1. Biochemistry and Chemical Biology
  2. Chromosomes and Gene Expression
Download icon

Nucleosomes influence multiple steps during replication initiation

  1. Ishara F Azmi
  2. Shinya Watanabe
  3. Michael F Maloney
  4. Sukhyun Kang
  5. Jason A Belsky
  6. David M MacAlpine
  7. Craig L Peterson
  8. Stephen P Bell  Is a corresponding author
  1. Howard Hughes Medical Institute, Massachusetts Institute of Technology, United States
  2. University of Massachusetts Medical School, United States
  3. Institute for Basic Science, South Korea
  4. Duke University Medical Center, United States
  5. Duke University, United States
Research Article
  • Cited 27
  • Views 2,638
  • Annotations
Cite this article as: eLife 2017;6:e22512 doi: 10.7554/eLife.22512

Abstract

Eukaryotic replication origin licensing, activation and timing are influenced by chromatin but a mechanistic understanding is lacking. Using reconstituted nucleosomal DNA replication assays, we assessed the impact of nucleosomes on replication initiation. To generate distinct nucleosomal landscapes, different chromatin-remodeling enzymes (CREs) were used to remodel nucleosomes on origin-DNA templates. Nucleosomal organization influenced two steps of replication initiation: origin licensing and helicase activation. Origin licensing assays showed that local nucleosome positioning enhanced origin specificity and modulated helicase loading by influencing ORC DNA binding. Interestingly, SWI/SNF- and RSC-remodeled nucleosomes were permissive for origin licensing but showed reduced helicase activation. Specific CREs rescued replication of these templates if added prior to helicase activation, indicating a permissive chromatin state must be established during origin licensing to allow efficient origin activation. Our studies show nucleosomes directly modulate origin licensing and activation through distinct mechanisms and provide insights into the regulation of replication initiation by chromatin.

https://doi.org/10.7554/eLife.22512.001

eLife digest

Each human cell contains more than two meters of DNA. To fit this length into a cell, remodeling enzymes compact the DNA by helping it to bind to specific proteins. This compaction has the side effect of making the DNA harder to access.

DNA replication is one process that requires access to the DNA. Replication occurs each time a cell divides, so that each newly formed cell receives a full set of genetic material. DNA replication starts simultaneously at hundreds of sites across the DNA. At each of these sites, cells assemble a protein called a replicative helicase. Helicases play a important role in many steps of DNA replication, but their most fundamental role is to separate the two DNA strands that make up the double helix; these strands then act as templates during replication.

A helicase is initially inactive when loaded at a replication start site. Additional proteins then bind to the helicase to activate it. Studies have shown that DNA compaction influences DNA replication, but it was not known exactly how compacted DNA affects helicase loading and activation.

To investigate the effects of compacted DNA during replication in more detail, Azmi et al. created different types of compacted DNA molecules using various remodeling enzymes. Some of the compacted DNAs directly prevented the binding of a protein that is required to load the helicase to the replication start site. In addition, the compaction reduced the number of sites on the DNA where replication could begin. Other types of compacted DNA allowed the helicase to be loaded normally, but inhibited the subsequent activation of the helicase. However, treating these DNA types with particular remodeling enzymes restored helicase activation to normal levels.

Overall, the findings presented by Azmi et al. suggest that cells can control helicase loading and activation independently by compacting DNA in different ways. Such control is important to ensure that each time a cell divides, it fully replicates its entire DNA.

https://doi.org/10.7554/eLife.22512.002

Introduction

The eukaryotic genome is packaged into a condensed form known as chromatin that presents a barrier to DNA-associated processes. Chromatin is primarily composed of nucleosomes, each of which consists of ~147 base pairs of DNA wrapped around a histone octamer. The location and modification state of nucleosomes is dynamic, regulates access to the DNA and partitions the genome into distinct chromatin states (Clapier and Cairns, 2009). Nucleosome positioning and modifications influence all DNA processes including replication, transcription, repair and recombination. Thus, maintaining appropriate chromatin states across the genome is critical for cellular viability (Hargreaves and Crabtree, 2011). Although there is a growing wealth of knowledge concerning the impact of nucleosomes on gene expression, significantly less is known about the role of nucleosomes in regulating DNA replication.

Proper eukaryotic DNA replication requires the temporal separation of two key events: origin licensing and origin activation (Li and Araki, 2013; Siddiqui et al., 2013). During G1, origin licensing is initiated by origin-recognition complex (ORC) binding to replication origin DNA. ORC then recruits Cdc6 and Cdt1 and these proteins load two inactive Mcm2-7 replicative DNA helicases around the origin DNA (Bell and Labib, 2016). Origin activation is temporally separated from origin licensing and occurs during S phase. S-phase cyclin-dependent kinases and the Dbf4-dependent Cdc7 kinase (DDK) drive recruitment of two helicase-activating proteins, Cdc45 and GINS, forming the active replicative helicase, the Cdc45/Mcm2-7/GINS (CMG) complex (Ilves et al., 2010; Tanaka and Araki, 2013). Recruitment of DNA polymerases and their accessory proteins to the CMG complex forms a bidirectional pair of replisomes. The majority of these events have been reconstituted in vitro using purified proteins and naked DNA (Yeeles et al., 2015, 2017).

Chromatin is proposed to influence multiple aspects of replication initiation including origin licensing, origin activation and the time of replication initiation within S phase. Origin DNA is nucleosome-free to allow ORC DNA binding and, once bound, ORC positions origin-proximal nucleosomes (Berbenetz et al., 2010; Eaton et al., 2010). Repositioning of origin-proximal nucleosomes reduces helicase loading and origin function (Lipford and Bell, 2001; Simpson, 1990), and loaded helicases appear to interact with these nucleosomes (Belsky et al., 2015). Local chromatin states have also been implicated in the activation of eukaryotic origins, each of which is predisposed to initiate earlier or later in S phase (Rhind and Gilbert, 2013). How chromatin modulates these events and whether specific chromatin regulators impact replication initiation events is unclear.

Chromatin-remodeling enzymes (CREs) play a major role in determining the chromatin landscape across the genome (Struhl and Segal, 2013). CREs are multi-protein complexes that use the energy of ATP binding and hydrolysis to assemble, move, slide or alter the composition of nucleosomes (Clapier and Cairns, 2009; Papamichos-Chronakis and Peterson, 2012). Four subfamilies of CREs are conserved from yeast to humans: ISWI, SWI/SNF, INO80, and CHD. Members of the ISWI and CHD subfamilies typically function in nucleosome assembly, and they can create regularly spaced nucleosomal arrays by ATP-dependent sliding of nucleosomes (Hamiche et al., 1999; Längst et al., 1999). Similarly, members of the SWI/SNF subfamily mobilize nucleosomes in cis, but these enzymes can also evict nucleosomal histones or eject entire nucleosomes (Clapier et al., 2016). Consequently, these CREs typically promote enhanced accessibility of nucleosomal DNA. Finally, members of the INO80 subfamily conduct the post-replicative removal of a particular histone within a nucleosome, and sequential replacement with either a canonical or a variant histone, a process termed nucleosome editing (Mizuguchi et al., 2004; Papamichos-Chronakis et al., 2011). Notably, some members of the INO80 subfamily can also catalyze nucleosome sliding (e.g. yeast INO80-C) (Shen et al., 2003). Although different CREs can exert a differential impact on nucleosomes, the current view is that each of these enzymes use ATP-dependent DNA translocation as a central mechanism for their activities.

Various CREs have been implicated in the regulation of DNA replication (MacAlpine and Almouzni, 2013). For instance, ISW1-containing remodeling complexes interact with replisome proteins (Poot et al., 2005) and Chd1 negatively regulate replication initiation (Biswas et al., 2008). Similarly, SWI/SNF stimulates replication initiation at specific yeast origins (Flanagan and Peterson, 1999) and is associated with a subset of human origins (Euskirchen et al., 2011).

Although elimination of different CREs influences DNA replication, whether these effects are direct or indirect and the specific events of replication that are impacted remain elusive. CREs impact multiple processes including transcription, histone modification, and nucleosome assembly (Clapier and Cairns, 2009) leaving open the possibility of indirect effects. In addition, cells express multiple members of each CRE family and overlapping functions of these enzymes could mask the effects of single CRE deletions (Tsukiyama et al., 1999). Although the simultaneous deletion of multiple CREs could overcome this issue, in many cases these are lethal events (Monahan et al., 2008; Tsukiyama et al., 1999).

Here we describe origin-dependent in vitro replication assays using nucleosomal DNA templates. To address how different nucleosomal states impact DNA replication, we investigated nucleosomal templates that were remodeled by different CREs. Consistent with in vivo studies, these templates showed distinct replication capacities. Most of the nucleosomal DNA templates permitted origin licensing, but ISW2- and Chd1-remodeled templates reduced the efficiency of this event by positioning nucleosomes over the origin DNA, decreasing ORC DNA binding and helicase loading. Although permissive for origin licensing, SWI/SNF- and RSC-remodeled templates showed reduced CMG formation and origin activation. Addition of specific CREs improved replication initiation from these templates but only if the CRE was added prior to CMG formation. Our findings show that local nucleosome status differentially modulates two steps during replication initiation and that specific CREs establish permissive and restrictive states for replication initiation.

Results

Reconstitution of Mcm2-7 helicase loading using nucleosomal DNA

To investigate the impact of chromatin on replication initiation, we first reconstituted origin licensing using nucleosomal DNA templates. To this end, we used purified ISW1a, Nap1 and budding yeast histone octamers (Figure 1—figure supplement 1A and B) to assemble nucleosomes on a 3.8 kb linear fragment of Saccharomyces cerevisiae DNA that included the ARS1 replication origin (Mizuguchi et al., 2012). We optimized the ratio of DNA to histone octamers to assemble regularly-spaced nucleosome arrays (Figure 1A and Figure 1—figure supplement 2A). After nucleosomes were remodeled, ISW1a, Nap1 and free histones were removed from the template (Figure 1—figure supplement 2B) to provide a defined nucleosomal DNA state by preventing additional nucleosome assembly and remodeling.

Figure 1 with 3 supplements see all
Mcm2-7 helicase loading onto nucleosomal DNA templates.

(A) Nucleosomes were remodeled with bead-coupled ARS1-containing linear DNA, ISW1a, yeast histone octamers and Nap1. Nucleosome assembly was assessed after partial MNase digestion. (B) Outline of the helicase-loading assay using nucleosomal DNA. (C). Comparison of helicase loading on naked DNA and on ISW1a-remodeled nucleosomal DNA. DNA templates were washed with high-salt (H) or low-salt (L) buffer after loading. Template-associated Mcm2-7, ORC and H2B was detected by immunoblot. (D) Helicase loading onto either wild-type (WT) or A-B2- (mut) (Heller et al., 2011) ARS1-containing DNA. As indicated, nucleosomal DNA was remodeled with ISW1a. Assays were performed in either 125 mM (to allow increased origin non-specific helicase loading) or 300 mM (origin specific helicase loading) potassium glutamate. After a high salt wash, DNA-associated Mcm2-7 was detected by immunoblot.

https://doi.org/10.7554/eLife.22512.003

Using purified ORC, Cdc6, Cdt1 and Mcm2-7 (Kang et al., 2014), we compared the ability of nucleosomal and naked DNA templates to participate in origin licensing as measured by loading of the Mcm2-7 helicase (Figure 1B). At the end of the reaction, DNA-beads were washed with a low- (L) or high-salt (H) containing buffer. The low-salt wash retains all DNA-associated proteins whereas the high-salt wash releases ORC, Cdc6, Cdt1 and incompletely-loaded Mcm2-7 but retains loaded Mcm2-7 complexes associated with successful origin licensing (Donovan et al., 1997; Randell et al., 2006). The amount of ORC DNA binding, helicase association (low-salt wash [L]) and helicase loading (high-salt wash [H]) were comparable between nucleosomal and naked DNA templates (Figure 1C). Thus, ISW1a-remodeled nucleosomes are permissive for origin licensing.

To address the effect of nucleosomes on origin selection, wild-type (WT) and mutant ARS1-containing DNA was assembled into nucleosomes and helicase loading was performed under lower-salt conditions that allow Mcm2-7 loading at non-origin sequences (compare upper and lower panels of (Figure 1D). Under these conditions, nucleosome assembly reduced non-specific Mcm2-7 loading onto mutant ARS1-containing DNA without altering helicase loading onto WT DNA (Figure 1D, top panel). Thus, nucleosomes reduced origin licensing at non-origin DNA sequences, consistent with previous in vivo studies implicating local nucleosomes in origin selection (Berbenetz et al., 2010; Eaton et al., 2010).

Origin-proximal nucleosome positioning influences helicase loading

To address how different local nucleosome landscapes influence replication initiation, we generated ARS1 origin DNA templates with distinct nucleosome patterns. To this end, we assembled nucleosomes onto origin DNA in the presence of seven different purified CREs: ISW1a, ISW1b, ISW2, INO80-C, Chd1, SWI/SNF and RSC (Figure 1—figure supplement 1B). The amount of CRE added was normalized according to their relative ATPase activity (Figure 1—figure supplement 3), (Smith and Peterson, 2005). After nucleosome assembly, the CRE, Nap1 and non-nucleosomal histones were removed (Figure 1—figure supplement 2B) to ensure that the nucleosomes deposited during assembly are not remodeled or moved during subsequent replication-initiation assays. First, we examined nucleosome assembly by partial MNase-digestion. ISW1a, ISW1b, INO80-C, ISW2 and Chd1 each resulted in regularly-spaced nucleosomes on the origin DNA, albeit with different spacings (Figure 2A). In contrast, SWI/SNF- and RSC-remodeled nucleosomes did not show evidence of uniformly-spaced nucleosomes, consistent with previous observations (Flaus and Owen-Hughes, 2003; Kassabov et al., 2003). It was possible that SWI/SNF and RSC treatment reduced or eliminated nucleosome assembly. To test this hypothesis, we compared the amount of DNA-associated H2B and H3 (Figure 2B and Figure 2—figure supplement 1A) and the amount of mono-nucleosomal DNA produced after extensive MNase treatment (Figure 2—figure supplement 1B). These studies showed that the presence of different CREs did not dramatically change the extent of nucleosome formation. For simplicity, we refer to the different nucleosomal DNA templates by the CRE present during their assembly (e.g. SWI/SNF template).

Figure 2 with 2 supplements see all
Comparison of helicase loading onto nucleosomal DNA templates remodeled with different CREs.

(A) Comparison of nucleosome assembly with different CREs. Nucleosomes were remodeled with the indicated CRE and assayed by partial MNase digestion. (B) Helicase loading onto nucleosomes remodeled with different CREs. After helicase loading, DNA was washed either with high-salt (H) or low-salt (L) buffer. Mcm2-7 and H2B DNA association was detected by immunoblot. (C) Comparison of origin-proximal nucleosome positioning established by different CREs. The positions of nucleosome dyads remodeled with the indicated CRE were analyzed by high-throughput MNase-Seq. Nucleosome dyad density (Y-axis) and the corresponding position of the dyad (X-axis) are plotted. Zero on the X-axis indicates the first nucleotide of the ARS1 consensus sequence (ACS). The elements of ARS1 (Marahrens and Stillman, 1992) are indicated above. (D) ORC association with nucleosomal DNA remodeled with different CREs. Template association of ORC was detected by immunoblot. (E) Addition of ORC during nucleosome assembly restores helicase loading on ISW2 and Chd1 templates. Nucleosomes were assembled onto ARS1 DNA with the indicated CRE in the presence or absence of ORC. Helicase loading was performed and analyzed as described in (B).

https://doi.org/10.7554/eLife.22512.007

We examined each of the different nucleosomal templates for origin licensing. SWI/SNF and RSC templates showed levels of Mcm2-7 loading similar to ISW1a templates (Figure 2B, Figure 2—figure supplement 1C and Figure 2—figure supplement 1—source data 1 and 2). ISW1b and INO80-C templates showed modest reductions in loaded Mcm2-7 and Chd1 and ISW2 templates showed progressively less loading. Thus, the CRE present during nucleosomal assembly impacted the extent of origin licensing.

Local nucleosomes reduce helicase loading by inhibiting ORC DNA binding

To investigate the cause of the differential origin licensing, we determined the position of origin-proximal nucleosomes for the ISW1a, ISW1b, INO80-C, Chd1 and ISW2 templates using MNase-seq (Cole et al., 2012; Eaton et al., 2010). The ISW1a template showed a nucleosome-free region (NFR) overlapping ARS1 with well-defined flanking nucleosomes (Figure 2C). In contrast, ISW2 template showed the appearance of a positioned nucleosome overlapping the origin (centered at −54 bp relative to ACS, Figure 2C). In addition, the flanking nucleosome on the opposite side of the origin was shifted towards the ACS (from +222 to +168) in the ISW2 templates. These data support a model in which encroachment of origin-proximal nucleosomes onto origin DNA directly inhibits origin licensing.

To determine whether the reduced origin licensing of the ISW2 and Chd1 templates was caused by decreased ORC DNA binding, we examined ORC association with these nucleosomal templates (Figure 2D). The extent of ORC binding to the ISW2 and Chd1 templates correlated with the amount of Mcm2-7 loading (Figure 2B and D). We asked if addition of ORC during nucleosome-assembly reactions restored Mcm2-7 loading. Importantly, when ORC bound DNA prior to Chd1- or ISW2-directed nucleosome assembly, loaded Mcm2-7 levels were restored to levels similar to ISW1a templates (Figure 2E). Together, these data indicate that nucleosome positioning over the origin reduces origin licensing by inhibiting ORC DNA binding and that ORC is not sufficient to move nucleosomes in the absence of a CRE.

To further investigate the role of ORC in the establishment of Mcm2-7-loading-competent chromatin states, we evaluated the role of the Orc1 bromo-adjacent homology (BAH) domain. BAH domains bind to nucleosomes (Yang and Xu, 2013) and elimination of the Orc1 BAH domain reduces initiation from a subset of replication origins in yeast (Müller et al., 2010). We purified ORC lacking the Orc1 BAH domain (ORC∆BAH, Figure 2—figure supplement 2A) and performed helicase-loading assays using ISW1a, ISW2 and INO80-C templates (Figure 2—figure supplement 2B–C). Consistent with the limited effect of deletion of the Orc1 BAH domain on ARS1 function in vivo (Müller et al., 2010), ORC and ORC∆BAH showed comparable levels of helicase loading onto all the nucleosomal templates.

We also examined whether the presence of the ARS1-binding protein, Abf1, influenced helicase loading in the presence of nucleosomes. Previous studies showed that Abf1 and ORC position nucleosomes on either side of ARS1 (Lipford and Bell, 2001) and that elimination of the Abf1 binding sites reduced ARS1 function (Marahrens and Stillman, 1992). Addition of purified Abf1 (Figure 2—figure supplement 2D) to either naked DNA or ISW1a templates did not improve helicase loading (Figure 2—figure supplement 2E). We also asked whether addition of Abf1 to the ISW2 nucleosome assembly would rescue the helicase-loading defects of ISW2 templates, as we observed for ORC (Figure 2E). In contrast to ORC, Abf1 did not improve helicase loading on the ISW2 template (Figure 2—figure supplement 2F).

Local nucleosomes impact replication events downstream of helicase loading

Next, we examined the effect of nucleosomes on replication-initiation events after origin licensing had occurred. To this end, we performed replication assays (Gros et al., 2014; Heller et al., 2011; On et al., 2014) by sequentially adding DDK and an S-phase extract to helicases loaded onto DNA templates with or without nucleosomes (Figure 3A). ISW1a templates showed comparable levels of replication products to that of naked DNA (Figure 3B). Nucleotide incorporation was Cdc6- (Figure 3B), DDK- (Figure 3C, and Figure 3—source data 1) and origin-sequence-dependent (Figure 3—figure supplement 1) indicating that the DNA synthesis observed was due to replication initiation and elongation (rather than DNA repair).

Figure 3 with 1 supplement see all
Replication initiation on nucleosome templates.

(A) Outline of nucleosomal DNA replication initiation assay using purified proteins and yeast S-phase cell extract. (B) ISW1a templates do not interfere with replication. Naked DNA or ISW1a templates were assayed in the presence or absence of Cdc6. Radiolabeled replication products were analyzed by alkaline agarose electrophoresis and autoradiography (top). Template-associated H2B was detected by immunoblot (lower). (C) Comparison of replication using ISW1a, ISW1b, ISW2, INO80-C and Chd1 templates in the presence and absence of DDK. Products of the extract-based replication assays were analyzed as in (B, top). H2B levels for each template are shown (middle). Quantification of replication products was performed as in Figure 2B. Error bars show the SD (n = 3, lower). (D) Comparison of replication of ISW1a, SWI/SNF and RSC templates in the presence or absence of DDK. Analysis of replication products, template-associated H2B and quantification (n = 3) as in (C).

https://doi.org/10.7554/eLife.22512.012
Figure 3—source data 1

Raw values used in the quantification of Figure 3C (n = 3).

https://doi.org/10.7554/eLife.22512.013
Figure 3—source data 2

Raw values used in the quantification of Figure 3D (n = 3).

https://doi.org/10.7554/eLife.22512.014

To compare replication initiation from nucleosome templates remodeled with different CREs, we carried out the same replication assay with each template. For ISW1a, ISW1b, ISW2, INO80-C and Chd1, the level of replication products closely matched the amount of helicase loading with the same templates (compare Figure 3C and Figure 2—figure supplement 1C). Thus, origin activation and replisome assembly were not further reduced by these nucleosomal templates. In contrast, the RSC and SWI/SNF templates showed a disconnect between the extent of origin licensing and the levels of replication initiation. SWI/SNF, RSC and ISW1a templates showed comparable levels of Mcm2-7 loading (Figure 2B and Figure 2—figure supplement 1C), but the amount of replication products from SWI/SNF and RSC templates was reduced ~5 fold relative to ISW1a templates (Figure 3D and Figure 3—source data 2). Thus, nucleosomal DNA templates remodeled by SWI/SNF and RSC inhibit one or more events downstream of origin licensing.

RSC and SWI/SNF templates impede CMG complex formation

The presence of multiple CREs in the S-phase extract led us to adapt a fully-reconstituted replication-initiation assay (Lõoke et al., 2017; Yeeles et al., 2015) to investigate the cause of the reduced replication of the SWI/SNF and RSC templates (Figure 4A). Compared to the S-phase-extract-based assay, ISW1a templates showed reduced replication using the fully-reconstituted assay (Figure 4—figure supplement 1A), most likely due to a lack of CREs and histone chaperones present in the S-phase-extract-based assay (Devbhandari et al., 2017; Kurat et al., 2017). Nevertheless, replication of the SWI/SNF and RSC templates was similarly reduced relative to their ISW1a-remodeled counterpart using the reconstituted assay (Figure 4B, Figure 4—figure supplement 1B and Figure 4—figure supplement 1—source data 1). Importantly, the reduced replication observed for the RSC or SWI/SNF templates was not simply because of a lack of uniformly-spaced nucleosomes. When we assembled nucleosomes in the absence of any CRE, the resulting nucleosomes were similarly non-uniformly spaced (Figure 4—figure supplement 2A) but the levels of replication from these templates were comparable to ISW1a templates (Figure 4—figure supplement 2B). Thus, the reduced replication capacity of the RSC and SWI/SNF templates requires the activity of the corresponding CRE.

Figure 4 with 2 supplements see all
SWI/SNF and RSC templates show reduced CMG formation.

(A) Outline of fully-reconstituted nucleosomal DNA replication initiation assay. The proteins added at each step are indicated. (B) Comparison of reconstituted nucleosomal DNA replication using ISW1a, SWI/SNF and RSC templates in the presence or absence of DDK. Analysis of replication products and H2B as in Figure 3B. (C) Comparison of Mcm2-7 phosphorylation by DDK on ISW1a, SWI/SNF and RSC templates. Phosphorylation of Mcm6 is indicated by reduced electrophoretic mobility and was analyzed by immunoblot (top). Template associated H2B is shown (lower). (D) Comparison of replication of ISW1a, RSC and SWI/SNF templates. Reactions were performed with or without DDK and replication products of the reconstituted replication reactions were analyzed as in Figure 3B (top). Template association of Mcm2-7, Cdc45, GINS and H2B was measured after a high-salt wash at the end of reconstituted replication assay by immunoblot (lower panels). (E) Comparison of CMG formation and activation using ISW1a, SWI/SNF and RSC templates. To prevent replication initiation, the only nucleotide present was ATP and Pol α was left out of the assay. Template association of Mcm2-7, Cdc45, GINS, Rfa1 and H2B were measured by immunoblot.

https://doi.org/10.7554/eLife.22512.016

To identify the replication event(s) that was reduced by SWI/SNF- and RSC-remodeled nucleosomes, we monitored different events of origin activation. First, we examined DDK phosphorylation of Mcm2-7 (detected by retardation of Mcm6 electrophoresis, Francis et al., 2009). This modification was either unchanged (SWI/SNF) or improved (RSC) relative to ISW1a templates (Figure 4C), indicating Mcm2-7 phosphorylation by DDK was not reduced. Next, we assessed CMG formation by examining Cdc45 and GINS template association after replication initiation and elongation (Figure 4D). Both SWI/SNF and RSC templates showed reduced Cdc45 and GINS template association compared to ISW1a templates. For these initial experiments, we measured template association at the end of the replication reaction. Thus, the decreases in Cdc45 and GINS template association could be due to inefficient CMG formation during initiation or increased CMG dissociation during elongation. To distinguish between these possibilities, we repeated the replication-initiation assays in the presence of ATP but without other rNTPs or dNTPs (Figure 4E). Under these conditions, the CMG can form and partially unwind DNA but replication cannot initiate (Yeeles et al., 2015). As in the previous assays, we observed reduced Cdc45 and GINS association with SWI/SNF and RSC templates compared to the ISW1a templates. Consistent with reduced active helicases and DNA unwinding, the amount of Rfa1 (a subunit of the eukaryotic single-stranded DNA binding protein RPA) association with RSC and SWI/SNF templates was also reduced (Figure 4E). Thus, the observed reduction in DNA replication products observed for the SWI/SNF and RSC templates in the complete assays was due to reduced CMG formation and helicase activation.

CRE-specific restoration of replication and CMG formation to RSC and SWI/SNF templates

Our previous replication assays were performed in the absence of CREs to address how different chromatin states impact replication initiation. In vivo, however, these enzymes could be present at origin-proximal chromatin during initiation. To address whether the continuous presence of a CRE during replication initiation altered our findings, we asked if the addition of ISW1a, RSC or SWI/SNF during replication-initiation assays improved replication initiation from the RSC and SWI/SNF templates. Adding ISW1a during the helicase-loading step (DL) of the assay (Figure 5A) restored CMG formation and increased DNA replication of the RSC templates (Figure 5B, compare lanes 4 and 5). In contrast, adding RSC during the helicase-loading step did not alter either the amount of replication or CMG formation (Figure 5B, lane 6). Similar experiments with SWI/SNF templates showed that the addition of ISW1a (but not SWI/SNF) improved replication of SWI/SNF templates (Figure 5—figure supplement 1A). In contrast to the ability of ISW1a to improve replication from the RSC and SWI/SNF templates, addition of RSC or SWI/SNF to ISW1a templates during helicase loading did not reduce replication levels (Figure 5—figure supplement 1B). Interestingly, consistent with its ability to reduce helicase loading, addition of Chd1 to ISW1a during helicase loading did reduce replication. Together these data indicate that the defects that we observe in CMG formation and replication for the RSC and SWI/SNF templates are not simply due to the lack of a CRE during the replication assay. Instead, our findings suggest that specific CREs create nucleosomal states that facilitate CMG formation and replication initiation.

Figure 5 with 4 supplements see all
Rescue of SWI/SNF and RSC template replication initiation.

(A) Schematic of ISW1a addition at various steps during the replication assay. (B) Addition of ISW1a at the helicase-loading step rescues replication initiation from RSC templates. Reconstituted replication assays were performed on ISW1a and RSC templates with or without DDK. ISW1a or RSC was added to the templates during helicase loading and not deliberately removed (DL) or upon addition of the helicase activation and elongation proteins (I/E) as indicated. The lane that show I/E is from the same gel as the rest of the panel. Replication products (top panel) and Mcm2-7, Cdc45, GINS and H2B template association (lower panels) were assayed as in Figure 4D. (C) Specific CREs improve RSC-template replication. Reconstituted replication assays were performed with RSC templates with or without DDK. ISW1a, ISW1b, INO80-C or ISW2 was added to RSC templates after helicase loading (AL). Replication products (top) and Mcm2-7, Cdc45, GINS and H2B template association (lower panels) were assayed as in Figure 4D. (D) Addition of ISW1a after nucleosome assembly facilitates replication and CMG formation of SWI/SNF templates. Reconstituted replication assays were performed with ISW1a or SWI/SNF templates with or without DDK. ISW1a was added to the templates either during helicase loading (DL), after helicase loading (AL) or upon addition of the helicase activation and elongation proteins (I/E) as indicated. The lane that show I/E is from the same gel as the rest of the panel. Replication products (top) and Mcm2-7, Cdc45, GINS and H2B template association (lower panels) were assayed as in Figure 4D. (E) ISW1a addition after helicase loading (AL) to RSC templates, but removed before helicase activation improves replication of and CMG complex formation on RSC templates. Reconstituted replication reactions were performed with the indicated templates with or without DDK. ISW1a was added to the RSC templates upon completion of helicase loading (AL). Replication products (top) and Mcm2-7, Cdc45, GINS and H2B template-association (lower panels) were assayed as in Figure 4D.

https://doi.org/10.7554/eLife.22512.020

Because we observed a connection between origin-proximal nucleosome positioning and origin-licensing capacity (Figure 2C), we asked if the reduced origin activation of SWI/SNF and RSC templates corresponded to a particular positioning of origin-proximal nucleosomes. Analysis of local nucleosome positioning by MNase-seq did not reveal a nucleosomal pattern that distinguished the RSC and SWI/SNF templates from ISW1a templates in origin proximal region (Figure 5—figure supplement 2). Consistent with the robust helicase loading observed for all three templates, they each exhibited a nucleosome-free region over the origin. The nucleosome pattern near ARS1 was similar between the ISW1a and SWI/SNF templates. In addition, treatment of SWI/SNF templates with ISW1a did not cause major changes in nucleosome positioning. The pattern of RSC-remodeled nucleosomes was substantially different from SWI/SNF and ISW1a templates on the right side of ARS1. ISW1a addition to RSC templates enhanced the positioning of one nucleosome centered at ~400 bp on the right side of the ARS1 ACS. Thus, unlike the situation for origin licensing (Figure 2), there was no apparent correlation between flanking nucleosome positions and the reduced origin activation observed for SWI/SNF and RSC templates.

Previous studies have reported a histone-binding motif in Mcm2 (Foltman et al., 2013; Huang et al., 2015), raising the possibility that nucleosome-Mcm2-7 interactions may facilitate replication initiation in a nucleosomal context. To test this possibility, we purified a mutant version of Mcm2-7 that lacks the Mcm2 histone-binding motif. Incorporation of this mutation did not alter helicase loading or DNA synthesis with or without nucleosomes (Figure 5—figure supplement 3A–B), suggesting that this interaction is not critical for replication initiation under the conditions of these assays. This is consistent with the lack of an obvious replication phenotype for this mutation (Foltman et al., 2013; Huang et al., 2015).

Given the redundant functions of chromatin remodelers in vivo, we asked if the ability to improve the replication of RSC and SWI/SNF templates was unique to ISW1a or if other CREs could perform the same function. As discussed above, addition of RSC or SWI/SNF to the corresponding nucleosomal templates during helicase loading did not improve CMG formation or replication (Figure 5B and Figure 5—figure supplement 1). Similarly, ISW2 addition to RSC templates resulted in only limited rescue of both CMG formation and replication (Figure 5C). In contrast, addition of ISW1b or INO80-C to RSC templates after helicase loading improved replication initiation and CMG formation to similar levels as ISW1a addition (Figure 5C). Thus, the ability to restore full replication competence to the RSC or SWI/SNF templates is limited to a subset of CREs, consistent with previous studies indicating that these complexes have distinct functionalities (Clapier and Cairns, 2009).

We also asked when ISW1a needed to be present during a specific replication event to improve replication of SWI/SNF and RSC templates (Figure 5A). When added only during the helicase-loading step (DL) or after helicase-loading but removed before DDK treatment (AL), ISW1a significantly improved replication of the SWI/SNF (Figure 5D) and RSC templates (Figure 5B and E). Addition of ISW1a to the SWI/SNF and RSC templates after the DDK-step (but before CMG formation and replication initiation) also improved replication and CMG formation (Figure 5—figure supplement 4). In contrast, addition of ISW1a only during the initiation/elongation (I/E) step of the replication reaction did not improve CMG formation or replication of RSC (Figure 5B) or SWI/SNF templates (Figure 5D). Thus, ISW1a can only improve the replication competence of RSC and SWI/SNF templates if it acts prior to the events of origin activation.

Discussion

The development of fully-reconstituted replication-initiation assays (Yeeles et al., 2015, 2017) has opened the way to biochemically investigate the interactions between the replication-initiation machinery and nucleosomes. Although previous in vivo studies have revealed evidence that local chromatin states impact replication initiation (Berbenetz et al., 2010; Eaton et al., 2010; Lipford and Bell, 2001; MacAlpine and Almouzni, 2013; Simpson, 1990), a molecular understanding of these interactions has been difficult to attain. Here, we have used origin-containing nucleosomal DNA templates assembled in the presence of different CREs to investigate how origin-proximal nucleosomes affect replication initiation. For many of our studies, we have deliberately excluded CREs during the replication assays to investigate the impact of different static nucleosomal states on replication initiation. We also performed replication initiation assays in the presence of CREs, a situation that is potentially more representative of the in vivo situation at promoter-proximal origins. Together these studies reveal that local nucleosomes directly impact two steps in replication initiation: ORC DNA binding and CMG formation. Our studies do not include a full representation of the events that have the potential to impact replication initiation in vivo as they lack histone modifications and higher-order chromatin structures. Nevertheless, these studies represent an important first step to understanding the molecular mechanisms by which chromatin states modulate replication initiation.

Specific CREs establish helicase-loading competent origin-proximal nucleosomes

Previous studies have shown that replication origins are included within nucleosome-free regions (NFRs) and this characteristic is important for origin activity (Berbenetz et al., 2010; Eaton et al., 2010; Lipford and Bell, 2001; MacAlpine et al., 2010; Simpson, 1990; Xu et al., 2012). Consistent with nucleosomes impacting origin selection, we found that assembly of DNA into nucleosomes reduced origin-independent initiation (Figure 1D). Given the redundancy of CREs in vivo, which CREs are capable of establishing NFRs at replication origins is unknown. Our findings demonstrate that CRE-dependent differences in local nucleosomes impact origin licensing. Only a subset of CREs positioned nucleosomes in a manner that allowed efficient origin licensing (Figures 2B, 6A and B). ISW2 templates showed the most inefficient Mcm2-7 loading compared to other templates and this reduction correlated with the encroachment of origin-proximal nucleosomes over origin DNA in a manner that inhibited ORC DNA binding (Figures 2C and 6C). This finding is consistent with studies showing that ISW2 slides nucleosomes towards the promoter-proximal NFR to suppress transcription at cryptic transcription-start sites (Whitehouse et al., 2007). Our findings suggest that ISW2 and perhaps Chd1 play a similar role in regulating origin usage. Interestingly, once ORC is bound to origin DNA, ISW2 is unable to displace ORC with a nucleosome (Figure 2D). Similarly, once ISW2 establishes nucleosome positioning at the origin, ORC is unable to bind (Figure 2E), suggesting ORC cannot displace interfering nucleosomes. These findings suggest that both the relative timing of ORC binding and histone deposition and the CRE present at this time will influence the use of a given site as an origin.

Nucleosomes remodeled by different CREs influence replication initiation differently.

Nucleosomes affect multiple steps of replication initiation using distinct mechanisms. Schematic of ATP-dependent nucleosome assembly with different CREs and their affect on replication initiation. Opacity of the nucleosome represents nucleosome density at each location. (A) Replication permissive nucleosomes are remodeled by ISW1a, ISW1b and INO80-C. These templates are competent for both origin licensing and origin activation. Nucleosome positioning is comparable in these templates. (B) SWI/SNF and RSC templates are origin-licensing competent but are inefficient for subsequent origin activation. We propose that the SWI/SNF and RSC templates have alternate/destabilized nucleosome structures indicated by their different color and that these nucleosomes are not conducive to origin activation. Although both reduce origin activation, SWI/SNF and RSC templates do not share similar nucleosome positioning. (C) ISW2 (and Chd1) templates have nucleosomes over the replication origin that reduce ORC DNA binding and, therefore, origin licensing.

https://doi.org/10.7554/eLife.22512.025

Neither the BAH domain of Orc1 nor the ARS1-binding protein Abf1 contributed to helicase loading in our experiments. The lack of a role for the ORC BAH domain is expected given the modest effect of deletion of the BAH domain on ARS1 replication initiation in vivo (Müller et al., 2010). Given the observation that other BAH domains recognize specific modified forms of nucleosomes (Yang and Xu, 2013), it is also possible that we did not observe a role for the Orc1 BAH domain due to the unmodified status of the histones used in these experiments. Abf1 binding positions nucleosomes on one side of ARS1 in vivo, however, ORC is able to perform this function in the absence of Abf1 at many origins (Eaton et al., 2010). One notable difference from the in vivo situation compared to our in vitro studies is that in vivo the TRP1 gene transcribes into ARS1. Thus, it is possible that Abf1 binding is important to position nucleosomes in the presence of this invasive transcription but not in the absence (such as in our experiments).

SWI/SNF and RSC templates reduce CMG formation

Our findings indicate that local nucleosomes also impact efficient CMG formation and, therefore, origin activation. In particular, SWI/SNF and RSC templates reduced this event. Our finding that the reduced replication of these templates is rescued by addition of other CREs (Figure 5) makes it clear that these effects are nucleosome-dependent. It remains to be determined how the RSC and SWI/SNF templates modulate this event. Given that these templates allow efficient helicase loading, simple steric inhibition by encroaching nucleosomes is unlikely to explain these effects. This conclusion is reinforced by the presence of a large nucleosome-free region overlapping the origin for both templates (Figure 5—figure supplement 2). Indeed, despite their different capacities for replication initiation and CMG formation, ISW1a and SWI/SNF templates had a similar pattern of surrounding nucleosomes and the SWI/SNF nucleosomes pattern was not changed by the same ISW1a addition that restored robust CMG formation and replication initiation to these templates (Figure 5D and Figure 5—figure supplement 2). Another possibility is that the lack of uniformly-spaced nucleosome arrays in the RSC and SWI/SNF templates reduces CMG formation. Of the seven CREs we tested, only SWI/SNF and RSC did not establish uniformly-spaced nucleosomes (Figure 2A). On the other hand, we observed a similar lack of uniformly positioned nucleosomes when we did not add any CRE to the nucleosome assembly reactions, and these templates showed much higher levels of replication initiation than the RSC and SWI/SNF templates (Figure 4—figure supplement 2).

Although it is possible that very subtle changes in nucleosome positioning are responsible, a more likely explanation is that the structure of the nucleosomes remodeled by RSC and SWI/SNF is different (Figure 6A and B). Previous studies have suggested that SWI/SNF family remodelers establish nucleosome structures that are different from canonical nucleosomes (Lorch et al., 1998; Schnitzler et al., 1998; Ulyanova and Schnitzler, 2005). Such altered nucleosomes could have distinct abilities to interact with the replication machinery. SWI/SNF and RSC are also known to remove H2A/H2B dimers from nucleosomes (Clapier et al., 2016). Although we do not observe a dramatic reduction in the relative amounts of H3 and H2A for these templates (Figure 2—figure supplement 1A) it is possible that a subset of nucleosomes assembled in the presence of RSC or SWI/SNF are lacking the full complement of H2A/H2B. One argument against this possibility is the ability of ISW1a addition to readily restore full replication initiation and CMG formation to these templates (Figure 5). Given that free histones are removed after initial nucleosome assembly, it is not clear how ISW1a addition could restore full nucleosomes to the RSC or SWI/SNF templates.

One interesting possibility to explain the different capacities of the templates to facilitate CMG formation is raised by recent studies suggesting that loaded Mcm2-7 interacts with adjacent nucleosomes (Belsky et al., 2015). It is possible that different positions/conformations of local nucleosomes (see below) impacts the ability of nucleosomes to interact with Mcm2-7 double hexamers. These interactions could directly or indirectly modulate access of helicase-activating proteins to loaded Mcm2-7 double hexamers. Addition of some CREs to RSC or SWI/SNF templates improved their replication but only if added before the helicase-activation proteins (Figure 5), suggesting that a positive interaction between loaded Mcm2-7 and nucleosomes must be established prior to CMG formation. Interestingly, we found that RSC and SWI/SNF could not reduce replication from an ISW1a template when added during helicase loading. Perhaps a subset of CREs produce nucleosomes that can interact with the replication machinery positively and once these interactions are established they prevent other CREs from inducing alternative conformations. Although previous studies have identified a histone-binding motif in Mcm2 (Foltman et al., 2013; Huang et al., 2015), incorporation of this mutation into the Mcm2-7 complex did not alter helicase loading or DNA synthesis with or without nucleosomes (Figure 5—figure supplement 3A–B). It is possible, however, that this mutation is not sufficient to eliminate Mcm2-7 interactions with adjacent nucleosomes.

CREs show different capacities to establish replication-competent chromatin states

Our studies show that the different CREs are not equivalent in their ability to establish replication-competent nucleosomes. These differences were observed both with regard to the initial deposition of nucleosomes on DNA (e.g. RSC templates inhibiting CMG formation, Figure 5) and when CREs were added after deposition (e.g. ISW1a addition rescuing the reduced CMG formation of RSC templates, Figure 5). The specificity of the different CREs in our assays suggests that the presence of different CREs at origins will impact origin usage. Localization of specific CREs to origin DNA through interactions with the replication machinery (Euskirchen et al., 2011; Papamichos-Chronakis and Peterson, 2008) or adjacent promoters/transcriptional machinery (Yen et al., 2012) could impact either origin licensing or activation. Our studies indicate that presence of a specific CRE at an origin during a particular cell cycle stage would have different consequences for DNA replication. A CRE present during S phase would impact the initial assembly of nucleosomes and more likely impact subsequent origin licensing/helicase loading. A CRE that is present during G1 is less likely to impact origin licensing and more likely to modulate subsequent CMG formation. Thus, simple deletion of a CRE or monitoring of CRE association with origin DNA in an asynchronous cell culture is unlikely to reveal their full impact on the events of DNA replication.

Interestingly, the RSC and SWI/SNF templates show hallmarks of late-initiating origins. Like late-initiating origins (Belsky et al., 2015; Santocanale et al., 1999; Wyrick et al., 2001), these templates showed efficient origin licensing but reduced/delayed replication initiation. In addition, replication timing is established in late M or early G1 phase (Dimitrova and Gilbert, 1999; Raghuraman et al., 1997) and replication timing can only be reprogrammed prior to S phase (Peace et al., 2016). Similarly, SWI/SNF and RSC templates can be remodeled to replicate efficiently if certain CREs are added prior to shifting the templates into helicase-activation conditions (Figure 5B), which is the biochemical equivalent of the G1-S phase transition. These similarities suggest that local nucleosome states influence replication timing.

Although our studies investigated DNA replication in the context of S. cerevisiae DNA-sequence-defined origins of replication, they are relevant to DNA replication in all eukaryotic organisms. Although most organisms do not use sequence-defined origins of replication, origin-proximal nucleosome-free regions are a common characteristic of origins in many organisms (Fragkos et al., 2015). Thus, our findings regarding the impact of local nucleosomes on origin licensing and selection are relevant to these origins as well. Indeed, the absence of specific sequences directing initiation of replication suggests that local chromatin states will have an even more important role in most organisms. Importantly, once helicases are loaded, specific origin sequences have little or no impact on subsequent origin activation (Gros et al., 2014, 2015). The assays described here lay the groundwork for future studies of the impact of nucleosome structure and histone modification on DNA replication, and can be extended to query DNA replication-dependent nucleosome assembly events and epigenetic inheritance mechanisms.

Materials and methods

Construction of yeast strains and plasmids

Request a detailed protocol

Yeast strains used in these studies are derivatives of W303 and are described in Supplementary file 1. Epitope tagging was performed by PCR-based homologous recombination as previously described (Longtine et al., 1998). Plasmids used in this study are described in Supplementary file 2 and were created by conventional molecular-cloning methods.

DNA template preparation

Request a detailed protocol

DNA templates were isolated from the pARS1-Nco-Nco plasmid (Kang et al., 2014). The plasmid was digested with BamHI, filled in with biotinylated-dATP, dGTP, dCTP and dTTP using Klenow enzyme (NEB) to facilitate attachment to beads. After spin column purification (Plasmid Miniprep Kit from Qiagen), the biotin labeled DNA was cut with Nsi I and Sac II followed by a second spin column purification. This creates a large 3.8 kb BamHI to Nsi I DNA fragment that is biotinylated at one end and that is entirely derived from native S. cerevisiae sequences surrounding the ARS1 origin of replication. A small biotinylated DNA (released by Sac II) is removed by the spin column and the remaining bacterial/vector DNA is not biotinylated and will not bind to streptavidin beads. The 3.8 kb biotinylated-DNA was immobilized on streptavidin-coated paramagnetic beads (Dynabeads M-280, ThermoFisher) according to manufacturer instructions and the non-biotinylated DNA fragment was washed away.

Protein purification

Yeast histones and hNap1 were purified using previously established methods (Vary et al., 2004). Mcm2-7/Cdt1, ORC, Cdc6 and DDK were purified as previously described (Heller et al., 2011; Kang et al., 2014). S-CDK, Sld3/Sld7, Sld2, Cdc45, Dpb11, GINS, Pol ε, Pol α/primase, RPA and Mcm10 were purified as described previously (Lõoke et al., 2017).

Chromatin remodeling enzymes (CREs)

Request a detailed protocol

Tandem affinity purification of ISW1a (TAP-Ioc3), ISW1b (TAP-Ioc2), Chd41 (TAP-Chd1), RSC (TAP-Rsc2), INO80-C (TAP-Ino80), and SWI/SNF (TAP-Swi2) was performed as described (Watanabe et al., 2015). ISW2 (FLAG-Isw2) was purified according to manufacturer's protocol (Sigma), except that E-buffer (20 mM HEPES [pH7.5], 350 mM NaCl, 10% glycerol, 0.1% Tween) was used during the entire purification. Purified proteins were concentrated with VIVASPIN concentrators (Sartorius, Sartorius, Germany) and dialyzed against E-Buffer with 1 mM DTT. The ATPase activity of each remodeling complex was determined as described (Smith and Peterson, 2005), and the concentration of each remodeling complex was estimated relative to a SWI/SNF standard.

ORC∆BAH

Request a detailed protocol

ORC∆BAH was purified as previously described (Frigola et al., 2013) with the following changes. After the Q-sepharose column, pooled fractions of ORC∆BAH were applied to a Superdex 200 column equilibrated with buffer H (50 mM HEPES-KOH [pH7.6], 1 mM EDTA, 1 mM EGTA, 5 mM MgOAc, 10% glycerol) with 0.3 M potassium glutamate.

Abf1

Request a detailed protocol

Abf1 was purified as previously described (Eaton et al., 2010).

Nucleosome assembly

Request a detailed protocol

Nucleosomes were assembled as previously described (Mizuguchi et al., 2012). Nucleosome formation was optimized using a 3.8 kb S. cerevisiae DNA fragment (see below), S. cerevisiae histone octamers (using an DNA:octamer ratio of 1:1.3 by mass) and varying human Nap1 (hNap1) concentration. After determining an optimal histone octamer:hNap1 ratio, nucleosome assembly was further optimized by varying ISW1a concentration. Finally, after optimizing the Nap1 and ISW1a concentrations, the nucleosome assembly reaction was optimized for the DNA:octamer ratio. Consequently, nucleosomes were assembled with ~137 nM yeast histone octamers, 267 nM hNap1, 10 nM CRE (ISW1a or ISW1b or ISW2 or INO80-C or Chd1 or SWI/SNF or RSC) and 120 fmol Dyna bead-immobilized 3.8 Kbp ARS1 DNA in a 20 µl reaction. Initially, hNap1, histone octamers were incubated in ExB 5/50 buffer (10 mM HEPES-KOH [pH7.6], 0.5 mM EGTA, 5 mM magnesium chloride (MgCl2), 50 mM potassium chloride (KCl), 10% glycerol, 0.1 mg/ml BSA) in for for 45 min followed by CRE addition and continued incubation for another 15 min., bringing the total reaction volume to 12.5 µl. 7.5 µl of ATP regeneration mix (5 mM ATP, 30 mM creatine phosphate and 1 μg/ml creatine kinase in 1x ExB 5/50 buffer) was added to the histone octamers, hNap1 and CRE reaction mix and immediately added to the bead-immobilized DNA and incubated at 30°C at 1400 rpm for 4.5 hr in a Thermomixer (Eppendorf, Hauppauge, NY). Nucleosomal-DNA beads were stored at 4°C and used for assays within 12–24 hr. Nucleosome assembly was analyzed by digesting 120 fmol nucleosomal DNA with limiting (0.04 U) MNase at 25°C at 1300 rpm for 15 min in a Thermomixer. The resulting DNA fragments were purified using spin columns (EZ Nucleosomal DNA prep Kit from Zymo Research) and separated on a 1.5% agarose gel and stained with ethidium bromide. Purification protocols for chromatin-remodeling enzymes, histones and hNap1 are described in Supplementary file 2.

Helicase-loading assay

Request a detailed protocol

Helicase loading was performed as previously described (Kang et al., 2014) for naked DNA templates with the following modifications. Mcm2-7/Cdt1, ORC, and Cdc6 were purified as previously described (Kang et al., 2014). The bead-coupled nucleosomal DNA was magnetically separated from unassociated or loosely bound proteins and the supernatant was removed. Nucleosomal DNA was washed twice with 20 µl buffer A-0.35 (25 mM HEPES-KOH [pH7.6], 0.5 mM EGTA, 0.1 mM EDTA, 5 mM MgCl2, 10% glycerol, 0.02% NP40, 0.1 mg/ml BSA and 0.35 M KCl) and once with 20 µl buffer A-0.3 KGlut (0.3 M potassium glutamate [KGlut] instead of 0.35 M KCl in buffer A-0.35). Helicase loading was initiated by the addition of 120 fmol ORC, 180 fmol Cdc6, and 360 fmol Mcm2–7/Cdt1 in a 20 μl reaction containing 60 fmol of bead-coupled 3.8 kb ARS1 DNA (with or without nucleosomes) in helicase-loading buffer (25 mM HEPES-KOH [pH 7.6], 12.5 mM magnesium acetate (MgAc), 300 mM KGlut, 20 μM creatine phosphate, 0.02% NP40, 10% glycerol, 3 mM ATP, 1 mM dithiothreitol (DTT), and 2 μg creatine kinase). The reaction were briefly vortexed or mixed by pipetting (if necessary) to remove any bead clumping. The reactions were incubated at 25°C at 1250 rpm for 25 min in a Thermomixer. Beads were washed three times with 150 µl Buffer H (25 mM HEPES-KOH [pH 7.6], 1 mM EDTA, 1 mM EGTA, 5 mM MgAc, 10% glycerol, and 0.02% NP40) containing 0.3 M KGlut for low salt wash experiments. Experiments with high-salt washes substituted buffer H with 0.5 M NaCl for the second of the three wash steps. DNA-bound proteins were eluted from the beads using 2x sample buffer (120 mM Tris [pH 6.8], 4% SDS and 20% glycerol). Eluted proteins were separated by SDS-PAGE and analyzed by immunoblotting.

Quantification of replication products

Request a detailed protocol

Replication products were measured by incorporation of 32P-dCTP into newly synthesized DNA. Incorporated 32P-dCTP was detected after denaturing gel electrophoresis using a phosphor-imager. For relative replication product quantification, nucleosomal templates assembled with the indicated CRE were quantified with ImageJ software. For each assay, three (n = 3) biological replicates were quantified. The mean value for ISW1a (reactions with DDK) was calculated and set as 100%. All the other values were calculated as a percentage of the mean value of the ISW1a experiment (always performed as part of the same experiment and separated on the same gel). Statistical analysis was performed using Prism software. Error bars indicate standard deviation (SD).

Replication initiation assays

Extract-based replication assay

Request a detailed protocol

Replication assays with extracts were performed as previously described (Kang et al., 2014). Nucleosomes were remodeled, washed and Mcm2-7 was loaded on 60 fmol of DNA in 20 µl reaction volume as described above. After helicase loading, the reaction mix was removed and loaded Mcm2-7 complexes were phosphorylated with 930 fmols DDK in DDK reaction buffer (50 mM HEPES-KOH [pH7.6], 3.5 mM MgAc, 225 mM KGlut, 0.02% NP40, 10% glycerol, 1 mM spermine, 1 mM ATP, and 1 mM DTT) in 30 µl. After DDK phosphorylation was completed, the reaction mix was removed from the beads and replication was initiated by adding 375 ng of S-phase yeast extract to replication buffer (25 mM HEPES-KOH [pH 7.6], 12.5 mM MgAc, 300 mM KGlut, 20 μM creatine phosphate, 0.02% NP40, 10% Glycerol, 3 mM ATP, 40 μM dNTPs, 200 μM CTP/UTP/GTP, 1 mM DTT, 10 μCi [α-P32] dCTP, and 2 μg creatine kinase) in a final volume of 40 µl and incubated for 1 hr at 25°C and 1250 rpm in a Thermomixer. Upon completion, the nucleosomal DNA beads were washed as described for the helicase-loading assay using a low-salt wash. Occasional clumping of nucleosomal DNA beads was eliminated by vortexing. DNA synthesis was monitored using 0.8% alkaline-agarose (in 30 mM sodium hydoxide) gel electrophoresis followed by detection of incorporated 32P-dCTP. DNA-bound proteins were released from the beads with 2X sample and analyzed by SDS-PAGE and immunoblotting.

Reconstituted replication assay for nucleosomal DNA

Request a detailed protocol

The reconstituted nucleosomal DNA replication assay was adapted from a reconstituted replication assay described for naked DNA (Lõoke et al., 2017). Helicase-loading reactions were performed with 60 fmol of nucleosomal DNA (washed with buffer A-0.35 and buffer A-0.3 KGlut as previously described in the text) with 500 fmol ORC, 500 fmol Cdc6 and 500 fmol Mcm2-7/Cdct1 in 10 µl helicase-loading buffer at 25°C at 1250 rpm for 25 min in a Thermomixer. Upon completion of helicase loading, the DNA beads were isolated and the supernatant was removed. Next, 900 fmol of DDK in 10 µl DDK reaction mix (described above) was added to the DNA beads to phosphorylate the loaded Mcm2-7 complexes for 20 min at 25°C at 1250 rpm in a Thermomixer. Upon completion of DDK phosphorylation, the DDK reaction mix was removed from the beads and the replication initiation/elongation step was carried out by adding the indicated amounts of the following proteins (0.5 pmol S-CDK, 0.1 pmol DDK, 0.5 pmol Sld3/7, 2.5 pmol Cdc45, 1 pmol Sld2, 1 pmol Dpb11, 2.5 pmol GINS, 80 fmol Mcm10, 0.93 pmol Polε, 1.25 pmol Polα and 1 pmol RPA) in 30 µl of replication-initiation buffer (25 mM HEPES-KOH [pH7.6], 12.5 mM MgAc, 300 mM KGlut, 20 μM creatine phosphate, 0.02% NP40, 0.04 mg/ml BSA, 10% Glycerol, 3 mM ATP, 40 μM dNTPs, 200 μM CTP/UTP/GTP, 1 mM DTT, 10 μCi [α-P32] dCTP, 2 μg creatine kinase, and 0.5X complete protease inhibitor [Roche]) to the DNA beads for 60 min at 25°C and 1250 rpm in a Thermomixer. At the end of the reaction, immobilized-DNA was washed with a high-salt wash as previously described. Replicated DNA and associated proteins were analyzed as previously described for extract-based replication assay. For assays which examined replication protein association prior to elongation, reactions were performed by leaving out dNTPs, rNTPs (except ATP) and Polα. For experiments in which a CRE was added during a specific step of the reconstituted replication assay, the indicated CRE was added to a final concentration of 10 nM. CRE addition was added at five different times during replication assay: (DL) the CRE was added during the helicase-loading step but was washed off at the end of this step; (AL) the CRE was added after helicase loading but washed away before DDK-phosphorylation; (After DDK) the CRE was added after DDK-phosphorylation but washed away before addition of helicase-activation and elongation proteins; (I/E) the CRE was added with elongation and helicase-activation proteins. When indicated, the CRE was removed by sequential 20 µl washes with A-0.35 KCl (1X) and A-0.3 KGlut (1X). Purification protocols for replication proteins are described in Supplementary file 2.

Quantification of helicase-loading

Request a detailed protocol

Quantification of relative Mcm2-7 loading (immunoblots for Mcm2-7) for nucleosomal templates assembled with the indicated CRE was determined with ImageJ software. For each assay three (n = 3) biological replicates were quantified. The mean value for ISW1a (reaction with high-salt wash) was calculated and set as 100%. All the other values were calculated as a percentage of the mean value of the ISW1a experiment (always performed as part of the same experiment and separated on the same gel). Statistical analysis was performed using Prism software. Error bars indicate standard deviation (SD).

Mono-nucleosome analysis

Request a detailed protocol

Nucleosomes were remodeled on 120 fmol of DNA as previously described (nucleosome assembly and analysis section) with ISW1a, SWI/SNF or RSC. Nucleosomal DNA was digested with 1 U of MNase for 30 min shaking at 25°C at 1350 rpm in Thermomixer. The released DNA was purified using spin columns (EZ Nucleosomal DNA prep Kit from Zymo Research). The purified DNA was separated on a 1.5% agarose gel electrophoresis and stained with ethidium bromide.

Nucleosome positioning analysis

Request a detailed protocol

Nucleosomal DNA was washed with buffer A-0.35 (2X) and digested with 1 U of MNase in 20 µl MNase-digestion buffer (12.5 mM HEPES-KOH [pH7.6], 0.5 mM EDTA, 5 mM MgCl2, 2 mM calcium chloride, 5% glycerol and 1 mM DTT) for 30 min at 25°C shaking at 1350 rpm in a Thermomixer. DNA was purified using spin columns. The purified DNA was separated on a 1.5% agarose gel electrophoresis and stained with ethidium bromide. DNA fragments in the size range that includes mono-nucleosomes (<160 bp) were extracted from the gel and purified using spin columns (Freeze 'N Squeeze DNA Gel Extraction Spin Columns, Bio-Rad). DNA samples were end-repaired and adaptor-ligated using the SPRI-works Fragment Library System I (Beckman Coulter Genomics) and indexed during amplification. Libraries were quantified using the Fragment Analyzer (Advanced Analytical) and qPCR before being loaded for paired-end sequencing using an Illumina HiSeq 2000 (MIT BioMicroCenter).

Paired-end sequencing reads were aligned to the plasmid sequence using Bowtie 1.1.2 (Langmead et al., 2009) with the following parameters: -n 2 -l 20 - -best - -strata. To obtain nucleosome occupancy profiles, the midpoint positions from all 125–175 bp read fragments were extracted. The nucleosome signal was smoothed by constructing a 20 bp Gaussian kernel around each midpoint position, and smoothed kernels were aggregated together to form a nucleosome signal track (Boyle et al., 2008). Signal was then normalized to read depth for each sample. Consensus nucleosome positions were determined by finding peaks (above a threshold of 2) in the nucleosome-density signal (Flores and Orozco, 2011).

References

    1. Rhind N
    2. Gilbert DM
    (2013) DNA replication timing
    Cold Spring Harbor Perspectives in Biology 5:a010132.
    https://doi.org/10.1101/cshperspect.a010132

Decision letter

  1. Robert Sclafani
    Reviewing Editor; University of Colorado School of Medicine, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "Nucleosomes Influence Multiple Steps During Replication Initiation" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Jessica Tyler as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

The proposed studies are clearly very early work on chromatin and initiation of DNA replication in vitro. The studies represent a very good first step in understanding a complex mechanism, even if the detailed molecular insight remains to be determined. It will be acceptable for publication provided they address the following important criticisms and adequately reply to these comments:

Essential revisions:

1) They demonstrate the activity of each CRE in an ATPase assay or on a mononucleosomal template.

2) They test whether the presence of a histone chaperone has an effect on the replication assays, where histones are presumably displaced.

3) They provide a better description of how the chromatin was prepared.

4) They discuss the caveat of why the specificity of initiation seen in vivo has not been duplicated using their in vitro system.

Other Comments:

1) The manuscript does not mention one of the main interaction points between the chromatin and the initiation machinery: the BAH domain of ORC1 that can bind nucleosomes. This domain has been shown in yeast to affect certain initiation events and in humans this domain has been implicated in both reading an origin "histone code" and potentially in Meier-Gorlin Syndrome. A mutant version of ORC could be produced with the BAH removed and the assays in this manuscript could be repeated. If the domain is not important for any activities described in the paper, this may suggest that post-translational modifications of the histones may be required for BAH-specific activity. Alternatively, the BAH domain may be necessary for ORC, when pre-bound to naked DNA, to resist the conversion of the chromatin organization by ISW2 to a non-permissive state.

2) SWI/SNF and RSC CREs generate chromatin organization that is not permissive to origin activation, yet it is not clear why this occurs. It is also not clear why subsequent treatment with Isw1a rescues activation. The nucleosome occupancy patterns in Figure 5—figure supplement 2 do not shed light on this. One possibility is that there are altered nucleosomes around the origin that are formed by SWI/SNF or RSC that are not adequately profiled by the MNase-Seq, which is described as sequencing only mono-nucleosomal fragments less than 160 bp. Alternatively, is there a possibility that the intensity of the nucleosomes is not easily comparable across different templates, due to internal normalization? The profiles of the whole 3.8 kb template for each remodeling condition may help to address this, or using some independent method to fragment the DNA and sequencing could address percent occupancy. Addressing these points would at the very least rule out certain caveats.

3) Why do you need to purify away the Nap1 and CREs? This is not discussed, but I would argue that you may need a histone binding protein to avoid that the evicted histones precipitate onto DNA again. How much of each CRE remains bound is a critical issue.

4) In Figure 1—figure supplement 3, are anti-CPB and anti-FLAG on the same membrane? If not, an internal reference should be used to validate the wash of the CREs. The homogenous removal of the CREs is essential in interpreting the results.

5) Clarify these statements:- "The results of the MNAse in presence of SWI/SNF and RSC (Figure 2A) shows that the "nucleosomes" in those samples are not stable". Could you perform the MNase-seq (Figure 5—figure supplement 2) and get a 147 bp protection in Figure 2—figure supplement 2? What is the difference between these experiments?

6) Figure 1D should include a Western blot for ORC as well.

7) Figure 2D would be improved by showing ORC occupancy in all CRE contexts.

8) The size range of all replication products run on denaturing agarose gels should be indicated with marker indicator.

9) Dissection of the steps after DDK treatment, using purified components such as Cdc45, would make this manuscript even more impactful.

10) Including a mutant MCM2 lacking the histone binding domain would increase the impact of the manuscript, similar to the ORC BAH mutant.

11) Including the Abf1 protein, which binds the B3 element of ARS1, would increase the impact of the manuscript. Previously it was proposed that Abf1 helps establish a nucleosome free region for MCM loading.

We have included the full reviews at the bottom of this letter, to further clarify these requests.

Reviewer #1:

This manuscript addresses the effects of chromatin on the complete mechanism of the initiation of DNA replication in a reconstituted system for the first time, using the well-established budding yeast model. The authors first show that a regularly-spaced nucleosome array made with recombinant histones is equally efficient at one of the first major steps in initiation, the formation of the pre-replication complex (pre-RC), which culminates in the loading of the MCM helicase rings. Next, they prepare nucleosomal templates using multiple chromatin remodeling enzymes that generate different patterns of nucleosome placement. These different nucleosome organizations result in different efficiencies of pre-RC formation and further analysis suggests that the positioning of the nucleosomes within the replication origin sequence is the cause of this reduction by limiting the association of the Origin Recognition Complex (ORC). The authors then test whether the chromatin templates are competent for origin activation and resulting DNA replication, both in an extract that provides the downstream components and then in a fully-purified system. The results in both systems further emphasize that a specific chromatin configuration is required for the most-efficient replication activity, highlighting a critical window for establishing this chromatin context before/during the helicase priming step by the Dbf4-dependent kinase (DDK), which is required for multiple downstream steps leading to initiation of replication.

The work in this manuscript provides an important step in dissecting how chromatin influences DNA replication initiation. The strengths lie in the ability to profile, in a controlled manner, multiple CREs to demonstrate that distinct chromatin organization affects specific substeps in the mechanism. The manuscript could be significantly improved by the addition of a small set of experiments and data. Most notably, incorporating experiments using an ORC1 mutant lacking the bromo-adjacent homology (BAH) domain, previously demonstrated to bind nucleosomes and influence replication initiation in vivo.

Major Comments:

1) The manuscript does not mention one of the main interaction points between the chromatin and the initiation machinery: the BAH domain of ORC1 that can bind nucleosomes. This domain has been shown in yeast to affect certain initiation events and in humans this domain has been implicated in both reading an origin "histone code" and potentially in Meier-Gorlin Syndrome. A mutant version of ORC could be produced with the BAH removed and the assays in this manuscript could be repeated. If the domain is not important for any activities described in the paper, this may suggest that post-translational modifications of the histones may be required for BAH-specific activity. Alternatively, the BAH domain may be necessary for ORC, when pre-bound to naked DNA, to resist the conversion of the chromatin organization by ISW2 to a non-permissive state.

2) SWI/SNF and RSC CREs generate chromatin organization that is not permissive to origin activation, yet it is not clear why this occurs. It is also not clear why subsequent treatment with Isw1a rescues activation. The nucleosome occupancy patterns in Figure 5—figure supplement 2 do not shed light on this. One possibility is that there are altered nucleosomes around the origin that are formed by SWI/SNF or RSC that are not adequately profiled by the MNase-Seq, which is described as sequencing only mono-nucleosomal fragments less than 160 bp. Alternatively, is there a possibility that the intensity of the nucleosomes is not easily comparable across different templates, due to internal normalization? The profiles of the whole 3.8 kb template for each remodeling condition may help to address this, or using some independent method to fragment the DNA and sequencing could address% occupancy. Addressing these points would at the very least rule out certain caveats.

Reviewer #2:

Main scientific findings:

The manuscript addresses the issue of how chromatin structure influences/ regulates replication initiation at two steps: (i) origin licensing by orc binding to origins and (ii) origin activation by recruitment and activation of the MCM2-7 helicase. The main conclusions are, derived from in vitro reconstitution of chromatin in a template DNA containing an ars (ori) using different chromatin remodeling complexes, are (i) chromatin structure enhances the specificity of orc binding to ori, (ii) although remodeling complexes inhibit MCM activity when added during the reaction, pre-incubation with remodelers to the template before the addition of MCM and other proteins of the pre-initiation complex rescued replication. Thus, the manuscript demonstrates, perhaps as expected, that nucleosomal landscape modulates replication just as it does of other DNA transactions such as transcription, DNA repair, replicative life span control.

Significance:

The authors get credit for biochemically addressing the relationship between chromatin structure and replication initiation. There are two other papers in Molecular Cell on this subject from Diffley's group from UK that has carried the work into chromatin structure and its regulatory role in replication initiation further. This reviewer is of the opinion that the work by the other group should not in any way diminish the impact of the present manuscript from the point of view of publication priority.

Critique:

The work presented in this manuscript is well done. However, the evidence for specific ori-dependent initiation in the opinion of this reviewer does not meet the gold standards established by Kornberg and others in oriC, plasmid replication reconstitution and SV40 replication. Without 2D gel analyses or EM, it was difficult to assess to what extent initiation fidelity was enhanced by the chromatin remodeling factors in effectively limiting initiation to the ars sequences. After all, in vivo no initiation occurs under normal conditions without the presence of an ars.

A second question is how does chromatin remodeling enhance the specificity of licensing and activation of the origins? Is it just by repositioning the nucleosomes around the region of the ars and thereby restricting orc binding to the origins and preventing its physiologically unwarranted binding to the non-ars sequences or is it more intricate? Diffley's group finding that FACT is involved in the process suggests a more elaborate reaction involving perhaps transcription. It was reported by Marahrens and Stillman that one essential element of the ars directs transcription and is needed for ars activity.

In summary, this is a good paper that explores the role of chromatin in replication and therefore has some novelty. As noted above, the specificity of initiation seen in vivo has not been duplicated in vitro and therefore any conclusions from the chromatin reconstitution work should be cautiously interpreted.

Reviewer #3:

The manuscript by Azmi et al. presents the in vitro reconstitution of DNA replication on chromatinized templates, pre-treated with different chromatin remodeling enzymes (CREs). The authors show how treatment with different CREs lead to differential effects to ORC binding or replication efficiency.

Although it is important to understand how nucleosomes affect DNA replication, I am skeptical about the impact of the data, as they are presented here. In my opinion, the manuscript represents a collection of observations, with insufficient mechanistic insight to warrant publication in eLife.

The conclusion that encroachment of the origin inhibits helicase loading is valid. However, the remaining conclusions are not supported by sufficiently controlled mechanistic insight. In particular, the chromatin templates (after remodeling) are insufficiently characterized to allow conclusions on the role of CREs and the nature of the nucleosomes at the origin. The CREs preparation should be tested in a bona fide CRE assay to ensure that they are all "equally" trustworthy/active.

1) The data presented seems to have more to do with the activities of the CREs than with replication. A gold standard CRE assay to validate the activity of preparations should be shown. Some introduction into the mechanistic differences between the different CREs (sliding, exchanging, etc.) have to be included both in the Introduction and in the Discussion.

2) What happens if SWI/SNF or RSC are added to an ISWI1a preparation? Do they overrule ISWI1a? In other words, once you have a good template, can you mess it up (after loading)? To have an impactful conclusion from these studies, a more controlled analysis of the cross-talk between CREs is required, as all CREs are present in the nucleus.

3) It is unclear how the chromatinized plasmid was prepared. In some part of the text, and for example in Figure 1—figure supplement 2, it is stated that the octamer:DNA ratio was 1:1.3, yet from the method section it appears that this is not accurate as the octamer/Nap1 complexes are in large excess to the DNA (180 nM octamer to 120 fmol 3.8kb DNA). This needs to be clarified.

4) Why do you need to purify away the Nap1 and CREs? This is not discussed, but I would argue that you may need a histone binding protein to avoid that the evicted histones precipitate onto DNA again. How much of each CRE remains bound is a critical point.

5) This is linked to point 2. In Figure 1—figure supplement 3, are aCPB and aFLAG on the same membrane? If not, an internal reference should be used to validate the wash of the CREs. The homogenous removal of the CREs is essential in interpreting the results.

6) I am confused by some results/statements:

- The results of the MNAse in presence of SWI/SNF and RSC (Figure 2A) shows that the "nucleosomes" in those samples are not stable, yet you could perform the MNase-seq (Figure 5—figure supplement 2) and you could get a 147 bp protection in Figure 2—figure supplement 2. What is the difference between these experiments?

– Related to this, the statement “Of the seven CREs we tested, only SWI/SNF and RSC did not establish regularly-spaced nucleosomes”, as RSC and SWI/SNF do not affect spacing, there are just no stable nucleosomes in those samples.

– The statement “These similarities suggests that local nucleosome positioning and/or states influence replication timing”: nucleosome positioning is not an issue with RSC and SWI/SNF, so I would rephrase this and remove the position component. I also think more controls are needed to make this statement.

7) I am left with the question whether CREs are essential. Can you just add histones and Nap1 to make nucleosomes? How would this "chromatin template" behave in your assays? Would it look like ISWI2/Chd2? This has not been shown, but it may be a good control to strengthen the relevance of CREs in the process.

https://doi.org/10.7554/eLife.22512.029

Author response

Essential revisions:

1) They demonstrate the activity of each CRE in an ATPase assay or on a mononucleosomal template.

We have included a new figure (Figure 1—figure supplement 3) showing ATPase activity of each CRE. Based on these measurements, we used equal ATPase activity for each CRE for all the assays.

2) They test whether the presence of a histone chaperone has an effect on the replication assays, where histones are presumably displaced.

We have included a new figure showing the effect of adding Nap1 or FACT to our assays. In each case we do not see stimulation of replication (see Author Response Image 1). This is likely because other factors are limiting the elongation phase of the assay (e.g. Mrc1, see (Yeeles et al., 2017)). We note that we do not intend this study to address the elongation stage of the reaction as other previous papers have (Devbhandari et al., 2016; Kurat et al., 2016). Instead, we are using the replication phase of our assays as a simple readout of the number of replication forks that are established. That this is the case, is reflected in the close correlation that we observe between the extent of CMG complex formation and the amount of replication products throughout our assays. Although we will include this data in the final manuscript if the reviewers feel strongly, we do not think it enhances our conclusions about the effect of chromatin on replication initiation events and may confuse the readers as to the goals of our study.

3) They provide a better description of how the chromatin was prepared.

The method we employed to prepare nucleosome DNA templates was previously published from Dr. Carl Wu’s lab (Mizuguchi et al., 2012) and optimization was carried out with close communication with Dr. Mizuguchi. The final ratio of DNA: octamer used in our assays is 1:1.3 by mass (296ng of DNA and 385ng of octamers), which is now indicated in the description of the template preparation (subsection “SWI/SNF and RSC templates reduce CMG formation”).

4) They discuss the caveat of why the specificity of initiation seen in vivo has not been duplicated using their in vitro system.

In contrast to the reviewer’s suggestion, our helicase loading and initiation reactions exhibit sequence specificity that is consistent with that observed in vivo. Contrary to studies from the Diffley and Remus labs (Devbhandari et al., 2016; Gros et al., 2014; Kurat et al., 2016), the conditions that we use for our helicase loading and replication initiation assays require an intact ARS1 sequences – in particular the highly-conserved ARS consensus sequence that binds ORC (Heller et al., 2011; Ticau et al., 2015). We show that this is the case for helicase-loading reactions with and without nucleosomes in Figure 1D, bottom row (all our reactions are normally performed at 300 mM K-glutamate). Consistent with the requirement of helicase loading for replication initiation, our replication reactions are also strongly dependent on an intact origin. We have added an additional experiment illustrating this dependence (Figure 3—figure supplement 1) which we describe in the subsection “Local nucleosomes impact replication events downstream of helicase loading”.

The only exception to this origin-dependence in our manuscript is that we deliberately promoted non-specific helicase loading on naked DNA by using lower-salt conditions (Devbhandari et al., 2016; Kurat et al., 2016; Remus et al., 2009) for Figure 1D (125 mM K-glutamate, top row). We did this to ask whether nucleosomes would enhance origin specificity under conditions in which origin selection is less specific. Indeed, we found that nucleosomes make helicase loading origin-specific even when naked DNA loading is origin-non-specific.

Other Comments:

1) The manuscript does not mention one of the main interaction points between the chromatin and the initiation machinery: the BAH domain of ORC1 that can bind nucleosomes. This domain has been shown in yeast to affect certain initiation events and in humans this domain has been implicated in both reading an origin "histone code" and potentially in Meier-Gorlin Syndrome. A mutant version of ORC could be produced with the BAH removed and the assays in this manuscript could be repeated. If the domain is not important for any activities described in the paper, this may suggest that post-translational modifications of the histones may be required for BAH-specific activity. Alternatively, the BAH domain may be necessary for ORC, when pre-bound to naked DNA, to resist the conversion of the chromatin organization by ISW2 to a non-permissive state.

To address the role of the Orc1 BAH domain, we have produced ORC lacking the BAH domain (ORC∆BAH) and studied its effect on helicase loading using nucleosomes assembled with several different CREs (new Figure 2—figure supplement 2). Our results indicate that ORC∆BAH fully substitutes for ORC in these conditions. In particular, we did not observe further decreases in helicase loading for nucleosome assembled by ISW2. These findings suggest that interactions between the Orc1 BAH domain and local nucleosomes are not required for helicase loading at ARS1. We note that previous studies indicate that deletion of the BAH domain in vivo has only a limited effect on ARS1 function (Müller et al., 2010). Although it would be interesting to explore the function of the BAH domains at other origins, this is beyond the scope of the current study. We agree with the reviewer that these findings may also indicate that the effects of the BAH domain require modified versions of histones not present in our current assays. We describe these new findings in subsection “Local nucleosomes reduce helicase loading by inhibiting ORC DNA binding” as well as in the Discussion section.

2) SWI/SNF and RSC CREs generate chromatin organization that is not permissive to origin activation, yet it is not clear why this occurs. It is also not clear why subsequent treatment with Isw1a rescues activation. The nucleosome occupancy patterns in Figure 5—figure supplement 2 do not shed light on this. One possibility is that there are altered nucleosomes around the origin that are formed by SWI/SNF or RSC that are not adequately profiled by the MNase-Seq, which is described as sequencing only mono-nucleosomal fragments less than 160 bp. Alternatively, is there a possibility that the intensity of the nucleosomes is not easily comparable across different templates, due to internal normalization? The profiles of the whole 3.8 kb template for each remodeling condition may help to address this, or using some independent method to fragment the DNA and sequencing could address percent occupancy. Addressing these points would at the very least rule out certain caveats.

We agree with the reviewer that MNase-seq may not fully reveal the differences between the origin-proximal nucleosomes for the ISW1a, SWI/SNF or RSC templates. Unfortunately, our original experimental procedure eliminated smaller fragments so we cannot explore this data to identify sub-nucleosomal fragments that might impact our findings. On the other hand, it is not clear how addition of ISW1a would remove or remodel subnucleosomal particles to eliminate this issue. With regard to comparing nucleosome occupancy, we see similar amounts of nucleosomes associated with the different templates (Figure 2 —figure supplement 1A-B) and the nucleosome occupancies that we observe in the MNase-Seq experiments are similar between ISW1a, RSC and SWI/SNF (Figure 5 —figure supplement 2). We agree that the current data does not provide a clear mechanistic understanding and there is much to be done to address these observations at a detailed mechanistic level. Nevertheless, our data clearly indicates that these differences observed between the different nucleosomal DNA templates is the result of different CREs establishing distinct nucleosomal states.

3) Why do you need to purify away the Nap1 and CREs? This is not discussed, but I would argue that you may need a histone binding protein to avoid that the evicted histones precipitate onto DNA again. How much of each CRE remains bound is a critical issue.

We choose to remove the CRE from templates primarily to prevent additional nucleosome assembly and remodeling during the assays. Our intent was to establish different chromatin states and ask how DNA replication initiation responds to them. Nap1 gets removed as a consequence of removing CREs, since removal of CRE requires a stringent wash. Additionally, chaperons are primarily involved in the replication elongation step (e.g. replication-coupled nucleosome removal and assembly (Burgess and Zhang, 2013)). In this manuscript, we focus on replication initiation step (e.g. we have not included all the factors required for lagging-strand synthesis in our assays) and thus we did not focus our efforts in understanding how would chaperones influence replication elongation. Nonetheless, we have performed a replication assay to examine whether Nap1 or FACT influence replication initiation in our assay and found no dramatic increase in replication (see major comments, point 2, above).

We address the amount of each CRE that remains bound to the template (Figure 1—figure supplement 1). In each case, little of no detectable CRE remains. We note that for the case of the RSC and SWI/SNF templates, the effects on replication cannot be due to residual association of these enzymes as adding these enzymes back to the reactions does not restore full levels of replication products (Figure 5B and Figure 5—figure supplement 1). We only observe rescue of the reduced replication capacities of these templates when we add back different CREs (Figure 5B-C and Figure 5—figure supplement 1) and only when they are added before the helicase-activating proteins (Figure 5D-E).

4) In Figure 1—figure supplement 3, are anti-CPB and anti-FLAG on the same membrane? If not, an internal reference should be used to validate the wash of the CREs. The homogenous removal of the CREs is essential in interpreting the results.

In Figure 1—figure supplement 2B, the anti-CPB and anti-FLAG immunoblots were performed on the same membrane. After the anti-CPB immunoblot was performed, the membranes were stripped and re-probed for the FLAG-epitope.

5) Clarify these statements:- "The results of the MNAse in presence of SWI/SNF and RSC (Figure 2A) shows that the "nucleosomes" in those samples are not stable". Could you perform the MNase-seq (Figure 5—figure supplement 2) and get a 147 bp protection in Figure 2—figure supplement 2? What is the difference between these experiments?

The reviewer is correct that we used the incorrect terminology here. What we should have stated is that the nucleosomes are not uniformly spaced. We have revised the text accordingly (subsection “CRE-specific restoration of replication and CMG formation to RSC and SWI/SNF templates“).

The difference in the experiment shown in Figure 2A and those shown for Figure 2—figure supplement 1B and Figure 5—figure supplement 2 is the extent of MNase digestion. Under low levels of digestion, (Figure 2A) we observe a relatively random assortment of fragments that indicate that the assembled nucleosomes are not precisely spaced relative to one another. When we add higher amounts of MNase, this random pattern of cleavage between nucleosomes collapses into similar amounts of protected ~147bp DNA for the RSC and SWI/SNF templates as we see for templates that show more uniformly spaced nucleosomes (Figure 2—figure supplement 1B). We also have performed MNase-seq with SWI/SNF and RSC templates and observe 147 bp protected fragments as shown in Figure 5—figure supplement 2.

6) Figure 1D should include a Western blot for ORC as well.

Figure 1D is a helicase loading reaction that is washed with high-salt. ORC is removed under these conditions and, therefore, would not be detected. We note that the key question in Figure 1D is whether Mcm2-7 has been loaded (the purpose of the high-salt wash is to remove any incompletely loaded Mcm2-7 – a standard method for the field). We also note that we include an ORC blot in Figure 1C, where a low-salt wash that does not remove ORC is performed and we see equivalent ORC binding with and without nucleosomes.

7) Figure 2D would be improved by showing ORC occupancy in all CRE contexts.

Mcm2-7 loading requires ORC. Given that we did not see defects in helicase loading for the remaining CREs we do not think that including them in Figure 2D will enhance the conclusion of the manuscript. We chose the subset shown to focus on addressing the defects in helicase loading observed for the Isw2 and Chd1 templates. We do show that there are not significant differences for two templates that exhibit the same helicase loading observed for naked DNA (Figure 2B) and showing additional ORC blots would not help to explain the defects in loading for Chd1 and Isw2 templates.

8) The size range of all replication products run on denaturing agarose gels should be indicated with marker indicator.

We have included this information for each figure.

9) Dissection of the steps after DDK treatment, using purified components such as Cdc45, would make this manuscript even more impactful.

We agree that further determination of the precise step impacted by different nucleosomal template in CMG formation would be of interest. Nevertheless, the critical issue is whether the CMG is formed or not and we have clearly documented that it is at this step that nucleosomes assembled in the presence of RSC and SWI/SNF are impacting initiation. Whether this occurs at the Cdc45 or GINS recruiting step will be one of several future directions for these studies.

10) Including a mutant MCM2 lacking the histone binding domain would increase the impact of the manuscript, similar to the ORC BAH mutant.

We had done this experiment previously and mentioned it in the Discussion of the previous version of the paper because we did not observe a defect for this mutant. We have now included this experiment comparing helicase loading and replication initiation using Mcm2-7 with and without the histone-binding domain in the Results section (subsection “CRE-specific restoration of replication and CMG formation to RSC and SWI/SNF templates” paragraph three; Figure 5—figure supplement 3).

11) Including the Abf1 protein, which binds the B3 element of ARS1, would increase the impact of the manuscript. Previously it was proposed that Abf1 helps establish a nucleosome free region for MCM loading.

To address a possible role for Abf1, we examined whether pre-binding of Abf1 to DNA prior to ISW2 nucleosome assembly rescued the defects in helicase loading observed (Figure 2—figure supplement 2E-F). Unlike ORC pre-binding to the DNA, we found that Abf1 is not sufficient to rescue the defect in helicase loading in the presence of ISW2. Additionally, when Abf1 is included during ISW1a nucleosome assembly, we do not see an increase or decrease in the ability of ISW1a templates to load Mcm2-7. Although prior genetic studies show that Abf1 binding enhances replication initiation (Marahrens and Stillman, 1992), this is in the context of a complete chromosome with the adjacent TRP1 gene being transcribed into the ARS1 origin. We suspect that Abf1 may become more important under these conditions.

We have included the full reviews at the bottom of this letter, to further clarify these requests.

Reviewer #2:

[…]

A second question is how does chromatin remodeling enhance the specificity of licensing and activation of the origins? Is it just by repositioning the nucleosomes around the region of the ars and thereby restricting orc binding to the origins and preventing its physiologically unwarranted binding to the non-ars sequences or is it more intricate? Diffley's group finding that FACT is involved in the process suggests a more elaborate reaction involving perhaps transcription. It was reported by Marahrens and Stillman that one essential element of the ars directs transcription and is needed for ars activity.

We believe that the effect of nucleosomes on helicase loading is simply due to nucleosome occlusion and this conclusion is supported by our studies of local nucleosome positioning (Figure 2C). The role of nucleosomes in the helicase activation process remains less clear. We provide strong evidence that the effects that we see are due to nucleosomes since they are readily modulated by CREs. Because our reactions do not include any components of the transcription apparatus (they are all performed with purified replication proteins – see (Looke et al., 2017) for gels of the proteins used) any effect of Abf1 or other added proteins on transcription would not be observed in our assays. The same is true for the Diffley lab experiment addressing the role of FACT in replication (Kurat et al., 2016). Instead, FACT is likely to be functioning in the Diffley studies to remove nucleosomes in front of the replication fork (as has been described for FACT’s role in transcription of chromatin templates (Belotserkovskaya et al., 2004)). Indeed, one of the FACT subunits was originally discovered due to its association with DNA Pol α (Miles and Formosa, 1992; Wittmeyer and Formosa, 1997).

Finally, the factor that the reviewer is referring to in the Marahrens and Stillman paper is Abf1(Marahrens and Stillman, 1992), whose function we now address (Figure 2—figure supplement 2E-F). Although this protein can act as a transcription factor in some situations, there is no evidence that it does so at ARS1. Instead, Abf1 is known to be involved in the establishment of a positioned nucleosome adjacent to ARS1 (Lipford and Bell, 2001). in vivo this may be particularly important since the TRP1 gene transcribes into the ARS1 region on the Abf1-side of the origin DNA.

Reviewer #3:

[…]

1) The data presented seems to have more to do with the activities of the CREs than with replication. A gold standard CRE assay to validate the activity of preparations should be shown. Some introduction into the mechanistic differences between the different CREs (sliding, exchanging, etc.) have to be included both in the Introduction and in the Discussion.

The first issue is addressed above: major comments, point 1. We have also added additional information about the mechanistic differences between the CREs used into the Introduction (paragraph four) and Discussion section.

2) What happens if SWI/SNF or RSC are added to an ISWI1a preparation? Do they overrule ISWI1a? In other words, once you have a good template, can you mess it up (after loading)? To have an impactful conclusion from these studies, a more controlled analysis of the cross-talk between CREs is required, as all CREs are present in the nucleus.

We have done this experiment and found that SWI/SNF or RSC cannot reduce the efficiency of a ISW1a template. This suggests that once a chromatin state that is active for CMG formation is formed, it is not easily removed. This could be due to the different activities of the CREs or because there are positive interactions formed (e.g. between Mcm2-7 and nucleosomes) that are difficult for RSC or SWI/SNF to eliminate. We have added this data to the Results section(subsection “CRE-specific restoration of replication and CMG formation to RSC and SWI/SNF templates”; Figure 5—figure supplement 1B) and discuss the implications of the findings (subsection “CREs show different capacities to establish replication-competent chromatin states.”).

3) It is unclear how the chromatinized plasmid was prepared. In some part of the text, and for example in Figure 1—figure supplement 2, it is stated that the octamer:DNA ratio was 1:1.3, yet from the method section it appears that this is not accurate as the octamer/Nap1 complexes are in large excess to the DNA (180 nM octamer to 120 fmol 3.8kb DNA). This needs to be clarified.

Addressed above, major comment 3. We note that the relative molarity of the plasmid DNA and the octamer is not the correct comparison, as different lengths of DNA would provide different total amounts of DNA in the reaction for the same molarity. Instead, one has to determine the relative total amounts of DNA and octamer (in our case this is a 1:1.3 ratio or 296ng of DNA and 385ng of octamers in a typical reaction).

4) Why do you need to purify away the Nap1 and CREs? This is not discussed, but I would argue that you may need a histone binding protein to avoid that the evicted histones precipitate onto DNA again. How much of each CRE remains bound is a critical point.

Addressed above: other comments, point 3.

5) This is linked to point 2. In Figure 1—figure supplement 3, are aCPB and aFLAG on the same membrane? If not, an internal reference should be used to validate the wash of the CREs. The homogenous removal of the CREs is essential in interpreting the results.

Addressed above: other comments, point 4.

6) I am confused by some results/statements:

- The results of the MNAse in presence of SWI/SNF and RSC (Figure 2A) shows that the "nucleosomes" in those samples are not stable, yet you could perform the MNase-seq (Figure 5—figure supplement 2) and you could get a 147 bp protection in Figure 2—figure supplement 2. What is the difference between these experiments?

Addressed above: other comments, point 5.

– Related to this, the statement “Of the seven CREs we tested, only SWI/SNF and RSC did not establish regularly-spaced nucleosomes”, as RSC and SWI/SNF do not affect spacing, there are just no stable nucleosomes in those samples.

We disagree that the inability of RSC and SWI/SNF to space nucleosomes indicates that there are no stable nucleosomes on these templates. Both the equivalent levels of 147 bp protected DNA (Figure 2—figure supplement 2), the similar patterns of origin proximal nucleosomes between ISW1a and SWI/SNF (Figure 5—figure supplement 2), and the equivalent levels of bound histones (e.g. Figure 3D, Figure 4B-E) all indicate that the RSC and SWI/SNF templates include stably-associated nucleosomes.

– The statement “These similarities suggests that local nucleosome positioning and/or states influence replication timing”: nucleosome positioning is not an issue with RSC and SWI/SNF, so I would rephrase this and remove the position component. I also think more controls are needed to make this statement.

We agree that local nucleosome states do not appear to be an issue with SWI/SNF vs. ISW1a, however, there are significant differences for RSC and ISW1a as well as for RSC vs ISW1a treated RSC templates. Thus, we think that including changes in nucleosome positioning in this statement is appropriate. It is not clear what other controls the reviewer is referring to. We note, however, that the statement at issue is only suggesting a possibility raised by our data, not making a conclusion.

7) I am left with the question whether CREs are essential. Can you just add histones and Nap1 to make nucleosomes? How would this "chromatin template" behave in your assays? Would it look like ISWI2/Chd2? This has not been shown, but it may be a good control to strengthen the relevance of CREs in the process.

We have assembled nucleosomes with only and histones and Nap1 and the MNase pattern looks similar to that of RSC or SWI/SNF, however, helicase loading and replication initiation is similar to that of ISW1a or naked DNA templates. Not surprisingly, when analyzed by limited MNase digestion, these templates show a similar lack of uniformly-spaced nucleosomes as we observe for RSC and SWI/SNF templates. Thus, the different replication capacities we observe for the RSC and SWI/SNF templates requires the action of these enzymes and is not simply a consequence of non-uniformly spaced nucleosomes. We have now included this experiment as Figure 4—figure supplement 2 and modified the text accordingly (subsection “RSC and SWI/SNF templates impede CMG complex formation” and “CRE-specific restoration of replication and CMG formation to RSC and SWI/SNF templates”).

https://doi.org/10.7554/eLife.22512.030

Article and author information

Author details

  1. Ishara F Azmi

    Department of Biology, Howard Hughes Medical Institute, Massachusetts Institute of Technology, Cambridge, United States
    Contribution
    IFA, Conceptualization, Resources, Formal analysis, Methodology, Writing—original draft, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6280-3149
  2. Shinya Watanabe

    Program in Molecular Medicine, University of Massachusetts Medical School, Worcester, United States
    Contribution
    SW, Resources, Methodology
    Competing interests
    The authors declare that no competing interests exist.
  3. Michael F Maloney

    Department of Biology, Howard Hughes Medical Institute, Massachusetts Institute of Technology, Cambridge, United States
    Contribution
    MFM, Resources, Methodology
    Competing interests
    The authors declare that no competing interests exist.
  4. Sukhyun Kang

    1. Department of Biology, Howard Hughes Medical Institute, Massachusetts Institute of Technology, Cambridge, United States
    2. Center for Genomic Integrity, Institute for Basic Science, Ulsan, South Korea
    Contribution
    SK, Resources, Methodology
    Competing interests
    The authors declare that no competing interests exist.
  5. Jason A Belsky

    1. Department of Pharmacology and Cancer Biology, Duke University Medical Center, Durham, United States
    2. Program in Computational Biology and Bioinformatics, Duke University, Durham, United States
    Contribution
    JAB, Formal analysis, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2945-6282
  6. David M MacAlpine

    Department of Pharmacology and Cancer Biology, Duke University Medical Center, Durham, United States
    Contribution
    DMM, Supervision, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.
  7. Craig L Peterson

    Program in Molecular Medicine, University of Massachusetts Medical School, Worcester, United States
    Contribution
    CLP, Supervision, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.
  8. Stephen P Bell

    Department of Biology, Howard Hughes Medical Institute, Massachusetts Institute of Technology, Cambridge, United States
    Contribution
    SPB, Conceptualization, Funding acquisition, Writing—review and editing
    For correspondence
    spbell@mit.edu
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-2876-610X

Funding

American Cancer Society (123700-PF-13-071-01-DM)

  • Ishara F Azmi

National Institute of General Medical Sciences (GM54096)

  • David M MacAlpine
  • Craig L Peterson

National Institute of General Medical Sciences (GM104097)

  • David M MacAlpine
  • Craig L Peterson

Howard Hughes Medical Institute (Investigator)

  • Stephen P Bell

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Manabu Gaku Mizuguchi and Carl Wu for advice on nucleosome assembly assays and members of the Bell laboratory for helpful discussions. We thank Megan Warner for comments on the manuscript. This work was supported the by NIH grants GM054096 (CLP) and R01-GM104097 (DMM). SPB is an investigator with the Howard Hughes Medical Institute. IFA was supported in part by American Cancer Society Postdoctoral Fellowship (123700-PF-13-071-01-DMC). This work was supported in part by the Koch Institute Support Grant P30-CA14051 from the NCI. We thank the Koch Institute Swanson Biotechnology Center for technical support, specifically the Biopolymers and Genomics cores.

Reviewing Editor

  1. Robert Sclafani, University of Colorado School of Medicine, United States

Publication history

  1. Received: October 19, 2016
  2. Accepted: March 20, 2017
  3. Accepted Manuscript published: March 21, 2017 (version 1)
  4. Version of Record published: April 21, 2017 (version 2)

Copyright

© 2017, Azmi et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,638
    Page views
  • 741
    Downloads
  • 27
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, Scopus, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Biochemistry and Chemical Biology
    2. Cell Biology
    Jingxiang Li et al.
    Research Article Updated

    Autophagy acts as a pivotal innate immune response against infection. Some virulence effectors subvert the host autophagic machinery to escape the surveillance of autophagy. The mechanism by which pathogens interact with host autophagy remains mostly unclear. However, traditional strategies often have difficulty identifying host proteins that interact with effectors due to the weak, dynamic, and transient nature of these interactions. Here, we found that Enteropathogenic Escherichia coli (EPEC) regulates autophagosome formation in host cells dependent on effector NleE. The 26S Proteasome Regulatory Subunit 10 (PSMD10) was identified as a direct interaction partner of NleE in living cells by employing genetically incorporated crosslinkers. Pairwise chemical crosslinking revealed that NleE interacts with the N-terminus of PSMD10. We demonstrated that PSMD10 homodimerization is necessary for its interaction with ATG7 and promotion of autophagy, but not necessary for PSMD10 interaction with ATG12. Therefore, NleE-mediated PSMD10 in monomeric state attenuates host autophagosome formation. Our study reveals the mechanism through which EPEC attenuates host autophagy activity.

    1. Biochemistry and Chemical Biology
    2. Structural Biology and Molecular Biophysics
    Bence Hajdusits et al.
    Research Article

    In Gram-positive bacteria, the McsB protein arginine kinase is central to protein quality control, labelling aberrant molecules for degradation by the ClpCP protease. Despite its importance for stress response and pathogenicity, it is still elusive how the bacterial degradation labelling is regulated. Here, we delineate the mechanism how McsB targets aberrant proteins during stress conditions. Structural data reveal a self-compartmentalized kinase, in which the active sites are sequestered in a molecular cage. The 'closed' octamer interconverts with other oligomers in a phosphorylation-dependent manner and, contrary to these 'open' forms, preferentially labels unfolded proteins. In vivo data show that heat-shock triggers accumulation of higher-order oligomers, of which the octameric McsB is essential for surviving stress situations. The interconversion of open and closed oligomers represents a distinct regulatory mechanism of a degradation labeler, allowing the McsB kinase to adapt its potentially dangerous enzyme function to the needs of the bacterial cell.