Ycs4 Subunit of Saccharomyces cerevisiae Condensin Binds DNA and Modulates the Enzyme Turnover

Abstract

Condensins play a key role in higher order chromosome organization. In budding yeast Saccharomyces cerevisiae, a condensin complex consists of five subunits: two conserved structural maintenance of chromosome subunits, Smc2 and Smc4, a kleisin Brn1, and two HEAT repeat subunits, Ycg1, which possesses a DNA binding activity, and Ycs4, which can transiently associate with Smc4 and thereby disrupt its association with the Smc2 head. We characterized here DNA binding activity of the non-SMC subunits using an agnostic, model-independent approach. To this end, we mapped the DNA interface of the complex using sulfo-NHS biotin labeling. Besides the known site on Ycg1, we found a patch of lysines at the C-terminal domain of Ycs4 that were protected from biotinylation in the presence of DNA. Point mutations at the predicted protein–DNA interface reduced both Ycs4 binding to DNA and the DNA stimulated ATPase activity of the reconstituted condensin, whereas overproduction of the mutant Ycs4 was detrimental for yeast viability. Notably, the DNA binding site on Ycs4 partially overlapped with its interface with SMC4, revealing an intricate interplay between DNA binding, engagement of the Smc2-Smc4 heads, and ATP hydrolysis and suggesting a mechanism for ATP-modulated loading and translocation of condensins on DNA.

Graphical Abstract

INTRODUCTION

Condensins are responsible for higher-order chromosome organization in organisms ranging from bacteria to humans. These multi-subunit proteins are involved in diverse cellular functions including chromosome condensation and segregation and DNA repair and recombination (reviewed in refs 3–5). The proteins consist of a dimeric core of SMC (structural chromosome maintenance) subunits, which dynamically interact with the regulatory non-SMC subunit. Two large families of condensins have been identified. The highly conserved SMC family of condensins is found in eukaryotes, archaea, and bacteria.⁶ In addition, many bacteria carry condensins from the MukBEF/MksBEF superfamily, which are characterized by low sequence conservation but are structurally and functionally similar to the SMC condensins.⁷

The mechanisms of condensins are based on several DNA remodeling activities. Condensins from diverse organisms act by stabilizing dynamic, ATP-controlled bridges between distant DNA segments.^8–12 These binding and bridging of DNA were proposed to employ an ATP-driven loop-extrusion mechanism^13–17 where a DNA loop trapped by the protein progressively elongates via an ATP-mediated translocation. On the biochemical level, however, condensins vary between the families. In the Escherichia coli condensin MukBEF, the DNA reshaping activity resides in its SMC subunit, whereas the non-SMC subunit alone does not directly interact with DNA.^18,19 In contrast, the SMC family condensins are stimulated by their non-SMC components,^1,20–22 which can bind DNA on their own.^1,2

In budding yeast Saccharomyces cerevisiae, condensins consist of a heterodimer of Smc2 and Smc4, a kleisin Brn1, and two HEAT repeat subunits Ycs4 and Ycg1 (Figure 1A).²³ HEAT repeats in general display structural flexibility²⁴ and high elasticity against external forces²⁵ and mediate protein–protein interactions.²⁶ In Ycg1, a stretch of HEAT repeats bends by 180°, creating a hairpin configuration capable of encircling DNA.² All five subunits of the yeast condensin are essential for cell viability^27–30 and faithful progression of both mitosis and meiosis.^3,31–33

Figure 1.

S. cerevisiae non-SMC subunits Ycs4 and Brn1 bind DNA. (A) Subunit geometry of yeast condensin. Smc2 and Smc4 form a V-shaped structure with two globular head domains that are connected by two long coiled coils with a hinge domain at the junction. Two HEAT-repeat non-SMC subunits (Ycs4 and Ycg1) bind SMC head domains via a kleisin subunit. SDS-PAGE analysis of purified complexes of Smc2-Smc4 (SMC2/4) and Ycs4-Brn1-Ycg1 (Ycs4-BY). (B) Gel shift analysis reveals the similar affinity of SMC2/4 and non-SMC subunits. (C) Binding affinity of the non-SMC subcomplex resides in Ycs4 and Brn1.

The functions of all condensin subunits have been largely determined. The heterodimer Smc2-Smc4 comprises the ATPase and DNA binding core of the complex.¹⁰ Its activity is modulated by the non-SMC subunit wherein each subunit has its own function.^12,1,2,22 The kleisin Brn1 bridges the head domains of Smc2 and Smc4, completing the tripartite ring. Brn1 also associates with Ycg1 and Ycs4, which do not directly interact with each other and require the kleisin to be brought into the holoenzyme.^1,2 Ycg1, in complex with Brn1, can entrap DNA, which presumably augments the DNA interaction with Smc2/4.² Ycs4, on the other hand, can bind the Smc4 head in a manner that precludes its ATP binding and association with Smc2.^22,34

We report here that Ycs4 has a similar affinity to DNA as the Ycg1-Brn1 complex. Using covalent lysine modification, we mapped the DNA binding site of Ycs4 to its C-terminal domain, adjacent to an HEAT-repeat cluster. Point mutations at the predicted protein–DNA interface reduced Ycs4 binding to DNA and the ATPase activity of the reconstituted condensin, whereas overproduction of the mutant Ycs4 was detrimental for yeast viability. Thus, the interaction of Ycs4 with DNA is essential for the yeast condensin function. Intriguingly, the DNA binding site of Ycs4 partially overlaps with its reported Smc4 binding site.²² We propose that Ycs4 binding to the DNA molecule precludes it from engaging Smc4 and thereby stimulates the ATPase cycle and resulting DNA condensation.

MATERIALS AND METHODS

Yeast Strains, Plasmids, and Growth Assays.

S. cerevisiae strains and plasmids used in this study are summarized in Table 1. A diploid S. cerevisiae strain InVSc1 (ThermoFisher Scientific) was used for coexpression of Smc2 and Smc4. Non-SMC subunits were expressed in a multiple-protease-deficient strain BCY123.^10,35 The strain BCY123 and 2 μm plasmid-based expression plasmids for SMC2 (Trp⁺), SMC4 (Ura⁺),^10,36 YCG1 (Trp⁺), YCS4 (Ura⁺), and BRN1 (Ade⁺) were gifts of Dr. Janet E. Lindsley. All expression plasmids used in this study are galactose inducible. A heterozygous BYcs4-ΔC strain ycs4(978–1175)::HIS3/YCS4 was constructed using a one-step gene integration method by replacing one of the YCS4 alleles of a diploid BY4743 strain with its C-terminally truncated version (amino acids 978–1175) linked to an HIS3 marker gene. BYcs4-x strains were constructed by reinserting the full-length Ycs4 gene harboring mutations denoted by the “-x” and linked to the URA3 gene to replace the truncated YCS4. Mutant Ycs4 proteins were purified from pYcs4his-x plasmids for DNA binding experiments. pYcs4-x plasmids without fusion tags were constructed for cell spotting experiments. Amino acid selection markers were used to maintain various plasmids. Cells were grown in a YNB (yeast nitrogen base) medium supplemented with appropriate amino acids and 2% glucose. The overproduction toxicity of mutant Ycs4 was observed by spotting cells containing pYcs4-x plasmids onto YNB-galactose plates containing a prototrophic marker uracil.

Table 1.

Plasmids and Strains Used in this Study

strains	genotype	reference or source
InVSc1	MATa/α his3D1/his3D1 leu2/leu2 trp1–289/trp1–289 ura3–52/ura3–52	ThermoFisher Scientific
BCY123	MATa pep4::HZS3 prbl::LEU2 bar1::HZSG lys2::GALlIlO-GAIA can1 ade2 trpl ura3 his3 leu23,112	10
BY4743	MATa/α his3Δ1/his3Δ1 leu2Δ0/leu2Δ0 met15Δ0/met15Δ0 ura3Δ0/ura3Δ0 lys2Δ0/lys2Δ0	GE Healthcare Dharmacon Inc.
BYcs4-ΔC	BY4743 MATa/α ycs4(978–1175)::HIS3/YCS4	this study
BYcs4-WT	BY4743 MATa/α YCS4::YCS4 URA3/YCS4	this study
BYcs4-x	BY4743 MATa/α YCS4::ycs4-x URA3/YCS4	this study
plasmids
pSJ50	GAL1 SMC4-intein/CBD tag URA3; purification of Smc4	10
pED1	GAL1 SMC2 TRP1; expression of Smc2	JE Lindsley
pSG170	GAL1 YCG1-His9 TRP1; purification of Ycg1	JE Lindsley
pJES22	GAL1 YCS4-His9 URA3; purification of Ysc4	JE Lindsley
pSG545	GAL1 BRN1 ADE2; expression of Brn1	JE Lindsley
pYcs4his-x	GAL1 YCS4-x- His9 URA3; purification of Ycs4-x	this study
pYcs4-x	GAL1 YCS4-x URA3; expression of Ycs4-x	this study

Random Spore Analysis.

The diploid BYcs4-x cells were grown at 30 °C up to the exponential phase in a pre-sporulation medium (10% glucose media: 0.8% yeast extract, 0.3% peptone, 10% glucose). Cells were spun down and washed with sterile water, sporulation was induced by resuspending the cells in a sporulation medium (1% potassium acetate), and incubation continued at 22 °C for 5–10 days. Sporulation efficiency was monitored by counting cells using light microscopy. Vegetative diploid cells that failed to sporulate were selectively lysed by incubating in the spheroblasting buffer (2 M sorbitol, 10 mM KPO₄ pH 7.2, Zymolyase T-100) for 40 min at 30 °C with gentle shaking. The ascus walls of sporulated cells were disrupted to release haploid spores using low-pulse sonication (Branson Sonifier Model 450, pulse: 3 × 10 s) with intermittent cooling. The haploid spores were plated on YPD, the colonies were then replica-plated onto YNB medium plates that either contained or lacked histidine or uracil, as appropriate, and the colonies grown under selective and unselective conditions were counted.

Protein Expression and Purification.

Smc2 and Smc4 proteins were co-expressed and purified as the Smc2/4 complex as previously described¹⁰ with minor modifications. Briefly, InVSc1 cells containing both SMC2 and SMC4 plasmids were grown in a YNB medium supplemented with amino acids and 2% glucose at 30 °C and collected by low-speed centrifugation at an A₆₀₀ of 1.2. Then, cells were equilibrated with a YP (2% yeast extract and 4% tryptone) medium, induced by the addition of 2% galactose, collected by centrifugation after 16 h, and washed in buffer A (25 mM HEPES of pH 8.1, 500 mM NaCl, 10% glycerol, 1 mM EDTA, 1 mM EGTA). The cells were first powdered in liquid nitrogen using mechanical pressure, and then buffer A was added to prepare the cell lysate. Smc2/4 proteins were purified using a chitin resin (New England Biolabs), Hitrap heparin column (Amersham Biosciences), and finally with Sepharose 300 gel filtration resin (Amersham Biosciences). Non-SMC subunits were purified as a complex from BCY123 cells containing Ycg1-His9, Ycs4, and Brn1 plasmids and washed with buffer A (25 mM HEPES of pH 7.7, 20 mM imidazole, 500 mM NaCl, 10% glycerol) using a nickel-charged His-Bind resin (Novagen), HiTrap heparin column (Amersham Biosciences), and Sepharose 300 gel filtration resin (Amersham Biosciences). All the purified proteins were finally dialyzed against 20 mM HEPES, 200 mM NaCl, 1 mM EDTA, 1 mM DTT, and 50% glycerol and stored at −20 °C. Single WT and mutant Ycs4 proteins were purified with a C-terminal His9 tag. Purified proteins from the His-bind resin were further purified using anion exchange (Q-Sepharose) columns. Expression levels of WT and mutant Ycs4 were checked by loading the same amount of total proteins on western blot using anti-His and anti-tubulin antibodies.

Protein Structure Predictions and Electrostatic Potential Calculation.

The secondary structure of Ycg1, Ycs4, and Brn1 was predicted with PSIPRED, and homology modeling was performed using the Phyre2 server (http://www.sbg.bio.ic.ac.uk/phyre2),³⁷ in which a protein sequence of interest is compared with crystal structure data of the respective proteins using one-to-one threading. The homology-modeled structure of S. cerevisiae Ycs4 was virtually identical (Figure S1) to the crystal structure of Chaetomium thermophilum Ycs4 (6QJ4). Ycs4 electrostatic potential was calculated using Swiss PDB Viewer (https://spdbv.vital-it.ch).³⁸

Protein Surface Mapping by Biotin Labeling and Mass Spectrometry.

Chemical modifications of lysines of the purified single Ycs4 and Ycs4-Brn1-Ycg1 (Ycs4-BY) complex were carried out using NHS-biotin in the presence and absence of the 900 bp dsDNA. Ycs4 alone or the Ycs4-BY complex was incubated with 900 bp dsDNA (DNA:protein = 50:1) in the binding buffer (20 mM HEPES, pH 7.7, 1 mM DTT, 20 mM NaCl, 5% glycerol) at 30 °C for 30 min, and then the surface-exposed lysines of the protein were modified by adding sulfo-N-hydroxysufosuccinimide biotin (NHS-biotin, Sigma/Pierce) for an additional 30 min at 30 °C. The reaction was quenched by the addition of 10 mM lysine followed by an extensive dialysis of the labeled protein/protein–DNA complex at 4 °C. All samples were digested with trypsin (Sigma, protein:trypsin = 1:10 w/w) in 50 mM Tris, pH 8, 150 mM NaCl, and 1 mM CaCl₂ overnight at room temperature. DNA bound complexes were further digested with DNase I (Invitrogen). The digested protein was loaded to HPLC/MS/MS for further analysis.

To design proper labeling conditions, purified single Ycs4 and non-SMC proteins were incubated with DNA before and after labeling with increasing amounts of sulfo-NHS biotin and the degree of labeling was detected using a gel shift assay following biotinylation quenching with an excess of lysine (Figure S2A). Biotin modification of lysine residues inhibited protein binding to DNA when the protein was incubated with sulfo-NHS biotin prior to the addition of the DNA. The complete inhibition of protein lysine residues occurred at the 2000:1 molar ratio of sulfo-NHS-biotin to the protein. However, if the binding site is occupied by DNA, then lysine residues remain unlabeled. When a protein was incubated with DNA prior to the labeling reaction, the excess sulfo-NHS biotin disrupted the protein–DNA complex at molar ratios above 2000:1. Therefore, we used the 2000:1 molar ratio for mapping the DNA footprint on Ycs4 and the non-SMC subunits.

Mass Spectrometry Analysis.

LC–MS/MS analysis was performed using a standard top 15 method on a Thermo Scientific Q-Exactive orbitrap mass spectrometer in conjunction with a Proxeon Easy-nLC II HPLC (Thermo Scientific) and Proxeon nanospray source. The digested peptides were reconstituted in 2% acetonitrile and 0.1% trifluoroacetic acid and loaded onto a 100 μm × 25 mm Magic C18 100 Å 5 U reversed phase trap where they were desalted online before being separated using a 75 μm × 150 mm Magic C18 200 Å 3 U reversed phase column. MS/MS data was collected using higher energy collision dissociation. Peptides were eluted using a gradient of 0.1% formic acid (A) and 100% acetonitrile (B). A 65 min gradient was performed with 5–35% B over 50 min, 35–80% B over 3 min, 80% B for 1 min, 80–5% B over 1 min, and finally held at 5% B for 10 min.

Database search.

Tandem mass spectra were extracted by Proteome Discoverer v 1.2. Charge state deconvolution and deisotoping were not performed. All MS/MS samples were analyzed using X! Tandem (The GPM, thegpm.org; version CYCLONE (2013.02.01.1)). X! Tandem was set up to search the Uniprot Saccharomyces cerevisiae (strain ATCC 204508/S288c) database (13,486 entries) plus an equal number of reverse decoy sequences and common laboratory contaminants (www.thegpm.org/crap) assuming the digestion enzyme trypsin. X! Tandem was searched with a fragment ion mass tolerance of 20 ppm and a parent ion tolerance of 20 ppm. ICAT-D of cysteine was specified in X! Tandem as a fixed modification. Glu to pyro-Glu modification of the N-terminus, ammonia loss of the N-terminus, Gln to pyro-Glu modification of the N-terminus, dioxidation of methionine and tryptophan, acetyl of the N-terminus, and biotin of lysine were specified in X! Tandem as variable modifications.

Criteria for protein identification.

Scaffold (version Scaffold_4.7.5, Proteome Software Inc., Portland, OR) was used to validate MS/MS-based peptide and protein identifications. Peptide identifications were accepted when they could be established at a greater than 61.0% probability by the Scaffold Local FDR algorithm. Protein identifications were accepted when they could be established at a greater than 16.0% probability to achieve a protein false discovery rate, FDR, of less than 5.0% and contained at least two identified peptides. These settings resulted in a 3.3% protein decoy FDR. Protein identification probabilities were assigned by the Protein Prophet algorithm.³⁹ Proteins that contained similar peptides and could not be differentiated based on MS/MS analysis alone were grouped to satisfy the principles of parsimony. Proteins sharing significant peptide evidence were grouped into clusters.

DNA Binding Assays.

For gel shift experiments, 10 ng of 200 bp dsDNA in reaction buffer was mixed with the specified amounts of the proteins and incubated for 30 min at 30 °C. The reaction mixtures were then placed on ice and analyzed by electrophoresis through a 0.7% agarose gel in 89 mM Tris borate, pH 8.3, for 80 min at 4 V/cm at 4 °C. To visualize DNA, the gels were stained with SYBR Gold (Molecular Probes, Inc., Eugene, OR).

ATPase Assays.

ATPase rates were measured using the EnzChek phosphate assay kit from Molecular Probes in 20 mM HEPES (pH 7.7), 40 mM NaCl, 2 mM MgCl₂, 1 mM DTT, 5% glycerol, and 1 mM MgATP buffer. Typically, 2 μg of Smc2/4 complex and pUC40 DNA (Smc2/4:DNA = 10:1) were added with an increasing ratio (Smc2/4:non-SMC) of WT or mutant non-SMC Ycs4-BY complex in a reaction mixture.

RESULTS

Ycs4, a Non-SMC Subunit of Budding Yeast Condensin Binds DNA.

We first purified various condensin subunits (Figure 1A) and analyzed their relative efficiencies in DNA binding (Figure 1B). Smc2/4, the non-SMC subunit (Ycs4-Brn1-Ycg1), Ycs4, or Ycg1 was incubated with 10 ng of 200 bp DNA and analyzed by gel electrophoresis. Smc2/4 and Ycs4-Brn1-Ycg1 (Ycs4-BY) were binding DNA with similar affinities, with the midpoint of transition observed at a 16 nM protein (Figure 1B). No DNA binding was observed for Ycg1 (Figure 1C). However, both Ycs4 and the Brn1-Ycg1 complex were binding to DNA with an apparent dissociation constant of 77 nM (Figure 1B,C). Thus, the DNA binding activity of the non-SMC subunit is shared by Ycs4 and the Brn1-Ycg1 complex.

Mapping the Protein–DNA Interface in Ycg1 and Brn1 Using Mass Spectrometry Footprinting.

We next mapped the interface of Ycs4 with DNA and the rest of the non-SMC subunit. To this end, we chemically modified solvent-accessible lysine residues of the protein using sulfo-NHS biotin (Figure 2A). The labeled lysines were then identified using mass spectrometry. From the comparison of the lysine accessibility maps on Ycs4 alone and its complex with Brn1-Ycg1 and DNA, the protein–protein and protein–DNA interaction maps were derived (Figure 2B).

Figure 2.

Mass spectrometric detection of biotin accessibility of Ycg1. (A) Schematic presentation of biotin labeling of non-SMC proteins with sulfo-NHS biotin. (B) LC–MS/MS identification of sulfo-NHS-biotin-labeled peptide. Biotinylated lysines were detected through their increased m/z value. The y-ion series includes fragments that contain the C-terminus of the intact peptide. Note the large mass increment between y3 and y4 corresponding to the biotinylated lysine. (C) Identification of protein–protein and protein–DNA interactions within the non-SMC subunits. (D) Ycg1 lysines that were labeled identically in Ycs4-Brn1-Ycs4 (Ycs4-BY) and Ycs4-BY-DNA complexes. Tick marks and labels indicate every other lysine in the protein. Lysines in forest green color signify the identically labeled lysines in the presence and absence of DNA. Lighter green-colored lysines represent identically unlabeled but detected lysines in both the complexes. (E) Differentially labeled Ycg1 lysines in Ycs4-BY and Ycs4-BY-DNA complexes. Ruby color denotes lysines labeled in the absence of DNA but unlabeled in the presence of DNA or with the biotinylation level reduced by more than 80%. Salmon color denotes lysines labeled in the absence of DNA but undetected in the presence of DNA. Warm pink lysines are the differentially detectable lysines that are inaccessible to biotinylation.

Biotin labeling was carried out at the 2000:1 molar ratio of sulfo-NHS-biotin to the protein. After such treatment, Ycs4 was unable to bind DNA (Figure S2), indicating that its interface with DNA was significantly modified. This treatment, however, was not too extensive to disrupt a pre-formed protein–DNA complex (Ycs4-DNA; Figure S2). Therefore, the chosen conditions were judged optimal for the detection of the protein–DNA interface.

The labeled proteins were digested with trypsin and, when appropriate, DNase I. The labeled lysines were then identified using mass spectrometry. Biotinylation of a single lysine produces an increase in an m/z of 226.4 Da, which allows a quantitative estimate of the extent of biotinylation (Figure 2C). All peptides that could be assigned to a single protein as well as their biotinylated forms were counted during the analysis. Lysines with a labeling efficiency greater than 5% (FDR < 1%) were considered biotinylated. Lysines were considered differentially labeled when their biotinylation levels under the conditions in question differed at least 5-fold. The measured biotinylation levels of all lysines in Ycs4, Brn1, and Ycg1 are summarized in Table S1.

We first analyzed interactions of Ycg1 and Brn1 with DNA. To generate the DNA footprint, the tripartite Ycs4-Brn1-Ycg1 complex was incubated with 50-fold excess DNA during biotin labeling. Under these conditions, we mapped identically labeled and differentially labeled lysines of Ycg1 (Figure 2D,E) and Brn1 (Figure 3A,B), respectively, in the presence and absence of DNA. In Ycg1, 15 lysines were undetected, 11 lysines were equally accessible to HNS-biotin in the presence and absence of DNA (Figure 2D), whereas 21 lysines of Ycg1 were differentially labeled (Figure 2E). Among 21 differentially labeled lysines, 18 lysines were protected in the presence of DNA, which are equally distributed throughout the Ycg1 structure. The other three differentially labeled lysines (Figure 2E, pink) were not the result of DNA footprint but the result of structural changes in the presence of DNA. Therefore, we did not consider these unlabeled lysines for protein–DNA surface mapping.

Figure 3.

Mass spectrometric detection of biotin accessibility of Ycg1 and Brn1. (A) Identically labeled (dark green) or unlabeled (light green) Brn1 lysines in Ycs4-BY and Ycs4-BY-DNA complexes. (B) Differentially labeled lysines of Brn1 in Ycs4-BY and Ycs4-BY-DNA complexes, including those protected by DNA binding (ruby red) and those whose accessibility was indirectly altered by DNA binding (salmon red, orange, and pink). (C) Mapping of the identically labeled and protected lysines of Ycg1 and Brn1 on the crystal structure of the Ycg1-Brn1-DNA complex. Ycg1 and Brn1 are shown in cyan and wheat colors, respectively. Blue color indicates lysine residues previously implicated in the DNA interaction.² The three patches of modified lysines described in the text are encircled.

The protected lysines of Ycg1 were spread throughout the proteins. In Figure 3C, we mapped them onto the crystal structure of the Ycg1-Brn1 complex.² The lysines that were labeled in the Ycs4-BY complex but remained protected in the presence of DNA were marked with ruby red and salmon red in the figure. The ruby red-colored lysines appeared unlabeled in the mass spectra of the Ycs4-BY-DNA complex, whereas the salmon red-colored lysines did not appear in the mass spectrometric measurement of the Ycs4-BY-DNA, presumably due to the poor ionization of the unbiotinylated form of respective peptides.⁴⁰ In principle, the differentially detectable peptides are not necessarily differentially labeled but are only likely to be. Therefore, the two types of lysines were marked in different colors in Figure 2 and thereafter. However, for the purpose of this analysis, this distinction is almost semantic since all detected and suspected differentially labeled lysines were later evaluated as potential DNA binding sites through genetic means.

Two of the protected residues (K69 and K798) were found in the close vicinity of bound DNA, as detected in the crystal structure.² Two more patches, I and II, stood out for DNA binding. Patch I includes several DNA protected lysines, K624, K650, K680, and K476, and can be indicative of another possible DNA binding site in the protein. Lysine patch II consisting of K39, K40, K44, K60, and K69, is close to the bound DNA, and therefore their protection in the biotin-labeling reaction can be the result of DNA-dependent structural changes. A few sparsely distributed protected lysines can be the result of conformational changes due to DNA binding.

In Brn1, 11 lysines were undetected, 11 lysines were equally labeled in the presence and absence of DNA (Figure 3A), and 28 lysines were differentially labeled (Figure 3B). Twenty-two out of these 28 lysines were protected in the presence of DNA. Three undetected lysines of Brn1 in the Ycs4-BY complex, K651, K652, and K688, were found labeled in the presence of DNA. This observation indicates a conformational transition of Ycg1 and Brn1 in the presence of DNA rather than site-specific DNA binding. Likewise, K573 was accessible to biotinylation in the presence but not absence of DNA. As the full crystal structure of Brn1 is not available, we mapped the DNA binding residues available in the short Brn1 peptide chain of the Ycg1-Brn1 crystal structure and marked the residues with a sphere presentation. One of such lysines resides close to the Ycg1 lysine patch II, which forms the DNA binding site along with Ycg1 patch II lysines. Another DNA binding lysine of Brn1 was also found very close to the patch I of Ycg1 and belongs to the third DNA binding site.

Ycs4 Interface with Brn1-Ycg1.

We next compared lysine accessibilities for Ycs4 alone and in complex with Brn1 and Ycg1. Out of 93 lysines of Ycs4, we detected 65, and 44 of them were biotinylated (Figure S3). In the Ycs4-BY complex, we detected 56 of the Ycs4 lysines, 39 of which were biotinylated. Figure 4A presents the residues that were either labeled or only detected (marked as identically labeled in Figure 4A) in both Ycs4 and the Ycs4-BY complex. These lysines were distributed along the entire length of the polypeptide and represent the surface of Ycs4 that was not affected by the formation of the complex. In contrast, the differentially labeled lysines were mostly found in four separate patches on the protein (Figure 4B). These lysines mark the area that was altered during complex assembly, either through a direct interaction or by an ensuing conformational change.

Figure 4.

Ycs4 interface with Brn1-Ycg1. (A) Identically labeled (green) or unlabeled (light green) lysines of Ycs4 alone and in the Ycs4-BY complex. (B) Differentially labeled Ycs4 lysines in the protein alone and in the complex with Brn1 and Ycg1 (Ycs4-BY). Light blue-colored lysines remained undetected in the Ycs4-BY complex but were labeled in Ycs4 alone. Lysines colored in dark blue are labeled in Ycs4 but remain unlabeled in the Ycs4-BY complex. Teal lysines were accessible in Ycs4 but not in the Ycs4-BY complex, while orange-colored lysines became exposed in the complex. (C) Surface mapping of lysines labeled identically (green) or differentially (blue, purple, teal, and orange) along with the previously¹ described lysines involved in the Ycs4-Brn1 interaction (yellow).

We generated an S. cerevisiae Ycs4 structure using homology modeling³⁷ of the C. thermophilum Ycs4 crystal structure²² (Figure S1) and mapped the identified lysines onto the homology-modeled structure. The Ycs4 protein consists of three globular domains with a tubular α-helical extension in the middle of the protein²² (Figure 4C). The equally accessible lysines (green colored) were spread throughout the surface of the protein as were some of the differentially accessible lysines. Lysines colored in blue and purple indicate protected lysines in the protein–protein complex, whereas teal-colored lysines designate differentially detectable lysines. The differentially accessible lysines formed four large patches, suggesting that these patches are likely to be the primary interface between Ycs4 and Brn1. Two out of the four lysine patches (patch II and III), in the middle and N-terminal domains of Ycs4, comprised several differentially labeled lysines. Two differentially labeled lysines K498 and K693 in patch III found in our surface mapping were implicated in the Ycs4-Brn1 interaction in a prior disuccinimidyl sulfoxide cross-linking study.¹ Patches II and IV were found close to the Brn1 binding site in the crystal structure²² of Brn1-Ycs4. Patch III located close to the α-helical protrusion present in the crystal structure of Ycs4 holds a large cluster of differentially labeled lysines and is indicative of another plausible Brn1 binding site. A separately located patch I, on the other hand, can be indicative of the structural changes of the protein due to Brn1 binding in the middle of the protein.

DNA Footprint on Ycs4.

The comparison of Ycs4 maps within the non-SMC subunit in the presence and absence of DNA revealed 27 lysines that were not affected by the presence of DNA (Figure 5A) and 33 that were (Figure 5B). Differential labeling of Ycs4 lysines in the Ycs4-BY and Ycs4-BY-DNA complexes demonstrates DNA foot printing on the Ycs4-BY complex. With the exception of K12, K498, K693, and K921, all differentially labeled lysines were more accessible to NHS-biotin in the absence of DNA, consistent with them being directly or indirectly protected by DNA. Most of the protected lysines were dispersed throughout Ycs4, and several of them were found in the areas implicated in Ycs4-Brn1 interactions (areas I, II, and III in Figures 5C and 4C). This pattern is more consistent with indirect effects of DNA on Ycs4 conformation than a site occlusion by DNA. In contrast to these, seven protected lysines of Ycs4 localized to a compact area at the C-terminal domain of the protein (Figure 5C). No lysines implicated in Ycs4-Brn1 binding were detected in their vicinity. Five of the seven lysines (K1048, K1104, K1108, K1109, and K1120) were close to each other on the same face of the domain, suggesting that this might be the primary area of DNA binding of Ycs4 (Figure 6A).

Figure 5.

Differentially labeled lysines in the Ycs4-DNA complex. (A) Identically labeled lysines of Ycs4 in Ycs4-BY and Ycs4-BY-DNA. (B) DNA-protected lysines of Ycs4 in Ycs4-BY (ruby red) and lysines with DNA-altered accessibility (salmon red and warm pink). (C) DNA-protected Ycs4 lysines mapped onto the structure of ScYcs4.

Figure 6.

Cell viability defects of Ycs4 mutants. (A) Location of mutated lysine and arginine residues. Residues found essential for viability or not are colored in ruby and red or orange, respectively. (B) Detection of Ycs4-induced defects in cell viability by plating cells in the presence and absence of the inducer galactose. (C) Immunoblot analysis of the Ycs4 abundance in cells that overexpress the WT and mutant Ycs4 variants. An anti-tubulin antibody was used for the loading control. (D) Colony formation by cells that overproduce various mutants Ycs4. Lysines K1048, K1050, K1104, K1108, K1109, and K1120 in various pairwise combinations showed viability defects. Asterisks mark strains with no detected colonies. (E) Spore viability of cells expressing mutant Ycs4.

To further corroborate these predictions, we examined the electrostatic map of Ycs4 using the Swiss PDB Viewer program. We found that Ycs4 is a negatively charged protein, with most of the charge located at its N-terminal and middle domains including the HEAT repeats. In contrast, the implicated patch on the C-terminal domain was positively charged (Figure S4).

DNA Binding Site of Ycs4 Resides at the C-Terminus and Is Essential for Its Function.

To validate the predicted DNA binding site, we carried out charge-inversion mutagenesis of the identified amino acids. Several mutant variants of Ycs4 (Figure 6A) were cloned under the control of a Gal1 promoter, purified, and tested for DNA binding. During these experiments, we noticed that some mutations in Ycs4 impair the growth of the host S. cerevisiae cells, sometimes dramatically (Figure 6B and Figure S5), although all mutants were expressed at similar levels (Figure 6C). We used this dominant negative phenotype as a proxy for the Ycs4 function to perform a more detailed mapping of the DNA binding site.

A simple pattern quickly emerged. Mutations in the C-terminal lysine patch were detrimental for cell viability, whereas the others were well tolerated. In particular, overproduction of the K1048E K1050E mutant protein reduced the colony formation by 4400-fold. A smaller, 9-fold effect was observed for a combination of mutations in lysines 1104, 1108, 1109, and 1120 (Figure 6D). This was not due to a lower expression of the functional variants of Ycs4 since all generated proteins were produced at similar levels following galactose induction (Figure 6C). Rather, the result suggests the importance of the identified amino acids for the activity of the protein.

The DNA binding activity of the purified mutant Ycs4 proteins mirrored their toxicity in vivo (Figure 7A,B). Mutations in lysines 1048 and 1050 completely abolished DNA binding, as did some combined mutations of lysines 1104, 1108, 1109, and 1120. In contrast, mutations in lysines and arginines K353, K432, K449, K677, K987, R1021, R1022, K1058, K1081, K1112, K1125, K1130, R1134, and K1137 (orange-colored lysines in Figure 6A) did not impact DNA binding. These results confirm our earlier conjecture that the primary DNA binding site is located at the C-terminal domain of Ycs4. Furthermore, we infer from this result that DNA binding at the predicted site is essential for the activity of Ycs4.

Figure 7.

DNA binding site of Ycs4 resides at the C-terminus. (A) Gel shift assay of WT and mutant Ycs4. (B) Apparent K_d values of Ycs4 mutants. Mutants 1104/08/09/20, 1048/50, 1108/09/20, 1104/08/09, and 1104/20 displayed a decline in DNA binding. (C) Titration of DNA for the stimulation of ATPase rate of the holoenzyme. Lines are spline curves connecting data points. (D) ATP hydrolysis of the holoenzyme condensin in the presence of DNA fit to a two-state binding model (shown with lines), which postulates that DNA binding to the non-SMC subunit activates the Smc2/4 ATPase rate, whereas the non-SMC subunit with mutant Ycs4 is impaired in DNA binding and therefore unable to stimulate the SMC ATPase rate. (E) Correlation between DNA binding, ATP hydrolysis, cell viability, and overproduction toxicity. (F) Sequence alignment of the DNA binding domain of Ycs4 from S. cerevisiae (Sc), Candida auris (Cu), C. albicans (Ca), C. thermophilum (Ct), S. pombe (Sp), Homo sapiens (Hs), and Mus musculus (Mm). Lysines implicated in DNA binding are labeled below the graph.

To verify the last conclusion, we performed a random spore analysis^41,42 to measure the impact of individual mutations on the ability of YSC4 to support viability in haploid cells, where the mutant allele is the only version present. We first constructed a diploid strain that was heterozygous at the YCS4 locus (MATa/α ycs4(978–1175)::HIS3 / YCS4; Table 1). Then, we systematically replaced the ycs4(978–1175)::HIS3 allele with WT or mutant versions of YCS4 linked to the URA3 marker. The generated diploid cells (MATa/α YCS4-x URA3/YCS4), where “-x” denotes the type of mutation in YCS4, were then sporulated to produce haploid spores bearing either the YCS4 or YCS4-x URA3 locus. The spores were individualized and plated on a non-selective medium, and the colonies were then replica-plated on selective (minus uracil) and non-selective media. Under selective conditions, only cells bearing the mutant allele of YCS4 are expected to produce colonies. Since half the spores inherit the mutant allele (and URA3), a 50% survival rate on the selective medium is expected if a given mutant is fully functional. We indeed observed 50% survival for the wild type YCS4 and several of the point mutants (Figure 6E). In contrast, in diploids that were heterozygous for the truncated version of YCS4 or the K1048E K1050E and K1104E K1108E K1109E K1120E mutants, less than 1% of colonies grew on the selective plates. Thus, the same YSC4 mutant alleles that had a dominant negative effect when overproduced were not capable of supporting growth when expressed at normal levels as the only source of Ysc4 protein in the cell.

The roles of the DNA binding lysines were unequal although the expression level for all the mutant proteins were about the same. K1048 and K1050 were always required for DNA binding as well as at least one of K1104, K1108, K1109, and K1120 (Figure 7). The lysines from the last group showed defects in DNA binding and cell viability only when used in pairs. For example, DNA binding was only modestly affected when a mutation in K1104 was combined with that in an unrelated K987 but substantially declined for mutants 1104/08/09, 1104/20, and 1104/08/09/20 (Figure 7B and Figure S5). Apparently, these two groups of amino acids represent two parts of the interface with DNA, both of which are necessary for the binding, whereas the lysines at the 1104/08/09/20 site are partially redundant with each other. Accordingly, we found only mild sporulation defects in the partial mutants (about 30% sporulation rate in mutants versus 50% in the fully viable parental strain), whereas the combined mutants 1104/08/09/20 and 1048/50 did not sporulate at all (Figure 6E).

DNA Binding by Ycs4 Stimulates ATPase Activity of the Holoenzyme.

The S. cerevisiae condensin holoenzyme is a slow ATPase, which is stimulated by DNA.¹ This stimulation does not happen for the core ATPase Smc2/4 subcomplex. This reveals a critical role of the Ycs4-BY/ DNA interaction in ATP hydrolysis by Smc2/4.

To test this idea, we co-expressed His-tagged mutant Ycs4 proteins together with unmodified Brn1 and Ycg1 proteins and purified the complexes using nickel-chelate chromatography. The purified non-SMC subunits were reconstituted with Smc2/4 to generate a holoenzyme. The reconstituted condensins were then supplemented with DNA and assayed for ATP hydrolysis. Optimal DNA stimulation required at least 200 nM DNA (Figure 7C). Therefore, we used 400 nM pUC40 DNA in all subsequent experiments. This DNA has previously been used by our DNA bridging experiments.⁹ Notably, the same amount of DNA was needed for stimulation of the endogenous and mutant proteins, indicating that other condensin subunits define its affinity for DNA.

ATPase rates were measured as a function of the amount of the non-SMC subunit used during reconstitution. In all cases, the rate showed a saturable increase with the amount of Ycs4-BY subcomplex (Figure 7D). The results were fit to a simple two-state model where the binding between the Smc2/4 and non-SMC subcomplexes is assumed to follow the Langmuir kinetics, and Smc2/4 and the holoenzyme hydrolyze ATP at their own rates (Figure 7D). We found that the stimulation can be reduced or abolished by mutations in the C-terminal domain of Ycs4.

This kinetic pattern suggests that the mutant proteins were partially defective in stimulating ATP hydrolysis but not the assembly of the holoenzyme. An overall correlation was found between cell viability, protein–DNA binding, and ATP hydrolysis for the tested mutants. Mutants defective in DNA binding were in general less efficient in ATPase stimulation and were more toxic to host cells upon overexpression (Figure 7E). Therefore, all three phenomena are likely to have the same mechanistic basis.

The identified DNA binding site was then aligned with the Ycs4 sequence from several organisms (Figure 7F). Lysines in positions 1048, 1050, and 1120 were highly conserved from fungi to primates. In addition, at least one basic residue was found in position 1108/09. In contrast, K1104 was conserved only among closely related species to be replaced by an acidic residue in fission yeast and higher eukaryotes. Notably, lysines 1104, 1108, and 1109 are located within 0.5 nm from each other (Figure S6). A single basic residue at this location might suffice for DNA binding. In contrast, lysines 1048, 1050, and 1120 are all at the interface with the ATP-free form of Smc4 (Figure S6) and are likely to experience a stronger evolutionary pressure.

DISCUSSION

We characterized here the DNA binding activity of Ycs4, a non-SMC subunit of yeast condensin. The high affinity of the non-SMC component for DNA is perhaps the greatest biochemical distinction of eukaryotic condensins from their bacterial counterparts. We mapped this DNA binding to two distinct sites, one on the Brn1-Ycg1 complex and the other on a patch of lysine residues in the C-terminal domain of Ycs4. The interaction of the Brn1-Ycg1 with DNA has been validated in a recent study.² The interaction between Ycs4 and DNA has not been known until now.

We mapped the DNA binding site using a DNA footprinting method, which is based on protection of the surface of the protein from chemical modification by small molecules. This is a zero-length technique, which reveals solvent-accessible residues on the protein and depends on the formation of a long-lived protein–DNA complex. Given these restrictions, the method tends to overlook transient interactions and reveals stable binding sites. This approach has been previously used to identify protein–DNA interfaces in HIV reverse transcriptase and replication protein A and protein Ku.^43–46

The DNA binding sites of the Brn1-Ycg1 complex appear as three separate patches located at the N-terminus, C-terminus, and in the middle part of the protein. Among these three distinctly separate patches, two patches of lysines at the N- and C-terminal domains, patches II and III, are close to a groove in the crystal structure of Ycg1-Brn1 that was implicated in DNA binding (Figure 4A of ref 2) and partially coincide with it (Figure 3C). Apparently, the contacts between DNA and specific amino acid residues on the protein are relatively short-lived. As a result, the longer DNA used in this study can shift position within the cavity formed by Brn1 and Ycg1 and maximize the number of protein–DNA contacts (PDB files 5OQP and 5OQO of ref 2). The third patch (patch I) found in this study is distinct from the other two and consists of K476, K624, K650, and K680 of Ycg1 and K519 of Brn1 (Figure 3C). This patch appears close to the base of the Ycg1 ring and might represent a binding site for DNA threaded through the ring. Alternatively, the altered lysine accessibility within this site could occur indirectly due to a large conformational transition in the protein. Such a conformational transition could conceivably happen as a result of DNA binding at sites II and III, which might be bringing the N- and C-terminal domains of Ycg1 together.

The DNA binding site of Ycs4 is located at the C-terminal globular domain of Ycs4 next to a cluster of HEAT repeats, which form a solenoid-like structure. Stable DNA binding required at least two contacts with DNA, one with lysines 1048/1050 and the other with any two out of lysines 1104, 1108, 1109, and 1120. These two groups of residues are located next to each other on two opposite tips of one of the HEAT repeats of the protein (Figure 6A and Figure S6). Mutations of other amino acids on both sides of this patch had no effect on DNA binding or activity of the protein (Figure S5). The proximity of the DNA binding site to an HEAT repeat cluster suggests a degree of flexibility in orientations of the captured DNA.

The interaction of Ycs4 with DNA proved to be essential for the function of the yeast condensin and for stimulation of ATP hydrolysis by the holoenzyme. It is interesting that our observed DNA binding site of Ycs4 overlaps with the Smc4 binding site of Ycs4.²² The association of Smc4 with Ycs4 prevents its binding to the Smc2 subunit and thereby the entire turnover of the holoenzyme. Our data indicate that the interaction between Smc4 and Ycs4, and indirectly with ATP, is modulated by DNA.

Figure 8 illustrates how this interaction could give rise to translocation of the protein along DNA⁴⁷ or loop extrusion.^13–15 The reaction cycle begins with DNA loading onto condensin (stage I). An opportunity for this arises when Brn1 dissociates from Smc2 following ATP hydrolysis and ADP release by the Smc2/4 head.²² The initial DNA binding is presumably accomplished by Ycs4 (this study) and Ycg1.^1,2 The DNA is then transferred from Ycs4 to Smc4 (stage II). In principle, Smc4 does not have to capture the same DNA stretch that has been previously bound to Ycs4, potentially resulting in DNA translocation (marked DNA translocation I in the figure). With DNA gone, Ycs4 becomes competent for binding to Smc4,²² whereas Brn1 restores its contact with the nucleotide-free form of Smc2³⁴ (stage II). The next step is prompted by ATP binding to Smc4, which induces a conformational switch in the protein that displaces Ycs4²² (stage III). Since no energy dissipation happens at this time, this step is reversible, unless the Smc4 binding site on Ycs4 is occluded by DNA (stage IV; this study). The diagram postulates that the DNA is transferred from Smc4, perhaps with a displacement, potentially leading to another translocation along DNA (DNA translocation II). However, DNA could conceivably be engaged by both Smc4 and Ycs4 at this step. Without interference from Ycs4, Smc4 can proceed to associate with Smc2³⁴ (stage V) followed by the binding of the second ATP and ATP hydrolysis, which displaces Brn1³⁴ (stage VI). Separation of the head domains of Smc2 and Smc4 completes the cycle, creating a conformation that is capable of both unloading DNA if it detaches from Ycs4 and Ycg1 or proceeding to the next round or reaction otherwise.

Figure 8.

Schematic diagram of condensin translocation along DNA. Details of the mechanism are outlined in the text.

The interaction between DNA and Ycs4 plays a dual role in this mechanism. First, by occluding the SMC4 binding site, DNA prevents Ycs4 from blocking the progression of the ATPase cycle. In this respect, Ycs4 serves as a sensor for low DNA availability that precludes futile cycling of the enzyme in the absence of its substrate. Second, Ycs4 offers an alternative landing site for DNA in proximity to its binding site on the Smc head. DNA transfer from Smc to Ycs4 and back could conceivably result in DNA translocation at one or both steps 1 and 3 in Figure 8.

Bacterial non-SMC subunits do not bind DNA on their own. However, they do stimulate the ATPase activity of the SMC core and thereby modulate SMC–DNA interactions.^19,48,49 Yet, the role of DNA in this stimulation differs for the proteins from MukBEF and SMC-ScpAB superfamilies. Specifically, the ScpAB subunit of the Bacillus subtilis condensin entraps DNA in a compartment produced by its binding to the SMC head.^21,49,50 In contrast, the binding of MukEF to MukB displaces the DNA.^48,51 In this light, the difference between the non-SMC subunits might reflect a key biochemical distinction of the two protein superfamilies.