Volume 15, Issue 3, Pages 437-451 (August 2004) hMSH4-hMSH5 Recognizes Holliday Junctions and Forms a Meiosis-Specific Sliding Clamp that Embraces Homologous Chromosomes  Timothy Snowden, Samir Acharya, Charles Butz, Mark Berardini, Richard Fishel  Molecular Cell  Volume 15, Issue 3, Pages 437-451 (August 2004) DOI: 10.1016/j.molcel.2004.06.040 Copyright © 2004 Cell Press Terms and Conditions

Figure 1 hMSH4-hMSH5 Binds Uniquely and Specifically to Holliday Junctions (A) hMSH4-hMSH5 uniquely binds to Holliday Junctions. hMSH4-hMSH5 heterodimer was incubated with 10 fmol of end-labeled substrate DNA: an 80-mer duplex DNA (dsDNA; circles), an 82-mer containing a single G/T mismatch (G/T; diamonds), a Y Junction (YoJo, triangles), or a Holliday Junction (HoJo; squares) in a 20 μl reaction. Concentrations of hMSH4-hMSH5 [nM] are indicated above each lane. The first lane of each substrate panel [0 nM] shows the labeled DNA substrate in the absence of protein. The percentage of DNA bound by hMSH4-hMSH5 was determined by phosphorimager and plotted. Error bars indicate the standard deviation of at least three independent experiments. (B) hMSH4-hMSH5 binding to Holliday Junctions is specific. hMSH4-hMSH5 heterodimer (75 nM) and 10 fmol end-labeled Holliday Junction substrate were incubated with increasing concentrations of either unlabeled duplex DNA (dsDNA, lanes 3–7; circles) or unlabeled Holliday Junction DNA (Hojo, lanes 10–14; squares) in a 20 μl reaction. Fold excesses of unlabeled competitor concentrations are shown above each lane. Lanes 1 and 8 show labeled HoJo substrate in the absence of protein. Quantitation was performed as described in (A). Unbound Holliday Junction substrate (UB) and specific hMSH4-hMSH5 gel shift (GS) are marked with arrows. (C) Filter binding analysis of hMSH4-hMSH5 binding to ssDNA, dsDNA, or Holliday Junction DNA. Increasing concentrations of hMSH4-hMSH5 heterodimer were incubated with 10 fmol end-labeled oligo-dT80 (ssDNA; triangles), duplex DNA (dsDNA; circles), or Holliday Junction DNA (Hojo; squares) in a 20 μl reaction. Protein-DNA complexes retained on the filter were counted by scintillation. Error bars indicate the standard deviation of at least three independent experiments. (D) Competition analysis of hMSH4-hMSH5 bound to Holliday Junctions. hMSH4-hMSH5 heterodimer (75 nM) and 10 fmol end-labeled Holliday Junction DNA (HoJo) were incubated with increasing concentrations of unlabeled competitor DNA: oligo-dT80 (ssDNA; triangles), duplex DNA (dsDNA; circles), or Holliday Junction DNA (Hojo; squares) in a 20 μl reaction. Reactions were performed and quantitated as described in (C). (E) hMSH4-hMSH5 binds to the core of Holliday Junctions. Four related Holliday Junction constructs, each containing one arm with a 5′ end label, were incubated with increasing concentrations of hMSH4-hMSH5, followed by digestion with DNase I in a final reaction volume of 20 μl. Protein concentrations are noted below each lane [nM]. Strand names are based upon the restriction sites closest to the 5′ end of the oligonucleotide containing the label (see Supplemental Data). For example, “A-X” = the 5′ end-labeled Asc-Xba 80-mer strand. The branch-immobile symmetric core of the Holliday Junction is designated by the asterisk. Maxim Gilbert sequencing of the B-S strand is shown for orientation. The red bars at the right of each panel indicate the areas of protection in the presence of hMSH4-hMSH5. The areas of protection from all strands are indicated by gray over color-coded strands in the diagram of the Holliday Junction substrate at the right of the figure. Scale bars are approximate nucleotide positions. Molecular Cell 2004 15, 437-451DOI: (10.1016/j.molcel.2004.06.040) Copyright © 2004 Cell Press Terms and Conditions

Figure 2 Holliday Junction-Dependent Activation of hMSH4-hMSH5 ATPase and ADP→ATP Exchange Activities (A) hMSH4-hMSH5 binds both ADP and ATP. Filter binding was used to calculate the equilibrium dissociation constant (KD) for ADP (diamonds, KD = 20 ± 1 μM) or ATPγS (triangles, KD = 5.9 ± 0.5 μM) in the absence of DNA in a 20 μl reaction. Error bars indicate the standard deviation of at least three independent experiments. (B) Adenosine nucleotide binding leads to distinct conformational changes in hMSH4-hMSH5. hMSH4-hMSH5 in the presence of magnesium alone (Mg), magnesium-ADP (Mg-ADP), or magnesium-ATPγS (Mg-ATPγS) was digested with increasing concentrations of the protease Endo Glu-C. Products were separated on a 10% SDS PAGE gel and visualized by silver staining. Arrows indicate bands that are unique or altered between the three conditions tested. (C) Holiday Junction DNA stimulates the hMSH4-hMSH5 ATPase. The ATPase activity of hMSH4-hMSH5 (200 nM) are plotted as a function of ATP concentration, and the curves were fit to the Michaelis-Menten equation (Gradia et al., 1997). ATPase activity was measured in the absence of DNA (diamonds), in the presence of duplex DNA (dsDNA, 600 nM; triangles), and in the presence of Holliday Junction DNA (Hojo, 600 nM; circles) in a 20 μl reaction. Error bars indicate the standard deviation of at least three independent experiments. (D) Single-round ATP hydrolysis by hMSH4-hMSH5 is DNA independent. hMSH4-hMSH5 (200 nM) was preincubated with [γ32P]-ATP (2.3 mM), followed by addition of magnesium and an excess of unlabeled ATP in the absence of DNA (diamonds), in the presence of duplex DNA (dsDNA, 600 nM; triangles), and in the presence of Holliday Junction DNA (Hojo, 600 nM; circles) in a final reaction volume of 20 μl. The rate of γ-phosphate release is indicative of a single round of hydrolysis of the prebound [γ32P]-ATP. Error bars indicate the standard deviation of at least three independent experiments. (E) Holliday Junction DNA provokes ADP→ATP exchange by hMSH4-hMSH5. hMSH4-hMSH5 was prebound to [3H]-ADP, followed by the addition of 500 μM unlabeled ATP in the absence of DNA (diamonds), in the presence of excess duplex DNA (dsDNA; triangles), and in the presence of Holliday Junction DNA (Hojo; circles) in a final reaction volume of 20 μl. Results are plotted as the fraction of total [3H]-ADP that remained bound over time. The standard deviation was calculated from the averages [3H]-ADP released at four different concentrations of hMSH4-hMSH5 (25, 50, 75, 100 nM) from at least three independent experiments each. (F) ATP binding following the release of ADP by hMSH4-hMSH5 is stimulated by Holliday Junction DNA. Reactions are as described in (D), except unlabeled ADP was preincubated with hMSH4-hMSH5, and the subsequent binding of labeled [35S]-ATPγS was measured over time in the absence of DNA (diamonds), in the presence of 3-fold excess duplex DNA (dsDNA; triangles), and in the presence of Holliday Junction DNA (Hojo; circles) in a final reaction volume of 20 μl. Error bars indicate the standard deviation of at least three independent experiments. Molecular Cell 2004 15, 437-451DOI: (10.1016/j.molcel.2004.06.040) Copyright © 2004 Cell Press Terms and Conditions

Figure 3 ATP Binding Induces Hydrolysis-Independent Release of hMSH4-hMSH5 from Holliday Junctions Containing Open Ends (A and B) ATP- and ATPγS-dependent release of hMSH4-hMSH5 prebound to Holliday Junction DNA containing open ends. hMSH4-hMSH5 (75 nM) binding to 10 fmol labeled, open-end Holliday Junction DNA in the presence of increasing concentrations of either (A) ATP or (B) ATPγS in a 20 μl reaction were examined by EMSA. Nucleotide concentrations are indicated above each lane. The first lane in each panel represents the migration of labeled Holliday Junction in the absence of protein. Unbound Holliday Junction substrate (UB) and specific hMSH4-hMSH5 gel shift (GS) are marked with arrows. The gel shift in the absence of ATP/ATPγS is indicated by [0 μM]. (C) Fraction of open-end Holliday Junction bound by hMSH4-hMSH5 in the presence of increasing concentrations of ATP (circles) or ATPγS (triangles). The percentage of DNA bound in (A) and (B) was determined by phosphorimager analysis. Error bars indicate the standard deviation at least three independent experiments. (D) DNase I footprint analysis of ATPγS-dependent release of hMSH4-hMSH5 prebound to Holliday Junction DNA containing open ends. DNase I footprinting of the open-end Holliday Junction with increasing concentrations of hMSH4-hMSH5 was performed as described in Figure 1E (left panels for each strand). Protein concentrations are shown below each lane [nM]. DNase footprint analysis of ATPγS-induced dissociation from open-ended Holliday Junction DNA (right panels of each strand) was performed with prebound hMSH4-hMSH5 (200 nM) and then introduced to increasing concentrations of ATPγS [mM]. The red bars at the right indicate the areas of DNase I protection by hMSH4-hMSH5 in the absence of ATPγS. A cartoon diagram of the Holliday Junction substrate is shown on the far right. Molecular Cell 2004 15, 437-451DOI: (10.1016/j.molcel.2004.06.040) Copyright © 2004 Cell Press Terms and Conditions

Figure 4 hMSH2-hMSH6 Binding to Holliday Junctions Is Distinctly Different than hMSH4-hMSH5 (A) hMSH2-hMSH6 displays high-affinity binding to DNA containing a G/T mismatch, Y Junction DNA, and Holliday Junction DNA. hMSH2-hMSH6 heterodimer was incubated with 10 fmol of end-labeled substrate DNA: an 80-mer dsDNA (dsDNA; circles), an 82-mer containing a single G/T mismatch (G/T; diamonds), a Y Junction (YoJo, triangles), or a Holliday Junction (HoJo; squares) in a 20 μl reaction (compare with Figure 1A). Increasing concentrations of hMSH2-hMSH6 are indicated above each lane. The first lane of each substrate panel indicates the DNA migration in the absence of protein. The fraction of DNA bound by hMSH2-hMSH6 was determined by phosphorimager analysis. Error bars indicate the standard deviation of at least three individual experiments. (B) Competition of hMSH2-hMSH6 bound to Holliday Junction results in multiple species that are distinctly different from hMSH4-hMSH5. hMSH2-hMSH6 heterodimer (75 nM) and 10 fmol end-labeled Holliday Junction substrate were incubated with increasing concentrations of either unlabeled duplex DNA (dsDNA, lanes 3–7; circles) or unlabeled Holliday Junction DNA (Hojo, lanes 10–14; squares) in a 20 μl reaction. Fold excess of unlabeled competitor concentrations are shown above each lane. Lanes 1 and 8 show labeled Holliday Junction substrate in the absence of protein. Quantitation was performed as described in (A). Unbound Holliday Junction substrate (UB), hMSH2-hMSH6 gel shift (GS), and partial competition species (PC) are marked with arrows. (C) ATP and ATPγS do not induce release of hMSH2-hMSH6 from open-end Holliday Junctions. EMSA contained hMSH2-hMSH6 (75 nM), 10 fmol labeled, open-end Holliday Junction, and increasing concentrations of either ATP or ATPγS in a 20 μl reaction. Nucleotide concentrations are shown above each lane. The first lane in each gel represents labeled Holliday Junction substrate in the absence of protein. Quantitation was performed as described in (A). Unbound Holliday Junction substrate (UB) and hMSH2-hMSH6 gel shift (GS) are marked with arrows. Molecular Cell 2004 15, 437-451DOI: (10.1016/j.molcel.2004.06.040) Copyright © 2004 Cell Press Terms and Conditions

Figure 5 hMSH4-hMSH5 Forms a Hydrolysis-Independent Sliding Clamp (A) Quadruple blocked-end Holliday Junction DNA retains hMSH4-hMSH5 in the presence of ATP and ATPγS. hMSH4-hMSH5 (75 nM) was prebound to 10 fmol end-labeled Holliday Junction DNA containing 3-biotin bound by streptavidin on all the available ends. ATP (top) or ATPγS (bottom) was added at concentrations indicated above each lane in a 20 μl reaction and subjected to EMSA. Unbound quadruple blocked-end Holliday Junction substrate (UB) and specific hMSH4-hMSH5 gel shift (GS) are marked with arrows. The first lane in each gel represents labeled Holliday Junction substrate in the absence of protein. The fraction of quadruple blocked-end Holliday Junction DNA bound was determined by phosphorimager. Error bars indicate the standard deviation at least three independent experiments. A cartoon diagram of the quadruple blocked-end Holliday Junction substrate used in this panel is shown at the far right. (B) Lateral opposing, double blocked-end Holliday Junction DNA retains hMSH4-hMSH5 in the presence of ATP and ATPγS. hMSH4-hMSH5 (75 nM) was prebound to 10 fmol end-labeled Holliday Junction DNA containing 3-biotin bound by streptavidin lateral opposing ends of the antiparallel stacked-X structure (see Supplemental Figure S2). ATP (top) or ATPγS (bottom) was added at concentrations indicated above each lane in a 20 μl reaction and subjected to EMSA. Unbound double blocked-end Holliday Junction substrate (UB) and specific hMSH4-hMSH5 gel shift (GS) are marked with arrows. The first lane in each gel represents labeled Holliday Junction substrate in the absence of protein. Quantitation was performed as described in (A). A cartoon diagram of the double blocked-end Holliday Junction substrate used in this panel is shown at the far right. (C) ATP and ATPγS induces the release of hMSH4-hMSH5 from single blocked-end Holliday Junction DNA. hMSH4-hMSH5 (75 nM) was prebound to Holliday Junction DNA (10 fmol) containing a single 3-biotin bound by streptavidin on one arm of the antiparallel stacked-X structure (see Supplemental Figure S2). ATP (top) or ATPγS (bottom) was added at concentrations indicated above each lane and subjected to EMSA. Unbound, single-end Holliday Junction substrate (UB) and specific hMSH4-hMSH5 gel shift (GS) are marked with arrows. The first lane in each gel shows the migration of the Holliday Junction substrate in the absence of protein. Quantitation was performed as described in (A). A cartoon diagram of the single blocked-end Holliday Junction substrate used in (C) is shown at the far right. (D) DNase I footprint analysis of ATPγS-dependent release of hMSH4-hMSH5 prebound to laterally opposing, double blocked-end Holliday Junction DNA. DNase I footprinting of the double blocked-end Holliday Junction with increasing concentrations of hMSH4-hMSH5 was performed as described in Figure 1E (left panels for each strand). Protein concentrations are shown below each lane [nM]. DNase footprint analysis of ATPγS-induced dissociation from double blocked-end Holliday Junction DNA (right panels of each strand) was performed with prebound hMSH4-hMSH5 (200 nM) and then introduced to increasing concentrations of ATPγS [μM]. The red bars at the right indicate consensus areas of DNase I protection by hMSH4-hMSH5 in the absence of ATPγS. Protection is expanded along the entire length in the presence of ATPγS. (E) The mechanics of hMSH4-hMSH5 recognition of Holliday Junctions. hMSH4-hMSH5 binding to a Holliday Junction provokes ADP→ATP exchange and the formation of a hydrolysis-independent sliding clamp that links two homologous duplex DNA arms. Dissociation of one hMSH4-hMSH5 sliding clamp from the Holliday Junction exposes the crossover to additional hMSH4-hMSH5 binding and clamp-formation events. Molecular Cell 2004 15, 437-451DOI: (10.1016/j.molcel.2004.06.040) Copyright © 2004 Cell Press Terms and Conditions

Figure 6 hMSH4-hMSH5 Recognizes and Specifically Binds to a Consensus Holliday Junction Progenitor (A) hMSH4-hMSH5 uniquely binds to a Holliday Junction progenitor. hMSH4-hMSH5 was incubated with 10 fmol of end-labeled substrate DNA: Holliday Junction (HoJo; squares) or Holliday Junction progenitor containing a single-stranded arm (see substrate cartoon at far right; Pro-HoJo; circles) in a 20 μl reaction. Concentrations of hMSH4-hMSH5 [nM] are indicated above each lane. The first lane of each substrate panel [0 nM] shows migration of the DNA substrate in the absence of protein. The percentage of DNA bound by hMSH4-hMSH5 was determined by phosphorimager. Error bars indicate the standard deviation of at least three independent experiments. (B) The specificity of hMSH4-hMSH5 binding to progenitor Holliday Junctions. hMSH4-hMSH5 (75 nM) and progenitor Holliday Junction substrate (10 fmol) were incubated with either unlabeled single-stranded DNA (ssDNA, lanes 3–7; triangles), duplex DNA (dsDNA, lanes 10–14; circles), or unlabeled Holliday Junction DNA (Hojo, lanes 17–21; squares) in a 20 μl reaction. Fold excess of unlabeled competitor is shown above each lane. Lanes 1, 8, and 15 are labeled Pro-HoJo substrate in the absence of protein. Quantitation was performed as described in (A). Unbound progenitor Holliday Junctions and Holliday Junction DNA substrates (UB) and specific hMSH4-hMSH5 gel shift (GS) are marked with arrows. (C) ATP and ATPγS induced the release of hMSH4-hMSH5 from progenitor Holliday Junction DNA. hMSH4-hMSH5 (75 nM) was bound to progenitor Holliday Junction DNA (10 fmol) in the presence of ATP (top panel) or ATPγS (bottom panel) in a 20 μl reaction and subjected to EMSA. Adenosine nucleotide concentrations are indicated above each lane. The first lane in each panel represents the migration of progenitor Holliday Junctions in the absence of protein. Unbound progenitor Holliday Junction substrate (UB) and specific hMSH4-hMSH5 gel shift (GS) are marked with arrows. The gel shift in the absence of ATP/ATPγS is indicated by [0 μM]. Quantitation was performed as described in (A). A cartoon diagram of the progenitor Holliday Junction substrate is shown at the right. Molecular Cell 2004 15, 437-451DOI: (10.1016/j.molcel.2004.06.040) Copyright © 2004 Cell Press Terms and Conditions

Figure 7 A Model for the Role of hMSH4-hMSH5 in Stabilizing and Preserving Holliday Junctions during Prophase of Meiosis I See text for discussion. Molecular Cell 2004 15, 437-451DOI: (10.1016/j.molcel.2004.06.040) Copyright © 2004 Cell Press Terms and Conditions