Mei5-Sae3 stabilizes Dmc1 nucleating clusters for efficient Dmc1 assembly on RPA-coated single-stranded DNA

Abstract

Interhomolog recombination in meiosis requires a meiosis-specific recombinase, Dmc1. In Saccharomyces cerevisiae, the Mei5-Sae3 complex facilitates the loading of Dmc1 onto the replication protein A (RPA)-coated single-stranded DNA (ssDNA) to form nucleoprotein filaments. In vivo, Dmc1 and Mei5-Sae3 are interdependent in their colocalization on the chromosomes. However, the mechanistic role of Mei5-Sae3 in mediating Dmc1 activity remains unclear. We used single-molecule fluorescence resonance energy transfer and colocalization single-molecule spectroscopy experiments to elucidate how Mei5-Sae3 stimulates Dmc1 assembly on ssDNA and RPA-coated ssDNA. We showed that Mei5-Sae3 stabilized Dmc1 nucleating clusters with two to three molecules on naked DNA by preferentially reducing Dmc1 dissociation rates. Mei5-Sae3 also stimulated Dmc1 assembly on RPA-coated DNA. Using green fluorescent protein-labeled RPA, we showed the coexistence of an intermediate with Dmc1 and RPA on ssDNA before RPA dissociation. Moreover, the displacement efficiency of RPA depended on Dmc1 concentration, and its dependence was positively correlated with the stability of Dmc1 clusters on short ssDNA. These findings suggest a molecular model that Mei5-Sae3 mediates Dmc1 binding on RPA-coated ssDNA by stabilizing Dmc1 nucleating clusters, thus altering RPA dynamics on DNA to promote RPA dissociation.

Graphical Abstract

Figure 1.

Mei5–Sae3 enhanced Dmc1-mediated homologous recombination. (A) Schematic drawing of the DNA strand-exchange assay. ssDNA was sequentially incubated with Dmc1 and Mei5–Sae3, followed by the addition of ³²P-labeled dsDNA to initiate strand-exchange reactions. The reactions were stopped and deproteinized by SDS and proteinase K. The asterisk denotes the ³²P-label. (B) The strand-exchange activities of Dmc1 alone (1.6 μM, lane 2) and in the presence of Mei5–Sae3 in the indicated concentrations (lanes 3–9) were compared as shown. Mei5–Sae3 only (lane 10) was included as a negative control. B (blank) refers to a reaction mix without proteins. (C) The time course of Dmc1-mediated (1.6 μM) strand exchange in the presence of Mei5–Sae3 (0.8 μM, lanes 3–7), as well as Dmc1 only (lane 2), was analyzed at the indicated time points. Reaction without ATP (lane 8) was included as a negative control. (D) Schematic drawing of the D-loop formation assay. Cy5-labeled ssDNA was sequentially incubated with Dmc1 and Mei5–Sae3, followed by the addition of supercoiled dsDNA to initiate the reaction. The reactions were stopped and deproteinized by SDS and proteinase K. The asterisk denotes the Cy5 label. (E) The D-loop formation of Dmc1 alone (1.2 μM, lane 2) and Dmc1 with Mei5–Sae3 in the indicated concentrations (lanes 4–7) was compared as shown. Mei5–Sae3 only (lane 3) and reaction without ATP (lane 8) were included as negative controls. The percentage of the product was quantitated in the bottom panel. Reported results were derived from three independent experiments. The annotation **** indicates P ≤ 0.0001, *** indicates P ≤ 0.001, ** indicates P ≤ 0.01 and ns indicates P > 0.05.

Figure 2.

Mei5–Sae3 stabilized small Dmc1 clusters on ssDNA by reducing Dmc1 dissociation rates during assembly. (A) Schematic setup for the nuclease protection assay. The asterisk denotes the ³²P label. (B) The protected ssDNA (top band) was analyzed for Dmc1 alone (1 μM, lane 3) and in the presence of the indicated concentrations of Mei5–Sae3 (lanes 4–6). All results were from at least three independent experiments. The annotation **** indicates P ≤ 0.0001. (C) A Dmc1 assembly FRET assay on dT13 substrate. The FRET histograms of the dT13 substrate after adding Dmc1 alone (D) or in the presence of 500 nM Mei5–Sae3 (G) at 600 s after adding Dmc1. The bold orange curves indicate the sums of the fitted Gaussian mixture distribution, while the dashed curves indicate the fitted individual Gaussian components. The gray-shaded areas cover the mean ± 2 standard deviations (SDs) of the fitted Gaussian peaks. Exemplary FRET time courses with Dmc1 alone (E) or Dmc1 with Mei5–Sae3 (H). Orange lines were inferred from the fitted Gaussian Hidden-Markov Model. Transition density plots (TDPs) for FRET transitions observed for Dmc1 alone (F) or with Mei5–Sae3 (I). (J) Bayesian information criterion (BIC) score of the fitted HMM given different numbers of hidden states. The BIC scores reached a plateau for k ≥ 5 hidden states (black arrow). (K) Bar charts quantifying Dmc1-bound fraction at different times. Error bars indicate the standard error of the mean (SEM) of three individual experiments. Differences between reactions were tested using Student’s t-test. (L) Mean dwell times of different FRET states (states 1–4) on dT13. Error bars indicate the SEM of four recordings, each containing >200 molecules. Differences were tested using Student’s t-test; only statistical significances with P ≤ 0.01 (**) were annotated. The annotation *** indicates P ≤ 0.001 and ns indicates P > 0.05.

Figure 3.

Mei5–Sae3 mediated Dmc1 assembly on RPA-coated ssDNA. (A) The D-loop formation percentages in the presence of RPA (0.11 μM)-coated ssDNA were compared between Dmc1 alone (1.2 μM, lane 3) and with additional Mei5–Sae3 (lanes 6–9) in the indicated concentrations. Mei5–Sae3 alone (lane 5) and reaction without ATP (lane 10) were included as negative controls. Controls without RPA for Dmc1 alone (lane 2) and Mei5–Sae3 alone (lane 4) were included. The product percentage is shown in the bottom panel. All results were derived from at least three independent experiments. The annotation **** indicates P ≤ 0.0001. FRET histograms of median FRET values and heatmaps of FRET values and GFP–RPA fluorescence intensities of RPA-coated dT12 + 21 (B, C), with 4000 nM Dmc1 (D, E) and with 4000 nM Dmc1 and 500 nM Mei5–Sae3 (M5S3; F, G) at 30 min after adding proteins. The RPA-only FRET was ∼0.51 ± 0.045, and the GFP fluorescence signal was >400, indicating the stable binding of GFP–RPA (B, C). With the addition of Dmc1 and Mei5–Sae3, ∼50% of DNA shifted to the lower FRET values (∼0.2–0.4), concomitant with the disappearance of GFP signals (<400), indicating the displacement of RPA by Dmc1 (F, G). (H) Fraction of RPA displacement fraction on dT12 + 21 substrates: 4000 nM Dmc1 alone (yellow circles) and with an additional 500 nM Mei5–Sae3 (blue circles). Shaded areas indicate the SEM of N = 5 independent experiments. The control experiment without Dmc1 but with 500 nM Mei5–Sae3 was included as the green triangle. (I) The RPA-displaced fraction at different times (5, 10, 20 and 30 min; data from panel H). Differences were evaluated using Student’s t-test. The annotation **** indicates P ≤ 0.0001 and ** indicates P ≤ 0.01.

Figure 4.

Dmc1 showed similar concentration dependence for RPA displacement and assembly on short ssDNA. (A) The D-loop formation product in the presence of RPA-coated ssDNA was compared without and with calcium ions for 1.2 μM Dmc1 (lanes 3 and 5). The D-loop product percentage is shown in the bottom panel. (B) The Dmc1 filament stability was assayed against Benzonase nucleases at the indicated Dmc1 concentrations with calcium (lanes 6–8) and without calcium (lanes 3–5). The percentage of protected DNA is shown in the bottom panel. All results were derived from at least three independent experiments. The annotation **** indicates P ≤ 0.0001. (C) RPA-coated dT12 + 18 resulted in a FRET state centered at ∼0.56 ± 0.060 (Supplementary Figure S9). After incubating 1800 nM Dmc1 for 600 s, the FRET histogram showed that fractions of RPA-coated DNA molecules formed Dmc1 filaments in the presence of calcium ions. The histogram can be fitted to three Gaussians, representing the RPA-coated (dashed curves) and the RPA-displaced states (red bold curves). (D) RPA displacement efficiency by Dmc1 on RPA-coated dT12 + 18 substrates at various Dmc1 concentrations. The curve was fitted to the Hill–Langmuir equation (dotted line), giving K_D= 1895 ± 31 nM and n = 5.77 ± 0.53. K_D is the apparent dissociation constant, while n is the fitted Hill coefficient. Error bars are the SD of N = 4 individual experiments. (E) Bare dT9 substrates resulted in a FRET state of ∼0.85 ± 0.046 (Supplementary Figure S10). After incubating 1800 nM Dmc1 for 600 s, the FRET histogram showed three lower FRET peaks (red bold curves), likely corresponding to different Dmc1-bound states. (F) Binding curves of Dmc1 assembling on dT7, dT9 and dT13 substrates at various Dmc1 concentrations. The curves were fitted to the Hill–Langmuir equation (dotted line), giving K_D= 1725 ± 77 nM and n = 2.83 ± 0.42 for dT7, K_D= 1402 ± 39 nM and n = 4.81 ± 0.56 for dT9, and K_D= 440 ± 32 nM and n = 6.91 ± 1.5 for dT13. Error bars are the SD of N = 3 individual experiments.

Figure 5.

Dmc1 bound to RPA-coated ssDNA before RPA dissociation from DNA. (A) Real-time monitoring of 4000 nM Dmc1 binding on the GFP–RPA-coated dT12 + 21 substrate in the presence of 500 nM Mei5–Sae3. The RPA-coated dT12 + 21 substrate returned a stable FRET value of ∼0.51 ± 0.045, and the GFP–RPA signal was >400, indicating the stable binding of RPA (yellow barcode, top). The decrease in FRET (top, t_FRET) at ∼217 s reflected Dmc1 binding between the FRET dye pairs near the 5′ junction while the GFP–RPA signal stayed (brown barcode). The disappearance of the GFP signal (bottom, t_GFP, intensity <400) at ∼239 s reflected RPA dissociation (blue barcode). The time between these two events (brown barcode) indicated the coexistence of Dmc1 and RPA before RPA dissociation. (B) The probability distribution (n = 59) of the t_GFP − t_FRET dwell time for dT12 + 21 with Mei5–Sae3. Dwell time >0 s indicates the Dmc1–RPA coexistence (82%). (C) Real-time monitoring of Dmc1 binding on dT12 + 18 substrate precoated by GFP-labeled RPA. The quick decrease in the FRET signal (top, t_FRET) at ∼111.5 s reflected Dmc1 binding between the two FRET dye pairs at the 5′ junction, and the disappearance (<400) of the GFP signal (bottom, t_GFP) at ∼116 s reflected the dissociation of RPA. (D) The probability distribution (n = 55) of t_GFP − t_FRET dwell time for dT12 + 18 in the presence of Ca²⁺ ions. Dwell time >0 s indicated the Dmc1–RPA coexistence (84%). (E) An exemplary trace of real-time monitoring of Dmc1 binding on dT21 + 12 3′ dye pair substrate precoated by GFP-labeled RPA. The RPA-coated substrate returned a stable FRET value of ∼0.70 ± 0.041, and the GFP–RPA signal was >400. With the addition of 2000 nM Dmc1 + 0.8 mM Ca²⁺, a slight drop in FRET (FRET ∼ 0.55) was observed at ∼30 s, followed by the disappearance of the GFP signal (<400, t_GFP) at ∼32 s. The FRET state rose to a high FRET state (FRET ∼ 0.8) almost simultaneously with the disappearance of GFP and entered the Dmc1 state (FRET < 0.4, t_FRET) at ∼42 s. (F) The probability distribution (n = 105) of the t_GFP − t_FRET dwell time for dT21 + 12 3′ dye pair in the presence of Ca²⁺ ions. Dwell time <0 s indicated that Dmc1 had not formed on the 3′ side of ssDNA when RPA dissociated (92%).

Figure 6.

Model of Mei5–Sae3 stimulating Dmc1 assembly on RPA-coated ssDNA by stabilizing Dmc1 clusters on short ssDNA exposed during the dynamic of RPA. RPA contains multiple OB domains that transiently expose short ssDNA segments near the ss/dsDNA junctions, allowing accessibility of other DNA binding proteins. Dmc1 alone cannot stably bind to such exposed short ssDNA segments. In the presence of Mei5–Sae3 or Ca²⁺, Dmc1 binding to such short ssDNA segment is stabilized, thus preventing the rebinding of A–B OB domains of RPA. Consequent elongation of Dmc1 eventfully leads to RPA dissociation, resulting in the formation of active Dmc1 filaments required for homologous recombination events.