Nucleosomes accelerate transcriptionfactor dissociationYi Luo1 Justin A North2 Sean D Rose2 and Michael G Poirier123

1Biophysics Graduate Program The Ohio State University Columbus OH 43210 USA 2Department of PhysicsThe Ohio State University Columbus OH 43210 USA and 3Department of Chemistry and BiochemistryThe Ohio State University Columbus OH 43210 USA

Received September 17 2013 Revised November 24 2013 Accepted November 26 2013

ABSTRACT

Transcription factors (TF) bind DNA-target siteswithin promoters to activate gene expression TFstarget their DNA-recognition sequences with highspecificity by binding with resident times of up tohours in vitro However in vivo TFs can exchangeon the order of seconds The factors that regulateTF dynamics in vivo and increase dissociationrates by orders of magnitude are not known Weinvestigated TF binding and dissociation dynamicsat their recognition sequence within duplex DNAsingle nucleosomes and short nucleosome arrayswith single molecule total internal reflection fluores-cence (smTIRF) microscopy We find that the rate ofTF dissociation from its site within either nucleo-somes or nucleosome arrays is increased by1000-fold relative to duplex DNA Our resultssuggest that TF binding within chromatin could beresponsible for the dramatic increase in TFexchange in vivo Furthermore these studies dem-onstrate that nucleosomes regulate DNAndashproteininteractions not only by preventing DNAndashproteinbinding but by dramatically increasing the dissoci-ation rate of protein complexes from their DNA-binding sites

INTRODUCTION

Initiation of eukaryotic gene expression involves tran-scription factor (TF) binding to DNA-target sites atgene promoters within chromatin (12) Chromatin iscomprised of a long array of nucleosomes each containing147 bp of DNA wrapped around a histone proteinoctamer core (34) TF-target sequences are often locatednear the DNA entryndashexit region of the nucleosome (5ndash8)so that the nucleosome structure sterically occludes TFoccupancy at its binding site However transient partialunwrapping fluctuations in addition to chromatin

remodeling (9) and nucleosome disassembly (10) providelimited access to TF-target sites (1112)TFs target specific genes by binding particular DNA

sequences with high affinities that are quantified by thedissociation constant KD= koffkon (Figure 1A) The KD

is the characteristic concentration for binding and can bedetermined experimentally by measuring S05 the concen-tration of TF at which 50 of the target DNA sequence isbound Under conditions where the DNA-target sequenceconcentration is significantly below the KD S05=KDThe KD is typically between nanomolar and picomolarfor DNA-binding TFs This is usually achieved byhaving relatively long resident times on and slow dissoci-ation rates from the target sequence In vitro TFs canhave residence times of about an hour at their DNA-recognition sequence (1314) which implies dissociationrates as low as 104s1 Surprisingly in vivo fluorescencerecovery after photobleaching measurements of TFdynamics find that TFs exchange on the scale of seconds(15ndash19) even though their resident times at their DNA-target site in vitro are much longer The mechanisms bywhich TF dissociation is dramatically accelerated remainunknown (2021)Previous studies have investigated the influence of

nucleosomes on site-specific DNA-binding proteins suchas TFs Restriction enzyme (RE) cleavage experimentshave demonstrated that target sites within nucleosomesare accessible to protein binding by partial DNAunwrapping from the histone octamer (HO) core (1222)Fluorescence resonance energy transfer (FRET) measure-ments of nucleosome unwrapping were then used to detectbinding of a model TF LexA to its target sequencebetween the 8th and 27th base pairs of the nucleosomeunder low ionic conditions (1123) They found thatLexA bound to its target sequence within a partiallyunwrapped nucleosome albeit the concentration to bind50 of the nucleosomes S05 was 100-fold higher relativeto duplex DNA More recently FRET measurementsdetermined that monovalent ionic conditions DNAsequence and histone post-translational modificationsinfluenced LexA occupancy (24ndash26) Previous studies

To whom correspondence should be addressed Tel +1 614 247 4493 Fax +1 614 292 7557 Email mpoiriermpsohio-stateedu

Nucleic Acids Research 2013 1ndash11doi101093nargkt1319

The Author(s) 2013 Published by Oxford University PressThis is an Open Access article distributed under the terms of the Creative Commons Attribution License (httpcreativecommonsorglicensesby30) whichpermits unrestricted reuse distribution and reproduction in any medium provided the original work is properly cited

Nucleic Acids Research Advance Access published December 17 2013 at O

HIO

STA

TE

UN

IVE

RSIT

Y L

IBR

AR

IES on D

ecember 18 2013

httpnaroxfordjournalsorgD

ownloaded from

have inferred the site exposure equilibrium constant fromthe ratio of the S05 for LexA binding to duplex DNArelative to binding to nucleosomes This assumed thatthe increase in S05 is due to sterically occluding TF andRE binding from their binding sites and that the dissoci-ation rate of the TF and RE are the same between duplexDNA and partially unwrapped nucleosomal DNAWe investigated the hypothesis that the nucleosome not

only suppresses TF binding but also influences TF dissoci-ation The dynamics of TF binding to and dissociationfrom its site within duplex DNA single nucleosomes andnucleosome arrays were quantified using single moleculetotal internal reflection fluorescence (smTIRF) microscopy(27) We detected binding of two TFs LexA and Gal4 totheir target sites within duplex DNA with Protein InducedFluorescence Enhancement (PIFE Figure 1A and D) (28)PIFE detects protein binding by attaching a fluorophoresuch as Cy3 adjacent to the TF-target sequence Upon TFbinding the fluorescence emission increases We quantifiedfluctuations in Cy3 fluorescence to determine the bindingand dissociation rates with duplex DNA We then separ-ately detected TF binding to its target site within mono-and dinucleosomes with FRET (Figure 1B C and D) (29)TF binding traps the nucleosome in a partially unwrappedstate This increases the distance between the Cy3 andCy5 fluorophores in the nucleosome and results in a

reduction in FRET efficiency By monitoring fluctuationsbetween high and low FRET efficiency we determined thebinding and dissociation rates of both LexA and Gal4within nucleosomes We find that nucleosomes not onlysuppress TF binding but enhance the rate of dissociationby up to three orders of magnitude These measurementsindicate that nucleosomes regulate TF occupancy notonly by blocking binding but by increasing the dissoci-ation rate Furthermore our results indicate that nucleo-somes can facilitate TF exchange and is a potentialmechanism for the measured difference in the rate of TFdissociation from chromatin in vivo and from duplex DNAin vitro

MATERIALS AND METHODS

Preparation of TFs

LexA protein was expressed and purified from pJWL288plasmid (generous gift from Dr Jonathan Widom) as pre-viously described (30) Briefly LexA was expressed inEscherichia coli BL21(DE3)pLysS cells (Invitrogen) byinducing with 02mM IPTG for 2 h Cells were harvestedby centrifugation and resuspended at 50ml per 1 l startingculture in Buffer A (50mM TrisndashHCl pH 80 200mMNaCl 1mM DTT 05mM EDTA 10 wv sucrose)The cells were lysed by lysozyme and cell debris were

Figure 1 DNA and nucleosome constructs Kinetic models of TF binding to (A) DNA (B) single nucleosomes and (C) nucleosome arrays(D) DNA constructs for single molecule TIRF measurements with Cy3 (green) Cy5 (magenta) biotin (black circle) 601 NPS (blue) and a Gal4-or LexA-target sequence (red) DNA molecules for making mononucleosome and dinucleosome array were labeled with Cy3 fluorophore as theFRET donor DNA molecules for PIFE experiments were labeled with Cy3 as the PIFE indicator and Cy5 to help locate the molecule during single-molecule experiments (E) Structure of the nucleosome (PDB 1KX5) that indicates the location of the TF-target sequence (red) the Cy3 fluorophorelocation (green) and the Cy5 fluorophore location (magenta)

2 Nucleic Acids Research 2013

at OH

IO ST

AT

E U

NIV

ER

SITY

LIB

RA

RIE

S on Decem

ber 18 2013httpnaroxfordjournalsorg

Dow

nloaded from

removed by centrifugation DNA was removed by precipi-tation with 35 polyethyleneimine and then LexA wasprecipitated twice by 40 ammonium sulfate LexA wasresuspended in Buffer B (20mM potassium phosphate pH70 05mM EDTA 10 vv glycerol) with 1mM DTTand 500mM NaCl and then dialyzed against the samebuffer overnight Dialyzed LexA was diluted 25-foldwith Buffer B plus 1mM DTT to give a final NaCl con-centration of 200mM before loading onto a cellulosephosphate column LexA was then eluted by a lineargradient of Buffer B from 200mM to 800mM NaClFractions containing LexA were then loaded onto ahydroxyapatite column and then eluted with a gradientof Buffer C (10 vv glycerol plus desired concentrationof potassium phosphate pH 70) from 50mM to 400mMpotassium phosphate Fractions containing high purityLexA were then dialyzed extensively against Buffer D(10mM PIPESndashNaOH pH 70 01mM EDTA 10 vvglycerol 200mM NaCl) and stored at ndash80C

The Gal4 expression vector was prepared by cloningthe Gal4 gene for residues 1ndash147 from Saccharomycescerevisiae genomic DNA (generous gift from Dr YvonneFondufe-Mittendorf) into pET3a at the NdeI and BamHIsites Gal4(1ndash147) was expressed in E coli Rosetta(DE3)pLysS cells (Millipore) by inducing with 1mMIPTG for 3 h Cells were harvested by centrifugation andresuspended at 50ml per 1 l starting culture in Buffer A[50mM Tris pH 80 200mM NaCl 1mM DTT 10 mMZnCl2 1mM phenylmethanesulfonyl fluoride (PMSF)]with 20 mgml leupeptin and 20 mgml pepstatin Thecells were lysed by sonication and cell debris wereremoved by centrifugation DNA was removed by precipi-tation with 35 polyethyleneimine and then Gal4(1ndash147)was precipitated by 40 ammonium sulfate (5)Gal4(1ndash147) was resuspended in Buffer A with 20 mgmlleupeptin and 20 mgml pepstatin and loaded onto aSephacryl 200HR gel filtration column (GE healthcare)(1) Fractions containing Gal4(1ndash147) were dialyzed intoBuffer B (20mM potassium phosphate pH 70 10glycerol 1mM DTT 10 mM ZnCl2 1mM PMSF) with200mM NaCl directly loaded onto a cellulose phosphatecolumn and then eluted by linear gradient of Buffer Bfrom 200mM NaCl to 800mM NaCl Fractions contain-ing Gal4(1ndash147) were dialyzed into Buffer C (25mM TrispH 75 1mM DTT 10 mM ZnCl2 1mM PMSF) with200mM NaCl and loaded directly onto a TSKgel SP5-PW (Tosoh biosciences) anion exchange column andeluted by a linear gradient of Buffer C with 200mMNaCl to 800mM NaCl Fractions containing high purityGal4(1ndash147) were dialyzed into Buffer D (10mM HEPESpH 75 200mM NaCl 10 glycerol 1mM DTT 10 mMZnCl2 1mM PMSF) and stored at ndash80C

Preparation of DNA molecules

DNA molecules for PIFE experiments and DNA mol-ecules for making mononucleosome and dinucleosomearray (Figure 1D) were prepared by PCR with Cy3Cy5biotin-labeled oligonucleotides from plasmid containingthe 601 nucleosome positioning sequence (NPS) with aconsensus LexA-binding site (TACTGTATGAGCATAC

AGTA) or Gal4-binding site (CCGGAGGACTGTCCTCCGG) at bases 8ndash27 (LexA) or bases 8ndash26 (Gal4)Oligonucleotides (Supplementary Table S1) were labeledwith Cy3 or Cy5 NHS ester (GE healthcare) at an aminogroup attached at the 50-end or to a modified internalthymine and then HPLC purified on a 218TP C18column (GraceVydac) Following PCR amplificationeach DNA molecule was purified by HPLC on a Gen-Pak Fax column (Waters)The dinucleosome DNA was synthesized by ligation of

two shorter PCR products PCR synthesized DNA mol-ecules containing a TspRI site and a 601 sequence or 601sequence with a LexA-binding site were digested by TspRIin NEB buffer 4 (New England Biolabs) Digestionproducts were purified by polyacrylamide gel electrophor-esis Purified two short DNA pieces with TspRI stickyends were mixed and ligated with T4 ligase (NewEngland Biolabs) in supplied buffer plus 2mM ATP andHPLC purified with a Gen-Pak Fax column to removeunligated fragments

Preparation of HOs

Xenopus laevis recombinant histones were expressed andpurified as previously described (31) Plasmids encodinghistones H2A(K119C) H2B H3 and H4 were generousgifts from Dr Karolin Luger (Colorado State University)and Dr Jonathan Widom Mutation H3(C110A) wasintroduced by site-directed mutagenesis (Stratagene)Each of the four histones were combined at equal molarratios refolded and purified as previously described (31)H2A(K119C)-containing HO was labeled with Cy5-maleamide (GE Healthcare) as previously described (25)

Preparation of nucleosomes

Nucleosomes were reconstituted from Cy3-labeled DNAand purified Cy5-labeled HO by salt double dialysis andpurified by sucrose gradient as previously described (25)Mononucleosome reconstitutions contained a molar ratioof 0851 of HO DNA dinucleosome reconstitutionscontained a mass ratio of 13 1 2 of HO templateDNA lambda DNA HO and DNA were mixed in 05TE pH 80 with 1mM Benzamidine hydrochloride (BZA)and 2M NaCl in a volume of 50 ml and then loaded into adialysis chamber The small dialysis chamber was thenplaced into a large dialysis tube containing 80ml of 05TE pH 80 with 1mM BZA and 2M NaCl and thendialyzed extensively against 05 TE pH 80 with 1mMBZA at 4C Dialyzed nucleosomes were loaded ontosucrose gradient and purified by centrifugation on anOptima L-90K Ultracentrifuge (Beckman Coulter) witha SW41 rotor Mononucleosomes were purified by agradient of sucrose from 5 to 30 wv anddinucleosome arrays were purified by a gradient ofsucrose from 5 to 35 wv Fractions containingnucleosomes were then collected and concentrated

Electromobility shift assays

DNA containing the LexA-target site was incubated at02 nM with 0ndash100 nM LexA in 05 TE pH 80 for2min at 20C and then resolved by Electrophoretic

Nucleic Acids Research 2013 3

at OH

IO ST

AT

E U

NIV

ER

SITY

LIB

RA

RIE

S on Decem

ber 18 2013httpnaroxfordjournalsorg

Dow

nloaded from

Mobility Shift Assay (EMSA) with a native 5 polyacryl-amide gel in 03 TBE DNA containing the Gal4-targetsite was incubated at 02 nM with 0ndash3 nM Gal4 in 10mMTrisndashHCl pH 80 130mM NaCl 01mgml BSA 10glycerol 0005 TWEEN20 1mM DTT and 5 ngmlpoly-dIdC (Sigma P4929) for 2min at 20C and thenresolved by EMSA with a native 5 polyacrylamide gelin 03 TTE

Ensemble PIFE measurements

TF binding to its target site within Cy3-labeled DNA wasdetermined by PIFE (28) where the Cy3 fluorescence in-creases upon protein binding Fluorescence spectra duringTF titrations were acquired with a Fluoromax4 (Horiba)using an excitation of 510 nm LexA titrations were donewith 02 nM fluorophore-labeled DNA in 10mM TrisndashHCl pH 80 130mM NaCl 10 Glycerol 0005TWEEN20 01mgml BSA and 1 BME Gal4 titrationswere done with 01 nM fluorophore-labeled DNA in10mM TrisndashHCl pH 80 130mM NaCl 10 Glycerol0005 TWEEN20 and 1 BME Fluorescence spec-trums were analyzed with Origin (OriginLab) to deter-mine the change in Cy3 fluorescence We carried outPIFE measurements of LexA and Gal4 titrations withdsDNA that did not contain their respective targetsequences (Supplementary Figure S1A and B) and didnot observe an increase in PIFE This implies that theincrease in PIFE is due to LexA and Gal4 binding totheir target sequences

Ensemble FRET measurements

TF binding and nucleosome site accessibility equilibriumconstants were measured with LexA binding to its targetsite buried within Cy3ndashCy5-labeled nucleosomes as previ-ously described (1125) TF binding to its target site trapsthe nucleosome into a partially unwrapped state (11) re-sulting in a partial reduction in FRET efficiency which weused to detect TF binding LexA titrations were done with5 nM nucleosomes in 50mM HEPES pH 75 130mMNaCl 10 Glycerol 0005 TWEEN20 01mgmlBSA 2mM Trolox (Sigma 238813) 00115 vvCyclooctatetraene (COT Sigma 138924) and 0012 vv3-Nitrobenzyl alcohol (NBA Sigma 146056) Gal4 titra-tions were done with 02 nM nucleosomes in 10mM TrisndashHCl pH 80 130mM NaCl 10 Glycerol and 0005TWEEN20 FRET efficiency measurements were deter-mine by the (ratio)A method (29) Fluorescence emissionspectra were measured as previously described (25) Wepreviously determined that nonspecific DNA binding ofLexA does not reduce the FRET efficiency and thatbinding of LexA to its target sequence within the nucleo-some does not induce dissociation of H2AndashH2Bheterodimers (25) We also carried out control titrationsof Gal4 with nucleosomes that do not contain the Gal4-binding site (Supplementary Figure S1C) As with LexAnon-specific binding of Gal4 did not reduce the FRETefficiency These control measurements imply that ourobserved reduction in FRET is due to TF binding totheir target sequence

Single molecule smTIRF microscope

The smTIRF microscope was built on an IX71-invertedmicroscope (Olympus) as previously described (27) 532and 638 nm diode lasers (Crystal Lasers) were used forCy3 and Cy5 excitation The excitation beams wereexpanded and then focused through a quartz prism(Melles Griot) at the surface of the quartz flow cellA 12NA water immersion objective (Olympus) wasused to collect the Cy3 and Cy5 fluorescence which wereseparately imaged onto a PhotonMax EMCCD camera(Princeton Instruments) with a Dualview (OpticalInsights) containing bandpass filters and a dichroic beamsplitter (Chroma Tech) Each image time series wasacquired with a PC using Winview (Roper Scientific)and analyzed as describe below

Flow cell preparation

Quartz microscope slides (G Finkenbeiner) werefunctionalized with poly-ethylene glycol (PEG LaysanBio MPEG-SVA-5000) and biotin-PEG (Laysan BioBiotin-PEG-SVA-5000) and assembled with glass cover-slips to make the flow cell Briefly quartz microscopeslides and glass coverslips were cleaned in toluene andethanol with sonication and then further cleaned inPiranha solution (3 1 mixture of concentrated sulfuricacid to 50 hydrogen peroxide) and washed in 1Msodium hydroxide The cleaned slides were treated with2 vv 3-aminopropyl-triethoxysilane (MP biomedicals215476680) in acetone and then with 10 wv PEG in01M potassium tetraborate pH81 (100 1 mass ratiomixture of mono-functional PEG to biotin-PEG)Functionalized quartz slides and coverslips wereassembled into microscope flow cells using parafilm withcut channels Before each experiment the flow cell istreated sequentially with 1mgml BSA 20 mgmlstreptavidin and biotin-labeled DNAnucleosomesamples to form surface tethers

Single molecule fluorescence measurements of TFbinding and dissociation

Biotinylated sample molecules (DNA mononucleosomeor dinucleosomes arrays) were allowed to incubate in theflow cell at room temperature for 5min and then washedout with imaging buffer containing the desired concentra-tion of TF protein The samples were first exposed to638 nm excitation to determine which molecules containedCy5 then the times of Cy3 and Cy5 emission was acquiredwhile exciting with 532 nm The imaging buffer for nucleo-somes contained 50mM HEPES pH 75 130mM NaCl10 vv glycerol 0005 vv TWEEN20 01mgml BSA2mM Trolox 00115 vv COT 0012 vv NBA whilethe imaging buffer for duplex DNA contained 10mMTrisndashHCl pH 80 130mM NaCl 10 vv glycerol0005 vv TWEEN20 01mgml BSA 1 vv BMEAn oxygen-scavenging system containing 16 wvglucose 450 mgml glucose oxidase (Sigma G2133) and22 mgml catalase (Sigma C3155) was also supplied withthe imaging buffer to suppress photobleaching of thefluorophores (2)

4 Nucleic Acids Research 2013

at OH

IO ST

AT

E U

NIV

ER

SITY

LIB

RA

RIE

S on Decem

ber 18 2013httpnaroxfordjournalsorg

Dow

nloaded from

Single-molecule time series were fit to two-state stepfunction by hidden Markov method using vbFRETMatlab program (13) provided by Dr Ruben Gonzalez(Columbia University) Idealized time series were furtheranalyzed using custom written Matlab programs to deter-mine the dwell-time distributions of the TF bound andunbound states The FRET efficiency from 70 of themononucleosomes and 70 of the dinucleosomesfluctuated in the presence of TF We included each ofthese time traces for determining the bound andunbound dwell-time distribution with nucleosomes TheCy3 fluorescence of 25 of DNA contained PIFE fluctu-ations in the presence of LexA We included the analysisof the times series of all of the molecules with PIFE fluc-tuations for determining the bounded and unboundeddwell-time distributions with DNA Weighted fits of thedwell-time distributions to single exponential-decay curveswere used to obtain the characteristic times and rate con-stants for the transition between bounded and unboundedstates

RESULTS

LexA has a resident time of 5 min at its target sequencewithin duplex DNA

We first investigated the binding dynamics of LexA a TFthat binds as a homodimer as do many high affinity TFs(32) LexA has been extensively studied with DNA (3334)and as a model TF that binds within nucleosomes(51123252635) We did single-molecule PIFE experi-ments to study the binding of LexA to its targetsequence within duplex DNA A Cy3 fluorophorewas attached adjacent to the LexA- target sequence Thefluorescence of Cy3 is enhanced upon LexA binding(Figure 2A) The enhancement is sensitive only toLexA binding to the specific recognition sequence(Supplementary Figure S1A) To confirm that the Cy3signal acquired was from immobilized DNA moleculesinstead of any possible background fluorescence in theflow cell a Cy5 fluorophore was also attached on the

same DNA further away from the LexA-target site(Figure 1D) so that only the signal from a Cy3 that co-localized with a Cy5 was selected for analysis From gt333molecules 2000-s time series of Cy3 fluorescence wereacquired at each LexA concentration (003 01 03 and1 nM) The Cy3 fluorescence enhancement upon LexAbinding allows us to determine the dwell time of eachbounded and unbounded DNA state The bounded andunbounded dwell-time histograms were determined foreach LexA concentration (Supplementary Figure S2)and fit to a single exponential decay Aexp(ndasht) where is the characteristic decay time The exponential decayfits of the bounded and unbounded dwell-time histogramsdetermine the characteristic bound time bound andunbound time unbound respectively The bound isconstant for increasing concentrations of LexA with avalue of (290plusmn20) s implying that koff=1bound=(34plusmn02) 103s1 The unbound decreases as A[LexA] where 1A= kon= (005plusmn001)s1 nM1

(Figure 2B) This yields a dissociation constantKD= koffkon= (007plusmn002) nM (Table 1) This KD isconsistent with ensemble measurements of S05 byEMSA and by ensemble PIFE measurements (Figure2C Supplementary Figure S3A) Furthermore wedetermined the fraction of time the DNA molecules werebound by LexA during the smTIRF measurements andplotted this versus LexA concentration These valuesagreed with the EMSA and ensemble PIFE measurementsof LexA binding to DNA (Figure 2C) This agreement ofensemble and smTIRF measurements with duplex DNAindicates that the surface tethering does not impact the TFbinding and dissociation dynamics

LexA has a resident time of 03 s at its target sequencewithin the nucleosome entryndashexit region

To determine the influence of the nucleosome on TFbinding and dissociation dynamics we investigated withsmTIRF LexA binding and dissociation at its target sitelocated between the 8-th and the 27-th bp of sucrosegradient purified nucleosomes (Supplementary Figure

Figure 2 Single molecule measurements of LexA binding and dissociation to DNA (A) Single molecule PIFE traces of LexA binding to its target sitein Cy3-labeled duplex DNA with 0 (top) 01 (middle) and 1 (bottom) nM LexA The histogram on the right shows the distribution of Cy3 fluorescencefor each trace (B) The unbound (magenta circles) and bound (blue squares) dwell times unbound with duplex DNA as a function of LexA concen-tration Each dwell time was determined from an exponential fit to the dwell-time histogram (Supplementary Figure S2) The LexA concentrationdependence of the unbound dwell times were fit to unbound=A[LexA] with A=(20plusmn6) s nM and the bound dwell times were fit to bound=con-stant= (290plusmn20) s respectively (C) The fraction of DNA bound by LexA as determined by EMSA (magenta triangles) ensemble PIFE measure-ments fit with a non-cooperative binding curve with a KD=(013plusmn006) nM (blue squares) and single molecule PIFE measurements (red circles)

Nucleic Acids Research 2013 5

at OH

IO ST

AT

E U

NIV

ER

SITY

LIB

RA

RIE

S on Decem

ber 18 2013httpnaroxfordjournalsorg

Dow

nloaded from

S3C) LexA binding to its target site traps the nucleosomein a partially unwrapped state causing a significant drop inFRET efficiency (11) The nucleosome is labeled on the50-end with Cy3 while H2A(K119C) is labeled with Cy5One of the two Cy5 fluorophores is within the Forsterradius of the Cy3 molecule for fully wrapped nucleosomeresulting in a high-FRET efficiency Upon LexA bindingwhich traps the nucleosomes in a partially unwrappedstate the distance between the Cy3 molecule and thenearby Cy5 molecule will increase and result in a reduc-tion in FRET efficiency While there are two Cy5fluorophores in the nucleosome the FRET efficiency canstill be estimated by the ratioA method (11) Nucleosomesspontaneously partially unwrap and rewrap (12) wherethe rewrapping occurs on the milisecond time scale (23)This is much faster than our 50-ms time resolution so weobserve a constant FRET efficiency of about 08 in theabsence of LexA (Figure 3A)Upon the addition of LexA we observe transient reduc-

tions in FRET efficiency to 02 (Figure 3A) the sameFRET efficiency observed in ensemble measurements atsaturating LexA concentrations (Supplementary FigureS4) The 200 s time series were acquired for gt183 mol-ecules at each LexA concentration (15 5 15 and 50 mM

Figure 3A) The dwell-time histograms of the nucleosomebound and unbound states were determined at each LexAconcentration and fit to single exponential decays todetermine the characteristic dwell times (SupplementaryFigure S5AndashH) The characteristic dwell time of theunbound state fit to A[LexA] with an effective bindingrate of kon=A1= (9plusmn2) 105s1 nM1 The dwelltime of the bound state was independent of LexA with abound= (031plusmn005) s implying an effective dissociationrate of koff=1bound= (33plusmn06)s1 (Table 1)

During the smTIRF experiment we confirmed theintegrity of the nucleosome tethered on the microscopeslide surface by examining the co-localization of FRETdonor Cy3 and acceptor Cy5 Complete nucleosomecomplex showed Cy5 signal when excited with 638-nmlaser and either a high FRET if in the LexA unboundstate or low FRET signal if in the bound state whenexcited with 532 nm For each LexA concentration 70of the complete nucleosome molecules showed fluctu-ations in the FRET efficiency between the high- andlow-FRET states All fluctuating molecules were used inthe dwell time histograms

Comparison of the rates of LexA binding to and dis-sociating from duplex DNA and nucleosomes implies

Figure 3 Single molecule measurements of LexA binding and dissociation to single nucleosomes and dinucleosome arrays (A) Single moleculeFRET traces of LexA-trapping nucleosomes in partially unwrapped states with 0 (top) 5 (middle) and 50 (bottom) mM LexA The histogram showsthe distribution of the FRET for each trace (B) The unbound (magenta circles) and bound (blue squares) dwell times with mononucleosomes andthe unbound (green triangles) and bound (red inverted triangles) dwell times with dinucleosome arrays as a function of LexA concentration Eachdwell time was determined from an exponential fit to the dwell time histogram (Supplementary Figure S5 and S6) The LexA concentrationdependence of the unbound dwell times were fit to unbound=A[LexA] with AmonoNuc= (11plusmn03) 10ndash4 s nM and AdiNuc= (11plusmn03) 10ndash4 snM and the bound dwell times were fit to bound=constant with bound(monoNuc)= (031plusmn005) s and bound(diNuc)= (029plusmn005) s (C) The relativechange in energy transfer efficiency versus the LexA concentration determined by both ensemble and single molecule measurements The ensemblemeasurements relied on analysis of fluorescence spectra (Supplementary Figure S4) by the (ratio)A method with mononucleosomes (red squares) anddinucleosome (blue squares) The fraction of time in the low- and high-FRET states were determined from single-molecule FRET time series for bothmononucleosomes (red circles) and dinucleosomes (blue circles)

Table 1 Binding and dissociation time constant and rate constant obtained from real-time single-molecule experiments

bound unbound koff kon KD

(s) (s nM) (s1) (s1 nM1) (nM)

LexAndashDNA 290plusmn20 20plusmn6 (34plusmn02) 10ndash3 005plusmn001 007plusmn002LexAndashmonoNuc 031plusmn005 (11plusmn03) 10ndash4 33plusmn06 (9plusmn2) 10ndash5 (4plusmn2) 104

LexAndashdiNuc 029plusmn005 (11plusmn03) 10ndash4 35plusmn03 (9plusmn2) 10ndash5 (4plusmn1) 104

Gal4ndashDNA 2000 ND (5 10ndash4) ND NDGal4ndashmonoNuc 50plusmn2 25plusmn01 0020plusmn0001 040plusmn002 0051plusmn0005

ND indicates not determined

6 Nucleic Acids Research 2013

at OH

IO ST

AT

E U

NIV

ER

SITY

LIB

RA

RIE

S on Decem

ber 18 2013httpnaroxfordjournalsorg

Dow

nloaded from

that the nucleosome reduces the binding rate by 500-foldand increases the dissociation rate by 1000-fold (Table 1)We determined the relative change in FRET from the singlemolecule and ensemble measurements (Figure 3C) andfind they depend similarly on the LexA concentrationindicating that the surface tethering does not significantlyalter the TF binding and dissociation dynamics Weobserve an increase of S05 by 105 when LexA binds tothe nucleosomes This is consistent with our previous meas-urements of LexA binding to nucleosome with 130mMNaCl (24) and is similar to RE measurements with nucleo-somes containing a variant of the 601 sequence (22)Additional reports of TF binding within nucleosomesobserve a smaller increase of S05 (11) However thesemeasurements were done in low-ionic conditions of1mM We have previously reported that these changesin ionic conditions dramatically impact the TF concentra-tion required to bind within nucleosomes (24)

We chose to label H2A because H2AndashH2B hetero-dimers dissociate before the H3ndashH4 tetramer (36)Therefore the fact that we observe significant energytransfer for each molecule implies that we are not detect-ing LexA binding to tetrasomes (DNA molecule onlybound to the H3ndashH4 tetramer) In addition wecompared the Cy5-labeling efficiency of 088 that wasmeasured by absorption spectrometry to the labeling effi-ciency predicted by the measured fraction of nucleosomesin the single-molecule FRET experiments with two Cy5fluorophores relative to one Cy5 fluorophore Thisallows for an estimate of the fraction of nucleosomesrelative to hexisomes (nucleosomes that are missing oneH2AndashH2B dimer) We determined for 763 molecules theemission intensity of Cy5 during direct excitation by638 nm ICy5_direct and the Cy5 emission intensity fromthe high-FRET-efficiency state ICy5_FRET during excita-tion by 532 nm The histogram of ICy5_directICy5_FRET hastwo distinct peaks since each the molecule can containeither one or two Cy5 fluorophores (SupplementaryFigure S5I) ICy5_direct was rescaled by ICy5_FRET to helpremove the variation in emission intensity that is due tothe spatial excitation variation of our smTIRF micro-scope The histogram was fit to a sum of two Gaussiandistributions For nucleosome with full HO the ratio ofthe areas under the Gaussian distributions are related tothe labeling efficiency by Atwo_Cy5Aone_Cy5=LP

22LP (1 ndashLP) where LP is the labeling efficiency and Atwo_Cy5 andAone_Cy5 are the areas under the Gaussian distributions fortwo and one Cy5 molecules respectively We measuredthe ratio of the areas to be 14 which implies a predictedlabeling efficiency LP of 074 The similarity between themeasured and predicted Cy5-labeling efficiencies indicatesthat most of the molecules measured are nucleosomes witha full HO

LexA binding and dissociation kinetics withindinucleosome arrays and mononucleosomes are similar

In vivo nucleosomes are imbedded into long chromatinmolecules Therefore we carried out similar LexA-binding experiments with dinucleosome arrays to deter-mine if the presence of an adjacent nucleosome influences

TF binding andor dissociation (Figure 3A and B andSupplementary Figure S6) As with the previous measure-ments we used ionic conditions of 130mM NaCl and nodivalent ions to mimic open euchromatin We find that theLexA-binding and -dissociation rates to dinucleosomesare identical to mononucleosomes (Table 1) In additionwe determined the relative change in FRET from thesingle molecule and ensemble measurements withdinucleosome arrays (Figure 3C) and find they dependsimilarly on the LexA concentration indicating that thesurface tethering does not significantly alter the TFbinding and dissociation dynamics These results suggestthat within open euchromatin a neighboring nucleosomedoes not impact TF binding and dissociation dynamics

Gal4 has a resident time much greater than 30 min at itsconsensus sequence within duplex DNA

We were concerned that the dramatic impact of thenucleosome on TF dissociation was unique to LexAFurthermore LexA is a prokaryotic TF and does notinteract with nucleosomes in vivo Therefore weinvestigated the influence of nucleosomes on the bindingand dissociation of the eukaryotic TF Gal4 Gal4 is amodel eukaryotic TF that recognizes its 19-bp consensusbinding site with a S05 10 pM (37) binds DNA as ahomodimer (3738) and upon binding activates the Gal110 genes in S cerevesiea (3940) As with previous studieswe used the first 147 amino acids of Gal4 which includesthe DNA recognition and dimerization domains (37)While the Gal4 dissociation rate has not been reportedother TFs such as Glucocorticoid Receptor and NF-kBwhich bind with picomolar dissociation constants dissoci-ate from duplex DNA on the hour time scale (1314)We used ensemble PIFE and EMSA to detect Gal4

binding to its recognition sequence within duplex DNA(Figure 4A Supplementary Figure S3B) The EMSAmeasurements which were done with 200 pM DNAdetermined a S05 of (242plusmn10) pM while the PIFE meas-urement which were done with 100 pM DNA determineda S05 of (42plusmn6) pM Both methods determined a S05

that was similar to the concentration of the DNA usedin the measurement indicating that the dissociationconstant KD is significantly less than 100 pM which isconsistent with previous measurements (37) and thatPIFE can be used to detect Gal4 binding to its consensussite Furthermore we confirmed PIFE did not increase inthe presence of Gal4 with Cy3-labeled DNA that did notcontain a Gal4-target sequence (Supplementary FigureS1B) This demonstrates that PIFE is only detectingGal4 binding to its target sequenceWe then carried out smTIRF measurements of Cy3ndash

Cy5-labeled duplex DNA (Figure 1D) with the Gal4-recog-nition sequenceWe acquired 600-s time series of Cy3 fluor-escence at Gal4 concentrations of 0 1 3 10 30 and 100 pMfor gt 300 molecules at each concentration We found thatonly a small fraction of DNA molecules displayed PIFEfluctuations So we acquired 2000-s time series with 30 and0 pMGal4 (Figure 4C) and found that the fractions of timeseries with one or more PIFE fluctuations were (7plusmn2)and (16plusmn06) respectively To confirm that Gal4 is

Nucleic Acids Research 2013 7

at OH

IO ST

AT

E U

NIV

ER

SITY

LIB

RA

RIE

S on Decem

ber 18 2013httpnaroxfordjournalsorg

Dow

nloaded from

binding to its DNA-target sequence during the smTIRFexperiments we plotted histograms of the Cy3 fluorescenceintensity from each Cy3ndashCy5-labeled DNA molecule ateach Gal4 concentration (Figure 4B) during the 600-stime series We observe a shift in the fluorescence distribu-tion at a concentration of 10 pM We fit each fluorescentdistribution as the sum of two Gaussian distributions thedistribution without Gal4 and the distribution with100 pM Gal4 We determine the relative area under thehigh distribution to the area under the full distributionand plotted the relative change in area of the high peakand find it fits to a binding curve with a S05 of (5plusmn2)pM which is similar to the previously report KD of10 pM (37) This S05 value is also less than our EMSAand ensemble PIFE measurement as expected since theseensemble measurements determined a S05 similar to theconcentration of the DNA These results indicate thatour smTIRF measurements are representative of Gal4binding in solution Our combined observations thatmost of the duplex DNA-target sites are bound by Gal4at 30 pM and that 90 of these bound molecules do notdissociate during the entire 2000-s acquisition implies thatGal4 remains bound to its site for much greater than 2000 sand that the dissociation rate is much less than 00005s

Gal4 has a resident time of 50 s at its target sequencewithin the nucleosome entryndashexit region

We investigated the influence of the nucleosome on Gal4binding with smTIRF as we did with LexA binding tonucleosomes We prepared sucrose gradient purified nu-cleosomes where the 20-bp LexA-recognition sequencewas replaced with the 19-bp Gal4 consensus sequenceChanges in FRET efficiency were used to detect bindingof Gal4 as it traps the nucleosome in a partiallyunwrapped state We recorded 400 s time series of Cy3and Cy5 fluorescence for gt529 single nucleosomes ateach Gal4 concentration (30 100 and 300 nM) We

observe fluctuations in FRET efficiency similar to theLexA measurements with nucleosomes (Fig 5A) Wequantified the dwell time of each bound and unboundnucleosome state determined the dwell-time histogramfor each Gal4 concentration and fit each to a single ex-ponential decay to determine the bound and unboundcharacteristic dwell times (Supplementary Figure S7)The characteristic dwell time of the unbound state fit toA[Gal4] with an effective binding rate of kon=A1=(040plusmn002)s1 nM1 (Figure 5B and Table 1) The char-acteristic dwell time of the bound state was independentof Gal4 with a bound= (50plusmn2) s implying an effectivedissociation rate of koff= (0020plusmn0001)s1 (Figure 5Band Table 1) This demonstrates that the Gal4-dissoci-ation rate is at least 100-fold greater from nucleosomesrelative to duplex DNA confirming the LexA resultsWe determined the relative change in FRET for thesingle molecule and ensemble measurements (Figure 5C)and find they depend similarly on Gal4 concentrationindicating that the surface tethering does not significantlyalter the TF binding and dissociation dynamics

DISCUSSION

We find that the rate of TF binding to a target site withinthe nucleosome entryndashexit region relative to duplex DNAis reduced by over two orders of magnitude while the rateof dissociation is enhanced by three orders of magnitudeThe site exposure equilibrium Keq will impact the TF-binding rate since LexA can only bind to the unwrappednucleosome state Under the assumption that the LexA-binding rate to its site within duplex DNA is the same asto a site unwrapped from the nucleosome the 500-foldreduction in the binding rate is equal to the reduction inprobability that the target site is exposed This regulationof TF binding by site exposure via transient unwrappingof nucleosomal DNA is well-established (12) Howeverthe finding that nucleosomes also regulate TF occupancy

Figure 4 Single molecule measurements of Gal4 binding and dissociation to DNA (A) The fraction of DNA bound by Gal4 with a cooperativebinding curve fit as determined by EMSA [magenta triangles KD=(240plusmn10) pM Hill coefficient=15] ensemble PIFE measurements [bluesquares KD=(42plusmn5) pM Hill coefficient=15] and the single molecule Cy3 fluorescence-intensity histograms from panel B [red circlesKD=(5plusmn1) pM Hill coefficient=2] (B) Fluorescence distributions from Cy3-labeled DNA molecules containing the Gal4-target site with 0 310 and 100 pM Gal4 (ordered from top to bottom respectively) The distributions were fit with the sum of two Gaussian distributions (black)the distribution without Gal4 (blue) and with 100 pM Gal4 (red) (C) Example Cy3 emission time traces of Cy3-labeled DNA containing theGal4-recognition site without Gal4 bound (top) a dissociation and binding event (middle) and bound with Gal4 (bottom)

8 Nucleic Acids Research 2013

at OH

IO ST

AT

E U

NIV

ER

SITY

LIB

RA

RIE

S on Decem

ber 18 2013httpnaroxfordjournalsorg

Dow

nloaded from

by dramatically increasing the rate of protein dissociationappears to be a new observation This provides an add-itional mechanism to regulate TF occupancy Given thatwe observe a dramatic increase in dissociation rate fortwo separate TFs this appears to a general feature ofTF binding within nucleosomes While regulating TF-binding rates will largely be influenced by nucleosomeproperties such as DNA sequence histone PTMs andlocation of the target sequence (242541) the regulationof TF dissociation rates will be influenced by both nucleo-some and TF properties

There are at least two non-exclusive models by whichthe nucleosome could enhance TF dissociation (i) Thenucleosome could directly influence the TF-dissociationrates by altering the structure of the recognition site thatis exposed for TF binding within the partially unwrappednucleosome relative to the duplex DNA structure suchthat the TF resident time is shortened This would bemanifested by a direct change in koff (Figure 1E) Inaddition (ii) partially unwrapped nucleosome statescould compete with partially bound TF states (Figure 6)For example the LexA dimer binds with nanomolaraffinities while the LexA monomer binds withmicromolar affinities (33) suggesting that the monomerhas a much larger dissociation rate If the inside portionof a TF dimer were to transiently dissociate the nucleo-some could partially rewrap blocking the rebinding of theinner portion of the TF With only the outer monomer ofthe TF dimer bound it is effectively bound as a monomerwith a significantly increased dissociation rate The partialrewrapping that blocks rebinding of the inner portion ofthe TF could result in a transient intermediate FRETstate which we do not observe However the lifetime ofthis state is likely to be on the scale for nucleosomerewrapping which occurs on the millisecond time scale(23) and is too fast for us to detect Future studies arerequired to determine the mechanism behind the increasein TF dissociation rate Interestingly the competitive

model is similar to how adjacent TFs bind within nucleo-somes cooperatively (4243)Our results also appear to be consistent with previous

reports of competitive binding between high affinityDNA-binding proteins A number of sequence independ-ent DNA-binding proteins have resident times of 1 h(4445) in the absence of soluble protein However uponthe addition of soluble protein (45) or mechanical strain(46) the resident times are reduced to minutes It wasproposed that DNA-binding proteins undergo lsquomicro dis-sociationrsquo where the protein partially or fully dissociatesfrom the DNA but remains within the screening length(1 nm) of the DNA and is much faster than the macro-scopic dissociation rate (45) These rapid brief excursionsof the bound protein from the DNA may allow solubleproteins to compete with rebinding thus increasing the

Figure 5 Single molecule measurements of Gal4 binding and dissociation to nucleosomes (A) Single FRET traces of Gal4 trapping nucleosomes inpartially unwrapped states with 0 (top) 30 (middle) and 300 (bottom) pM Gal4 The histogram shows the distribution of the FRET for each trace(B) The unbound (magenta circles) and bound (blue squares) dwell times with single nucleosomes as a function of Gal4 concentration Each dwelltime was determined from an exponential fit to the dwell-time histogram (Supplementary Figure S7) The Gal4 concentration dependence of theunbound and bound dwell times were fit to unbound=A[Gal4] with A=(25plusmn 01) s nM while the bound dwell times were fit to bound=con-stant= (50plusmn2) s (C) The relative change in energy transfer efficiency from mononucleosome versus the Gal4 concentration was determined by bothensemble and single molecule measurements The ensemble measurements relied on analysis of fluorescence spectra (Supplementary Figure S4) by the(ratio)A method (blue squares) The relative change in FRET was determined from single-moelcule FRET time series (red circles) by determining thefraction of time each time series is in the low- and high-FRET states

Figure 6 Model of competitive binding between nucleosome wrappingand TF binding TFs could partially dissociate where part of the TFtransiently releases from the DNA-target site and then rapidly fullyrebinds again However if the site is located within the nucleosomesand the part of the TF further into the nucleosome transiently releasesthan the nucleosome could rewrap preventing the TF from fully rebind-ing which could increase the rate at which the TF fully dissociates

Nucleic Acids Research 2013 9

at OH

IO ST

AT

E U

NIV

ER

SITY

LIB

RA

RIE

S on Decem

ber 18 2013httpnaroxfordjournalsorg

Dow

nloaded from

macroscopic dissociation rate Similarly it appears thatnucleosome rapid unwrapping and rewrapping could simi-larly function to compete with TFs that undergo microdissociation to increase the dissociation rateWhile eukaryotic TFs can bind their target sequences

with picomolar dissociation constants they can be atnanomolar concentrations or higher within the cell(4748) Under these conditions the placement of a nucleo-some relative to the TF-target sequence will influence boththe TF occupancy and dynamics Gene promoters withTF-target sites positioned outside of a nucleosome willbe occupied by the TF and remain bound for hours re-sulting in a constituently activated gene Under these con-ditions the dissociation rate is so slow that it will take onthe order of the cell-cycle time to reach equilibriumTherefore the rate at which the TF can bind will deter-mine the rate of gene activation implying that it is kinet-ically controlled Gene promoters with TF-target siteswithin the entryndashexit region of the nucleosome will alsobe occupied by the TF at nanomolar concentrations suchas Gal4 but will exchange on the second to minute timescale allowing for rapid regulation of the gene Here theequilibrium between the bound and the unbound state willdetermine the occupancy which can be tuned by TF con-centration In contrast gene promoters with TF-targetsites near the nucleosome dyad symmetry axis will rarelybe exposed and therefore will not likely be accessible forTF binding resulting in an inactive gene unless acted uponby chromatin modifying and remodeling complexesInterestingly bursts of TF localization and mRNA pro-duction has been reported to be on the minute time scaleby single cell measurements (49ndash51) suggesting that theinfluence of nucleosomes on TF dissociation could play aregulatory role of transcriptional burstsNumerous other proteins bind DNA within nucleo-

somes including DNA repair and DNA replicationcomplexes The influence of the nucleosome on thebinding and dissociation kinetics of other DNA-bindingcomplexes may play a regulatory role of their functionsFuture studies of the binding dynamics of other DNA-binding complexes to sites within chromatin will be im-portant for determining if this feature of the nucleosomeplays a regulatory role in other types of DNA processing

SUPPLEMENTARY DATA

Supplementary Data are available at NAR Online

ACKNOWLEDGEMENTS

The authors thank Ralf Bundschuh Cai Chen JenniferOttesen and members of the Poirier Lab for helpfuldiscussions We are grateful for the advice from RubenGonzalez on setting up the smTIRF system flow cellpreparation and data analysis

FUNDING

National Institutes of Heath (NIH) [GM083055 toMGP] Burroughs Wellcome Fund (career award in

the Basic Biomedical Sciences to MGP) AmericanHeart Association [0815460D predoctoral fellowship toJAN] Pelotonia Fellowship (to JAN) Funding foropen access charges NIH [GM083055]

Conflict of interest statement None declared

REFERENCES

1 SerikawaY ShirakawaM MatsuoH and KyogokuY (1990)Efficient expression and Zn(II)-dependent structure of the DNAbinding domain of the yeast GAL4 protein Protein Eng 3267ndash272

2 RasnikI McKinneySA and HaT (2006) Nonblinking andlong-lasting single-molecule fluorescence imaging Nat Methods 3891ndash893

3 LugerK MaderAW RichmondRK SargentDF andRichmondTJ (1997) Crystal structure of the nucleosome coreparticle at 28 A resolution Nature 389 251ndash260

4 RichmondTJ and DaveyCA (2003) The structure of DNA inthe nucleosome core Nature 423 145ndash150

5 LinYS CareyMF PtashneM and GreenMR (1988) GAL4derivatives function alone and synergistically with mammalianactivators in vitro Cell 54 659ndash664

6 AlbertI MavrichTN TomshoLP QiJ ZantonSJSchusterSC and PughBF (2007) Translational and rotationalsettings of H2AZ nucleosomes across the Saccharomycescerevisiae genome Nature 446 572ndash576

7 MacIsaacKD WangT GordonDB GiffordDKStormoGD and FraenkelE (2006) An improved map ofconserved regulatory sites for Saccharomyces cerevisiaeBMC Bioinform 7 113

8 JiangC and PughBF (2009) A compiled and systematicreference map of nucleosome positions across the Saccharomycescerevisiae genome Genome Biol 10 R109

9 ClapierCR and CairnsBR (2009) The biology of chromatinremodeling complexes Annu Rev Biochem 78 273ndash304

10 ParkYJ and LugerK (2008) Histone chaperones in nucleosomeeviction and histone exchange Curr Opin Struct Biol 18282ndash289

11 LiG and WidomJ (2004) Nucleosomes facilitate their owninvasion Nat Struct Mol Biol 11 763ndash769

12 PolachKJ and WidomJ (1995) Mechanism of protein access tospecific DNA sequences in chromatin a dynamic equilibriummodel for gene regulation J Mol Biol 254 130ndash149

13 BronsonJE FeiJ HofmanJM GonzalezRL Jr andWigginsCH (2009) Learning rates and states from biophysicaltime series a Bayesian approach to model selection andsingle-molecule FRET data Biophys J 97 3196ndash3205

14 ZabelU and BaeuerlePA (1990) Purified human I kappa B canrapidly dissociate the complex of the NF-kappa B transcriptionfactor with its cognate DNA Cell 61 255ndash265

15 McNallyJG MullerWG WalkerD WolfordR andHagerGL (2000) The glucocorticoid receptor rapid exchangewith regulatory sites in living cells Science 287 1262ndash1265

16 RayasamGV ElbiC WalkerDA WolfordR FletcherTMEdwardsDP and HagerGL (2005) Ligand-specific dynamics ofthe progesterone receptor in living cells and during chromatinremodeling in vitro Mol Cell Biol 25 2406ndash2418

17 BosisioD MarazziI AgrestiA ShimizuN BianchiME andNatoliG (2006) A hyper-dynamic equilibrium between promoter-bound and nucleoplasmic dimers controls NF-kappaB-dependentgene activity EMBO J 25 798ndash810

18 SharpZD ManciniMG HinojosCA DaiF BernoVSzafranAT SmithKP LeleTP IngberDE andManciniMA (2006) Estrogen-receptor-alpha exchange andchromatin dynamics are ligand- and domain-dependent J CellSci 119 4101ndash4116

19 KarpovaTS KimMJ SprietC NalleyK StasevichTJKherroucheZ HeliotL and McNallyJG (2008) Concurrentfast and slow cycling of a transcriptional activator at anendogenous promoter Science 319 466ndash469

10 Nucleic Acids Research 2013

at OH

IO ST

AT

E U

NIV

ER

SITY

LIB

RA

RIE

S on Decem

ber 18 2013httpnaroxfordjournalsorg

Dow

nloaded from

20 MisteliT (2001) Protein dynamics implications for nucleararchitecture and gene expression Science 291 843ndash847

21 HagerGL McNallyJG and MisteliT (2009) Transcriptiondynamics Mol Cell 35 741ndash753

22 AndersonJD and WidomJ (2000) Sequence and position-dependence of the equilibrium accessibility of nucleosomal DNAtarget sites J Mol Biol 296 979ndash987

23 LiG LevitusM BustamanteC and WidomJ (2005) Rapidspontaneous accessibility of nucleosomal DNA Nat Struct MolBiol 12 46ndash53

24 NorthJA ShimkoJC JavaidS MooneyAM ShoffnerMARoseSD BundschuhR FishelR OttesenJJ and PoirierMG(2012) Regulation of the nucleosome unwrapping rate controlsDNA accessibility Nucleic Acids Res 40 10215ndash10227

25 ShimkoJC NorthJA BrunsAN PoirierMG andOttesenJJ (2011) Preparation of fully synthetic histone H3reveals that acetyl-lysine 56 facilitates protein binding withinnucleosomes J Mol Biol 408 187ndash204

26 SimonM NorthJA ShimkoJC FortiesRAFerdinandMB ManoharM ZhangM FishelR OttesenJJand PoirierMG (2011) Histone fold modifications controlnucleosome unwrapping and disassembly Proc Natl Acad SciUSA 108 12711ndash12716

27 RoyR HohngS and HaT (2008) A practical guide tosingle-molecule FRET Nat Methods 5 507ndash516

28 HwangH KimH and MyongS (2011) Protein inducedfluorescence enhancement as a single molecule assay with shortdistance sensitivity Proc Natl Acad Sci USA 108 7414ndash7418

29 CleggRM (1992) Fluorescence resonance energy transfer andnucleic acids Methods Enzymol 211 353ndash388

30 LittleJW KimB RolandKL SmithMH LinLL andSlilatySN (1994) Cleavage of LexA repressor MethodsEnzymol 244 266ndash284

31 LugerK RechsteinerTJ and RichmondTJ (1999) Preparationof nucleosome core particle from recombinant histones MethodsEnzymol 304 3ndash19

32 WolbergerC (1996) Homeodomain interactions Curr OpinStruct Biol 6 62ndash68

33 KimB and LittleJW (1992) Dimerization of a specificDNA-binding protein on the DNA Science 255 203ndash206

34 LittleJW MountDW and Yanisch-PerronCR (1981)Purified lexA protein is a repressor of the recA and lexA genesProc Natl Acad Sci USA 78 4199ndash4203

35 PoirierMG OhE TimsHS and WidomJ (2009) Dynamicsand function of compact nucleosome arrays Nat Struct MolBiol 16 938ndash944

36 AndrewsAJ ChenX ZevinA StargellLA and LugerK(2010) The histone chaperone Nap1 promotes nucleosomeassembly by eliminating nonnucleosomal histone DNAinteractions Mol Cell 37 834ndash842

37 CareyM KakidaniH LeatherwoodJ MostashariF andPtashneM (1989) An amino-terminal fragment of GAL4 bindsDNA as a dimer J Mol Biol 209 423ndash432

38 MarmorsteinR CareyM PtashneM and HarrisonSC (1992)DNA recognition by GAL4 structure of a protein-DNAcomplex Nature 356 408ndash414

39 KewOM and DouglasHC (1976) Genetic co-regulation ofgalactose and melibiose utilization in Saccharomyces J Bacteriol125 33ndash41

40 HopperJE BroachJR and RoweLB (1978) Regulation of thegalactose pathway in Saccharomyces cerevisiae induction ofuridyl transferase mRNA and dependency on GAL4 genefunction Proc Natl Acad Sci USA 75 2878ndash2882

41 AndersonJD LowaryPT and WidomJ (2001) Effects ofhistone acetylation on the equilibrium accessibility ofnucleosomal DNA target sites J Mol Biol 307 977ndash985

42 AdamsCC and WorkmanJL (1995) Binding of disparatetranscriptional activators to nucleosomal DNA is inherentlycooperative Mol Cell Biol 15 1405ndash1421

43 PolachKJ and WidomJ (1996) A model for the cooperativebinding of eukaryotic regulatory proteins to nucleosomal targetsites J Mol Biol 258 800ndash812

44 SkokoD WongB JohnsonRC and MarkoJF (2004)Micromechanical analysis of the binding of DNA-bendingproteins HMGB1 NHP6A and HU reveals their ability to formhighly stable DNA-protein complexes Biochemistry 4313867ndash13874

45 GrahamJS JohnsonRC and MarkoJF (2011)Concentration-dependent exchange accelerates turnover ofproteins bound to double-stranded DNA Nucleic Acids Res 392249ndash2259

46 McCauleyMJ RueterEM RouzinaI MaherLJ 3rd andWilliamsMC (2013) Single-molecule kinetics reveal microscopicmechanism by which High-Mobility Group B proteins alter DNAflexibility Nucleic Acids Res 41 167ndash181

47 LaughonA and GestelandRF (1982) Isolation and preliminarycharacterization of the GAL4 gene a positive regulatorof transcription in yeast Proc Natl Acad Sci USA 796827ndash6831

48 GilbertW and Muller-HillB (1966) Isolation of the lacrepressor Proc Natl Acad Sci USA 56 1891ndash1898

49 LarsonDR (2011) What do expression dynamics tell us aboutthe mechanism of transcription Curr Opin Genet Dev 21591ndash599

50 CaiL DalalCK and ElowitzMB (2008) Frequency-modulatednuclear localization bursts coordinate gene regulation Nature455 485ndash490

51 ChubbJR TrcekT ShenoySM and SingerRH (2006)Transcriptional pulsing of a developmental gene Curr Biol 161018ndash1025

Nucleic Acids Research 2013 11

at OH

IO ST

AT

E U

NIV

ER

SITY

LIB

RA

RIE

S on Decem

ber 18 2013httpnaroxfordjournalsorg

Dow

nloaded from