A non-catalytic role for RFC in PCNA-mediated processive DNA synthesis

SUMMARY

The ring-shaped sliding clamp PCNA enables DNA polymerases to perform processive DNA synthesis during replication and repair. The loading of PCNA onto DNA is catalyzed by the ATPase clamp loader RFC. Using a single-molecule platform to visualize the dynamic interplay between PCNA and RFC on DNA, we unexpectedly discovered that RFC continues to associate with PCNA after loading, contrary to the conventional view. Functionally, this clamp-loader/clamp (CLC) complex is required for processive DNA synthesis by polymerase δ (Polδ), as the PCNA-Polδ assembly is inherently unstable. This architectural role of RFC is dependent on the BRCT domain of Rfc1, and mutation of its DNA-binding residues causes sensitivity to DNA damage in vivo. We further showed the FEN1 flap endonuclease can also stabilize the PCNA-Polδ interaction and mediate robust synthesis. Overall, our work revealed that, beyond their canonical enzymatic functions, PCNA-binding proteins harbor non-catalytic functions essential for DNA replication and genome maintenance.

Graphical Abstract

In Brief

The RFC clamp loader remains associated with PCNA beyond the loading step, supporting processive DNA replication by stabilizing the inherently unstable PCNA-polymerase δ complex.

INTRODUCTION

Genome replication across all domains of life requires DNA sliding clamps that couple with DNA polymerases to achieve processive DNA synthesis^1-3. In eukaryotes, this activity is fulfilled by proliferating cell nuclear antigen (PCNA), a homotrimer that forms a ring-shaped complex encircling double-stranded (ds) DNA^4,5. The closed-ring architecture of PCNA creates a topological challenge for its loading onto DNA, and this challenge is overcome by a clamp-loader complex known as replication factor C (RFC), a heteropentameric ATPase⁶. In the prevailing model, RFC utilizes its ATPase activity to open the PCNA ring and deposit it on a primer-template junction with a 3’ recessed end; upon ATP hydrolysis, RFC is then released from DNA while PCNA is poised to interact with a polymerase for DNA synthesis⁷. However, RFC has also been shown to mediate PCNA loading onto nicked DNA substrates^8-11, suggesting the RFC-PCNA interaction is more nuanced than previously thought.

Besides RFC, PCNA also interacts with many other proteins that play diverse functions ranging from DNA damage repair to chromatin assembly and sister chromatid cohesion^12,13. Further, the homotrimeric architecture of PCNA in principle allows it to bind up to three partners simultaneously, leading to the proposal of a “toolbelt” model. In one rendition of this model, PCNA binds DNA polymerase δ (Polδ), flap endonuclease I (FEN1), and DNA ligase I (Lig1), three enzymes involved in Okazaki fragment maturation during lagging strand synthesis. Nevertheless, whether PCNA binds these proteins simultaneously or sequentially is still under debate^14,15. In general, how PCNA recruits, exchanges, and coordinates different proteins and their activities remains poorly understood.

In this study, we used single-molecule correlative force and fluorescence microscopy to visualize the dynamic behavior of PCNA and RFC on DNA. This approach led us to the surprising finding that these two complexes frequently remain associated on DNA even after PCNA loading. The sustained association of the PCNA-RFC assembly with the replication machinery is essential for processive DNA synthesis in vitro. Mutating the domain in RFC required for this association results in impaired genome maintenance in vivo. This work thus reveals an unexpected non-catalytic function of RFC, beyond its canonical clamp-loading activity, and thereby extends the possible functionality of clamp loaders.

RESULTS

PCNA and RFC form a long-lived complex that diffuses on dsDNA

We purified Saccharomyces cerevisiae PCNA and RFC (Figure S1A) and fluorescently labeled each of them with LD655 and Cy3 fluorophores, respectively. Through force manipulation of a 19-kilobase (kb) dsDNA with two pre-engineered nicks by optical tweezers, we generated a gapped DNA substrate containing a 3-kb single-stranded (ss) DNA region flanked by 10-kb and 6-kb dsDNA arms (Figures S1B and S1C). This gapped DNA, containing both 3’ and 5’ recessed DNA ends, was used to visualize the behavior of PCNA via scanning confocal fluorescence microscopy (Figure 1A). To mimic the in vivo scenario, we also included an excess of unlabeled RPA, the eukaryotic ssDNA binding protein. We were able to observe clear PCNA signals on the DNA in the presence of RFC and ATP (Figures 1B and S2A), whereas PCNA binding was completely abolished when either RFC or ATP was omitted (Figures S2B and S2C). These results are in agreement with the current understanding of the PCNA loading mechanism and thus validate our experimental platform.

Figure 1. Clamp-loader/clamp complexes bind and slide on dsDNA.

(A) Schematic of the experimental setup. A single DNA molecule containing a 3-kb ssDNA gap flanked by 10-kb and 6-kb dsDNA arms was tethered between a pair of optically trapped beads through biotin-streptavidin linkage. The tether was moved to a channel containing LD655-PCNA, Cy3-RFC, RPA, and ATP, and the behavior of PCNA and RFC was imaged via dual-color scanning confocal fluorescence microscopy.

(B) Two representative kymographs showing gapped DNA tethers incubated with 8 nM LD655-PCNA, 8 nM Cy3-RFC, 100 nM RPA, and 4 mM ATP. The red laser was turned off briefly to confirm the presence of the RFC fluorescence signal. Arrows denote the events when PCNA was engaged with the 3’ recessed end of the ssDNA gap.

(C) Distribution of the dwell times of PCNA binding events at the 3’ recessed end of the ssDNA gap. Bars represent mean and SEM (N = 61 events from 6 independent tethers).

See also Figures S1 and S2.

Surprisingly however, our single-molecule data revealed a more complex picture of the PCNA behavior on DNA than the prevailing model: rather than stably residing at 3’ recessed DNA ends, PCNA was observed to transiently engage with the 3’ end (white arrows in Figure 1B), yielding an average dwell time of 5.4 ± 0.9 s (mean ± SEM) (Figure 1C). Most of the time, PCNA stayed on either of the flanking dsDNA arms and displayed a one-dimensional sliding behavior (Figures 1B and S2A). No PCNA signal was observed within the ssDNA gap region, indicating that RPA-coated ssDNA serves as an efficient barrier against PCNA binding and sliding.

The other striking and unexpected observation from our data was that PCNA and RFC frequently remained associated on DNA and slid together as a complex, evidenced by diffusive trajectories containing both Cy3-RFC and LD655-PCNA signals (Figures 1B and S2A). This observation stands in contrast to the widely held notion that RFC is immediately released from DNA upon PCNA loading. We hereafter refer to these long-lived RFC-PCNA assemblies as Clamp-Loader/Clamp (CLC) complexes.

CLCs can be topologically bound to DNA

With the gapped DNA substrate, we frequently observed that the dual-color CLC signal first appeared within a dsDNA region (e.g. Figures 1B and S2A) rather than at a recessed 3’ or 5’ DNA end. To test whether CLC binding to DNA relies on recessed DNA ends at all, we performed single-molecule experiments using DNA tethers derived from the 48.5-kb bacteriophage λ genomic dsDNA, which lacks ssDNA gaps. We observed that CLC complexes can readily bind and slide on this substrate in an RFC- and ATP-dependent manner (Figures 2A, S2D, and S2E), a fraction of which presumably entered via nicks that may form during purification of the long duplex DNA. Thus, PCNA does not require a recessed end for DNA binding, consistent with previous biochemical data showing that RFC can load PCNA at a nicked site^8-10. Of note, we found that CLC complexes exhibited a longer residence time at nicks compared to on dsDNA regions where they slid rapidly (Figure S2F). We also found that RFC alone can also slide on dsDNA in the presence of ATP (Figure S2G, S2H).

Figure 2. Clamp-loader/clamp complexes can be topologically bound to DNA.

(A) Representative kymograph showing a double-stranded λ DNA tether incubated with 8 nM LD655-PCNA, 8 nM Cy3-RFC, 100 nM RPA, and 4 mM ATP. Red laser was turned off briefly to confirm the presence of green RFC signals.

(B) Kymograph showing a λ dsDNA tether first incubated in a channel containing 8 nM LD655-PCNA, 8 nM Cy3-RFC, 100 nM RPA, and 4 mM ATP, and then moved into a separate channel containing a high-salt buffer (500 mM NaCl) with ATP but no free proteins.

(C) Fraction of clamp-loader/clamp complexes (CLCs) that remained bound to or fell off dsDNA when challenged with high salt (N = 54 from 17 independent tethers). Error bars represent the 95% confidence interval using bootstrapping.

See also Figure S2.

Next, we asked whether these long-lived CLC complexes contain a topologically closed or open PCNA ring. To this end, we challenged CLC complexes bound to the λ DNA with 500 mM NaCl and found that a substantial fraction of them (~50%) persisted on DNA and continued to slide in this high-salt buffer (Figures 2B and 2C). We propose that this high-salt-resistant population of CLC complexes corresponds to those containing a topologically closed PCNA ring encircling the dsDNA.

CLC formation requires the BRCT domain of Rfc1

The BRCA1 C-terminal homology (BRCT) domain on the Rfc1 subunit, a region flexibly attached to RFC’s main AAA+ domain core, makes extensive contacts with dsDNA (Figure 3A). Therefore, we considered whether it may contribute to the persistent association of the CLC with DNA after PCNA loading. To test this, we purified and fluorescently labeled an RFC complex lacking the Rfc1 BRCT domain (RFC^ΔBRCT) (Figure S1A) and examined its behavior with PCNA on the gapped DNA substrate. We found PCNA can be efficiently recruited to DNA by RFC^ΔBRCT and then undergo diffusive movements on either the 5’ or 3’ dsDNA arm of the ssDNA gap (Figure 3B). However, in contrast to the full-length RFC (Figure 1B), RFC^ΔBRCT was much less frequently observed to colocalize with PCNA (Figure 3C), suggesting that it was quickly evicted from DNA after loading PCNA. The lack of RFC^ΔBRCT fluorescence signals was not due to a low labeling efficiency, as LD555-RFC^ΔBRCT can be readily detected within the RPA-coated ssDNA region (Figure S2I). Of note, RFC^ΔBRCT binds ssDNA in a largely static manner, in contrast with full-length RFC, which binds minimally to ssDNA but exhibits a diffusive behavior on dsDNA (Figures 1B and S2A). Moreover, we found that PCNA loaded by RFC^ΔBRCT exhibited a much greater diffusion coefficient than those loaded by full-length RFC (Figure 3D), which may be explained by the difference in molecular weight between CLC and PCNA alone or their different modes of DNA interaction. On the other hand, the diffusion coefficient for RFC alone is similar to that for CLC (Figure S2H), suggesting that the RFC-DNA contacts, possibly via the BRCT domain, exert a dominant effect on CLC’s diffusivity. Together, these results show that Rfc1’s BRCT domain is critical for the stable formation of CLCs on DNA.

Figure 3. The BRCT domain of Rfc1 promotes CLC formation on DNA.

(A) Domain structure of Rfc1 and Rfc1^ΔBRCT. The BRCT domain is highlighted in green.

(B) Two representative kymographs showing a gapped DNA tether incubated with 8 nM LD655-PCNA, 8 nM LD555-RFC^ΔBRCT, 100 nM RPA, and 4 mM ATP. The PCNA trajectory appears discontinuous because of its fast movement on DNA and the finite confocal line scanning rate. Inset shows a portion of the extracted PCNA trajectory using our line-tracking algorithm.

(C) Fraction of the trajectories on gapped DNA containing CLCs or PCNA only when RFC (N = 46 from 6 independent tethers) or RFC^ΔBRCT (N = 33 from 8 independent tethers) was used. Error bars represent 95% confidence intervals using bootstrapping.

(D) Diffusion coefficients (D) for PCNA trajectories sliding on dsDNA when loaded by RFC (N = 22 from 5 independent tethers) or RFC^ΔBRCT (N = 6 from 6 independent tethers). Bars represent mean and SEM. Significance was calculated using a two-tailed unpaired t-test with Welch’s correction.

See also Figures S1 and S2.

CLC assembles with Polδ to facilitate DNA synthesis

We next explored the functional relevance of the CLC complex during DNA synthesis. To this end, we included Polδ, the DNA polymerase mainly responsible for lagging strand synthesis, in our single-molecule assay using the gapped DNA substrate. This allowed us to observe DNA fill-in synthesis by Polδ across the ssDNA gap in real time. We first performed this assay using AlexaFluor488-labeled RPA in a distance-clamp mode where the positions of the two traps were fixed. In the presence of PCNA, RFC, Polδ (all unlabeled), and dNTPs, we observed progressive clearance of the RPA signal from the 3’ to the 5’ end of the gap (Figure 4A), confirming that the proteins used were active. To visualize the behaviors of PCNA and RFC during the fill-in synthesis, we used unlabeled RPA and instead followed the reaction from the force readings (Figure 4B). The conversion from ssDNA to dsDNA causes tether contraction at a starting force above 6 pN and thus a concomitant increase in the DNA tension¹⁶. We monitored the fluorescence signals from both PCNA and RFC and found that they frequently remained associated and traveled together continuously across the gap as the fill-in synthesis progressed (Figure 4B).

Figure 4. Real-time observation of fill-in synthesis by Polδ.

(A) (Top) Schematic of the experimental setup. A single gapped DNA molecule was tethered between a pair of optically trapped beads held at a constant distance. The tether was moved to a channel containing Polδ, PCNA, RFC, and AF488-RPA with ATP and dNTPs. The RPA signal was monitored using scanning confocal fluorescence microscopy. (Bottom) Kymograph showing the clearance of RPA during fill-in of the ssDNA gap when incubated with 20 nM Polδ, 5 nM PCNA, 5 nM RFC, 20 nM AF488-RPA, 4 mM ATP, and 130 μM dNTPs.

(B) (Top) Schematic of the experimental setup where the distance between beads was held constant at a starting force of 9 pN, and fill-in was monitored by an increase in the force as the tether shortened. (Bottom) Kymograph and associated force changes showing fill-in of a gapped DNA tether incubated with 20 nM Polδ, 5 nM LD655-PCNA, 5 nM Cy3-RFC, 100 nM RPA, 4 mM ATP, and 130 μM dNTPs. Red laser was turned off briefly to confirm the presence of green RFC signals. PCNA and RFC fluorescence channels are also separately shown to confirm the presence of each signal. Green box denotes the time period prior to the start of fill-in, and purple box denotes the completion of fill-in.

To circumvent complications with data interpretation due to the varying forces, we then conducted the experiments in a force-clamp mode where one of the trap positions was adjusted via feedback throughout the reaction to maintain constant DNA tension. Changes in the trap position were converted to the number of nucleotides filled in to track the progression of DNA synthesis (Figure 5A). Importantly, RFC was observed along the path of fill-in together with PCNA in the majority of reactions (9 out of 14 tethers). These results obtained from different data collection modes demonstrate that CLCs frequently assemble with Polδ during DNA synthesis.

Figure 5. Stable CLC formation promotes processive DNA fill-in synthesis.

(A) (Top) Schematic of the experimental setup where fill-in was monitored by the movement of one trap relative to the other via a feedback mechanism to maintain a constant force on the tether. (Bottom) An example kymograph and the associated trap position as a function of time at a constant force of 8 pN showing fill-in of a gapped DNA tether incubated with 20 nM Polδ, 5 nM LD655-PCNA, 5 nM Cy3-RFC, 100 nM RPA, 4 mM ATP, and 130 μM dNTPs. Red and green lasers were turned off briefly. PCNA and RFC fluorescence channels are also separately shown to confirm the presence of each signal. Green box denotes the period prior to the start of fill-in, and purple box denotes the completion of fill-in.

(B) An example for the same assay as in (A) except RFC^ΔBRCT was used instead of full-length RFC. Pink boxes denote inactive periods when fill-in synthesis stalled. Lefthand inset shows zoom-in of an instance where PCNA left the 3’ end, which coincided with one of the inactive periods. Righthand inset shows an instance where PCNA diffused back to the 3’ end, which coincided with restart of the synthesis.

See also Figures S3 and S4.

The finding that RFC remains part of the replication machinery begs the question of whether it contributes to DNA synthesis. To answer this question, we took advantage of the RFC^ΔBRCT variant, which we showed above to be proficient in loading PCNA onto DNA but deficient in forming CLC complexes. We performed the single-molecule fill-in assay with RFC^ΔBRCT in the force-clamp mode. In striking contrast to full-length RFC, RFC^ΔBRCT was never observed to co-travel with PCNA along the DNA fill-in path (0 out of 20 tethers). PCNA alone was still observed to translocate across the ssDNA gap but frequently slid back onto the dsDNA (Figures 5B and S3A). Interestingly, whenever PCNA left the path of synthesis (i.e. 3’ end of the filled gap), we observed a concomitant pause in the synthesis activity (pink boxes in Figures 5B and S3A), confirming that PCNA presence at a 3’ terminus is required for processive Polδ activity.

We quantified the fill-in synthesis activity in the presence of RFC^ΔBRCT or full-length RFC. The fraction of time that the fill-in trajectories spent in an inactive period was significantly higher for RFC^ΔBRCT than for full-length RFC (Figure 6A). Moreover, a substantial fraction of the fill-in trajectories when PCNA was loaded by RFC^ΔBRCT did not complete the entire 3-kb ssDNA track within the allotted experimental window (350 s), whereas virtually all trajectories completed the reaction when full-length RFC was used (Figure 6B).

Figure 6. FEN1 rescues deficient fill-in associated with unstable CLC formation.

(A) Fraction of time a fill-in trajectory spent in an inactive state when RFC (N = 14 from 14 independent tethers), RFC^ΔBRCT (N = 20 from 20 independent tethers), or RFC^ΔBRCT with FEN1 (N = 14 from 14 independent tethers) was used.

(B) Number of nucleotides filled in per tether for reactions incubated with RFC (N = 14 from 14 independent tethers), RFC^ΔBRCT (N = 20 from 20 independent tethers), or RFC^ΔBRCT with FEN1 (N = 14 from 14 independent tethers).

In (A) and (B), box boundaries represent 25^th to 75^th percentiles, middle bar represents median, and whiskers represent minimum and maximum values. Significance was calculated using two-tailed unpaired t-tests with Welch’s correction.

(C) An example kymograph and the associated trap position changes at a constant force of 8 pN showing fill-in of a gapped DNA tether incubated with 20 nM Polδ, 5 nM LD655-PCNA, 5 nM LD555-RFC^ΔBRCT, 15 nM FEN1, 100 nM RPA, 4 mM ATP, and 130 μM dNTPs. PCNA and RFC^ΔBRCT fluorescence channels are separately shown. Pink boxes denote inactive periods when the fill-in synthesis stalled, and purple box denotes the completion of fill-in.

See also Figures S4 and S5.

To assess whether the inactive periods coincident with PCNA departure from the 3’ end were due to Polδ stalling or its dissociation from the DNA altogether, we conducted experiments where we first allowed the fill-in reaction to start in a channel containing Polδ, RFC^ΔBRCT and PCNA; once we observed a pause in the synthesis activity, we moved the tether to a different channel lacking free Polδ in solution (Figure S3B). In this workflow, fill-in would only resume if the original Polδ remained at the 3’ terminus. However, the fill-in reaction never resumed (0 out of 7 tethers) even though PCNA repeatedly engaged with the 3’ end (e.g. Figure S3B). In contrast, when free Polδ was provided in solution, the fill-in reaction was able to continue once PCNA arrived back at the DNA 3’ end (e.g. Figures 5B and S3A). These different patterns indicate that the original Polδ is lost from the DNA 3’ end once PCNA leaves, and a new Polδ must be recruited to continue synthesis. Together, these results suggest that besides its catalytic function in loading PCNA onto DNA, RFC also fulfills an architectural function in stabilizing the PCNA-Polδ interaction—an activity that is abolished in the RFC^ΔBRCT variant—thereby enhancing the DNA synthesis efficiency.

FEN1 and RFC share a non-catalytic role in DNA synthesis

The differential ability of RFC and RFC^ΔBRCT to support processive fill-in synthesis points to the inherent instability of the PCNA-Polδ complex at the growing 3’ end. Given the many binding partners of PCNA¹³, we surmised that other factors may also fulfill the same function as RFC in stabilizing PCNA-Polδ interaction. The flap-processing endonuclease FEN1 has been shown to form a ternary complex with PCNA and Polδ in the human system¹⁷. We therefore sought to examine whether FEN1 can fulfill a similar function as RFC in stabilizing PCNA-Polδ interaction and promoting processive synthesis. We performed the single-molecule fill-in experiments in the presence of FEN1 and RFC^ΔBRCT, the latter of which is required for PCNA loading but cannot sustain stable CLC formation on DNA. Remarkably, we found that FEN1 largely rescued the deficiency of RFC^ΔBRCT in supporting processive synthesis, increasing the completion rate and decreasing the fraction of inactive periods to levels comparable to those observed in the presence of full-length RFC (Figures 6A and 6B). Indeed, when FEN1 was present, PCNA remained on the fill-in path and rarely slid back onto the dsDNA even without CLC formation (Figure 6C), and the processive fill-in trajectories phenocopied those obtained by using full-length RFC (compare Figures 6C and 5A).

These observations suggest that, even though RFC and FEN1 each have their own catalytic functions in lagging strand synthesis, they also share a non-catalytic role in stabilizing the PCNA-Polδ assembly. To gain further insight into this redundant structural function, we used AlphaFold3 to generate a predictive model of the PCNA-RFC-Polδ ternary complex with a 3’ recessed DNA (Figure S4A). The predicted model showed PCNA and Polδ in similar positions on the DNA as in a cryoelectron microscopy (cryo-EM) structure of the binary complex poised for synthesis¹⁸. It further showed that RFC occupies the space between the polymerase and one protomer of PCNA. Intriguingly, FEN1 was observed to occupy the same space in the cryo-EM structure of the PCNA-FEN1-Polδ complex (Figure S4B)¹⁷, and the region of Polδ that engages with FEN1 is predicted to interact with RFC (insets in Figures S4A and S4B). These predictions are consistent with the notion that FEN1 and RFC adopt redundant structural roles in stabilizing the fill-in machinery.

Next, we asked whether FEN1 and RFC can compete for PCNA binding. We first used mass photometry to measure the distribution of complex sizes with different combinations of proteins. When PCNA was incubated with RFC alone, a predominant mass peak corresponding to the CLC complex was observed (Figure S5A). When an excess of FEN1 was added, the CLC peak diminished and complexes corresponding to PCNA bound to one or more copies of FEN1 were observed (Figure S5B). We further performed single-molecule experiments to evaluate CLC formation on DNA in the presence of FEN1. We found that the likelihood of PCNA associating and co-sliding with RFC on dsDNA was substantially reduced by the addition of excess FEN1 (Figures S5C and S5D). These results indicate that FEN1 promotes RFC eviction from PCNA both on and off DNA.

DNA-binding residues in the BRCT domain are essential for genome maintenance

Our single-molecule data uncovered a structural role of RFC in supporting DNA synthesis via the formation of CLC complexes and further suggest the importance of Rfc1’s BRCT domain in this activity. To corroborate these findings, we examined the effects of BRCT mutants on genome maintenance in vivo. We generated S. cerevisiae cells wherein the endogenous RFC1 gene was replaced with a mutant allele in which the sequence encoding the BRCT domain was deleted (rfc1-BRCT∆). We also generated cells containing a mutant rfc1 allele in which 11 residues in the BRCT domain that have been identified to contact dsDNA^8-10 were mutated into alanine (rfc1-11A). The wild-type Rfc1, Rfc1-BRCT∆, and Rfc1-11A proteins were tagged with a TAP tag at their C-termini to assess protein levels. Neither mutant protein showed a significantly different expression level compared to the wild-type, although Rfc1-BRCT∆ did exist at higher levels than Rfc1-11A (Figure S6A). When cells were challenged with different DNA damaging agents, rfc1-BRCT∆ cells showed sensitivity to the DNA-methylating agent methyl methanesulfonate (MMS), but not to hydroxyurea (HU) that blocks dNTP synthesis or to camptothecin (CPT) that inhibits the DNA topoisomerase 1 (Figure 7A)¹⁹. The unique sensitivity to MMS is consistent with previous studies using plasmid-born rfc1 mutants with the BRCT domain or a larger N-terminal region deleted^10,20. The rfc1-11A mutant showed even more pronounced sensitivity to MMS (Figure 7A). As a control, wild-type Rfc1 tagged with TAP did not show sensitivity to MMS (Figures 7A and S6B), suggesting that the mutants’ defects were not caused by tagging. The sensitivity profiles of rfc1-BRCT∆ and rfc1-11A resemble those caused by deficiencies in proteins involved in lagging strand synthesis or base excision repair^21-23, implying that the BRCT domain and its DNA-binding sites are involved in these processes.

Figure 7. Mutating Rfc1’s BRCT domain sensitizes yeast cells to genotoxins and the deletion of FEN1 or Rad51.

(A) Cells harboring rfc1-BRCT∆ or rfc1-11A were grown in the absence or presence of methanesulfonate (MMS), hydroxyurea (HU), or camptothecin (CPT). Ten-fold serial dilutions of cells were used.

(B) Genetic interactions of rfc1-BRCT∆ and rfc1-11A with FEN1 (Rad27) deletion. Diploid cells with indicated genotypes were dissected and two representative tetrads, each containing 4 spore clones for each diploid are shown. Symbols denote spore clones with the indicated genotypes. Wild-type spore clones are not marked. Size comparison among spore clones was conducted within the same plate.

(C) Genetic interactions of rfc1-BRCT∆ and rfc1-11A with Rad51 deletion. Experiments were done and data are presented as in (B).

(D) Models for replicative clamp loading. (Top) In the conventional model, RFC binds PCNA in the presence of ATP, loads it onto a primer-template junction with a 3’ recessed end, and releases itself from the DNA upon ATP hydrolysis, leaving PCNA on the DNA to interact with a polymerase for DNA synthesis. (Bottom) Based our findings in this work, we propose a revised model for clamp loading. RFC loads PCNA onto DNA and frequently remains associated, which is dependent on an intact Rfc1 BRCT domain. The clamp-loader/clamp (CLC) complex can slide on duplex DNA until engaging with a polymerase to initiate synthesis. RFC can be replaced by FEN1 for PCNA binding. The PCNA-Polδ interaction is inherently unstable but can be stabilized by RFC or FEN1 via the formation of ternary complexes, which promotes processive fill-in synthesis over long ssDNA tracks. As such, RFC and FEN1 fulfill a non-catalytic role in DNA synthesis besides their canonical functions in clamp loading and flap processing, respectively.

See also Figure S7 and Table S1.

Prompted by these genetic data, we purified the RFC complex containing the Rfc1-11A mutant subunit (referred to as RFC^11A) to characterize its biochemical properties. Using a bulk assay, we found that RFC^11A loads PCNA onto DNA as efficiently as the wild-type RFC (Figure S6C). Using the single-molecule setup, we observed that PCNA loaded by RFC^11A diffused at a rate similar to those loaded by RFC^ΔBRCT and indeed much faster than those loaded by full-length RFC (Figures S6D and S6E), indicating the lack of CLC formation between PCNA and RFC^11A. These results suggest that the DNA-binding residues within Rfc1 BRCT are dispensable for PCNA loading, but essential for CLC formation. Given the similar in vitro behavior between RFC^ΔBRCT and RFC^11A, the more severe cellular phenotype associated with the latter mutant could be attributed to its reduced protein level.

FEN1 deletion shows negative genetic interactions with RFC BRCT mutations

To further explore the biological pathways that the Rfc1 BRCT domain is involved in, we examined the genetic interactions of rfc1-BRCT∆ and rfc1-11A with a panel of deletion mutants that impact different genome maintenance pathways. Considering FEN1 could rescue deficient fill-in caused by RFC^ΔBRCT in our single-molecule experiments, we first deleted the RAD27 gene that encodes FEN1 in yeast and performed tetrad analyses of diploid cells heterozygous for rfc1-BRCT∆ or rfc1-11A and for rad27∆. We found both rfc1 mutants exhibited negative interactions with rad27∆. While rfc1-BRCT∆ reduced the sizes of the rad27∆ spore clones, rfc1-11A was synthetically lethal with rad27∆ (Figure 7B). As a control, the TAP-tagged wild-type Rfc1 did not affect the growth of rad27∆ spore clones (Figure 7B, top). In conjunction with our single-molecule results, the simplest interpretation for these genetic results is that the Rfc1 BRCT domain and FEN1 both affect lagging strand DNA synthesis.

Next, we examined the genetic interaction between BRCT mutations and deletion of the gene encoding for the Rad51 recombinase, which mediates gap repair when lagging strand synthesis is impaired²⁴. Similar to their interaction with rad27∆, both rfc1-BRCT∆ and rfc1-11A substantially slowed the growth of rad51∆-containing spore clones, with rfc1-11A exerting a more severe effect (Figure 7C). Furthermore, we deleted the gene encoding the DNA damage checkpoint kinase Mec1 that can be activated by lagging strand synthesis defects²⁵. The SML1 gene was also deleted to support mec1∆ cell viability as mec1∆ is lethal²⁶. We found that rfc1-11A showed a strong negative interaction with mec1∆ sml1∆ (Figure S7A, left). Finally, we deleted genes encoding the ribonucleases RNase H1 and RNase H2, which remove DNA:RNA hybrids that can form during lagging strand synthesis²⁷. We found the double mutant (rnh1∆ rnh201∆), but not single mutants, had a synthetic lethal relationship with rfc1-11A (Figure S7A, right). The BRCT deletion mutant rfc1-BRCT∆ also caused a reduction in the size of spore clones containing mec1∆ sml1∆ or rnh1∆ rnh201∆, although to a lesser degree compared to rfc1-11A (Figure S7A). These growth defects became even more evident upon MMS treatment (Figure S7B). Collectively, these genetic profiling studies provide evidence for the importance of the Rfc1 BRCT domain, particularly its DNA binding sites, in genome maintenance related to lagging strand synthesis.

DISCUSSION

The conventional model postulates that RFC predominantly deposits PCNA at a recessed 3’ DNA end and gets evicted from the DNA soon after (Figure 7D, top). Contrary to this model, through direct real-time visualization, our single-molecule data reveal that RFC-mediated PCNA loading on DNA does not require a recessed end, echoing recent biochemical and structural studies^8-10. Moreover, we made the surprising finding that RFC tends to remain associated with PCNA after DNA loading, together forming a clamp-loader/clamp (CLC) complex that can be topologically linked to and slide on dsDNA. We further found evidence for the functional importance of CLC complexes in vitro and in vivo, specifically in stabilizing the replication machinery during DNA synthesis. By staying engaged with PCNA and the polymerase as a ternary complex on DNA, RFC fulfills a non-catalytic architectural role after accomplishing its ATP-dependent clamp-loading function (Figure 7D, bottom). Notably, it was recently reported that under replication stress when excess Okazaki fragments are produced, free pools of both PCNA and RFC are depleted^28,29. This result would be difficult to rationalize if RFC solely acted as a catalytic enzyme and were turned over after each loading event, while depletion of free RFC in cells could be potentially explained by our finding that CLCs form and accumulate on DNA.

Our work sheds new light on the role of the Rfc1 BRCT domain in RFC function. It has been shown that BRCT deletion has little effect on the PCNA loading activity of RFC, but results in shorter Okazaki fragment lengths in vitro and elevated sensitivity to certain DNA damage agents such as MMS^9,20. Our single-molecule results suggest that these defects are in least in part due to the inability of RFC^ΔBRCT to form stable CLC complexes with PCNA, which diminishes the processivity of DNA synthesis. The genotoxic sensitivity profile that we obtained for yeast cells lacking DNA-binding-competent Rfc1 BRCT mimics those for mutants deficient in lagging strand synthesis^10,21,22,30. Consistent with this observation, we detected negative genetic interactions between the BRCT mutants and the deletion of proteins involved in lagging strand synthesis (e.g. FEN1, RNase H1/2) or those responsive to lagging strand synthesis defects (e.g. Rad51, Mec1). It will be interesting to assess the contribution of the Rfc1 BRCT and the CLC complex to DNA synthesis occurring during other genome maintenance processes, such as the base excision repair pathway wherein PCNA-mediated DNA synthesis is also at work²⁰. In addition, given the BRCT truncation of the human RFC1 protein was found in patients with Hutchinson-Gilford progeria syndrome³¹, the structural role of RFC via its BRCT in DNA synthesis may be implicated in the pathology of this disease.

How does RFC help stabilize the PCNA-mediated DNA synthesis machinery? Although further experimental analysis is needed to answer this question, our AlphaFold structural prediction of the PCNA-RFC-Polδ ternary complex does offer some clues. The model shows extensive contacts between RFC and Polδ that include but extend well beyond the Rfc1 BRCT domain (Figure S4A). The BRCT domain in the ternary complex also does not appear to engage with DNA as it does in the PCNA-RFC binary complex shown by previous structural data⁸. We surmise that upon the CLC reaching a 3’ terminus where the local DNA geometry is permissible to Polδ recruitment, Polδ engages with PCNA and DNA, causing a conformational rearrangement of the BRCT domain to detach from the DNA. Meanwhile, RFC forms additional contacts with Polδ, which stabilize the ternary complex during fill-in. However, we did not find evidence that RFC and Polδ by themselves form a stable complex in solution. Therefore, we propose that the primary role of RFC in the ternary complex assembly is to tether CLC to DNA via the DNA-binding residues within the BRCT domain until Polδ engagement. Furthermore, the single-molecule platform developed here can be readily used to examine other sliding clamp-clamp loader/unloader pairs including Ctf18, Elg1, and Rad24-RFC³². Additionally, it will be interesting to examine whether the persistent formation of CLC can enhance the processivity of other DNA polymerases, including the leading-strand polymerase Polε and various translesion polymerases, especially where the fill-in of long ssDNA gaps is required³³.

Finally, we showed that RFC is not the only PCNA-binding factor that can stabilize the PCNA-Polδ assembly: FEN1 proficiently rescues the replication defect caused by Rfc1 BRCT mutation and promotes processive synthesis. Given that the interaction with PCNA is known to stimulate FEN1’s nuclease activity³⁴, the replacement of RFC for FEN1 during fill-in synthesis simultaneously maintains the structural integrity of the replication machinery and poises FEN1 for flap processing, thus coupling its non-catalytic and catalytic functions. In yeast cells, RFC and FEN1 exist at roughly equal abundance (~100 nM)³⁵, whereas in human cells, FEN1 exists in ~7-fold excess (~1,500 nM vs. ~200 nM)³⁶. Assuming the biochemical properties of the yeast and human proteins are comparable, it is reasonable to envision that FEN1 plays a more significant role in stabilizing PCNA-Polδ interaction during fill-in synthesis in human cells. It should be noted that the CLC would still be an obligatory intermediate because PCNA needs RFC to be loaded onto DNA in the first place. Nevertheless, the non-catalytic architectural function of RFC still appears to be relevant in human cells, particularly under conditions of replication stress^28,29. Similarly, other PCNA-binding partners may also serve a structural role besides their canonical functions in DNA replication, repair, and chromatin maintenance³⁷. This work demonstrates an approach to dissecting their dynamic interaction and exchange with PCNA.

Limitations of the Study

Our study provides evidence for ternary complex assembly between PCNA, RFC, and Polδ during fill-in synthesis. The structural configuration of this ternary complex is based on AlphaFold prediction and thus remains speculative. Given the functional importance of this ternary complex, it will be valuable to experimentally determine its atomic structure, which will guide mutagenesis studies to further delineate the critical residues and interfaces for the non-canonical activity of RFC. The single-molecule experiments in this study utilized a simplified reconstitution of lagging strand synthesis and did not include the other components of the native machinery such as the Pol α-primase. Although our in vivo genetic data corroborate the in vitro findings, future single-molecule studies incorporating the full set of the replisome components will shed additional light on the dynamics of CLCs during coordinated leading and lagging strand synthesis.

STAR Methods

Experimental model and study participant details

Experimental source materials

E. coli DH5α cells were transformed with plasmids for propagation or protein mutagenesis. E. coli BL21(DE3) cells were transformed with plasmids for protein overexpression of PCNA, Polδ, FEN1, and RPA and were grown as described in the Method Details section. A C600 S. cerevisiae strain OY001 (ade2-1 ura3-1 his3-11, 15 trp1-1 leu2-3, 112 can1-100 bar1Δ MATa pep4∷KANMX6), a strain constructed from W303 (a gift from A. Tackett and B. Chait, Rockefeller University), was modified to express RFC by inserting the five RFC genes into the genome under control of the Gal1/10 promotor using standard genetic procedures as outlined in the Method Details section. A S. cerevisiae strain W303 was used for all genetic experiments as detailed in Table S1.

Yeast strains

All S. cerevisiae strains used for genetic analyses are listed in the Key Resources Table and Table S1 where only one strain is listed per each genotype and at least two independent isolates of each genotype were used in experiments.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Bacterial and virus strains
Escherichia coli BLR(DE3)	Novagen	Cat# 69053
Escherichia coli BL21-CodonPlus (DE3)-RIL	Agilent	Cat# 230245
Biological samples
Chemicals, peptides, and recombinant proteins LD555-NHS	Lumidyne Technologies	Cat# 1
Alexa Fluor 488 NHS Ester	ThermoFisher	Cat# A20000
LD655-CoA	Lumidyne Technologies	Custom synthesis
Cy3-CoA	Lumidyne Technologies	Custom synthesis
Cyclooctatetraene	Sigma-Aldrich	Cat# 138924
4-Nitrobenzyl alcohol	Sigma-Aldrich	Cat# N12821
Bacterial protocatechuate 3,4-dioxygenase	OYC Americas	Cat# 46852004
3,4-Dihydroxybenzoic acid	Sigma-Aldrich	Cat# 37580
Streptavidin Coated Polystyrene Particles (4.0-4.9 μm)	Spherotech	Cat# SVP-40-5
Biotinylated Lambda DNA	LUMICKS	Cat# 00001
Biotinylated DNA Hybrid DNA	LUMICKS	Cat# 00027
Nb.BbvCI	New England Biolabs	Cat# R0631S
NheI-HF	New England Biolabs	Cat# R3131S
BsrGI-HF	New England Biolabs	Cat# R3575S
BsiWI-HF	New England Biolabs	Cat# R3553S
BsaI-HF®v2	New England Biolabs	Cat# R3733S
T4 DNA Ligase	New England Biolabs	Cat# M0202S
SphI-HF	New England Biolabs	Cat# R3182S
SpeI-HF	New England Biolabs	Cat# R3133S
ATP solution	ThermoFisher	Cat# R0441
Deoxynucleotide (dNTP) Solution Mix	New England Biolabs	Cat# N0447S
TCEP-HCl (Tris(2-carboxyethyl)phosphine hydrochloride)	Goldbio	Cat# TCEP1
DTT (Dithiothreitol) (> 99% pure) protease free	Goldbio	Cat# DTT10
Nt.BspQI	New England Biolabs	Cat# R0644S
PCNA	Bauer and Burgers, 198857	N/A
PCNA-S6	This paper	N/A
RFC	Finkelstein et al., 200340	N/A
RFC-S6	This paper	N/A
RFCΔBRCT	Johnson et al., 200641	N/A
RPA	Henricksen et al., 199458	N/A
Polδ	Langston et al., 200842	N/A
Rad27 (FEN1)	This paper	N/A
SFP synthase	Yin et al., 200659	N/A
IPTG	Goldbio	Cat# I2481C
Glutathione Sepharose™ 4B GST-tagged protein purification resin	Cytiva	Cat# 17075601
Q Sepharose™ Fast Flow ion exchange resin	Cytiva	Cat# 17051004
Cytiva™ Protein Select™ resin (FLAG affinity column)	Cytiva	Cat# 17542103
HisTrap™ HP His tag protein purification column	Cytiva	Cat# 17524801
HiTrap™ Heparin HP affinity columns	Cytiva	Cat# 17040701
Protein Kinase A from bovine heart	Sigma-Aldrich	Cat# P5511; CAS# 9026-43-1
Bio-Gel A-1.5m Gel, Fine	Bio-Rad	Cat# 151-0450
Critical commercial assays
Deposited data
Experimental models: Cell lines
Experimental models: Organisms/strains
S.cerevisiaestrain W303	Lab collection	N/A
Oligonucleotides
AAAAAGCTAGCCTCAGCATATTAGCTAAAACTAAAA GTGGTAAAACG	Integrated DNA Technologies	N/A
AAAAATGTACAGCTGAGGCGGAATCGGTAGTAAGTT ATAGC	Integrated DNA Technologies	N/A
dual-biotin-CAAGCCGCCATTCCACTCTGCCTA	Integrated DNA Technologies	N/A
dual-biotin-CAAGCCGCCATTCCACTCTGCCTA	Integrated DNA Technologies	N/A
dual-biotin-CTTGCTCGTAGTCAATGCGTCAC	Integrated DNA Technologies	N/A
phosphate-GGCAGTGACGCATTGACTACGAGCAAG	Integrated DNA Technologies	N/A
GCTTGTTAAGAAAGAAGAGG	Integrated DNA Technologies	N/A
GGCTAATTGGGAAAATTCAAAG	Integrated DNA Technologies	N/A
GGCTCAAGATGTACTAGATA	Integrated DNA Technologies	N/A
GTTCAGACTCGGCTTCGAAT	Integrated DNA Technologies	N/A
GTTCAGCAAAGGCATCCCTT	Integrated DNA Technologies	N/A
CTGTACGTACAACCGTCGTT	Integrated DNA Technologies	N/A
Recombinant DNA
Sc FLAG-S6-PCNA-pET	This paper	Addgene# 239471
Sc FLAG-S6-RFC1-pRSFDuet	This paper	Addgene# 239472
Sc Rad27-pET24	This paper	Addgene# 239204
Sc RFC[2+5}-pRS403/GAL	This paper	Addgene# 239474
Sc RFC[3+4]-pRS402/GAL	This paper	Addgene# 239475
Sc FLAG-S6-RFC1-pRSFDuet	This paper	Addgene# 239472
Software and algorithms
Bluelake	LUMICKS	https://www.lumicks.com/dynamic-single-molecule/solution/c-trap
MATLAB R2024b	MathWorks	https://www.mathworks.com/products/matlab.html
FIJI	Schindelin et al., 201260	https://imagej.net/software/fiji/
C-Trap .h5 File Visualization GUI	Watters, 202461	https://github.com/lumicks/harbor/tree/main/Visualization/C-Trap%20.h5%20File%20Visualization%20GUI
Ctrapviewer	Heller lab	http://www.nat.vu.nl/~iheller/download.html
Prism 10	GraphPad	https://www.graphpad.com/features
AcquireMP version 2.4.0	Refeyn	https://refeyn.com/about-mass-photometry/?v=0b3b97fa6688
Other

Method details

Protein purification and labeling

PCNA

S. cerevisiae S6-tagged PCNA was expressed and purified using a previously described protocol³⁸. Briefly, a 3X FLAG tag, followed by an S6-tag was added to the N-terminus of S.c. PCNA encoded by Pol30. The S.c. FLAG-S6-PCNA encoding plasmid (deposited to Addgene # 239471) was transformed into BLR(DE3) cells which were grown to OD 0.6 at 37 °C in LB media. Cells were then cooled to 15 °C and induced for expression using 1 mM IPTG for 18 h at 15 °C before harvesting by centrifugation. Cells were disrupted by French Press using buffer A (20 mM Tris-Cl pH 7.5, 1 mM DTT, 0.1 mM EDTA, 10% glycerol) and the cell lysate was clarified by centrifugation. The cell lysate was applied to a FLAG affinity column equilibrated in buffer A, washed extensively with buffer A + 500 mM NaCl, then eluted using buffer A + 300 mM NaCl and 2 μM FLAG-peptide. Eluted fractions were dialyzed against buffer A + 50 mM NaCl then purified from excess peptide using a Q Sepharose column eluted with a 100-600 mM NaCl gradient in Buffer A. Fractions containing FLAG-S6-PCNA were pooled, flash frozen and stored at −80 °C.

RFC complexes

S. cerevisiae RFC containing full-length Rfc1 having an N-terminal 3XFLAG tag followed by an S6 labeling sequence was cloned into pRS405 (gal) (deposited to Addgene as # 239472) and integrated into a C600 yeast strain³⁹. The strain was first modified by integration of plasmids encoding the Rfc2-5 subunits under control of the Gal1/10 promotor (i.e. S.c. RFC[2+5}-pRS403/Gal [deposited to Addgene # 239474] and S.c. RFC[3+4]-pRS402/Gal [deposited to Addgene # 239475]). A 1 L overnight culture of cells were grown in the appropriate drop out media at 30 °C and then inoculated into 24 2L flasks containing 1L YP media and grown overnight at 30 °C until the cell culture reached an O.D. = 0.6. Cells were then induced with Gal, allowed to incubate a further 6 h at 30 °C, then harvested by centrifugation. Harvested cells were resuspended in a minimal volume of buffer A and frozen dropwise into liquid nitrogen. Cells were then lysed using a cryogenic spex mill, the lysate was allowed to thaw, and insoluble material was removed by centrifugation. The supernatant was applied to a column containing immobilized FLAG antibodies, followed by extensive washing with buffer A + 500 mM NaCl, then eluted using buffer A containing 300 mM NaCl + 2 μM FLAG peptide. To remove excess FLAG peptide, FLAG-S6-RFC was serially concentrated on a centricon 30 in buffer A containing 300 mM NaCl, then aliquoted, snap frozen in liquid nitrogen and stored at −80 °C.

RFC containing a truncated Rfc1 that lacks the N-terminal residues 1-282 (RFC^ΔBRCT) and contains an N-terminal His/kinase tag, were co-expressed as described^40,41, and RFC containing the truncated Rfc1 (tRFC) was purified essentially as described³⁸. Briefly, E. coli BLR(DE3) cells with tRFC expression plasmids were grown under antibiotic selection at 30 °C to an OD₆₀₀ of ~0.6, brought to 15 °C by swirling in ice water, then 1 mM IPTG was added and cells allowed to incubate a further 18 h at 15 °C. Cells were collected by centrifugation, resuspended in an equal volume of 50 mM Tris-HCl pH 7.5, 0.5 % EDTA, 5 mM DTT, 10% sucrose, 500 mM NaCl, then lysed by French press. Insoluble material was discarded by centrifugation at 4 °C. The supernatant was diluted using buffer A to bring conductivity equal to 150 mM NaCl, then applied to SP-Sepharose, washed extensively with buffer A + 150 mM NaCl and then eluted with a 300 mL gradient of buffer A + 150 mM NaCl and buffer A + 600 mM NaCl. Peak fractions eluted at ~365 mM NaCl and were pooled, diluted with Buffer A to a conductivity of 150 mM NaCl and were applied to Q-Sepharose pre-equilibrated with Buffer A + 150 mM NaCl, then eluted with a gradient of Buffer A + 150 mM NaCl to 600 mM NaCl. Peak fractions eluted at ~300 mM NaCl and were pooled, aliquoted, and stored in −80 °C.

Polδ

S. cerevisiae Polδ was expressed and purified as described⁴². Briefly, two expression vectors containing yeast pol3, pol31, and GST-pol32 plasmids were co-transformed into E. coli BL21(DE3) cells. 12 L of cells (1 L per 2L shaker flask) were grown at 37 °C to OD₆₀₀ ~0.7 in the presence of ampicillin, kanamycin, and streptomycin. Flasks were then removed and the temperature of cultures were rapidly reduced to 15°C by swirling on ice (monitoring temperature with a thermometer) and induced upon adding 1 mM IPTG, the incubated with shaking for a further 8 hr at 15 °C. Cells were harvested by centrifugation, resuspended in a minimal volume of 50 mM Tris-HCl pH 7.5, 1 mM EDTA, 10% glycerol, 500 mM NaCl, 2 mM PMSF. Resuspended cells were lysed in a French Press, then 30 mM spermidine was added, and the lysate was clarified by centrifugation at 23,700 × g for 1 h at 4 °C. The clarified lysate was applied to glutathione-Sepharose 4B (GE Healthcare) pre-equilibrated with 50 mM Tris-HCl pH 7.5, 10% sucrose, 500 mM NaCl. After extensive washing with the same buffer, the protein was eluted with 40 mM reduced l-glutathione, and peak fractions were pooled, mixed with 40 units of PreScission Protease to remove the GST tag from the pol32 subunit, and dialyzed overnight against 4 L of Buffer A + 300 mM NaCl. Pooled fractions were diluted with Buffer A to a conductivity equal to 80 mM NaCl then applied to a Mono-Q column equilibrated with Buffer A + 80 mM NaCl. The column was washed with 10 column volumes of Buffer A + 80 mM NaCl and was eluted with gradient of Buffer A + 80 mM NaCl to Buffer A + 200 mM NaCl. Peak fractions of the stoichiometric Polδ heterotrimer were pooled, concentrated, aliquoted, and stored in −80 °C.

Rad27/Fen1 flap endonuclease

S.c. Rad27 (Fen1) with a 6xHis tag at the C-terminus was cloned into pET24 (deposited to Addgene # 239204). The ScRad27-pET24 plasmid was transformed into E. coli BL21(DE3) cells. Cells were grown in 3L of LB with 50 μg/ml of kanamycin and grown at 37 °C to OD₆₀₀ = 0.6, then induced with 1 mM IPTG for 3 h at 37 °C. Induced cells were harvested at 5,000 X g for 15 min at 4 °C, then resuspended in 100 ml 20 mM Tris-HCl pH 7.5, 200 mM NaCl, 10% glycerol. The resuspended cells were then lysed with a French Press and cell debris was removed by centrifugation at 48,000 X g for 40 min. Next, 5 mM imidazole was added to the clarified supernatant and then the supernatant was loaded onto a 1 ml HisTrapTM HP column equilibrated in 20 mM Tris-HCl pH 7.5, 200 mM NaCl, 10% glycerol, 5 mM imidazole. The column was washed with 3 column volumes of the same buffer, then washed in steps containing either 20, 60, 100, or 150 mM imidazole in 20 mM Tris-HCl pH 7.5, 200 mM NaCl, and 10% glycerol. S.c. Rad27 (Fen1) was eluted with 6 mL 20 mM Tris-HCl pH 7.5, 200 mM NaCl, 10% glycerol containing 200 mM imidazole, diluted with 2 volumes of 20 mM Tris-HCl pH 7.5, 10% glycerol and loaded onto a 1 mL HiTrapTM Heparin HP column. Rad27 (Fen1) was eluted using a 100 to 600 mM NaCl gradient in 20 mM mM Tris-HCl pH 7.5, 10% glycerol. Fractions containing the bulk of Fen1 were pooled, concentrated to 8 mg/ml, flash frozen in liquid nitrogen and stored at −80 °C.

RPA

RPA was purified as described⁴³. Plasmids that encode RFA1, 2 and 3 were cloned to be controlled by the IPTG inducible promotor. These plasmids were then transformed into BL21-DE3-codon plus cells followed by induction using IPTG for 16 h at 15 °C. Cells were collected by centrifugation at 4 °C and then were disrupted by a French Press. The resulting cell lysate was clarified by centrifugation for 1 h at 48,000g for 1 hr. The clarified lysate was applied to an Affi-Gel Blue column in 20 mM Tris-HCl pH 7.5, 0.8M KCl, 10% glycerol and washed with 0.8 M KCl and 1.5 M NaSCN, then NaSCN was removed using a hydroxyapatite column in 80 mM potassium phosphate. Fractions containing RPA were determined by analysis by PAGE. Peak fractions containing RPA were pooled, dialyzed using the same buffer but containing 200 nM KCl, then loaded onto a MonoQ column in 20 mM Tris-HCl pH 7.5, 10% glycerol 200 mM KCl. The column was then developed with a 10 column volume gradient of 20 mM Tris-HCl pH 7.5, 10% glycerol 200 mM KCl. Fractions were analyzed by PAGE and fractions containing >95% pure protein were pooled, aliquoted, flash frozen and stored at −80 °C.

Fluorescent labeling

To obtain site-specific fluorescently labeled S6-PCNA and S6-RFC, SFP synthase (4’-phosphopantetheinyl transferase) was utilized to transfer the CoA-functionalized moieties to the single serine residue contained in the tag. SFP synthase was expressed and purified as previously described⁴³. S6-PCNA or S6-RFC, SFP synthase, and LD655-CoA (Lumidyne Technologies) or Cy3-CoA were incubated at a 1:2:5 molar ratio for 1 h at room temperature in the presence of 10 mM magnesium chloride. Excess dye and SFP were removed by centrifugation through a 50 kDa molecular weight cutoff Amicon Ultra Centrifugal Filter (Millipore) using Buffer A containing 300 mM sodium chloride. The resulting labeled protein was aliquoted and stored in −80 °C.

To obtain fluorescently labeled RFC^ΔBRCT, LD555-NHS ester (Lumidyne Technologies) was used to nonspecifically label the primary amines of RFC^ΔBRCT. Preferential N-terminal labeling was performed by labeling at low pH = 7.0⁴⁴. RFC^ΔBRCT was dialyzed out of a Tris-based buffer and into Buffer B (50 mM HEPES pH 7.0, 300 mM NaCl, 1 mM DTT, and 0.25 mM EDTA). The protein was then incubated with LD555-NHS ester at a 1:5 molar ratio for 1 h at room temperature followed by overnight at 4 °C. The reaction was quenched with 25 mM Tris-HCl pH 6.8 for 5 min. Excess dye was removed by centrifugation through a 50 kDa molecular weight cutoff Amicon Ultra Centrifugal Filter (Millipore) using Buffer B, and the labeled protein was aliquoted and stored in −80 °C.

Fluorescently labeled AF488-RPA was obtained as described previously⁴³. Labeling efficiencies for all constructs ranged between 70-90%.

DNA substrate preparation

PCR of plasmid pLW58 was performed using primers PriF and PriR to amplify a 3kb AT-rich fragment and introduce two distal Nb.BbvCI nicking sites as well as cloning restriction sites. This was cloned into the 15.9-kb plasmid pRGEB32 (Addgene plasmid # 63142⁴⁵) via NheI/BsrGI (3-kb PCR product) and NheI/BsiWI (pRGEB32) to create the 18.9-kb plasmid pJF3KB1.

Purified pJF3KB1 was linearized by restriction digest with BsaI-HF v2 (NEB) and the resulting 20bp fragment was removed by PEG precipitation. Biotinylated oligo pairs (Oli1+Oli2, Oli3+Oli4) were annealed and ligated (T4 DNA ligase, NEB) to the two BsaI overhangs of the linearized plasmid. Ligase was then heat inactivated. The DNA was nicked on the same strand at the two nicking sites at either end of the 3kb AT-rich region (Nb.BbvCI, NEB). The endonuclease was then heat inactivated and the DNA was purified away from excess oligos by PEG precipitation. Aliquots of the intermediate linearized pJF3KB1 and the final biotinylated/nicked substrate were digested with SphI and SpeI (NEB) to check for successful BsaI restriction, PEG purification, and introduction of the oligo pairs at both ends of the DNA. The substrate was stored in TE buffer at 4 °C. All synthetic oligonucleotides were ordered from Integrated DNA Technologies:

PriF: AAAAAGCTAGCCTCAGCATATTAGCTAAAACTAAAAGTGGTAAAACG (italic = NheI, underline = Nb.BbvCI)

PriR: AAAAATGTACAGCTGAGGCGGAATCGGTAGTAAGTTATAGC (italic = BsrGI, underline = Nb.BbvCI)

Oli1: 5’ dual-biotin-CAAGCCGCCATTCCACTCTGCCTA

Oli2: 5’ dual-biotin-CAAGCCGCCATTCCACTCTGCCTA

Oli3: 5’ dual-biotin-CTTGCTCGTAGTCAATGCGTCAC

Oli4: 5’ phosphate-GGCAGTGACGCATTGACTACGAGCAAG

Single-molecule experiments

Experimental setup

Single-molecule experiments were performed on a LUMICKS C-Trap instrument at room temperature. This instrument combines three-color confocal fluorescence microscopy with dual-trap optical tweezers⁴⁶. Data was acquired using LUMICKS Bluelake software version 1.6.16. Rapid optical trap movement was achieved using a computer-controlled stage in a flow cell containing 5 separate channels. Channels 1-3 were separated by laminar flow and were used to tether DNA between two 4-μm streptavidin-coated polystyrene beads (Spherotech) held in optical traps. In channel 1, a single bead was caught in each trap. The traps were next moved to channel 2, and biotinylated DNA was tethered between the two beads as denoted by an increase in the force reading. The tether was then moved into channel 3, and the flow was stopped. The presence of a single DNA tether was confirmed by the generated force-distance curve.

To generate a gapped DNA tether, the ds gapped DNA construct was pulled to high forces (~60 pN) in a no-salt buffer (Tris-HCl pH 8.0). Flow was turned on (~0.1-0.2 bar) for 5 s and the tether was relaxed (Figure S1B). Flow was turned off, and successful generation of the gapped tether was confirmed by the force-distance curve (Figure S1C).

Channels 4 and 5 were loaded with proteins as described for each assay. Flow was turned off when collecting data and visualizing protein behavior. For fill-in assays, either the force or distance between the beads was held constant, and the resultant change in either the trap position or the force was recorded as a measurement of fill-in rate and completion.

Fluorescence detection

Cy3, LD655, and AF488 fluorophores were excited by three laser lines at 532, 638, and 488 nm respectively. Confocal line scanning at 100 ms/line through the center of each bead was used to generate kymographs. For experiments tracking PCNA loaded by FC^ΔBRCT or RFC^11A, a line scanning rate of 50 ms/line was used to capture more continuous trajectories. Occasionally, individual lasers were turned off to confirm the presence of fluorescently labeled proteins.

To investigate the behavior of LD655-PCNA and Cy3-RFC or LD555-RFC^ΔBRCT, optical traps tethering a gapped, nicked (LUMICKS), or λ DNA molecule (LUMICKS) under 6 or 1 pN of constant tension respectively were moved into channel 4 of the microfluidic flow cell containing 8 nM of each protein plus 100 nM of RPA in Image Buffer (25 nM Tris-HCl pH 7.5, 100 nM sodium chloride, 10 mM magnesium acetate, 3 mM DTT, 2 mM TCEP, 1% glycerol, 40 μg/uL BSA, 0.1 mM EDTA, 4 mM ATP, 1 mM cyclooctatetraene [Sigma], 1 mM 4-nitrobenzyl alcohol [Sigma], and oxygen scavenging reagents 10 nM protocatechuate-3,4-dioxygenase [OYC Americas] and 2.5 mM 3,4-dihydroxybenzoic acid [Sigma]). For high salt challenge experiments, the tether with bound proteins was moved to channel 5 with Image Buffer containing 500 mM NaCl.

For fill-in experiments, a gapped DNA tether was moved to channel 4, which contained 20 nM Polδ, 5 nM LD655-PCNA, 5 nM Cy3-RFC or LD555-RFC^ΔBRCT, and 100 nM RPA in Image Buffer supplemented with 115 μM of all deoxynucleotides and held at the specified tension. When RPA was visualized, 20 nM of AF488-RPA was used along with 5 nM of dark PCNA and RFC in lieu of the fluorescently labeled constructs. For experiments with FEN1, 15 nM of FEN1 was added to each sample.

Structural prediction

To predict the structure of the PCNA-RFC-Polδ complex, the AlphaFold webserver featuring AlphaFold3 was used (alphafoldserver.com). Three copies of S. cerevisiae PCNA protein sequence (POL30) (Saccharomyces Genome Database [SGD] #: YBR088C), one copy of RFC1-5 (SGD #: YOR217W; YJR068W; YNL290W; YOL094C; YBR087W), and one copy of each of the Polδ subunits, POL3 (SDG #: YDL102W), POL31 (SDG #: YJR006W), and POL32 (SDG #: YJR043C) were inputted as well as two complementary single-stranded DNA sequences (Primer strand 5’ to 3’: AGCTATGACCATGATTACGAATTGC; Template strand 5’ to 3’: CTGCACGAATTAAGC AATTCGTAATCATGGTCATAGCT)¹⁸. The highest ranked model was chosen for visualization and atoms were hidden if they had a B-factor less than 0.65.

Mass photometry

Data were collected using a OneMP mass photometer (Refeyn). Calibration of the instrument was performed using bovine serum albumin (66 kDa), beta amylase (224 kDa), and thyroglobulin (670 kDa). Movies were acquired for 6,000 frames (60 s) using AcquireMP software (version 2.4.0) under the default settings. Protein concentrations were empirically chosen to achieve ~75 binding events per second. Experiments were performed using 50 nM PCNA, 50 nM RFC, and 10 mM ATPγS with the specified molar ratio of FEN1. Peak mass values are typically within ~5% error of the true values. Raw data were converted to frequency distributions using Prism 9 (GraphPad) using a bin size of 10 Da.

PCNA-loading gel filtration assay

³²P-PCNA contains an N-terminal 6 residue kinase site and was purified and radiolabled using γ³²-ATP, and ATP dependent PKA kinase (Sigma) to a specific radioactivity of 81 cpm/fmol, as described^47,48. ³²P-PCNA was assembled onto nicked pUC19 plasmid DNA using Nt.BspQ1 that nicks a single site in this plasmid. The reaction contained 1 pmol (5 nM) Nt. BspQ1 nicked pUC19 DNA, 3 pmol (15 nM) ³²P-PCNA, 1 pmol (5 nM) RFC, and either wt or 11A mutant RFC in 200 ml of Buffer A containing 20 mM NaCl and 1 mM ATP. The reaction was incubated at 30 °C for 10 minutes before being applied to a 5 ml Bio-Gel A15m gel filtration column and eluted with Buffer A containing 150 mM NaCl. Seven drop fractions (~200 ml) were collected. ³²P-PCNA radioactivity was measured by liquid scintillation from which the concentration of PCNA-DNA complex was determined from the known specific radioactivity of the ³²P-PCNA. The DNA-bound ³²P-PCNA peak elutes first (fractions 11-16) followed by a second peak of free PCNA (fractions 19-31).

Genetic methods

Yeast strains

Strains used in this study are listed in the Key Resource Table and are isogenic to a RAD5 derivative of W303 (MATa ade2-1 can1-100 ura3-1 his3-11,15, leu2-3, 112 trp1-1 rad5-535)⁴⁹. At least two strains per genotype were examined for each assay and only one is listed in the Key Resources Table. Mutant construction and protein tagging were conducted using standard PCR-based method and genetic engineered loci were verified by sequencing. Standard procedures were used for cell growth, media preparation, genetic crosses, tetrad analyses, and genotoxin treatment. Cells were grown at 30 °C in YPD media for both tetrad analyses and genotoxin treatment experiments.

Primers used in the yeast genetics experiments:

Rfc1 1789 F: 5’-GCTTGTTAAGAAAGAAGAGG

Rfc1 1753 F: 5’-GGCTAATTGGGAAAATTCAAAG

1315 F: 5’-GGCTCAAGATGTACTAGATA

1290 F: 5’-GTTCAGACTCGGCTTCGAAT

Rfc1 Far DR: 5’-GTTCAGCAAAGGCATCCCTT

Rfc1 DR: 5’-CTGTACGTACAACCGTCGTT

Protein level examination

To detect the wild-type and mutant Rfc1 proteins C-terminal tagged with the TAP tag, protein extracts were made using a TCA (Trichloroacetic acid) method as previously described⁵⁰. In brief, cells pellets were resuspended in 20% TCA and homogenized using glass beads in a FastPrep-24 bead beating instrument (MP Biomedicals). The lysate was centrifuged to remove supernatant. The precipitated proteins were dissolved in Laemmli buffer (65 mM Tris-HCl pH 6.8, 2% SDS, 10% glycerol, 5% β-mercaptoethanol, and 0.025% bromophenol blue) with 2 M Tris to neutralize the lysate. Prior to loading, samples were boiled for 5 min and spun down to remove insoluble materials. Samples were separated on NuPAGE 4-12% Bis-Tris gels (ThermoFisher) for immunoblotting to detect the TAP-tagged Rfc1 proteins.

Quantification and statistical analysis

Single-molecule data analysis

Raw position data of righthand bead (Trap 1) collected at 100 Hz was first smoothed with a Savitzky-Golay smoothing filter of polynomial order 3 and frame length of 501 data points⁵¹. To infer polymerase trajectory from bead position data, the DNA fill-in reaction was modeled using a linear combination of the freely-jointed chain model for the single-stranded segment and the extensible worm-like chain model for the double-stranded segment^52,53. Based on reported data, RPA at the concentrations used in our experiments confers a negligible effect on the extension of ssDNA⁵⁴. From this linear combination, we predict the change in tether length per base pair filled in as a function of force applied. This relation was used to convert the bead trajectory in nanometers to a fill-in trajectory in base pairs. The instantaneous velocity was extracted from the fill-in trajectory with the gradient function, which returns the one-dimensional numerical gradient of the input vector. Pauses in the trajectory were identified using an empirical velocity cutoff of +15 bp/s. To avoid over-counting from local rapid changes in velocity, we only identify pauses at least 1 second long, and if pauses are separated by intervals of less than 2 seconds, the pauses are joined together. All analyses were performed in MATLAB R2024b.

Kymographs were processed and analyzed using a custom script (https://github.com/lumicks/harbor/tree/main/Visualization/C-Trap%20.h5%20File%20Visualization%20GUI) that incorporates tools from the lumicks.pylake Python library and other Python modules (Numpy, Matplotlib and Pandas) to generate tracked lines using the kymotracker greedy algorithm. To determine the MSD, the tracked lines were smoothed using a third-order Savitzky–Golay filter with a window length of 11 tracked frames, and the MSD was calculated from each smoothed trajectory. The diffusion coefficient (D) was calculated by fitting the MSD curve for each trajectory to the equation for 1D diffusion where MSD = 2Dtα (α is the exponential term used to characterize normal diffusion [α = 1], subdiffusion [α < 1] or superdiffusion [α > 1]). The first 1 s segment of each MSD curve was used for the fit⁵⁵. The fit was discarded if the R2 value of the fit was less than 0.8. Trajectories with an α value between 0.7 and 1.3 (over 50% of all trajectories) were included for further analysis.

Statistical analysis for the specified single-molecule experiments was performed using a bootstrapping approach. For experiments with N DNA tethers, a bootstrap sample of size N was generated by randomly selecting DNA tethers from the original sample with replacement. The desired statistic was calculated for this bootstrap sample. A total of 10,000 bootstrap samples were generated to create a distribution of sample statistics where the percentiles of such distribution define the limits for a confidence interval. Two-sample hypothesis testing with bootstrap was performed as outlined in⁵⁶.

Statistical analysis

Errors reported in this study represent standard deviation (SD) or standard error of the mean (SEM). P values were determined from two-tailed unpaired t tests (with Welch’s correction as specified in each figure caption) for comparison between two conditions using Prism 10 (GraphPad).

Supplementary Material

1

Figure S1. Protein and DNA substrate preparations, related to Figures 1 and 3.

(A) SDS-PAGE gel showing each purified protein used in this work with individual subunits labeled. Asterisk denotes a likely proteolytic product of Rfc1. (B) Schematic of the protocol to generate a gapped DNA tether in situ in an optical tweezers setup. A single dsDNA molecule containing a 3-kb, AT-rich segment flanked by two nicks and a 10-kb and 6-kb GC-rich arm was pulled to high forces (~60 pN). A no-salt buffer was flowed in gently to remove the untethered AT-rich ssDNA strand. The tether was then relaxed, and the flow was turned off, leaving a 3-kb ssDNA gapped DNA tether for subsequent single-molecule imaging. (C) Force-distance curves for dual-nicked dsDNA (grey) and 3-kb gapped DNA (blue) tethers. The differences allow the assessment of successful gap generation.

2

Figure S2. Additional experiments showing PCNA, RFC, and RFC^ΔBRCT behaviors on gapped, duplex, or nicked DNA, related to Figures 1, 2, and 3.

(A) Representative kymograph showing a gapped DNA tether incubated with 8 nM LD655-PCNA, 8 nM Cy3-RFC, and 100 nM RPA with 4 mM ATP. Red laser was turned off briefly to confirm the presence of green RFC signals. (B) Representative kymograph showing a gapped DNA tether incubated with 8 nM LD655-PCNA, 8 nM Cy3-RFC, and 100 nM RPA in the absence of ATP. (C) Representative kymograph showing a gapped DNA tether incubated with 8 nM LD655-PCNA and 100 nM RPA with 4 mM ATP in the absence of RFC. (D) Representative kymograph showing a ds λ DNA tether incubated with 8 nM LD655-PCNA, 8 nM Cy3-RFC, and 100 nM RPA in the absence of ATP. (E) Representative kymograph showing a ds λ DNA tether incubated with 8 nM LD655-PCNA and 100 nM RPA with 4 mM ATP in the absence of RFC. (F) A representative kymograph showing a dual nicked dsDNA tether incubated with 8 nM LD655-PCNA, 8 nM Cy3-RFC, and 100 nM RPA with 4 mM ATP. The lefthand schematic denotes locations of the two nick sites. (G) Representative kymograph showing a ds λ DNA tether incubated with 8 nM Cy3-RFC (without PCNA) and 100 nM RPA with 4 mM ATP. (H) Diffusion coefficients (D) for individual RFC trajectories sliding on dsDNA when bound to PCNA as a CLC (N = 22 from 5 independent tethers) or alone (N = 20 from 5 independent tethers). Bars represent mean and SEM. Significance was calculated using a two-tailed unpaired t-test with Welch’s correction. (I) A representative kymograph showing a gapped DNA tether incubated with 8 nM LD655-PCNA, 8 nM LD555-RFC^ΔBRCT, and 100 nM RPA with 4 mM ATP.

3

Figure S3. Additional data showing PCNA and RFC^ΔBRCT behaviors during fill-in synthesis, related to Figures 4 and 5.

(A) Kymograph and associated trap position changes at a constant force of 8 pN showing fill-in reaction of a gapped DNA tether incubated with 20 nM Polδ, 5 nM LD655-PCNA, 5 nM LD555-RFC^ΔBRCT, and 100 nM RPA with 4 mM ATP and 130 μM dNTPs. PCNA and RFC fluorescence channels are separately shown to confirm the presence of each signal. Pink boxes denote periods of inactive synthesis, green box denotes the period prior to the start of fill-in, and purple box denotes the completion of fill-in. (B) Kymograph and associated trap position changes at a constant force of 8 pN where the fill-in reaction initiated on the gapped DNA tether (denoted by the vertical dotted lines) in a channel containing the same components as (A). The partially filled-in tether was then moved to a different channel containing the same components except Polδ. The fill-in reaction failed to resume, indicating the requirement of free Polδ in solution for continued synthesis.

4

Figure S4. Structural models of PCNA-Polδ assemblies with an additional PCNA-binding protein, related to Figures 4, 5, and 6.

(A) AlphaFold3 prediction of a complex containing PCNA, RFC, Polδ, and a 3’ recessed DNA. Inset shows a zoom-in view of the RFC-Polδ interaction. (B) Cryo-EM structure of the PCNA-FEN1-Polδ ternary complex with a 3’ recessed DNA (PDB: 6tnz) ¹⁷. Inset shows a zoom-in view of the FEN1-Polδ interaction.

5

Figure S5. FEN1 and RFC compete for PCNA binding, related to Figure 6.

(A) Mass distribution for PCNA and RFC incubated in solution with ATPγS. Predicted complexes for the mass peaks are indicated. (B) Mass distribution for PCNA and RFC incubated in solution with ATPγS in the presence of excess FEN1. (C) Representative kymograph showing a gapped DNA tether incubated with 8 nM LD655-PCNA, 8 nM Cy3-RFC, 15 nM FEN1, and 100 nM RPA with 4 mM ATP. (D) Fraction of trajectories on gapped DNA corresponding to CLC (colocalized PCNA and RFC signals) versus PCNA only (no RFC signal) in the absence (N = 46 from 6 independent tethers) or presence (N = 87 from 11 independent tethers) of FEN1. Error bars represent 95% confidence intervals using bootstrapping.

6

Figure S6. Genetic analyses of Rfc1 BRCT mutants and biochemical and single-molecule characterization of the RFC^11A mutant, related to Figure 7.

(A) Immunoblotting (left) and quantification (right) of protein levels for Rfc1-BRCT∆ (deleted residues: 154-230) and rfc1-11A (mutated residues: T166A, R174A, R187A, T189A, K190A, S191A, S193A, S194A, K195A, K208A, K209A) compared to their wild-type counterpart. All examined proteins were C-terminally tagged with a TAP tag. Pgk1 was used as a loading control. Quantification was derived from at least two spore clones per genotype. The relative protein levels of Rfc1 over Pgk1 were calculated and normalized to the values for the wild-type. Bars represent mean and SD. Significance was calculated using two-tailed unpaired t-tests. (B) Wild-type Rfc1 tagged with TAP did not lead to genotoxic sensitivity of S. cerevisiae cells. The dotted line separates cells located in different sections of the same spot assay plate. (C) Gel filtration profile of 15 nM ³²P-PCNA loaded with 5 nM wild-type RFC or RFC^11A onto 5 nM nicked pUC19 plasmid in the presence of 1 mM ATP. Profile shows an earlier peak (fractions 7-17) constituting DNA-bound PCNA and a later peak (fractions 8-31) composed of free, unbound PCNA. (D) Representative kymograph showing PCNA diffusion along the dsDNA arm of a gapped DNA tether loaded by RFC^11A. (E) Diffusion coefficients (D) for PCNA trajectories sliding on DNA when loaded by RFC (N = 22 from 5 independent tethers), RFC^ΔBRCT (N = 6 from 6 independent tethers), or RFC^11A (N = 7 from 5 independent tethers). Bars represent mean and SEM. Significance was calculated using two-tailed unpaired t-tests with Welch’s correction.

7

Figure S7. Additional genetic analyses of Rfc1 BRCT mutants, related to Figure 7.

(A) Genetic interactions of rfc1-BRCT∆ and rfc1-11A with mutants lacking the Mec1 protein or the RNase H1 and RNase H2 proteins using tetrad analyses. The dotted line separates tetrads located in different sections of the same dissection plate. (B) Genetic interactions of rfc1-BRCT∆ with Mec1 or RNase H1 and RNase H2 mutants in the absence or presence of MMS. Both WT and mutant RFC1 strains used contain a TAP tag at their C-termini.

Highlights.

• RFC frequently remains associated with PCNA after loading it onto DNA

• The PCNA-RFC complex co-travels with DNA polymerase δ during fill-in synthesis

• The Rfc1 BRCT domain is important for PCNA-RFC association and Polδ processivity

• FEN1 and RFC can both play a structural role in processive DNA synthesis