Single-molecule imaging reveals control of parental histone recycling by free histones during DNA replication

Parental histone dynamics at the replication fork are altered by free histones, leading to controlled eviction or transfer.


Extract-induced chromatinization of fluorescent l nucleosomes in single-molecule replication assays
Xenopus egg extracts have a well-characterized ability to chromatinize naked DNA (Banaszynski et al Methods 2010;DOI: 10.1016DOI: 10. /j.ymeth.2009. In our single-molecule replication assays, upon introduction of HSS, additional dark nucleosomes are assembled on fluorescent l nucleosomes from endogenous histones, present in extracts at micromolar concentrations ( fig. S3A). The extent of extract-induced chromatinization is different for singly-and doubly-tethered molecules in a microfluidic chamber.
In the case of singly-tethered fluorescent l nucleosomes, the only limit to chromatinization is the amount of naked DNA available. Upon introduction of HSS, all naked DNA regions are prone to being occupied with dark nucleosomes and other DNA binding proteins present in extracts. In our assays, extract-induced chromatinization of singly-tethered fluorescent l nucleosomes leads to a complete compaction of individual molecules (to a diffraction-limited fluorescent spot), typically within 10-15 sec from the moment extract reached the chamber. Singly-tethered molecules compacted in extracts do not stretch under the flow of native buffers. Naked l DNA (48.5 kbp) can accommodate approximately 240 nucleosomes in total, assuming 0.2 kbp per nucleosome.
In the case of doubly-tethered, stretched l nucleosomes, extract-induced chromatinization is additionally limited by the slack within the molecule (only the 'loose' DNA can be chromatinized). Based on the measured average contour length for naked l and low-density l nucleosomes, we estimate that ~30 % of l DNA is available for nucleosome assembly. This suggests that ~70 nucleosomes (both fluorescent and dark; assuming 0.2 kbp per nucleosome) can be present on low-density fluorescent nucleosome templates in our assays (used in experiments presented in Fig. 3-5). As a point of reference, we typically see between 5 and 15 fluorescent nucleosomes per DNA molecule, which leaves room for about 55-65 dark nucleosomes. Please note that these calculations do not take into account DNA compaction caused by binding of other proteins from extracts. Indeed, upon introduction of HSS, dark histone loading inevitably competes with other binding events, including pre-replication complexes assembly. The actual number of dark nucleosomes may therefore deviate from the above estimate.

Loss of fluorescent histones during licensing and replication in extracts
Fluorescent l nucleosomes are stable in buffer. During licensing in HSS, histone-associated fluorescence is lost at a significantly faster rate (Fig. 2) than the measured rate of photobleaching in buffer ( fig. S2). We attribute this loss to the exchange of fluorescent histones with their native, unlabelled counterparts, present in HSS at ~1-6 µM concentration ( fig. S3A). This process is likely mediated by histone chaperones and chromatin remodelers (Onikubo and Shechter Int J Dev Biol 2016; DOI: 10.1387/ijdb.130188ds). Histones H2A and H2B exchange approximately three-times faster than histones H3 and H4 (Fig. 2). For low density nucleosome l templates (containing ~5-15 fluorescent nucleosomes upon immobilization in buffer), about a half of fluorescent histones H3 and H4 are lost, on average, during the 15-minute licensing reaction (Fig. 2).
Upon introduction of NPE, further loss of histone fluorescence is observed due to replication-independent (histone loss away from the replication fork) and replication-induced (histone eviction at the replication fork) processes. The two processes contribute more or less equally to the overall loss of histone fluorescence during replication ( fig. S4F) but nevertheless can be readily distinguished in real-time replication data for doublytethered l nucleosomes (for example, see Fig. 4E). Only replication fork-associated histone eviction events are counted in the collision outcome statistics presented in Fig. 5. It is possible that some of these evictions might not be caused by the replisome arrival but other replication-independent processes (e.g. photobleaching or exchange of fluorescent histones with their native counterparts from extracts). However, we estimate the likelihood of such a coincidence to be low, and not significant enough to affect the overall statistical analysis of collision outcomes. For example, in regular extracts, we counted 107 and 54 histone eviction events at the replication forks for nucleosomes containing H4-E63C A647 and H3-K36C Cy5 (Fig. 5), respectively, while the number of replicationindependent losses was 128 and 63, respectively (for the same sets of data). Given that replication-dependent histone eviction and replication-independent histone loss occur at equal probabilities ( fig. S4F), this comparison  (D) Bulk replication assay for naked pBRII plasmid and pBRII plasmid containing nucleosomes labelled at H3-K36C Cy5 or H4-E63C A647 . For each replicated template, a negative control is also presented (+Geminin). Quantification of the replication efficiency, as measured by the radioactivity signal, is presented in the lower panel.

DT-Bckg-Norm
Doubly-tethered Singly-tethered together (merge; right panels) are presented. Time and size scales are indicated. (D) Plot showing the mean loss of H3-K36C Cy5 fluorescence for doubly-(blue squares) and singly-tethered (magenta circles) l nucleosomes during replication under unrestricted firing conditions. 110 molecules were analyzed for each data set. Individual fluorescence decay traces were normalized to background ('0') and maximum values of fluorescence ('1'). A mean fluorescence value and standard deviation were calculated and plotted for each time point. We observed no difference in the loss of H3-K36C Cy5 fluorescence between the doubly-and singly-tethered l nucleosomes. (E) Plot showing the mean increase of Fen1-KikGR fluorescence for doubly-(blue squares) and singly-tethered (magenta circles) l nucleosomes during replication under unrestricted firing conditions. 110 molecules were analyzed for each data set. Individual fluorescence decay traces were normalized to minimum ('0') and maximum values of fluorescence ('1'). A mean fluorescence value and standard deviation were calculated and plotted for each time point. We observed no difference in the firing timing between the doubly-and singly-tethered l nucleosomes. (F) Plot showing the mean loss of H3-K36C Cy5 fluorescence for 1:1 mixture of doubly-and singlytethered l nucleosomes during replication under unrestricted firing conditions (red circles) versus non-replicating control (black squares; +Geminin). 36 fields of view were analyzed for each data set. Decay traces for individual fields of view were normalized to background ('0') and maximum values of fluorescence ('1'). A mean fluorescence value and standard deviation were calculated and plotted for each time point. Data sets were fitted to a linear regression model (see table S2 for fitting parameters) and the statistical significance of the differences between the slopes was estimated as <0.0001. The loss of histone-associated fluorescence is significantly faster for replicating nucleosomes than for nucleosomes incubated in non-replicating extracts (see Supplementary Text for further details and panel G for the replication efficiency comparison). (G) Plot showing the mean increase of Fen1-KikGR fluorescence for 1:1 mixture of doubly-and singly-tethered l nucleosomes during replication under unrestricted firing conditions (red circles) versus non-replicating control (black squares; +Geminin). 36 fields of view were analyzed for each data set. A mean fluorescence value and standard deviation were calculated and plotted for each time point.  termination in this assay, leads to non-specific sticking of the immobilized DNA molecules to the surface, which appear bent rather than straight lines under the microscope. Note that experiments presented in panels A and B were run in parallel using the same extract mixture, which was split and supplemented with either Fen1-KikGR for real-time replication visualization (A) or dig-dUTP for post-replication detection (B). After 30 minutes, replication was stopped by flowing in buffer containing 20 mM Tris pH 7.5, 10 mM EDTA and 0.5 M NaCl. Under these conditions, extracts were washed out from the flow cell but nucleosomes remained intact on the immobilized DNA and were imaged with 640-nm laser. Nascent DNA, containing dig-dUTP incorporated during replication, was visualized through immunostaining with an anti-digoxigenin antibody labelled with fluorescein (anti-dig Ab Fluor ; excited with 488-nm laser). In addition, non-and replicated DNA was stained with SYTOX Orange and visualized using 561-nm laser.   Tables   Table S1. Fitting parameters for histone dynamics data in HSS. Corresponds to Fig. 2, D and E. N indicates the number of individual decay traces used to generate plots of the mean loss of fluorescence. Mean loss of fluorescence data were fitted to a one phase decay model, where Y = (Y0 -Plateau)*exp(-K*X) + Plateau. Y0 is the Y value when X (time) is zero, K is the rate constant and t0.5 is the half-life, calculated as ln (2)