Okazaki fragment maturation: DNA flap dynamics for cell proliferation and survival

Abstract

Unsuccessful processing of Okazaki fragments leads to accumulation of DNA breaks, which are associated with many human diseases, including cancer and neurodegenerative disorders. Recently, Okazaki fragment maturation (OFM) has received renewed attention in how unprocessed Okazaki fragments are sensed and repaired and how inappropriate OFM impacts genome stability and cell viability, especially in cancer cells. Here, we provide an overview of the highly efficient and faithful canonical OFM pathways and their regulation of genomic integrity and cell survival. We also discuss how cells induce alternative error-prone OFM processes to promote cell survival in response to environmental stresses. Such stress-induced OFM processes may be important mechanisms driving mutagenesis, cellular evolution, and resistance to radio-/chemotherapy and targeted therapeutics in human cancers.

OFM and its importance in genome integrity and survival

DNA replication is a semi-discontinuous process. Leading strand DNA synthesis, which is mediated by DNA polymerase epsilon (Pol ε), is continuous, whereas lagging strand DNA synthesis is discontinuous. During lagging strand DNA synthesis, a complex of DNA polymerase alpha (Pol α) and primase synthesize an RNA primer (7–14 nt) and a short stretch (10–20 nt) of DNA [1]. The Pol α/primase complex is then replaced by DNA polymerase delta (Pol δ), which extends the RNA–DNA primers into a series of discrete DNA fragments called Okazaki fragments [2]. The Okazaki fragment maturation (OFM) (see Glossary) process, which produces an intact lagging strand DNA from thousands to millions of Okazaki fragments, involves highly coordinated sequential reactions catalyzed by polymerases, nucleases, and DNA ligases [3, 4]. First, the Pol δ-synthesized nascent DNA of the upstream Okazaki fragment extends into the downstream Okazaki fragment, producing a 5′ flap structure consisting of the RNA–DNA primer. The 5′ flap structure is then cleaved by members of the RAD2 structure-specific endonuclease family, typified by flap endonuclease 1 (FEN1), creating a ligatable DNA nick, which is joined by DNA ligase I (Lig I) [5]. In addition, since Pol α has no proofreading function, Pol α-synthesized DNA fragments, which account for 1.5% of the genome [6], have a high frequency of errors [7]. To avoid DNA mutations, Pol α errors must be eliminated. A Pol α error may be endonucleolytically removed during 5′ flap cleavage. When Pol α errors occur downstream of the 5′ flap, they can be further removed by the 5′ exonuclease activity of FEN1 in complex with MutSα, followed by Lig I-directed DNA ligation [8]. In yeast, a Pol α error may also be recognized by the Msh2–Msh6 complex, which subsequently promotes Pol δ-mediated strand displacement and flap cleavage by Rad27 (the budding yeast homolog of FEN1 [9]) to remove the error [10].

Efficient and faithful OFM is critical for genome integrity and cell survival. Deletion of genes encoding key enzymes, regulatory proteins, or protein-modifying enzymes of OFM result in un-ligated Okazaki fragments and incomplete DNA replication, thus leading to cell death. In this review, we will first summarize current advancements in understanding the distinct roles of various 5’ nucleases in RNA–DNA primer removal and the molecular mechanisms by which various OFM enzymatic reactions are coordinated through dynamic post-translational modifications and protein-protein interactions. We will then discuss various DNA intermediate structures caused by faulty OFM processes and induction of DNA damage checkpoints, and corresponding consequences in cell fate. In particular, we will review the most recent studies on how cells transform unprocessed RNA–DNA primers to induce alternative pathways for OFM and survival. In addition, over the last several years, increasing evidence has emerged to indicate that prolonged existence of DNA single-strand breaks (SSBs) is a major cause of cell death in human cancer cells, which are frequently deficient in one or more DNA damage response and repair proteins. We discuss new discoveries that implicate OFM enzymes as targets to specifically kill human cancer cells via synthetic lethality.

Multiple nuclease-driven pathways for proper RNA–DNA primer removal

The removal of RNA–DNA primers lies at the core of the OFM process. In eukaryotic cells, several different pathways have evolved to remove the RNA portion of Okazaki fragments, allowing subsequent DNA ligation [3]. Originally, it was thought that RNA–DNA primer removal mainly depended on RNase H and FEN1 [5]. RNase H degrades the ribonucleotides, leaving the last one in the RNA–DNA hybrid duplex. The last ribonucleotide is then excised by FEN1 to generate a ligatable nick [5]. However, subsequent genetic studies revealed that RNase H1 or RNase H2 are not required for DNA replication and survival in yeast, suggesting that other mechanisms might serve as the predominant pathways for the removal of RNA–DNA primers[11]. It is now known that RNA–DNA primers are primarily removed through 5′ flap formation and cleavage (Figure 1). Two distinct 5′ flap cleavage pathways, the short and long flap pathways, have been reported to remove RNA–DNA primers [12, 13]. In the short flap pathway (Figure 1), Pol δ in complex with proliferating cell nuclear antigen (PCNA) catalyzes strand displacement DNA synthesis, producing a short 5′ flap (2–10 nt). FEN1 effectively recognizes and cleaves the short 5′ flap at the junction of single-stranded DNA (ssDNA) and double-stranded DNA (dsDNA). Nick translation by Pol δ and FEN1 then continues until the RNA portion is completely removed and a ligatable nick is generated [14]. The function of FEN1 in cleaving the short 5′ flap can be supported by 5′ exonuclease 1 (EXO1) [15], which is a member of the RAD2 structure-specific endonuclease family. Short 5′ flaps that escape cleavage by FEN1 may result in the formation of longer 5′ flaps (up to 400 nt) [16]. Meanwhile, with the help of Pif1, a 5′ to 3′ helicase [17], Pol δ-driven strand displacement DNA synthesis may give rise to a relatively long 5′ flap [18]. Removal of long 5′ flaps during OFM requires DNA2 nuclease (Figure 1). Previous genetic and biochemical studies in yeast suggested that during long 5′ flap processing, the replication protein-A complex (RPA) binds to the long flap structure and inhibits FEN1 [13]. Meanwhile, RPA recruits DNA2 nuclease, which cleaves the long 5′ flap in the middle and generates a short 5′ flap of less than 10 nt [13]. FEN1 or EXO1 then cleaves the short 5′ flap, creating a ligatable nick. Consistent with this mechanism, a more recent study showed that Dna2 deletion resulted in an accumulation of Pif1-created long 5′ flaps in yeast cells [19].

Figure 1. Two canonical OFM processes via formation and cleavage of short or long 5’ flaps and nick ligation.

Pol δ-mediated Okazaki fragment extension encounters an RNA–DNA primer in a downstream Okazaki fragment and displaces the primer portion to form a 5′ flap. In most cases, the 5′ flap is 2–10 nt in length (short flap) and is primarily cleaved by FEN1, creating a DNA nick for ligation. EXO1 also degrades a relatively small percentage of such short 5′ flaps and serves as a backup of FEN1. The displacement-cleavage cycle continues until the whole RNA–DNA primer is degraded, which also removes any Pol α errors (dark dots in the RNA–DNA primer). The created ligatable DNA nick is then ligated by Lig I. This process is called short flap OFM. If a short RNA–DNA flap escapes cleavage by FEN1 or EXO1, continuous DNA strand displacement may result in a long 5′ flap of up to 300 nt in length. On the other hand, if PIF1 helicase is recruited to the replisome, DNA strand displacement is enhanced, leading to the formation of a long 5′ flap. The ssDNA binding protein RPA binds to the long RNA–DNA flap, inhibiting FEN1 cleavage of the 5′ flap. However, RPA recruits DNA2, which cleaves the long 5′ flap in the middle and creates a short 5′ flap (less than 10 nt). The short flap is then cleaved by FEN1 or EXO1, and the DNA nick is ligated by Lig I. The process that removes the long 5′ flap by the sequential actions of DNA2 and FEN1 (or EXO1) is called long flap OFM. In both short flap and long flap OFM, if the Pol α errors are not removed by 5′ flap cleavage, they can be edited out by the exonuclease activity of FEN1.

An important but not fully answered question is what each of these three nucleases contributes during OFM. To address this issue, a deep DNA sequencing approach was used to analyze Okazaki fragments in mutant yeast strains that were conditionally depleted of Rad27, Dna2, or Exo1, either individually or in combination [20]. Consistent with the results of previous yeast genetic studies, Rad27 was found to be responsible for processing the majority of the 5′ flaps. Exo1 also made a significant contribution to 5′ flap cleavage, even in the presence of Rad27, and served as a backup in the absence of Rad27. Surprisingly, sequence analysis of Okazaki fragments in wild-type and Dna2-deleted yeast cells revealed little difference in Okazaki fragment processing, suggesting that Dna2 plays a very limited role in 5′ flap cleavage [20]. Furthermore, this study discovered that Pol δ-driven strand displacement was mostly inhibited in cells lacking Rad27 and/or Exo1, suggesting that the long flap is rarely formed and that the long flap pathway is rarely used to remove RNA–DNA primers across the genome, even in yeast cells with Rad27 or Exo1 deficiency. This observation is in contrast to findings that long flaps exist in Rad27- or Dna2-deficient yeast cells and that deletion of Pif1 suppresses the lethality phenotype due to Dna2 knockout [16, 18, 19]. A possible explanation is that the DNA2-mediated long flap pathway is only critical for processing RNA–DNA primers at certain loci that are important for cell cycle progression and survival. Supporting this concept, human DNA2 was found to be a primary player during OFM at centromeric regions but not non-centromeric regions [21].

Dynamic post-translational modifications and protein complexes for efficient OFM

Efficient and proper OFM is essential for preventing prolonged exposure of ssDNA gaps, which may be converted into the most lethal and mutagenic double-strand breaks (DSBs), on the lagging strand template. This process requires the active enzyme activities of Primase, Pol α, Pol δ, FEN1, and Lig I, as well as the precise coordination of their sequential actions at Okazaki fragment sites. Key mediators have been identified for dynamic protein–protein interactions and protein post-translational modifications, which coordinate the access of different OFM enzymes to replication sites. One important role during Okazaki fragment synthesis is limiting the function of Pol α due to its low processivity and lack of proofreading capability. Replacing Pol α in the replisome with Pol δ, a processive and highly faithful polymerase, is critical for restricting unwanted Pol α action. The clamp loader replication factor C (RFC) is essential for polymerase switching and for restricting Pol α from rebinding to Okazaki fragments even in the absence of Pol δ [22]. In addition, RFC also restrains the involvement of Pol ε on the lagging strand, which has much lower strand displacement DNA synthesis activity relative to Pol δ [22]. Such restraint is enforced even under the condition of Pol δ deficiency [23].

OFM involves a series of events that include Pol δ-driven strand displacement DNA synthesis, flap cleavage by FEN1 or by FEN1 and DNA2, and DNA ligation by Lig I [3, 4]. To effectively complete these sequential reactions, cells have developed a precise mechanism for the orderly recruitment of these enzymes to the replication site to execute their functions and promptly dissociate from DNA substrates, allowing downstream enzymes to access DNA substrates. Physical interactions between PCNA and the core enzymes Pol δ, FEN1, and Lig I are essential for the recruitment of these core enzymes to the replication fork [24–26]. A key question is how these enzymes interact with PCNA in an orderly fashion. Previous studies on the dynamic interactions between FEN1 and PCNA during DNA replication offer some insights (Figure 2). It was shown that arginine methylation of FEN1 ensures its interaction with PCNA [27]. As FEN1 completes flap cleavage, it is demethylated by an arginine demethylase such as JMJD1B [28]. Demethylation allows cyclin-dependent kinase 1 (CDK1) to bind to and phosphorylate FEN1 [29]. Phosphorylated FEN1 then dissociates from PCNA [29]. Phosphorylated FEN1 undergoes further SUMO3 modification and ubiquitination, which mediate FEN1 degradation [30]. The dissociation from PCNA and subsequent degradation allow Lig I to interact with PCNA and join the DNA nicks (Figure 2).

Figure 2. Sequential post-translational modifications of FEN1 mediate its dynamic interactions with PCNA and timely exchange of Okazaki fragment enzymes.

During lagging strand DNA replication in S phase, FEN1 is symmetrically methylated at arginine residues, primarily R192 (R192me2s FEN1). The R192 methylation of FEN1 promotes its interaction with PCNA for recruitment to 5′ flaps. Once methylated FEN1 completes its RNA–DNA primer removal function, it is demethylated by an arginine demethylase, typified by JMJD1B. R192 demethylation of FEN1 allows its binding to CDK1, which catalyzes S187 phosphorylation of FEN1. S187 phosphorylates FEN1, dissociates from PCNA and undergoes sequential type 3 SUMO modification (K168) and ubiquitination (K354), leading to FEN1 degradation. Dissociation from PCNA as well as degradation of FEN1 ensure the efficient binding of Lig I to PCNA for DNA ligation.

One fundamental question regarding OFM is how strand displacement DNA synthesis is terminated to avoid Okazaki fragment overextension. Previous studies using genomic approaches have provided mechanistic insights into this question [6, 31]. It has long been known that DNA replication is closely coupled with histone deposition and nucleosome assembly to ensure efficient duplication of genetic and epigenetic information [32]. Chromatin assembly factor 1 (CAF-1), which mediates the assembly of histone proteins, is recruited to daughter-strand DNA by interacting with PCNA [32]. A close interplay was observed between OFM and histone assembly [6, 31]. A finding that the junction of two Okazaki fragments is near the midpoint of the nucleosome suggests that Okazaki fragment junctions are dictated by histone deposition [6]. Histone deposition acts as a barrier, terminating Pol δ-mediated strand displacement DNA synthesis and preventing long flap formation. On the other hand, premature deposition of histones may lead to persistent Pol α-generated errors, resulting in DNA mutations. In addition, 5′ flaps within nucleosomes that have high-affinity interactions between DNA sequences and histones are poorly cleaved by FEN1 [33]. Therefore, premature histone deposition has a negative impact on OFM, and the timing of histone assembly must be tightly regulated through the actions of Pol δ and FEN1. Interactions of the histone deposition factor CAF-1 and PCNA are mediated through PCNA SUMOylation in human cells [34]. Therefore, PCNA SUMOylation, which is catalyzed by TRIM28 [35], may serve as a key mediator in coordinating OFM and histone assembly.

Cytotoxic intermediate DNA structures from faulty OFM

Deficiency or inhibition of the core enzymes or disruption of the functions of the accessory proteins for OFM can lead to generation of various cytotoxic and mutagenic intermediate structures (Figure 3). Pol δ-driven strand displacement DNA synthesis initiates the OFM process. Because Pol δ is the sole polymerase for strand displacement DNA synthesis, Pol δ deficiency or mutations that affect Pol δ polymerase activity cause defects in the extension of OFM, leading to extensive ssDNA gaps [23]. Meanwhile, the replisome may encounter difficulty in replicating and may frequently form secondary structures such as G-quadruplex (G4) DNA, which frequently form in GC-rich regions in cells [36, 37]. Furthermore, environmental insults, endogenous metabolites, and genotoxic cancer therapeutic drugs can bind to or cross-link DNA. The resulting DNA lesions block Pol δ-driven Okazaki fragment extension, leading to ssDNA gaps [38, 39]. In the case of FEN1, its deficiency results in ssDNA gaps or nicks [26, 40], presumably with a short 5′ flap. Unligated ssDNA gaps/nicks with or without a 5′ flap may stall or collapse DNA replication forks during the next round of DNA replication or be cleaved by endonucleases, leading to a DSB (Figure 3), which is the most lethal and mutagenic type of DNA damage. Deficiency in Pol δ, Rad27, or Cdc9 (yeast Lig I) in yeast cells significantly enhances mitotic recombination, an indicator of DSBs [23, 41, 42]. FEN1 or DNA2 mutations in mammalian cells have been shown to cause endogenous DSBs [43, 44].

Figure 3. DNA lesions, damage responses, and cellular consequences of OFM defects.

Genetic and epigenetic alterations and various environmental insults can give rise to OFM defects or failure, which can result in ssDNA nicks or gaps with or without a short or long 5′ flap and ssDNA gaps with a secondary structure such as G4 opposite to the gap. DNA nicks or gaps can be further converted into DSBs by endonuclease cleavage or due to the collapse of replication forks in the next round of DNA replication. DNA damage sensor proteins, including RPA, the 9-1-1 complex, and PARP1 (in mammalian cells only), recognize these DNA lesions and activate the DNA damage response kinases ATM and ATR, which phosphorylate CHK2 and CHK1, respectively. Further downstream, these induce permanent damage, cellular senescence, and death or cell cycle arrest, DNA repair, and cell survival, respectively. Death or survival depends on the balance of ATM- or ATR-triggered pro-death and pro-survival signaling, and this balance hinges on the type and level of lesions in the cells. While SSBs are mostly not lethal, DSBs caused by fork collapse frequently lead to cell death. In mammalian cells, the cell fate decision made in response to various types and levels of DNA lesions is controlled by ATM/CHK2- or ATR/CHK1-mediated phosphorylation of the transcription factor p53, which induces the expression of both pro-death and pro-survival genes. Distinct p53 phosphorylation statuses determine the balance among pro-death and pro-survival p53 target genes, leading to cell death or cell survival.

When PIF1 is present in the replisome, Pol δ-driven strand displacement DNA synthesis creates long 5′ flaps. In addition, there are Pol δ mutations that enhance its processivity and strand displacement activity [45], potentially resulting in long DNA flaps. Because termination of strand displacement DNA synthesis is controlled by histone deposition [6], defects in the histone deposition process may also cause unwanted strand displacement and the formation of long DNA flaps. Intriguingly, a long DNA flap itself, if not removed, is more cytotoxic than a short flap, at least in yeast; DNA2 knockout, which causes a failure to remove long 5′ flaps, results in cell death [13]. However, deletion of PIF1, which is in complex with Pol δ for the formation of long 5′ flap structures in yeast, rescues the lethality caused by DNA2 knockout, supporting a model in which DNA2 knockout-induced lethality is at least partially due to the accumulation of long flaps during OFM [18]. Likewise, deletion of Pol32, a subunit of yeast Pol δ responsible for its processivity, also makes Dna2 dispensable for cell survival [46].

DNA damage responses in OFM

Various intermediate structures that arise during OFM are detected by DNA lesion sensor proteins to activate DNA damage response pathways and trigger cell cycle arrest, DNA repair, cell senescence, or apoptosis. Accumulation of ssDNA regions on the template strand or the nascent DNA strand is sensed by RPA, which leads to ATR activation and subsequent CHK1 phosphorylation [47]. ATR/CHK1 activation is critical for the survival of FEN1- or DNA2-mutant cells, as CHK1 inhibition causes cell death, possibly through chromosome catastrophe [21, 48]. ssDNA gaps with or without 5′ flaps may also be sensed by the DNA damage response protein poly(ADP-ribose) polymerase 1 (PARP1), which binds to various DNA lesions and catalyzes the polymerization of ADP-ribose at damage sites to facilitate DNA repair [49]. In human cells, chemical compounds that inhibit FEN1 nuclease activity promote recruitment of PARP1 to lagging strand DNA replication sites [50]. PARP1 localized to replication sites then recruits the SSB repair protein XRCC1, which facilitates DNA ligation of nicks by Lig3 [51]. PARP1 inhibition in FEN1-deleted DT40 chicken cells caused the mutant cells to accumulate additional ssDNA nicks and gaps [52]. A very recent study showed that DNA gaps in lagging strand DNA caused by Pol δ blockage by G4 may be recognized by the 9-1-1 complex. Binding of the 9-1-1 complex to un-ligated Okazaki fragments prevents EXO1-mediated over-resection [53]. Alternatively, an unprocessed long 5′ flap can directly serve as a signal to activate DNA damage checkpoints. A long 5′ flap may be bound by RPA to activate ATR and CHK1; it may also activate the 9-1-1 complex, which facilitates ATR activation [54]. A previous study showed that inactivation of the Rad9 checkpoint in yeast suppressed cell death induced by DNA2 deficiency [55]. Previous studies have shown that DSBs may be sensed by the MRE11/RAD50/NBS1 (MRN) complex [56], the KU70/KU80 complex [57], and/or PARP1 [49, 58], contributing to the activation of ATM, which phosphorylates CHK2 and other downstream DNA damage response proteins [59].

Activation of DNA damage signaling pathways leads to cell cycle arrest and induces various DNA repair pathways that are critical for preventing the potential detrimental effects of SSBs and DSBs. In yeast cells, RAD27 knockout and DNA2 mutation are synthetically lethal with deletion of MEC1 (ATR) or TEL1 (ATM), respectively [46]. Impairment of DNA damage response pathways by ATM or ATR inhibitors promotes cell death in FEN1- or DNA2-mutant cells. In addition, inhibition of FEN1 and DNA2 has strong synergy with PARP1 inhibition in inducing cell death in human cells [50, 52, 60, 61]. However, persistent activation of DNA damage checkpoints is also the cause of cellular senescence or apoptosis in cells with OFM defects. A long 5′ flap in DNA2 knockout yeast cells activates Rad9, and deletion of RAD9 rescues the lethal phenotype of dna2Δ cells [55]. Activation of CHK1 and subsequent p53 phosphorylation was associated with cellular senescence or apoptosis in FEN1-mutant MEF cells, but inactivation of CHK1 and epigenetic suppression of the p53 signaling pathway allowed these cells to escape from cellular senescence or apoptosis [48]. In mammalian cells, activation of the ATM/ATR kinases by DNA damage leads to p53 phosphorylation, which induces the expression of both pro-survival and pro-death genes [62]. The balance of these p53-target genes, which depends on the level of DNA damage, determines the fate of a cell. It is plausible to postulate that similar mechanisms control the fate of cells with defects in OFM. Low levels of SSBs, 5′ flaps, and DSBs due to faulty OFM transiently activate ATM/ATR, resulting in p53 protein phosphorylation (at Ser15, Ser20, and/or Ser33) [62] and the expression of downstream genes primarily involved in transient cell cycle arrest, DNA repair, and cell survival. On the other hand, continuous OFM and DNA replication failure, as in FEN1 homozygous deletion mouse embryos [63], persistently activates ATM/ATR, leading to different p53 protein phosphorylation (Ser46) [62] that mainly drives the expression of pro-senescence and pro-death genes (Figure 3).

Alternative 3′ flap OFM processes for survival under stress

DNA replication machinery frequently encounters environmental insults, endogenous metabolites, and genotoxic cancer therapeutics. Lagging strand DNA synthesis is particularly vulnerable to such stressors. Yeast cells carrying mutations in OFM genes such as RAD27 and DNA2 may be viable under normal growth conditions but inviable under conditions of stress, including at restrictive temperatures [9, 64, 65]. This suggests that stress may impair OFM processes. To survive under stress, cells have evolved alternative pathways to promote Okazaki fragment processing. It was previously suggested that an unprocessed 5′ flap may anneal to the DNA sequence on the template strand for DNA ligation (Figure 4). In this way, 5′ flaps, especially long 5′ flaps, may be resolved, albeit with the consequence of classic duplications. More recently, activation of Mec1-Rad53-Dun1 signaling and induction of a 3′ flap OFM were observed in a subpopulation of rad27Δ yeast cells in response to restrictive temperature [66]. In this 3′ flap OFM process (Figure 4), unprocessed 5′ flaps were transformed into 3′ flaps, possibly via flap equilibrium or helicase-driven 3′ end displacement. The 3′ nuclease activity of Pol δ and other 3′ nucleases, such as MRE11 and MUS81, may cleave the 3′ flaps, generating a nick for ligation into an intact DNA duplex. 3′ flaps, if they escape from 3′ nuclease cleavage, may anneal to a homologous sequence in sister chromatin, leading to Holliday junction (HJ) structures, which may be resolved via HJ resolvases (Figure 4). However, at certain regions, 3′ flaps may fold back and form secondary structures that facilitate 3′ end extension but not degradation. The 3′ flaps may mistakenly anneal to microhomology sequences and extend themselves. Ligation of an extended 3′ flap with a downstream Okazaki fragment produces a tandem duplication with a short spacer sequence (alternative duplication) (Figure 4) [66]. 3′ flap OFM is likely to be conserved in mammalian cells as well, given that 3′ flap OFM-related alternative duplications are frequently observed in human and mouse cancers. In addition, FEN1 mutations increased the frequency of alternative duplications [66].

Figure 4. Flap dynamics lead to a 3′ flap for alternative OFM pathways and cell survival.

Under normal physiological conditions, unprocessed 5′ flaps may anneal to the template strand, creating DNA nicks for ligation and classic duplications. Under conditions of stress, cells activate additional DNA damage response pathways, leading to transformation of 5′ flaps into 3′ flaps. Nucleolytic degradation of 3′ flaps results in DNA nicks for Okazaki fragment ligation, called 3′ flap OFM. In 3′ flap OFM, Pol α-introduced errors are not removed, potentially causing base substitutions. Meanwhile, in certain regions, 3′ flaps may form a fold-back structure or anneal to nearby regions of microhomology. Extension and ligation of such 3′ flaps result in alternative duplications. Alternatively, the 3′ flap may invade the sister chromatin and initiate the template switching process for OFM. This process requires the involvement of HDR proteins and leads to sister chromatin exchange.

Template switching pathway for OFM at “difficult-to-replicate” DNA regions

Across the genome, there are many “difficult-to-replicate” DNA regions; for example, those that may form secondary structures or those that are tightly bound by proteins [67]. One stable unusual DNA structure is the G4 structure [36, 37]. Although G4 RNA is rare in cells [68], G4 DNA structures frequently form at telomeric and non-telomeric regions in human cells [36]. During lagging strand DNA synthesis, such structures block Pol δ and the canonical OFM process (Figure 5). Normally, DNA helicases are recruited to resolve secondary structures and remove DNA-bound proteins to resume the movement of Pol δ for strand displacement DNA synthesis [69, 70]. However, in some cases, the action of helicases may be suppressed. For instance, G4 stabilizers may inhibit FANCJ and other DNA helicases that unwind G4 [71]. Under such situations, cells may process Okazaki fragments and bypass the secondary structures using homology-directed gap repair (Figure 5) [38, 72]. During this process, the small gap is first enlarged by nuclease-driven resection. RNase H, 5′ nucleases such as EXO1, and 3′ nucleases such as MRE 11 may be involved in this process [73–75]. Once the gap is enlarged, the 5′ to 3′ helicase, typified by PIF1, may bind the ssDNA gap and unwind the DNA duplex in the 5′ to 3′ direction to generate a 3′ flap, which then invades into the sister chromatin to initiate homology-directed recombination [76]. Ubiquitinated PCNA and Rad51 protect the ssDNA 3′ flap from nuclease cleavage to facilitate this process [77, 78]. RNA coated with Rad51-associated protein 1 was found to open the sister chromatin within the fork to form a bubble structure called an R-loop [79], which may be important for 3′ ssDNA flap annealing to homologous DNA sequences in sister chromatin during template switching.

Figure 5. The template switching process facilitates OFM in difficult-to-replicate regions.

Stable secondary structures such as G4 frequently form in difficult-to-replicate regions, including telomeres and centromeres, blocking Pol δ-driven Okazaki fragment extension and OFM and resulting in a gap on the nascent strand, with a secondary structure (G4) on the template strand. During or after DNA replication, cells utilize the template switching process to repair gaps and bypass stable secondary structures. This process includes sequential nuclease-mediated gap resection, helicase (PIF1)-mediated duplex unwinding to create a 3′ flap, and 3′ flap invasion into the sister chromatin. This process creates a Holliday junction (HJ) structure, which requires HJ resolvase to resolve. The bypassed G4 structure in the genome may be resolved by DNA2-mediated G4 cleavage or removed by nucleotide excision repair.

OFM genes as targets for chemoprevention and chemotherapy

Given the importance of OFM for cell survival and the fact that key enzymes and accessory proteins such as FEN1, Lig I, and PCNA are frequently overexpressed in human cancers, OFM proteins have been considered targets for chemoprevention and chemotherapy [61, 80]. Recent studies on the mechanisms of DNA replication in mammalian and human cells suggest that DNA replication and repair processes are interconnected. Indeed, many OFM proteins are important for processing OFM-associated intermediates as well as repairing OFM-associated DSBs. Thus, inhibiting these enzyme activities would presumably result in simultaneous enhancement of DSB induction and replication-associated DSB accumulation, and would thus effectively induce apoptosis, as DSBs are the most lethal type of DNA lesion in human cells [58, 81].

Nevertheless, how to achieve cancer specificity when targeting the OFM process has been a long-standing issue, given that the process is essential for both normal and cancer cells. Cancer cells frequently bear genetic alterations that disrupt the function of DNA damage response and repair proteins. Exploiting such genetic alterations in tumors offers an effective strategy to achieve cancer specificity via synthetic lethality (SL), in which the combined disruption of two genes or pathways leads to cell death [82]; in contrast, normal cells are well protected from SL by normal checkpoints and robust backup DNA repair pathways. PARP inhibitors (PARPis) represent a successful example of exploiting SL in cancer [83]. Partial inhibition of OFM is sufficient to cause persistent SSBs [84]. Because PARP1 is important for repairing SSBs due to OFM inhibition, combined inhibition of OFM enzymes and PARP1 synergistically drives the accumulation of SSBs, with the potential for conversion into DSBs as well as for stable binding of the PARP1-PARP1 inhibitor complex (also called PARP trapping), which may block replication forks (Figure 6). Meanwhile, the OFM template switching pathway and DSB repair require homology-directed repair (HDR) proteins. In yeast, rad27Δ was found to be SL with HDR proteins rad51Δ and rad52Δ [42], suggesting SL between OFM deficiency and HDR deficiency. In human cells, HDR depends on BRCA1/2 [85]; therefore, inhibition of OFM enzymes would also result in SL with BRCA1/2 deficiency (Figure 6). Supporting this model, a low dose of FEN1 and DNA2 inhibitors induces SL with BRCA1/2 deficiency and with PARPi in human cancers [61, 80, 86]. Additionally, stress-induced alternative OFM supports cell survival in yeast and likely supports cell survival in human cancer cells [66]. Because cancer cells suffer substantially higher replication stress than normal cells due to aberrantly high replication and cell division, it is likely that cancer cells, but not normal cells, rely on the stress-induced OFM pathway for survival. Thus, suppression of stress-induced OFM is an attractive new strategy for achieving cancer cell-specific cell death.

Figure 6. OFM inhibition creates multiple layers of synthetic lethality (SL) with PARP1 inhibition or BRCA1/2 deficiency in cancer cells.

Inhibition of OFM enzymes such as RNasH2, FEN1, DNA2, or EXO1 in human cancer cells leads to the formation of un-ligated Okazaki fragments, which recruit PARP1 for SSB repair. Inhibition of OFM and PARP1 synergistically drives the accumulation of SSBs that can be converted into DSBs and/or stalled/collapsed replication forks. DSB repair at least partly requires the BRCA1/2-mediated HDR pathway. Thus, OFM inhibition creates SL with BRCA1/2 deficiency. In addition, OFM enzymes, including DNA2 and EXO1, work together with other nucleases including CtIP and MRE11 to resect DNA end for DSB repair via the HDR pathway. Thus, the inhibition of OFM enzymes may impair HDR as well, contributing to SL with PARP1 inhibitors.

Concluding remarks

Efficient and accurate OFM is key to DNA replication, which is fundamental to all cells. Mutations that lead to defects in OFM can cause accumulation of SSBs and Pol α-incorporated errors, subsequently leading to accumulation of mutations, genomic instability, and cell death, and contributing to human diseases including cancer, developmental disorders, and neurodegeneration [26, 87–90]. Faithful OFM requires the removal of RNA primers and mismatches introduced by Pol α-synthesis, which together account for approximately 1.5% of the genome. RNA primers may be removed via strand displacement DNA synthesis by Pol δ, 5′ flap cleavage either by FEN1 alone (the short flap pathway) or by DNA2 and FEN1 (the long flap pathway), and the subsequent ligation of two Okazaki fragments by Lig I. These ordered enzymatic reactions are coordinated by PCNA, which sequentially interacts with Pol δ, FEN1, Lig I, and CAF-1. PCNA complex dynamics are controlled by post-translational modifications of OFM proteins. Despite these advancements in understanding, especially regarding the function and regulation of core OFM proteins, much less is known about the mechanism that regulates the choice of pathway for RNA primer removal and how cells counteract unprocessed 5′ flaps that escape from nuclease cleavage to support cell survival. The biological significance of the long flap pathway remains far from clear. It is unknown whether it functions in the removal of errors made by Pol α/primase.

It is now generally accepted that the FEN1-mediated cleavage of short RNA primer flaps is the primary pathway for RNA primer removal in both yeast and human cells, the two most common models for studying OFM. While the long flap pathway is rare under normal physiological conditions, it may become relatively frequent due to genetic or epigenetic alterations. For example, Pol δ mutations or defects in histone deposition, which terminates strand displacement, can cause unwanted strand displacement DNA synthesis. In addition, helicases such as PIF1 may bind to unprocessed short flaps and displace them into long flaps. Regardless of the physiological significance of the long flap pathway, unprocessed long flaps are presumably more mutagenic and cytotoxic than short flaps, as unprocessed long flaps have a greater likelihood of forming secondary structures and furthermore can act as a signal for DNA damage checkpoint activation that may lead to cell cycle arrest or cell death.

Because unprocessed 5′ flaps prevent the ligation of Okazaki fragments and are therefore potentially harmful, cells have evolved various molecular mechanisms to cope with deficiencies in 5′ flap processing capacity due to mutations or insufficient expression of FEN1, DNA2, or EXO1. In human cells, unprocessed 5′ flaps are recognized by PARP1, which catalyzes PARylation at replication sites and recruits DNA repair proteins such as XRCC1. PARP1 has also been shown to catalyze PARylation of nucleases, including FEN1 and DNA2. It remains unclear whether PARylation of these nucleases enhances flap cleavage. On the other hand, an unprocessed 5′ flap may be converted into a 3′ flap via a 5′ and 3′ flap equilibrium process. The 3′ flap is then degraded by the 3′ flap nuclease activity of Pol δ or other 3′ flap nucleases, generating nicks for ligation. Because PARP1 and/or poly(ADP-ribose) chains can interact with helicases and 3′ nucleases, PARP1 binding to unprocessed 5′ flaps may recruit helicases and 3′ nucleases for 3′ flap formation and processing during OFM. Future studies are needed to investigate this process and address other fundamental issues regarding OFM and its biological significance to cell survival (See Outstanding Questions).

Outstanding questions.

What is the composition of the stress-induced 3’ flap based OFM machinery, particularly, the helicase(s) that transform 5′ flaps into 3′ flaps and 3′ flap nuclease(s) that remove 3′ flap?

Whether stress conditions, including oncogenesis-induced stress and anticancer drug treatment, change the pattern of posttranslational modifications of Okazaki fragment maturation factors and cause the 5′ flap to escape from nucleolytic cleavage?

Whether and how posttranslational modifications of Pol δ regulate its strand displacement DNA synthesis activity, which dictates the choice of the short or long flap pathway for RNA primer removal?

What factors other than PARP1 contribute to sensing the unprocessed 5′ flap and inducing the alternative processing of RNA primers, including the 3′ flap-based Okazaki fragment maturation?

Whether the induction of 3′ flap-based Okazaki fragment maturation pathways in tumor cells promotes their survival in addition to producing genetic variations for molecular evolution, contributing to malignant progression and the development of drug resistance during therapeutic treatment?

Highlights.

Okazaki fragment maturation (OFM), the most frequently occurring DNA metabolic process, is efficient, faithful, and highly regulated, and it is critical for maintaining genome integrity and cell survival.

Unligated Okazaki fragments are a major source of DNA single-strand breaks (SSBs) in eukaryotic cells.

Cells activate DNA damage response checkpoints to induce cell cycle arrest and repair unprocessed Okazaki fragments under normal physiological conditions.

Stress conditions induce error prone OFM pathways, especially in cells with defects in the processing of Okazaki fragments. Induction of error prone pathways supports cell survival but also causes genetic alterations.

Inhibition of OFM enzymes may be targeted for specifically killing cancer cells via synthetic lethality.

Glossary

“Difficult-to-replicate” DNA

Genomic regions that are intrinsically hard to be copied by DNA polymerases.

DNA damage response

A complex process in which DNA damage sensor proteins recognize and bind DNA lesions to activate downstream signaling cascades for repairing DNA lesions and preventing their deleterious effects.

Okazaki fragment maturation (OFM)

The processes to remove RNA–DNA primers and edit out DNA polymerase α incorporation errors in the discrete nascent lagging strand DNA fragments (Okazaki fragments) during DNA replication.

Template switching

A process in which the 3’ end of a stalled nascent DNA anneals to DNA sequences in the sister chromatid and extends using the sister chromatid as the template.