAs), consisting solely of introns; exonic-intronic circRNAs (EIciRNAs), comprising both exons and introns; and tRNA intronic circRNAs (tricRNAs), which originate from splicing pre-tRNA introns. Currently, the majority of identified circRNAs belong to the exonic circRNA subtype. To date, several models have been proposed on the mechanisms by which circRNAs are generated (Figure 5). These models include a) lariat-driven circularization, b) intron pairing-driven circularization, and c) RNA binding protein (RBP)-driven circularization.

Figure 5:

Proposed models for circRNAs formation mechanisms: a) lariat-driven circularization following a covalent joining of the 3’ splice donor and the 5’ splice acceptor sites of precursor mRNA; b) intronic base pairing-driven circularization; c) RNA binding protein (RBP)-driven circularization where RBPs bridge intronic regions closer together; and d) intron cyclization, which depends on conserved RNA motifs.

4.1. Lariat-driven circularization model

Exonic circRNAs can be generated by both lariat-driven circularization and back-splicing mechanism. Here, the spliceosome-mediated splicing joins the 3′ hydroxyl end acceptor of the upstream exon with the 5′ phosphate end donor of the downstream exon to form a covalent linkage, producing a lariat precursor model. This intermediate lariat subsequently undergoes internal splicing, whereby the introns between the exons are removed to form exonic circular RNAs (ecircRNAs) (Barrett et al., 2015; Jeck et al., 2013). Therefore, circRNA biogenesis relies also on the spliceosome machinery, and it is increasingly becoming clear that back-splicing events leading to circRNA production competes with the canonical splicing. Indeed, Liang and Wilusz showed that depletion of the SF3A/B core-spliceosome complexes increased the efficiency of back-splicing and circRNAs over the linear mRNA production (Liang et al., 2017; Liang & Wilusz, 2014). This raises the question of how PARP1 plays a role in this process. We showed earlier that PARP1 associates with the SF3B1, a spliceosome subunit of the U2 snRNA complex, to possibly mediate splicing decisions (Matveeva et al., 2016). Additionally, PARP1 is known to PARylate and activate splicing factors (Ji & Tulin, 2009, 2013). In so doing, PARP1 might impact not only alternative splicing but back-splicing as well. Aside from splicing factors bringing the upstream splice acceptor and downstream donor in proximity during splicing, gene architecture may facilitate this process. In support of this idea, synthesis of long downstream introns favors back-splicing by increasing the elongation time of RNAPII along the gene body, making a downstream acceptor/linear splicing partner unavailable in comparison to an upstream donor/back-splicing partner (Santos-Rodriguez et al., 2021; Zhang et al., 2014). In support of PARP1’s role for this process, we recently discovered that depletion of PARP1 resulted in the accumulation of RNAPII downstream of the back splice donor site, possibly extending the time required for the linear acceptor to be produced to facilitate back-splicing (Eleazer et al., 2023). While our research is beginning to elucidate the role of PARP1 in circRNA production, more studies are needed to investigate all conceivable pathways by which PARP1 mediates the lariat-driven circularization model.

4.2. Intron pairing-driven circularization model

In this model, intronic complementary sequences, acting in cis, facilitate the circularization of gene transcripts (Cao, 2021; X. Li et al., 2017; Yu & Kuo, 2019). These complementary sequences flank the introns, allowing for RNA base pairing generating either EIciRNAs or ecircRNAs. These sequences could be non-repetitive (Zhang et al., 2014) or repetitive elements such as the transposable Alu element (Chen & Carmichael, 2008; Gruhl et al., 2021). Alu sequences pair with proximal and reversely oriented repeats to form double-stranded RNAs, correlating with circRNA formation from the human ZKSCAN1 gene (Liang & Wilusz, 2014). Supporting this finding, the insertion of Alu sequences into introns changes the gene output from a linear mRNA to a circular RNA (or vice versa) (Wilusz, 2015). Recent studies have also implicated and identified epigenetic/chromatin features that, together with repetitive elements, promote circRNA formation (Cardamone et al., 2023; Zhang et al., 2020). For instance, Su et al. (2014) showed genome-wide that Alu elements are usually bound by two well-phased nucleosomes containing the active histone marks H3K4me2 and H3K4me1 (Barski et al., 2007) to the exclusion of repressive histone marks (J.-B. Chen et al., 2018; Su et al., 2014). Another study using predictive models found H3k36me3, H3k79me2, and H4k20me1 correlated with low circRNA expression (Zhang et al., 2020). How would PARP1, therefore, impact histone PTM deposition to regulate circRNA expression? PARylation of KDM5B, the H3Kme3 demethylase, inhibits its chromatin binding, thus preventing its ability to demethylate H3K4me3 (Krishnakumar & Kraus, 2010). In addition, monoubiquitylation of histone H2B is coupled with RNAPII elongation (Pavri et al., 2006; Zhang & Yu, 2011) and promotes transcription by either facilitating histone H3K4 and H3K79 methylation (Kim et al., 2009; Sun & Allis, 2002) and/or cooperating with FACT (Pavri et al., 2006), a PARP1 interactome (Huang et al., 2006). PARP1 will therefore be involved in this process as it PARylates FACT (Huang et al., 2006) and EZH2 histone methyltransferase (Caruso et al., 2018; Masutani et al., 2003) to modify their enzymatic activities. Consequently, PARP1 might indirectly regulate the expression of circRNA (Figure 6a). In summary, these studies suggest that chromatin features/histone modifications may act as enhancers for circularization and that Alu elements harboring these features may play a critical role in shaping the transcriptome.

Figure 6:

a) PARP1 PARylates KDM5B to induce its disassociation from chromatin and reduce its demethylation activity, thereby decreasing the transcription of Alu elements crucial for the synthesis of circRNAs. B) ADAR1 catalyzes the conversion of adenosine to inosines within repetitive elements to lessen the efficiency of circRNA formation. By binding to Alu elements within the RNA sequence, ADAR1 introduces secondary structures recognized by HuR. PARP1 can then bind and PARylate HuR to promote its binding stability.

Alu elements are also recruitment hubs for adenosine deaminase acting on RNA-1 (ADAR1) proteins. ADAR1 is the master RNA editor, catalyzing the deamination of adenosine-to-inosine. Knockdown of ADAR1 upregulates the expression of specific circRNAs (Ivanov et al., 2015). This was further confirmed in human renal cells, where ADAR1 depletion increased the expression of circHIPK3 (Omata et al., 2022). ADAR1 could act directly or indirectly to regulate circRNA biogenesis. Directly, ADAR1 edits adenosines within Alu repeats, resulting in less efficient RNA pairing across flanking introns crucial for circularization (Wilusz, 2015). Indirectly, ADAR1 could induce secondary structures at the Alu sequences recognized by the Human antigen R (HuR) (Wang et al., 2013) to enhance transcript stability (Wang et al., 2013). Oddly, PARylation of HuR by PARP1 enhances oligomerization of HuR, resulting in dissociation of the microRNA-induced silencing complex, with consequences in target mRNA stabilization and gene expression (Ke et al., 2017). By enhancing HuR’s RNA binding, PARP1 could indirectly aid in increasing the stability of RNA transcripts (Figure 6b). Therefore, factors aiding in the formation and/or stability of the secondary structures created by Alu sequence may regulate the formation of circRNA. Since PARP1 regulates the ability of ADAR1 and HuR to associate with RNA, it may also regulate circRNA biogenesis through this pathway. While there is yet no evidence confirming that PARP1 mediates circRNA expression via HuR-ADAR1 axis, circRNAs also directly bind HuR to sequester this protein. This then modulates its binding to mRNA, thus suppressing the translation of mRNA transcript (Abdelmohsen et al., 2017). Future studies are therefore, critical in improving our understanding of how PARP1-HuR-ADAR1 interaction could potentially drive circRNA synthesis.

4.3. RNA binding protein (RBP)-driven circularization model

While repetitive Alu elements appear to be important for the biogenesis of circRNAs, flies lack Alu repeats (Westholm et al., 2014). Therefore, there must be another way for circularization that is independent of Alu elements, or dependent on other genomic cis elements (Kramer et al., 2015; McCullers & Steiniger, 2017). Additionally, RNA pairing across flanking introns increases the efficiency of circRNA formation but may not be sufficient to promote back-splicing (Zhang et al., 2014). In the RBP-driven circularization model, RBPs bind to flanking introns and via protein–protein interactions, bring the flanking introns to proximity to promote back-splicing and circularization (Ashwal-Fluss et al., 2014). Within a single gene, a set of RBP sites can add to the complexity of circRNA biogenesis. These binding proteins can cross talk in a combinatorial manner to generate distinct circRNAs in response to different environmental signals. For instance, splicing factors (Yu & Kuo, 2019), heterogeneous ribonucleoproteins (hnRNPs) and the serine/arginine-rich proteins (SRPs) have been shown to promote back-splicing and circularization (Kramer et al., 2015; Talukdar et al., 2011). While it is still yet to be investigated how PARP1 mediates back-splicing, PARP1 through PARylation regulates the function of these RBPs. Multiple hnRNPs bind tightly to PARylated PARP1 or free PAR molecules in a non-covalent manner, leading to the dissociation of hnRNPs from RNA (Gagne et al., 2005; Ji & Tulin, 2009; Jungmichel et al., 2013). In addition, SRPs bind to PAR moieties through a specific poly (ADP-ribose) recognition motif. This binding of PAR to SRPs not only competes with RNA binding (Teloni & Altmeyer, 2016), but also inhibits the phosphorylation of SRPs, altering their functional association with other proteins (Stamm, 2008). Indeed, PARylation of DNA topoisomerase I inhibited phosphorylation of the alternative splicing factor/splicing factor 2 (ASF/AF2), a member of the SRP (Malanga et al., 2008). Other RBPs implicated in circRNA expression include: 1) the splicing factor quaking homolog protein (QKI) (Hafner et al., 2010), 2) Muscleblind (MBNL1) (Ashwal-Fluss et al., 2014), and fused in sarcoma (FUS) (Errichelli et al., 2017; Lasda & Parker, 2014). Interestingly, all these proteins are part of the PARP1 interactome. In DNA damage response, activated PARP1 recruits and PARylates FUS (Singatulina et al., 2019) and QKI (Novikov et al., 2011) to sites of DNA damage. Given the effects of FUS and QKI in promoting back-splicing, it is not far-fetched to speculate that these RBPs cooperate with PARP1 to regulate circRNA expression during transcription. Finally, our recent finding shows that in the presence of PARP1 and/or PARylation, flanking introns of circRNAs are enriched in motifs recognized by these RPBs that are critical for circRNA biogenesis (Eleazer et al., 2023). Collectively, these studies provide supportive evidence that PARP1 and/or its PARylation activity adds to the complexity of circRNA processing pathways.

5. INTERPLAY BETWEEN PARP1 AND CIRCRNA FUNCTIONS

CircRNAs have been implicated functionally in several biological processes acting as sponges/decoys to microRNAs and RBPs, epigenetic regulators, and transcription and splicing modulators (Wilusz, 2018; X. Zeng et al., 2021; X. Zhang et al., 2019; Zhang et al., 2023). Intriguingly, recent studies now show that some circRNAs can be translated to proteins (N. Chen et al., 2018; Du et al., 2016; He et al., 2021; Y. Li et al., 2017; Li et al., 2015; Wang et al., 2022). Additionally, the dysregulation of circRNA biogenesis and functions have been implicated in many diseases (Gu et al., 2023; Y. Zeng et al., 2021).CircRNAs are widely expressed in mammalian cells, showing both cell type- or tissue-specific expression patterns (Misir et al., 2022). While strides have been made in profiling the expression of circRNAs, our understanding of their mechanistic function in developmental and cell-type speciation is still very much limited. Therefore, gaining a comprehensive grasp of the factors that facilitate circRNA biogenesis and their functions is crucial for comprehending cellular processes. Given PARP1’s involvement in every step of RNA biogenesis, PARP1’s possible role in circRNA synthesis may also add another layer of control to the functional pathways in which circRNAs participate. Below, we delve into the potential interaction between the functions of PARP1 and circRNAs.

RNA translation:

We discuss two ways by which PARP1 and circRNAs might regulate protein translation. 1. circRNAs can be translated, resulting in proteins that modulate the activity and stability of PARP1. For instance, circDIDO1 is translated into a novel death-inducer obliterator protein isoform that interacts with PARP1 in its DNA binding domain (DBD) and catalytic domain (CAT) to influence PARP1 activity (Xiao et al., 2022; Zhong et al., 2019). 2. both circRNA and PARP1 can work together to govern the regulation of protein translation. For instance, during translation, microRNAs are loaded onto Argonaute (Ago) which serve as guide molecules in microRNA-induced silencing complexes (miRISCs) for target-specific gene silencing (Ameres & Zamore, 2013). Several competing mechanisms are in play; a. Ago binds to microRNA resulting in mRNA degradation (Riley et al., 2012). b. circRNAs sequestering Ago proteins (Hansen et al., 2013; P. F. Zhang et al., 2019), which could impact the Ago–microRNA binding platform critical for gene expression, resulting in enhanced mRNA stability. c. PARylation of HuR by PARP1 interferes with miRISC activity, enhancing mRNA transcript stability (Y. Ke et al., 2021). d. CircRNA sequesters HuR, preventing its interaction with mRNA, leading to reduced translation target gene (Abdelmohsen et al., 2017). The identification of these competing mechanisms was strengthened by whole genome mapping studies that showed that most mRNAs with Ago-binding sites in their 3′UTRs also have HuR-binding sites (Y. Ke et al., 2021).

These discoveries suggest that the interplay between the Ago-microRNA-circRNA nexus and the PARP1-HuR nexus (Figure 7) may collaboratively regulate the equilibrium between mRNA stability and degradation (Y. Ke et al., 2021; Mukherjee et al., 2011; Mukherjee et al., 2009). Additionally, adding another layer to the intricate relationship between PARP1 and circRNA, a novel circRNA, Hsa_circ_0001944, has been identified as a mediator in modulating the interaction between PARP1 and HuR, thereby promoting apoptosis in hydroquinone-induced leukemia and malignant transformed cells (Zhong et al., 2023).

Figure 7:

In the cytoplasm, PARP1 PARylates HuR and promotes its ability to compete with the microRNA-induced silencing complexes at Ago binding sites, causing the alleviation of translational repression of mRNA target. CircRNAs can also alleviate microRNA-mediated repression by scaffolding the Ago–microRNA complex.

CircRNAs targeting PARP1 function:

Several studies have shown that circRNA transcripts can directly bind different PARP1 domains to alter PARP1 activity. In colorectal cancer cells, circ_0007142 was found to target PARP1 activity via the miR-103a-2–5p axis, promoting proliferation, migration, and invasion (Zhu et al., 2019). Another study showed that the binding of circTFDP2 binds to the DNA-binding domain of PARP1, inhibits its capase-3-dependent cleavage and activity, promoting prostate cancer progression (L. Ding et al., 2023). These results suggest that circRNAs in binding directly to PARP1, modulate its activity, with consequences in gene expression and disease.

Host gene expression:

PARP1 and circRNAs might work together to mediate the expression of circRNA host genes in cis or in trans (Bose & Ain, 2018; Qu et al., 2015). Li et al. (2015) showed that exon–intron circRNAs such as circEIF3J and circPAIP2 bind U1 small nuclear ribonucleic proteins (snRNPs) and RNAPII at the promoters of EIF3J and PAIP2 to enhance the expression level of their host genes. In support of this pathway, depletion of the intronic circRNAs, ci-ankrd52 and ci-sirt7, suppressed the expression of their parental genes (Zhang et al., 2013). These studies suggest that circRNAs can positively or negatively regulate host gene expression. Circular RNAs (CircRNAs) can influence RNA splicing not only through their interactions with splicing factors but also by contributing to the modulation of RNA polymerase II (RNAPII) pausing (Ashwal-Fluss et al., 2014; Li et al., 2020; Lupey-Green et al., 2018). For instance, circRNAs generated from the 6^th exon of the Arabidopsis SEPALLATA3 (SEP3) gene forms stable R-loops (RNA-DNA hybrids) causing a pause in RNAPII elongation (Conn et al., 2017). This pause coincides with the recruitment of splicing factors, consequently increasing the expression of alternatively spliced mRNA transcripts from the host gene (Alexander et al., 2010; Wongsurawat et al., 2012). While several circRNAs have been implicated in the formation and function of R-loops, PARP1 and/or its PARylation activity resolve R-loops (N. Chen et al., 2018; Laspata et al., 2023). Hence, it is plausible that PARP1 and its PARylation activity control the equilibrium of circRNA-mediated R-loops and function. Lastly, circRNAs might also directly regulate the expression of PARP1 as shown for PARP9, where circPRKC1 sponged miR-186–5p, resulting in PARP9 upregulation and carcinogenesis (Ma et al., 2020).

CircRNAs have also been shown to regulate the expression of host genes through epigenetic mechanisms. For instance, the exonic circFECR1 binds the promoter of its host gene (FLI1) and recruit TET1 to demethylate this promoter, with consequences in enhanced gene expression (N. Chen et al., 2018). On another hand, circFECR1 was shown to recruit DNMT1, resulting in promoter DNA methylation and repression of DNMTs (Li et al., 2020). This concordant coordinated regulation of circFECR1 on both TET1 and DNMT1 stimulates tumor cell invasion in breast cancers (Li et al., 2020). It is therefore possible that PARP1’s regulation of the occupancy and function of TETs and DNMTs (Ciccarone et al., 2015) may regulate how some of these circRNAs drive epigenetic regulation and specific gene expression patterns.

In summary, if PARP1 regulates circRNA biogenesis, it is also plausible that there is a functional network between PARP1 and circRNA in all the above processes. However, since this is a relatively new area of study, there is still much to be gleaned from circRNAs, the regulation of their biogenesis, their function, and the role of PARP1 in these processes.

Conclusion

This is an exciting time in PARP1 research because of the latest new discoveries into the role of this multifaceted protein. PARP1’s role in DNA damage repair and stress response is well-known, yet PARP1’s involvement in RNA biogenesis and metabolism is only now emerging. The discovery of PARP1 as an RNA binding protein (Melikishvili et al., 2017), together with its already known chromatin-binding property, has increased the complexity of its role in gene expression and regulation. Several pieces of evidence and our recent findings point to the possibility that PARP1 mediates circRNA formation. In this review, we have explored each step of circRNA biogenesis that PARP1 may regulate. Specifically, we examined PARP1’s role in transcriptional initiation (Matveeva et al., 2016), RNAPII elongation kinetics (Matveeva, Al-Tinawi, et al., 2019), and co-transcriptional splicing (Eleazer et al., 2023; Matveeva et al., 2016; Matveeva, Al-Tinawi, et al., 2019), all of which are known to influence circRNA biogenesis. Although many studies have advanced our understanding into the varied roles of PARP1 in RNA biogenesis and regulation, many questions remain. Understanding how PARP1 impacts circRNA synthesis will further our knowledge of PARP1 function in RNA biogenesis. Importantly, unraveling the role of PARP1 in circRNA expression will also provide critical insight into the pathologies many of these circRNAs are known to be involved in, such as cancers. Finally, not only will this information be critical in comprehending PARP1 biology, but it will also serve as a platform for disease diagnostics, biomarkers and in the development of therapeutics linked to PARP1 dysregulation.