A second DNA binding site on RFC facilitates clamp loading at gapped or nicked DNA

  1. Xingchen Liu
  2. Christl Gaubitz
  3. Joshua Pajak
  4. Brian A Kelch

  • Curated by eLife

    eLife logo

    Evaluation Summary:

    Replication Factor C (RFC) is known to play a role in both DNA replication and DNA repair by loading a protein clamp called PCNA onto DNA junctions with a 3'-recessed end. The current paper elegantly demonstrates that RFC has a second DNA binding site that recognizes a single strand-double strand DNA with a 5'-recessed junction. The paper reports a series of interesting structures and confirms binding to both short gapped DNA and nicked DNA by RFC, causing local unwinding DNA at the ssDNA/dsDNA junctions. The paper, which is of interest to colleagues studying DNA replication and repair, should be improved through a few clarifications.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1, Reviewer #2 and Reviewer #3 agreed to share their name with the authors.)

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Clamp loaders place circular sliding clamp proteins onto DNA so that clamp-binding partner proteins can synthesize, scan, and repair the genome. DNA with nicks or small single-stranded gaps are common clamp-loading targets in DNA repair, yet these substrates would be sterically blocked given the known mechanism for binding of primer-template DNA. Here, we report the discovery of a second DNA binding site in the yeast clamp loader replication factor C (RFC) that aids in binding to nicked or gapped DNA. This DNA binding site is on the external surface and is only accessible in the open conformation of RFC. Initial DNA binding at this site thus provides access to the primary DNA binding site in the central chamber. Furthermore, we identify that this site can partially unwind DNA to create an extended single-stranded gap for DNA binding in RFC’s central chamber and subsequent ATPase activation. Finally, we show that deletion of the BRCT domain, a major component of the external DNA binding site, results in defective yeast growth in the presence of DNA damage where nicked or gapped DNA intermediates occur. We propose that RFC’s external DNA binding site acts to enhance DNA binding and clamp loading, particularly at DNA architectures typically found in DNA repair.

Article activity feed

  1. Version published to 10.7554/elife.77483 on eLife
  2. eLife

    Author Response

    Reviewer #2 (Public Review):

    A 2 amino-purine fluorescence-based DNA melting assay is used to show that both 3' and 5' recessed DNA molecules (with ATPgS) exhibit an increase in fluorescence interpreted to mean that melting occurred (Supplementary Figure 1.4). Given the structure-based finding that the first molecule must bind first to enable binding of the second molecule, is it surprising that the 5' recessed molecule on its own is bound and melted (i.e., without the 3' recessed molecule binding first)?

    We thank the reviewer for the support. Our model (Figure 6) suggests that the non-primer duplex (second molecule, with a recessed 5’ end) binds and melts first at RFC’s external DNA binding site. Our 2AP experiments showing that the 5’ recessed molecule binds and is melted on its own (Figure 4—figure supplement 2) agrees with this model.

  3. eLife

    Evaluation Summary:

    Replication Factor C (RFC) is known to play a role in both DNA replication and DNA repair by loading a protein clamp called PCNA onto DNA junctions with a 3'-recessed end. The current paper elegantly demonstrates that RFC has a second DNA binding site that recognizes a single strand-double strand DNA with a 5'-recessed junction. The paper reports a series of interesting structures and confirms binding to both short gapped DNA and nicked DNA by RFC, causing local unwinding DNA at the ssDNA/dsDNA junctions. The paper, which is of interest to colleagues studying DNA replication and repair, should be improved through a few clarifications.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1, Reviewer #2 and Reviewer #3 agreed to share their name with the authors.)

  4. eLife

    Reviewer #1 (Public Review):

    DNA replication and DNA repair require the loading of proteins onto DNA that in turn recruit other proteins that either synthesize new DNA or repair gaps in DNA. One of these proteins is the Proliferating Cell Nuclear Antigen, an autoantigen in some autoimmune disorders, that is loaded onto DNA by Replication Factor C. The PCNA clamp is loaded by RFC onto ssDNA/dsDNA junctions with a 3'-recessed end during DNA replication, but how RFC/PCNA recognize double-stranded DNAs with a single strand nick or a short single-strand gap has not been known. The current paper nicely confirms that RFC has a second DNA binding site that is different from the well-known DNA binding site that recognizes ssDNA/dsDNA junctions with a 3'-recessed end. The new site binds to ssDNA/dsDNA junctions with a 5'-recessed end and the paper shows that RFC can locally unwind DNA at the ssDNA/dsDNA junctions. These structural studies reveal how RFC/PCNA can recognize DNA with a short gap or a nick for subsequent DNA repair.

  5. eLife

    Reviewer #2 (Public Review):

    Sliding clamps and clamp loaders are essential macromolecular complexes that enable DNA synthesis, as well as a diverse collection of other DNA maintenance reactions. Sliding clamps are toroidal ensembles (in eukaryotes, they are trimers termed PCNA) that bind to DNA via topology (in addition to biochemical interactions). The closed protein PCNA clamp rings require specialized catalysts, known as a clamp loader for assembly onto DNA. Clamp loaders are five-membered AAA+ ATPase machines that couple ATP binding and hydrolysis to crack open the clamp, select and insert the correct DNA substrates into the opened ring. The clamp loader then resets itself via ATP hydrolysis and release of the closed clamp. Remarkably, the structure and function of sliding clamps and clamp loaders are conserved in all domains of life.

    The eukaryotic clamp loader is comprised of five unique subunits (termed: RFC1, RFC2, RFC3, RFC4, and RFC5). Each feature shared as well as unique domains. Each RFC subunit contains the AAA structural module (referred to as AAA-ATP and AAA-lid). The heteropentameric RFC clamp loader is held together, so to speak, by a rigid collar formed out of helical domains contributed by each subunit. The eukaryotic sliding clamp is a trimer of identical PCNA subunits. The RFC clamp loader has been previously found to prefer 3' recessed primer-template junctions onto which the PCNA trimer is loaded. Primer-template junctions are appropriate substrates for supporting, enabling, and stimulating DNA synthetic processes. Remarkably, however, the RFC loader is not the only clamp loader described in eukaryotes. It has been known for nearly 20 years that the substitution of RFC1 for another RFC1-like subunit creates a new entity with apparently disparate biochemical properties. For example, the substitution of Rad24, an RFC1-like protein, for RFC1 creates a loader ensemble with properties distinct from the archetypical RFC loader. Unlike the RFC loader, the RAD24 loader binds to 5' recessed DNA, but not 3' recessed DNA, for loading the 911 clamp, but not the PCNA clamp.

    Insightful and thoughtful studies by many groups over the past decades have painted a reasonably detailed picture of the structure and function of clamps and clamp loaders. Notably, Kelch and coworkers produced a series of structures of yeast RFC bound to PCNA and primer-template DNA (eLife, 2022). These structures show, inter alia, the details of the primer-template DNA within the loader and the double-stranded portion of the DNA within the PCNA ring. Also, Remus, Hite and co-workers and O'Donnell/Li and co-workers recently published analyses of the Rad24 clamp loader bound to the 911 clamp. An important overall finding of the Rad24 studies was the discovery of a novel DNA binding mode. The Rad24 loader bound the 5' recessed DNA, not within the loader and the clamp, but 'above' (in a specific reference pose) the 911 ring.

    As a follow-up to the 2022 study by Kelch, Liu et al made the fortuitous observation that a second primer-template DNA molecule binds to RFC, in addition to the primer-template molecule that binds within the loader. Remarkably, the second primer-template molecule binds above the PCNA ring in a manner reminiscent of the position occupied by DNA in the Rad24 loader.

    To be sure, biochemical evidence did exist that the picture of DNA binding by RFC provided by prior structural biology was incomplete. Notably, Susan Hardin and coworkers (1998) and Gregg Siegel and coworkers (2006/2010) had investigated the BRCT domain found at the N-terminus of the RFC1 subunit (Drosophila and Human) and had found that, in isolation, this domain bound to one DNA molecule with a preference for a recessed 5' DNA structure. The finding that RFC had binding sites for 3' and 5' recessed DNA was not accommodated in the extant crystal and EM structures.

    To review this manuscript, I considered the narrative, 6 figures, 2 supplementary tables, 11 supplementary figures, 4 sets of coordinates and PDB validation reports, and 4 sets of accompanying EM density maps.

    This manuscript encompasses the following items:

    1. 4 atomic models of the RFC-clamp loader determined using cryo-EM
      a. RFC bound to PCNA, ATPgS, and two primer-template DNA molecules (3.4 Å)
      b. RFC-PCNA-ATPgS-5-nt-gap (3.0 Å);
      c. RFC-PCNA-ATPgS-6nt-gap (3.0 Å);
      d. RFC-PCNA-ATPgS-singly-nicked-DNA 3.7 Å.

    Supplementary table 1 and the PDB validation reports speak to the high-quality cryo-EM analyses performed by Liu and co-workers.

    The fortuitous finding of two primer-template molecules in the RFC structure also revealed the structure and role of the BRCT domain found at the N-terminus of RFC1. The first primer-template molecule binds with the RFC loader and extends into the central chamber of the PCNA clamp.

    On the other hand, the second primer-template molecule binds above the PCNA ring (in a certain RFC pose). The binding site of the second molecule is situated entirely on the RFC1 subunit. Contacts to the duplex portion of the primer template arise from the 'top' of the AAA-ATP, the AAA-lid, and the collar domains of RFC1 as well as the ordered BRCT domain. This is the first time that the BRCT domain has been visualized in any RFC structure. The ssDNA segment of the primer-template molecule is directed into the so-called A-gate and into the RFC loader.

    In addition to a previously described 'separation pin' within the RFC loader that separates the very 3' end of the primer-template junction, the authors identify a second 'separation pin' that appears to separate the DNA strands of the second molecule; this second pin is located on the collar domain.

    Comparison to the previously determined yeast RFC structures from the Kelch lab led the authors to conclude that the internal site must bind DNA first; binding triggers a series of structural changes that create the binding site for the second DNA molecule.

    1. A 2 amino-purine fluorescence-based DNA melting assay is used to show that both 3' and 5' recessed DNA molecules (with ATPgS) exhibit an increase in fluorescence interpreted to mean that melting occurred (Supplementary Figure 1.4).

    Given the structure-based finding that the first molecule must bind first to enable binding of the second molecule, is it surprising that the 5' recessed molecule on its own is bound and melted (i.e., without the 3' recessed molecule binding first)?

    1. The finding that RFC binds to two DNA molecules leads the authors to 'link' the two molecules and explore the substrate requirements of the resulting single DNA molecule, which features a series of gaps or a single nick. Using ATPase and 2 AP fluorescence assays, the authors examine the relationship between gap size and ATPase and base-melting activity. This series of experiments suggest that a gap size of 6 might be the appropriate length to link the internal and external DNA binding sites, but that a gap size of 4 experienced the greatest extent of base melting.

    2. To gain further insights into the RFC-PCNA-extended DNA complex, the authors performed three additional structure determinations this time with a single DNA scaffold, but which included 5, 6 nt gaps, and a single nick. The structures with the 5/6 gapped DNA molecule resembled the two-DNA RFC-PCNA complexes, with an additional ssDNA segment that linked the DNAs in the two sites. The finding that both structures featured a 6 NT gap implied that the 5 NT gap structure had experienced melting of a base pair. Together, these structures provide crucial support for the idea that RFC, PCNA can natively bind an extended DNA molecule and that molecules with small gaps were physiological substrates for RFC.

    The structural effort that encompassed the nicked DNA goes on to probe the idea that melting takes place at the second, external, DNA binding site. The nick-containing DNA structure overall resembles the others from an overall perspective. However, this structure also featured a five NT segment that appeared to be stretched relative to its counterparts in the gapped DNA structures. Moreover, both the internal and external DNA molecules show signs of having been melted by the internal and external separation pins; 3 bp at the internal site and 1 bp at the external are disrupted; the sum (4 nt) leaves one nucleotide unseen in the density maps. The DNA melted at the internal site is directed towards a channel near/underneath the collar domain.

    1. The authors start the process of testing their structural models by examining the response of yeast whose RFC1 lacks the BRCT domain. The growth of such a yeast strain is challenged by three DNA damaging agents (MMS, HU, and UV light) with distinct mechanisms of action. Of these, only the DNA methylating agent methylmethane sulphonate (MMS) reveals any type of yeast growth defect. The other agent: hydroxyurea, which damages DNA by compromising nucleotide pools and UV light appears to show no growth defects. This finding points to the involvement of RFC in aspects of base excision repair (BER), which repairs methylation damage. This study is undoubtedly the first of many by Kelch and other groups.

    Liu et al synthesize their current and prior findings, and those of other groups, into a structure-based model to explain how RFC could recognize and load PCNA onto nicked DNA. Their structural work also establishes a previously unrecognized link to the DNA binding mode exhibited by the Rad24 clamp loader.
    Overall, the data in the manuscript are of high quality and the accompanying narrative and figures are well presented.

  6. eLife

    Reviewer #3 (Public Review):

    This review is carried out with the caveat that I am not a structural biologist and therefore, cannot judge the correctness of the structures presented in this paper. Several cryoEM structures are presented of yeast RFC with PCNA and ATPgammaS. Many clamp-clamp loader structures are already available, including several from these authors in a 2022 eLife paper. However, the focus of this paper is on the BRCT domain that resides in the N-terminal region of the Rfc1 subunit. The authors used gapped DNA substrates with 5 or 6 nucleotides of single-stranded DNA, in which PCNA is found in a post-loading closed conformation with the 3'-junction buried inside the clamp loader. The 5'-junction and the double-stranded DNA beyond the junction are bound by the BRCT domain. Interestingly, when the gap is zero, i.e. a nick, the DNA at both the 3'- and 5'-junction is melted out such that the intervening ssDNA is 5 nucleotides.

    These are fascinating structures that are suggestive of a potential role of the RFC NTD in DNA repair, such as base excision repair. Indeed, a study in yeast cells shows that a BRCT deletion of Rfc1 results in sensitivity to an alkylating agent. This paper would have had a higher impact if more directed genetic experiments were carried out, which identified the pathway benefitting from the presence of the BRCT motif.

    My major concern with this study is in the choice of DNA substrate. Previous biochemical studies from several labs have shown that binding of 5'-junction DNA to an isolated BRCT domain strongly depends on the presence of the 5'-phosphate. Other single BRCT domains that bind DNA, e.g. from Rev1, also show a strong dependence on the 5'-phosphate. DNA repair intermediates, such as base excision repair products after incision by Apn1/2, carry 5'-phosphates. Very surprisingly, the DNA substrates used in this study lack the physiologically relevant 5'-phosphate. The only experiment in the paper that indirectly addresses the issue is in Fig. 4A; it shows that the melting of the 5'-nucleotide occurs independently of the presence of the phosphate. There is no discussion of why the authors chose the unphosphorylated DNA substrate. If the phosphate indeed is an important feature, it would benefit the authors to determine cryoEM studies with the proper DNA.

  7. Version published to 10.1101/2022.01.31.478526v1 on bioRxiv