An emerging paradigm in epigenetic marking: coordination of transcription and replication

ABSTRACT

DNA replication and RNA transcription both utilize DNA as a template and therefore need to coordinate their activities. The predominant theory in the field is that in order for the replication fork to proceed, transcription machinery has to be evicted from DNA until replication is complete. If that does not occur, these machineries collide, and these collisions elicit various repair mechanisms which require displacement of one of the enzymes, often RNA polymerase, in order for replication to proceed. This model is also at the heart of the epigenetic bookmarking theory, which implies that displacement of RNA polymerase during replication requires gradual re-building of chromatin structure, which guides recruitment of transcriptional proteins and resumption of transcription. We discuss these theories but also bring to light newer data that suggest that these two processes may not be as detrimental to one another as previously thought. This includes findings suggesting that these processes can occur without fork collapse and that RNA polymerase may only be transiently displaced during DNA replication. We discuss potential mechanisms by which RNA polymerase may be retained at the replication fork and quickly rebind to DNA post-replication. These discoveries are important, not only as new evidence as to how these two processes are able to occur harmoniously but also because they have implications on how transcriptional programs are maintained through DNA replication. To this end, we also discuss the coordination of replication and transcription in light of revising the current epigenetic bookmarking theory of how the active gene status can be transmitted through S phase.

DNA as a template for two biochemical processes

As initially described in the central dogma of molecular biology, DNA is the substrate for two critical biochemical processes: DNA replication and RNA transcription [1]. Since both of these processes require access to the DNA template, it has long been a question of what coordination is necessary in order to avoid conflicts between these two pivotal processes. Despite the theorized displacement of RNA polymerase during replication, the memory of whether a gene is active or repressed must be maintained through passage of the replisome in order for a cell to sustain its transcriptional output of lineage specific-genes during the cell cycle.

The cell cycle consists of four stages that cells pass through during a round of cellular division. Beginning in Gap 1 (G₁) phase cells then move into S phase wherein all of the DNA within the nucleus is duplicated during the process of DNA replication. Cells move from S phase into Gap 2 (G₂) phase, which is followed by mitosis in M phase. Cells then exit M phase back into G₁. Transcription was initially believed to be ceased during the whole several hours of S phase and take place only during the gap phases [2,3]. However, replication of a relatively short region from each replicon, which may contain the entire coding region of a gene, occurs relatively fast, and it is not clear what may prevent resumption of transcription of this gene within the continuing S phase. In line with this, more recent studies have shown that transcription does occur during S phase [4,5] and even M phase [6,7], despite the structural reorganization of chromatin during these phases of the cell cycle due to passage of the replisome [8,9] and mitotic condensation [10,11], respectively. During DNA replication, this requires a mechanism by which these two processes can occur together without causing widespread issues and genomic instability. In this review, we discuss various studies and theories about how these processes can occur harmoniously. We focus primarily on eukaryotic systems, since studies in prokaryotes have been extensively discussed elsewhere, and since bacteria may have their own approach to coordinating these processes, which largely involves orienting transcriptional units to be co-directional with replication [12–15]. Eukaryotic systems, however, contain chromatinized DNA and their gene expression profiles may vary vastly depending on the particular set of genes expressed in a particular cell lineage. Taken together, this suggests that coordination of these processes is more complicated in eukaryotic systems than in bacterial systems.

DNA replication

During S phase, DNA replication initiates asynchronously from origins of replication located throughout the genome, which are bound by the origin recognition complex (ORC) [16]. The ORC then recruits CDC6, CDT1, and the replicative DNA helicase, MCM2–7. GINS and CDC45 bind to MCM2–7 (forming the CMG helicase complex) which unwinds the DNA double helix creating two replication forks which move bi-directionally from one another [17]. The CMG complex subsequently recruits the replicative DNA polymerases [18,19]. Humans have 16 DNA polymerases [20], but the three that are critical during DNA replication are as follows: DNA polymerase alpha (Pol α), DNA polymerase delta (Pol δ), and DNA polymerase epsilon (Pol ε). DNA-dependent DNA polymerases require a double-stranded template in order to catalyze the addition of nucleotides, so the Pol α/primase complex creates an RNA primer on each of the two strands [21]. In addition to an RNA primer, Pol δ and ε both require a sliding clamp processivity factor, proliferating cell nuclear antigen (PCNA), in order to efficiently synthesize DNA [22]. PCNA is a ring-shaped homotrimer that is loaded onto DNA by the clamp loading complex replication factor C (RFC), which opens the PCNA ring and encloses it around DNA [23]. The complex comprising the CMG helicase, in addition to DNA polymerases, PCNA, RFC, and other critical replication proteins is often referred to as the replisome.

DNA replication involves incorporating nucleotide triphosphates into the nascent daughter DNA strand in a 5’ to 3’ direction. Because of the antiparallel nature of the DNA double helix, this means that the two strands at a replication fork must be synthesized in different manners. The leading daughter strand is synthesized continuously, whereas the lagging daughter strand must be synthesized in short stretches called Okazaki fragments which move in the opposite direction of progression of the replication fork [24,25]. This division of labor involves different DNA polymerases for each of the two strands with the leading and lagging strands being synthesized by Pol ε and Pol δ, respectively [26,27]. Okazaki fragments in eukaryotes are ~200 nucleotide stretches of nascent DNA [17], each of which begins with an RNA primer catalyzed by Pol α/primase. Once Pol δ reaches the primer of the previous Okazaki fragment, it displaces the RNA primer creating a nucleotide flap hanging off of the DNA which is then cleaved by flap endonuclease 1 (FEN1) [28]. The two Okazaki fragments are then stitched together by DNA ligase I [29]. The coordinated synthesis of the leading and lagging daughter strands allows for faithful duplication of the entire genome within a cell.

RNA transcription

Eukaryotic cells contain three nuclear DNA-dependent RNA polymerases: RNA polymerase I (Pol I), RNA polymerase II (Pol II), and RNA polymerase III (Pol III) [30]. Pol I is responsible for transcription of the 47 S pre-ribosomal RNA (rRNA) which is cleaved to form the 18 S, 5.8 S, and 28 S rRNAs [31,32]. Pol III transcribes all transfer RNAs (tRNAs), the 5 S rRNA, and numerous other small RNAs including several small nuclear RNAs (snRNAs) and small nucleolar RNAs (snoRNAs) [28,33]. Pol II transcribes all other RNAs, including messenger RNAs (mRNAs), as well as numerous types of non-protein coding RNAs (ncRNAs) including micro RNAs (miRNAs) [34], piwi-interacting RNAs (piRNAs) [35], long noncoding RNAs (lncRNAs) [36], snRNAs, and snoRNAs [37]. Importantly, Pol I and Pol III transcribe the same genes regardless of cell lineage because expression of these genes is required by all cells in an organism. Pol II, on the other hand, plays a pivotal role in differential gene expression and must be properly directed to and maintained at target gene loci that are specific to a particular cell type [38–41]. This makes Pol II critical when it comes to understanding epigenetic inheritance because all epigenetic programs must converge on faithful restoration of Pol II at its target loci throughout the cell cycle.

The core Pol II holoenzyme is a 12-factor complex, which contains RPBs 1 through 12 (including several homodimers of these products). The largest components, RPB1 and RPB2, form a cleft containing the active enzymatic domain of the complex responsible for incorporation of ribonuclear triphosphate precursors into the growing nascent RNA strand [42,43]. Additionally, RPB1 contains a heptad-repeat in the carboxy-terminal domain (CTD) with 26 repeats in yeast and 52 repeats in humans, whose consensus sequence is Tyr1-Ser2-Pro3-Thr4-Ser5-Pro6-Ser7 [44]. Although the CTD of Pol II contains multiple residues capable of being phosphorylated, the two residues most well characterized are particularly critical for the transcription cycle: Ser5 and Ser2, whose phosphorylation corresponds to transcriptional initiation and elongation, respectively [45,46]. In order to initiate transcription, numerous general transcription factors (GTFs) must be assembled at the gene promoter in a particular order, ultimately culminating in the formation of an active transcription bubble [47–50].

In vitro assessment of replication and transcription coordination

Some understanding of what happens when replication and transcription collide is based on in vitro experiments. These studies have primarily been carried out utilizing bacterial and viral systems. The question of what happens when RNA polymerase encounters DNA replication machinery was initially addressed by the Alberts group in the early 1990s. They showed that in a reconstituted in vitro system consisting of the T4 bacteriophage replisome and the E. coli RNA polymerase, both head-on and co-directional collisions of these enzymes did not displace RNA polymerase or cause it to stop synthesizing its nascent transcript [51–53]. This observation is supported by an in vitro investigation of collisions between the B. subtilis RNA polymerase and the phage Φ29 DNA polymerase, which showed that during head-on collisions, the enzymes are able to bypass one another as long as the transcriptional complex was not stalled [54]. Using the same system, Elías-Arnanz and Salas found that co-directional collisions (with the Φ29 DNA polymerase trailing RNA polymerase) resulted in a decreased rate of replication but not displacement of either complex from DNA [55].

More recent studies utilizing a different reconstituted in vitro system showed that the E. coli replisome is able to displace the T7 bacteriophage transcriptional machinery [56,57]. Another interesting in vitro study looked at the consequence of T7 RNA polymerase encountering the ORC, MCM double hexamer, or the ORC bound to MCM-CDT1 and found that the polymerase could push these complexes down the DNA, and even frequently bypassed them [58,59]. This indicates that RNA polymerase is capable of bypassing these replication proteins.

All of these studies, while biochemically informative, utilized reconstituted in vitro systems, which lack full cellular biological context. Furthermore, they focus on bacterial and viral enzymes, which may have no connection to how eukaryotic systems function, particularly because of the need to overcome chromatin barriers present on eukaryotic DNA. Additionally, members of the eukaryotic replication fork are continuing to be discovered [60], which could play a crucial role in resolving interactions between RNA and DNA polymerases. Thus, any in vitro system would likely lack the full complement of molecules that normally function in DNA replication. Given both the conflicting nature of these in vitro studies and their limitations, it is difficult to extrapolate their results on in vivo coordination of these processes, especially in higher eukaryotes.

Placement of origins near active gene promoters/order of replication

The S phase is often divided into early and late S phase. It has been well established that actively transcribed genes are replicated early in S phase, whereas repressed genes are replicated in late S phase [61–64]. Subsequent studies showed that origins of replication (ORIs), which fire during early S phase, are close to the promoters of actively expressed genes [65,66], as was initially demonstrated for several genes, including MYC (c-MYC) [67] and HBB (β-globin) [68]. This indicates not only that there is something about these regions that makes them more prone to firing early during replication but also that the selection of which ORIs fire and when they fire during S phase is dependent on the particular cell type [69,70].

This coordination between ORIs selected to fire and the transcriptional status of genes also indicates that there is some biological advantage to replicating these regions early in S phase. It has been shown that transcription factors associated with Pol II and Pol III can positively regulate DNA replication by their association with the budding yeast autonomously replicating sequence (ARS) [71,72]. While higher eukaryotes do not have a consensus sequence for ORC binding, data suggest that the ORC requires a nucleosome-free region (NFR) in order to efficiently load the MCM2–7 helicase complex [73]. Taken as a whole, sequences to load the ORC vary between eukaryotes, and do not appear to exist in metazoans, which likely require specific chromatin structures for ORC loading [74], a property that might be advantageous to coordinating replication with transcription in multicellular organisms. Overall, the existence of ORIs near active promoters implies a coordination between replication and transcription and that this coordination can be tailored to particular cell lineages and transcriptional profiles in higher order eukaryotes.

Transcription-replication conflict theory

Given that both transcription and replication require access to the same regions of DNA, understanding what happens when and if they come into contact with one another has been an active field of inquiry. Transcription-replication conflicts (TRCs) have been a main theory in the field and have been extensively covered in other review articles [12,75–80]. This theory holds that these processes are inherently disruptive to one another and collisions between these machineries can cause a number of outcomes including fork stalling, fork reversal, and/or formation of an R-loop [81], which is a RNA-DNA hybrid duplex, in addition to the other displaced DNA strand.

Resolution of TRCs ultimately requires removal of either Pol II or the replication machinery in order for replication to resume. It is of note, though, that these studies generally rely on stressing one or the other process in order to determine what happens when transcription and replication machinery collide due to using the same DNA template. Specifically, they use exogenous stressors including pharmacologic inhibition of replication [82,83], or transcription [84,85], or genetic disruption of the endogenous replication [86,87] or transcription [88,89] processes. This is an important consideration, because it suggests that these conflicts may not arise in cells where transcription and replication are occurring unimpaired. In addition, a recent study showed that these types of perturbations result in firing of non-canonical ORIs, which can lead to the resulting phenotype of fork stalling and/or collapse [90]. This suggests that the TRC phenotype observed may not be the result of normal transcription and replication.

Reports on the collisions between DNA and RNA polymerases have also been suggested to result in DNA replication that occurs outside of S phase of the cell cycle. Initial studies found that following DNA damage, replication could occur during mitosis (MiDAS) [91,92] and it was subsequently shown that normally cycling cells may not replicate some transcriptional start sites (TSSs) until the cell reaches G2/M phase of the cell cycle, in a process termed G-MiDAS [5]. While striking that normally dividing cells were replicating their DNA after S phase, the authors only observed this in one-fifth of their cell population, and of those cells it only occurred at approximately 400 TSSs. Thus, it does not appear likely that this is a prominent feature due to collisions between DNA and RNA polymerases even in stressed cells.

Another interesting investigation of TRCs is from a study on fragile site instability in long genes [93]. This study showed that not only are some particularly long genes transcribed through the entire S phase but that completing their transcription takes even longer than the entire length of the cell cycle in human B-lymphoblasts. This indicates that Pol II must finish transcribing these long genes following passage of the replisome. These observations led the authors to propose that there must be some mechanism to recruit Pol II bound to its immature RNA transcript back to these long genes in order to finish transcribing them, and the authors hypothesized that the immature transcript might function to recruit Pol II back to DNA [94]. Additionally, a study of the very long Ubx gene in Drosophila embryos, which have very short cell cycle, showed that its transcription begins shortly after the completion of mitosis and ends when the next mitotic phase begins [95]. Interestingly, another study also suggests that RNA may not be dissociated during DNA replication by use of an in vivo approach which knocked in the MS2 protein-binding sequence into the 3’ UTR of CCND1 in the presence of MS2-GFP expression, allowing visualization of the transcription dynamics of a single allele of a gene in real time [96]. When the gene was replicated the authors did not observe a loss of the mRNA signal during replication and instead saw the signal duplicate, suggesting that RNA, potentially in complex with Pol II, is transferred to both two daughter strands [96].

These studies imply that transcription of such long genes has to be only transiently interrupted by DNA replication. Taken together, these studies suggest that Pol II has the ability to recognize the unfinished nascent RNA transcript and complete its synthesis after passage of the replication machinery. These results imply a model of close coordination of replication and transcription, rather than TRCs, wherein Pol II must be removed from chromatin in order for replication to proceed.

Spatial separation of transcription and replication in factories

Contrary to the theory that replication and transcription machineries collide, numerous studies investigating these processes suggest that each occurs in distinct nuclear domains, termed replication factories and transcription factories, respectively [97–101]. This suggests a coordination of these two processes by physically separating them within the nucleus. The factory theory originates from studies using immunofluorescence of fixed samples where staining for transcription and replication proteins resulted in discrete foci throughout the nucleus [102–104]. However, rarely these foci may overlap, suggesting that transcription and replication may occasionally come into spatial or physical contact with one another, and that these processes may not occur entirely independently of one another [105].

These foci are termed factories because they contain multiple polymerases, on average eight for transcription factories [106] and up to 40 replication forks with numerous polymerases each within replication factories [107]. A principle component of the factory theory that distinguishes it from other conceptualizations of replication and transcription is that it posits that the enzymes are immobilized and DNA is spooled through them, as opposed to the enzymes traversing down DNA. This has long been an accepted theory for DNA replication [108–110]; however, it has only more recently become suggested for transcription [111,112]. Thus, conflicts between replication and transcription could only occur if the two are pulled together and collide.

There are several pieces of evidence supporting the factory model. For one, early data showed that components of the replisome associate with the nuclear lamin, indicating that these proteins are physically affixed to this nuclear scaffold [97,108,113]. Moreover, transcription factories have shown to remain even after the removal of chromatin, indicating that these structures are also bound to something within the nucleus apart from chromatin [113,114]. Second, in vitro studies have shown that both DNA polymerase [115] and RNA polymerase [103,116,117] are capable of spooling their DNA template through them as they catalyze their respective biochemical reactions. Thus, it is possible that replication and transcription take place within factories of immobilized enzymes, which would suggest one way to avoid direct physical collisions between the corresponding protein complexes.

Current theories of epigenetic bookmarking during S phase

Up to this point we have discussed the coordination of replication and transcription in a biochemical context, but it is important to also consider this from a conceptual standpoint. Maintenance of transcriptional programs through DNA replication is important for transmission of the status of genes as active or repressed. Since it is believed that transcription machinery leaves DNA during replication, this is thought to require epigenetic bookmarking as the replisome passes over differentially expressed regions of the genome. There are currently two main epigenetic theories during S phase. The predominant theory, is the reader-writer histone epigenetic hypothesis, where in modified histones are transferred during DNA replication from nucleosomes on the parental DNA to the newly formed nucleosomes on the two daughter strands (Figure 1(a)) [9,118]. These modified histones are able to re-recruit histone-modifying complexes via histone reader domains within the complexes, and these complexes then deposit modifications onto newly formed nucleosomes via histone writer domains. Recruitment of these complexes and propagation of histone modifications leads to the restoration of the chromatin landscape throughout the genome after DNA replication and is thought to be essential for recruitment of Pol II and other components of the transcriptional machinery to the promoters of active genes.

Figure 1.

Proposed models of epigenetic marking of active genes during DNA replication.

(a) Reader-writer epigenetic theory. During DNA replication, the replisome displaces most chromatin-bound proteins, including Pol II and chromatin-modifying complexes, such as TrxG proteins. Modified histones from parental nucleosomes are transferred semi-conservatively to nascent DNA into the newly assembled nucleosomes on daughter strands. Chromatin-modifying complexes are recruited to replicated DNA by interacting with these modified histones through “reader” protein components of these complexes. Deposition of active histone marks by “writer” protein components on all newly incorporated histones completes chromatin restoration and culminates in recruiting Pol II back to active genes.

(b) Chromatin-modifier epigenetic theory. During DNA replication, chromatin-modifying complexes are transferred from parental DNA to newly synthesized DNA, but modified histones are displaced. Unmodified histones are incorporated into nascent DNA and are post-translationally modified by the retained chromatin-modifying complexes in active regions. Once the structure of chromatin is restored, Pol II is recruited to replicated DNA at active genes in order to restore their transcriptional status.

(c) Pol II epigenetic theory. During DNA replication, Pol II in complex with immature RNA is retained close to the replisome via protein–protein interactions with PCNA, allowing it to rebind to nascent DNA immediately following passage of DNA polymerase. This occurs both at promoters and in the gene body. While other factors are likely bound to nascent DNA during this timeframe, potentially including nucleosomes and chromatin-modifying complexes, the presence of Pol II on DNA immediately after replication alleviates the proposed necessity of any other factor in order to epigenetically bookmark active genes. Therefore, nucleosomes and chromatin-modifying complexes were not included in this model.

The second theory, which is better supported by experimental data, holds the converse, namely, that chromatin-modifying complexes are transferred during DNA replication and are able to deposit their marks on unmodified histones that are loaded onto newly synthesized DNA during replication (Figure 1(b)) [119–123]. For active genes, this means the transfer of trithorax group (TrxG) proteins to nascent DNA which can then deposit active histone marks on newly deposited histones. These findings questioning whether modified histones are necessary for transmitting epigenetic information during S phase were further supported by a study showing that modified histones in active regions are not maintained during DNA replication [124]. Current thinking in the field has shifted such that, at least in active regions, modified histones are not believed to be the mediators of epigenetic bookmarking through replication [125–127]. Based on this, active regions are marked by factor(s) other than modified histones, potentially including chromatin-modifying complexes, as suggested [119,120].

An additional important caveat to the reader-writer theory is a question of whether modified histones in active regions are the drivers of gene activation or are the result of transcription. It appears that H3K4me3, the primary modified histone associated with active gene promoters, may be the result of transcription rather than the cause of transcription [128]. Moreover, the loss of H3K4me3 does not impact transcription levels [129,130], further undercutting its putative role as an epigenetic mark. A more recent study showed that rapid depletion of SET1/COMPASS caused loss of H3K4me3 at gene promoters without impacting transcription initiation but did cause a decrease in transcription elongation [131]. These findings are in line with other evidence that suggests methylation of H3K4 and H3K79 is the result of transcription, not the driver of it [132]. The authors of this study show that the transcription elongation component PAF-1 recruits SET1/COMPASS and DOT1P, which in turn methylate H3K4 and H3K79, respectively. Taken as a whole, it appears that chromatin modifications in active regions are not likely to be marks which drive transcription, but rather are deposited as the result of transcription by Pol II. Moreover, it is well established that in very early development of Drosophila, cells lack major histone methylation marks, including H3K4me3 and H3K27me3 [119,133], yet transcription begins during early development, suggesting that these marks are not required to initiate transcription. Thus, it is questionable as to whether modified histones are capable of bookmarking active genes during DNA replication.

Epigenetic bookmarking by RNA Pol II during S phase

Both of these theories indicate that certain factors are required to bookmark chromatin during replication in order to restore the chromatin landscape, which will ultimately dictate which genes are repressed or active and therefore where Pol II should bind. Importantly, both theories assume that most proteins, including Pol II, are evicted by the replisome and must be recruited back to replicated DNA at a much later time point. This may not be the case given some of the in vitro data discussed above. Moreover, two recent studies in budding yeast have suggested that Pol II can rebind DNA shortly following replication [134,135]. The first study integrated knowledge of the order of ORI firing in budding yeast along with Pol II chromatin immunoprecipitation followed by sequencing (ChIP-seq) datasets to speculate that Pol II binds to both promoters and within gene bodies within 3 min following passage of the replisome [134]. However, given their approach, they were only able to postulate that Pol II displacement is transient within this system without fully confirming that theory. The second study in budding yeast adapted a technique called nascent chromatin avidin pull-down (NChAP) [136] by combining it with chromatin immunoprecipitation (ChIP). This approach, called ChIP-NChAP involves EdU labeling of nascent DNA followed by ChIP for a protein. The immunoprecipitated material is then subjected to click chemistry to conjugate biotin to EdU and is subsequently captured on streptavidin and analyzed by next-generation sequencing. Using ChIP-NChAP for Pol II the authors speculate that Pol II is able to quickly bind to replicated DNA [135]. Importantly, ChIP-NChAP is limited by a minimum of 20 min EdU labeling times and cannot assess events that happen immediately during passage of the replication fork. Nevertheless, both of these studies provided evidence suggesting that in budding yeast Pol II may only be removed from DNA during replication for a short period of time.

Similar results were obtained in a study in mouse embryonic stem cells, which showed that initiating Pol II S5P is absent from nascent DNA in 10 min, but rebinds DNA within 30 min after replication by chromatin occupancy after replication (ChOR-seq), an approach very similar to ChIP-NChAP [137]. A more recent study in human cells showed that total Pol II and Pol II S2P bind to DNA within 15 min of replication using ChOR-seq [138,139]. Given the inherent limitations of these techniques based on the minimal timings of EdU labeling, ChIP-NChAP and ChOR-seq are mostly informative to assess chromatin maturation from 15 to 20 min onward post-replication. Thus, they cannot be used to track proteins on DNA immediately following replication.

To assess this principal uncertainty in association of Pol II with the immediate nascent DNA, we recently investigated the timing of Pol II re-association with DNA after replication in human cells [139–141]. For this study, we utilized several techniques to assess events happening immediately following passage of the replication machinery. Using our previously developed chromatin assembly assay (CAA) [119], in which nascent DNA is EdU-labeled and subsequently conjugated with biotin and then a proximity ligation assay (PLA) is performed between biotin and a protein of interest, we found that initiating and elongating Pol II bind to DNA within 5 min of replication in human cells [140]. We then adapted our CAA approach to multiplex it, wherein two PLA reactions are performed within the same cell, in this case between biotin and Pol II in parallel with biotin and a replication protein, such as PCNA (both strands), FEN1 (lagging strand), or Pol ε (leading strand). Measuring the distance between foci in our multiplex CAA allowed us to determine that initiating and elongating Pol II bind nascent DNA within several hundred bp behind the replication fork on both the leading and lagging strands. These results were confirmed by a sequential ChIP (re-ChIP) approach [119], where Pol II ChIP material was re-immunoprecipitated using antibodies against replication proteins PCNA, FEN1, or Pol ε. These assays detected Pol II at promoters and within the gene body on both the leading and lagging strands within 100–200 bp after passage of the replication machinery. The same results were obtained with several general transcription and elongation factors, suggesting that both the initiating and elongating transcriptional complexes may quickly re-associate with nascent DNA [140].

In the same work, by using the only currently available method to experimentally assess resumption of transcription following DNA replication, a PLA-based approach called the RNA-DNA interaction assay (RDIA) [4,142], we also found that new transcription begins within 20 min after DNA replication. Surprisingly, using RDIA and adapting multiplex CAA, we also found that immature transcripts in the process of being synthesized before replication remain bound together with Pol II through DNA replication [140].

Taken as a whole, these results indicate that RNA-bound Pol II and other general transcription factors may only transiently be displaced from DNA during replication and rebind nascent chromatin throughout the gene body immediately following passage of the replisome (Figure 1(c)) [140]. Additional results of this study also suggested that new RNA synthesis may resume by completing synthesis of immature RNAs, as was implied by previous studies [93–96]. This indicates that a widely held assumption that the transcriptional complex may need some additional epigenetic marks during replication may not necessarily be the case.

Association of general and lineage specific transcription factors with nascent DNA

Recruitment of Pol II to a gene promoter involves the coordination of lineage-specific transcription factors (TFs) in addition to assembling the pre-initiation complex which consists of numerous GTFs [143]. In addition to Pol II, it is important to understand what is happening with these factors during DNA replication. Our data showed that the TATA-binding protein (TBP), a core GTF, is bound to DNA within several hundred bp of replisome passage [140]. We also detected other transcriptional components, including CDK9 and Cyclin T (components of the pTEF-b complex) as well as SPT4 and SPT5 (which form the DSIF complex) bound to nascent DNA within this timeframe [140]. Takentogether, this suggests that the core transcriptional apparatus as a whole may not be displaced during DNA replication.

Data regarding the presence of TFs on nascent DNA, however, are more limited, and whether these factors bind immediately after passage of the replication machinery remains to be fully understood. Current knowledge of these factors is limited to various DNA labeling-based approaches. For example, using nascent ChIP-seq (nasChIP-seq), an approach analogous to ChOR-seq, it was shown that the boundary element protein CTCF binds to DNA within 20 min of replication in human embryonic stem cells [144,145]. Additionally, proteomic assays, such as NCC (nascent chromatin capture) and adaptations of iPOND (isolation of proteins on nascent DNA) followed by mass spectrometry (MS), have been utilized to assess the protein composition of nascent chromatin shortly after DNA replication. Both NCC and iPOND label and capture nascent chromatin, and studies using these assays have shown that various TFs, along with some GTFs, core Pol II components and Pol II accessory factors are bound to nascent DNA labeled from 5 to 15 min [60,146–149]. This suggests that TFs and GTFs rebind DNA shortly after replication, although the precise kinetics of their binding remains to be determined.

Whether TFs are bound to nascent DNA has also been assessed through the use of various nuclease accessibility assays combined with the knowledge of where these TFs bind. These assays utilize micrococcal nuclease (MNase) in order to ascertain whether an area is protected from digestion, which would suggest it is bound by a nucleosome or TF based on the size of the protected fragment. By pairing this strategy with EdU labeling, biotinylation, and streptavidin capture, the structure of nascent chromatin has been investigated in yeast, Drosophila, and mouse ESCs [134,150–152]. These strategies gave varying results of whether factors are bound or not to nascent chromatin. These methods rely on assumptions about fragments protected from MNase digestion rather than directly assessing protein association with DNA, and they are limited by EdU labeling. Therefore, other methods, which allow assessing events occurring much closer to the replication fork, such as multiplex CAA and re-ChIP [119,140], need to be used in order to fully assess what is happening to TFs during DNA replication.

Potential mechanisms of Pol II retention on nascent DNA

In order for Pol II to rebind DNA after passage of the replisome, it likely needs to interact with component(s) of the replication machinery to facilitate its retention. We found that Pol II S5P and S2P interact with PCNA, suggesting a way to keep Pol II close to DNA during replication [140]. These results were functionally validated by affecting PCNA trimerization [140]. Another paper confirmed the existence of this interaction between PCNA and Pol II and their data suggest that it is mediated by the RPB1 component of Pol II [153]. Interestingly, other interactions between Pol II and replication proteins have been observed. The Pol II holoenzyme interacts with several MCM proteins and this interaction is mediated by the CTD of RPB1 [154,155]. Furthermore, MCM proteins were shown to interact with the Integrator complex, suggesting additional links between these helicase components and transcription proteins [89]. Interaction between MCM7 and the yeast TFIIH complex have also been reported [156]. Additionally, hyperphosphorylated Pol II interacts with the leading strand DNA polymerase, Pol ε, but not Pol δ or Pol α [157]. Taken as a whole, these two complexes interact via numerous components of each.

The mechanism of retention and transcription resumption midway through the gene is still not fully resolved. In order for the resumption of transcription to occur, as suggested for some genes [93,94], Pol II must reengage DNA and form a transcription bubble. The mechanism for this is unclear, but potentially involves recruitment of helicase(s), such as RECQ5, Senataxin (SETX), and/or RTEL1, which have all been observed to function during DNA replication. RECQ5 has been shown to be both associated with Pol II during transcription, as well as during DNA replication [158–161]. Moreover, RECQ5 has been implicated for its role in resolving TRCs, by promoting PCNA ubiquitination and potentially allowing Pol II passage [162]. The RNA-DNA helicase SETX has also been reported to be involved in transcription elongation and to associate with replication forks [163,164]. SETX is believed to play a role in allowing for replication progression by acting on R-loops formed during transcription [165,166]. Finally, the helicase RTEL1, in complex with SLX4, associates with Pol II and to bind to nascent DNA [167], suggesting that it may also play a role in facilitating Pol II function on nascent DNA. Taken as a whole, any of these helicases may function in assisting Pol II to resume transcription after DNA replication.

Conclusions

Given the focus in the field on TRCs, our recent findings that transcription may only be transiently displaced by the replisome is striking. That being said, that Pol II bound to an immature transcript is transferred during replication is the only way to explain how some transcripts, such as long ones described by Helmrich et al. [93], or transcripts synthesized during short cell cycles in early development [95] are able to be completed. Moreover, ChOR-seq for Pol II S2P in human cells found that Pol II was bound in the distal regions of genes that take more than 15 min to transcribe, further suggesting that the elongating transcriptional apparatus is likely not ejected during DNA replication [138]. These conclusions also suggest a way in which higher level eukaryotes, are able to replicate DNA in a vast number of different cell types, each of which contains different transcriptional profiles and utilizes cell-specific ORIs.

Given our recent finding that in human cells Pol II is retained in close vicinity of the replication fork and quickly re-associates following passage of the replisome, several future questions need to be addressed. Numerous studies have indicated that replication proteins including PCNA, several MCM proteins, and DNA pol ε have been shown to interact with the RNA Pol II complex, and a better understanding of the nature of these interactions is needed, particularly the specific domains making contact. This would better explain the mechanism by which RNA Pol II is able to re-bind DNA immediately following passage of the replisome. Additionally, Pol II would need to reform the transcription bubble on nascent DNA and this would likely require a helicase. As discussed, there are several candidates, RECQ5, SETX, and RTEL1, whose function could be targeted pharmacologically or genetically in order to assess what happens to RNA Pol II detection on nascent DNA.

Further work will need to be done to understand the nature of these interaction and the detailed mechanism by which Pol II is able to be transferred during DNA replication. Given that RNA polymerase travels faster than DNA polymerase different scenarios might play out depending on whether the polymerases are positioned co-directionally or head-on, since RNA polymerase would approach a different component of replication machinery depending on the direction [141]. Besides the intricacies in interactions between transcription and replication complexes, such studies would provide new insights into a previously unknown mechanism of epigenetic bookmarking during S phase.

Funding Statement

This work was supported by NIH F31GM128300 to T.K.F. and NIH R01GM075141 to A.M.

Disclosure statement

No potential conflict of interest was reported by the author(s).