The Replication Checkpoint Prevents Two Types of Fork Collapse without Regulating Replisome Stability

Abstract

The ATR replication checkpoint ensures that stalled forks remain stable when replisome movement is impeded. Using an improved iPOND protocol combined with SILAC mass spectrometry, we characterized human replisome dynamics in response to fork stalling. Our data provide a quantitative picture of the replisome and replication stress response proteomes in 32 experimental conditions. Importantly, rather than stabilize the replisome, the checkpoint prevents two distinct types of fork collapse. Unsupervised hierarchical clustering of protein abundance on nascent DNA is sufficient to identify protein complexes and place newly identified replisome-associated proteins into functional pathways. As an example, we demonstrate that ZNF644 complexes with the G9a/GLP methyltransferase at replication forks and is needed to prevent replication-associated DNA damage. Our data reveal how the replication checkpoint preserves genome integrity, provide insights into the mechanism of action of ATR inhibitors, and will be a useful resource for replication, DNA repair, and chromatin investigators.

Figure 1

iPOND-SILAC-MS identifies replisome and replication stress response protein complexes. (A) Each HU-treated sample was compared to an untreated sample. EdU remained in the growth media during the HU time course. (B) Cells were treated with HU or HU and ATRi for between 15 minutes and eight hours. Each HU/ATRi-treated sample was compared to a HU-treated reference sample. (C and D) Unsupervised hierarchical clustering of proteins. The first column (Ctl) is a control comparing EdU-labeled vs. unlabeled cells. Green in this column indicates the protein is enriched on chromatin. Green in the second column (chase) indicates proteins that are enriched at replication forks compared to bulk chromatin. The other columns indicate proteins that are either increased (red) or decreased (green) in abundance upon HU-treatment or ATR inhibition. (Grey = not observed) See also Figures S1 and S2, Tables S1-S6.

Figure 2

ZNF644 forms a complex with G9A at active replication forks. (A) Unsupervised hierarchical clustering of the iPOND-SILAC-MS data identified ZNF644 as a protein that clusters with the G9a/GLP. (B) A profile plot of all the proteins in the dataset indicates that ZNF644 (red) abundance on nascent DNA is highly correlated with G9a, GLP, and WIZ (blue). (C) Coimmunoprecipitation of Flag-ZNF644 with G9a from HEK293T cells. (D) ZNF644 is a paralog of WIZ. Locations of zinc finger domains (ZnF) are indicated along with the percent amino acid sequence identity and location of the G9a/GLP binding motif in WIZ. (E) ZNF644 co-localizes with PCNA at sites of DNA replication. (F-H) U2OS cells were transfected with the indicated siRNAs (NT, non-targeting; si644, ZNF644). (F) Cell proliferation was measured using alamar blue. (*p<0.001) (G) The viability of cells treated with 0.2mM hydroxyurea (HU) for 72 hours was compared to untreated cells. (*p<0.05) (H) Cells were labeled with 10μM EdU for 30 minutes and then left untreated or treated with 2mM HU for 1 hour. The graph depicts the γH2AX intensity in each EdU positive nucleus. (*p<0.001) Data are presented as mean +/− SE.

Figure 3

ATR signaling does not regulate replisome stability but does control fork stability. (A-E) Each HU sample was compared to an untreated reference sample while each HU/ATRi sample was compared to HU alone. The Log2 of the average abundance ratio of all subunits in selected complexes is depicted for simplicity when the subunits behave similarly. (GINS = GINS1-4; CAF1 = CAF1A and CAF1B; POLε = POLE and POLE2; POLδ = POLD1-3; POLα-PRIM = POLA1, POLA2, PRIM1, PRIM2; RNAseH = RNASEHA, B, & C; FACT = SUPT6H and SSRP1; (RPA = RPA1-3; MRN = MRE11-RAD50-NBS1). See also Figures S3-S5.

Figure 4

The replication stress response includes changes in ATR, USP1 and PARG that regulate post-translational modifications needed for protein recruitment. (A-D) The abundance changes of select replication stress and DNA damage response proteins and protein complexes are diagrammed.

Figure 5

HU-treated cell populations exhibit slow replication elongation and minimal new replication initiation. (A-C) Cells were labeled with IdU for 15 minutes, then CldU in the presence of 3mM HU for increasing times between 15 minutes and 24 hours. (A) Representative DNA fiber images. There was no significant change in green fiber length in any sample. (B) Quantitation of CldU fiber length. At least 200 DNA fibers were measured per time point. (C) Mean fiber length for each time point is depicted. A rate of elongation in the HU-treated samples was calculated from the slope of the fitted line and converted to base pairs per minute. (D) iPOND-SILAC-MS was utilized to compare cell populations treated with CDC7i to untreated cells at the two and eight hour HU-time points. (MCM = MCM2-7; GINS = GINS1-4; CAF1 = CAF1A and CAF1B; POLε = POLE and POLE2; POLδ = POLD1-3; POLα-PRIM = POLA1, POLA2, PRIM1, PRIM2) (E) The percentage of CldU only fibers as a measure of new origin firing is depicted as a percentage of total fibers. See also Figure S6.

Figure 6

Blocking origin firing in ATR-inhibited cells reveals two distinct populations of collapsed forks. (A-C) ATRi+HU vs. HU abundance ratios (blue) are compared to ATRi+HU+CDC7i vs. HU+CDC7i abundance ratios (red). (D) Cells were labeled with IdU for 20 minutes, washed, then treated for 80, 120, or 240 minutes with HU, HU+ATRi, or HU+ATRi+CDC7i. These drugs were removed prior to the addition of CldU for 30 minutes. DNA fibers were analyzed and the percentage of fibers with both IdU and CldU staining indicating fork restart were quantitated. At least 300 fibers were analyzed. Data is presented as mean +/−SD. (E) A neutral comet assay was performed on cells treated with the indicated drugs for four hours. At least 100 tail moments were quantitated with CometScore software. One-way ANOVA analysis for the samples (p<0.0001) was followed up by Bonferroni’s multiple comparison tests. (n.s = not significant; *p<0.0001). See also Figure S7.

Figure 7

RPA overexpression delays fork collapse but does not alter the abundance of DNA damage response proteins on nascent DNA in checkpoint-deficient cells. (A) Immunoblot showing RPA32 overexpression compared to untransfected (Unt.) cells. (B) Fork recovery assays were performed as in Figure 6D in cells overexpressing RPA. (C) Abundance values of proteins from iPOND-SILAC-MS analysis comparing fork proteomes in RPA overexpressing to untransfected cell populations that were treated with HU and ATRi for the indicated times.