609

ISSN 0026-8933, Molecular Biology, 2018, Vol. 52, No. 4, pp. 609–620. © Pleiades Publishing, Inc., 2018.Original Russian Text © S.L. Grokhovsky, 2018, published in Molekulyarnaya Biologiya, 2018, Vol. 52, No. 4, pp. 00000–00000.

Ultrasonic FootprintingS. L. Grokhovsky

Engelhardt Institute of Molecular Biology, Russian Academy of Sciences, 119991 Moscow, Russiae-mail: grok@eimb.ru

Received November 14, 2017; in final form, January 10, 2018

Abstract—Ligand binding influences the dynamics of the DNA helix in both the binding site and adjacentregions. This, in particular, is reflected in the changing pattern of cleavage of complexes under the action ofultrasound. The specificity of ultrasound-induced cleavage of the DNA sugar-phosphate backbone was stud-ied in actinomycin D (AMD) complexes with double-stranded DNA restriction fragments. After antibioticbinding, the cleavage intensity of phosphodiester bonds between bases was shown to decrease at the chromo-phore intercalation site and to increase in adjacent positions. The character of cleavage depended on thesequences f lanking the binding site and the presence of other AMD molecules bound in the close vicinity. Acomparison of ultrasonic and DNase I cleavage patterns of AMD–DNA complexes provided more detail onthe local conformation and dynamics of the DNA double helix in both binding site and adjacent regions. Theresults pave the way for developing a novel approach to studies of the nucleotide sequence dependence ofDNA conformational dynamics and new techniques to identify functional genome regions.

Keywords: actinomycin D, footrpinting, ultrasound, DNA cleavage, DNA conformationDOI: 10.1134/S0026893318040064

INTRODUCTIONGene expression is regulated at several levels. A

substantial contribution to the process is made bytranscription factors, which recognize specific DNAsequences. The recognition is mainly based on thedirect interactions between DNA reaction centers andprotein amino acid residues, including hydrogenbonds and electrostatic and Van der Waals interac-tions. Apart from this coding type, which is possible toterm the direct recognition or digital coding, analogcoding, or an indirect recognition, is possible withDNA [1, 2]. Many proteins do not utilize direct con-tacts with DNA to recognize their functional sites, butrely on the DNA spatial structure variations, such asthose in conformation of the sugar-phosphate back-bone, the widths of the DNA grooves, or helix bend-ing. Similar changes in the conformation of the sugar-phosphate backbone may arise in regions with differ-ent nucleotide sequences. Local conformationaldynamic properties of DNA depend not only on thenucleotide sequence, but also on the temperature,ionic strength, pH, and other environmental factors,as well as on the presence of proteins and other ligandsbound in the vicinity. With these variations in condi-tions, one regulatory protein may recognize differentnucleotide sequences in DNA.

In 1975, Wells and colleagues [3, 4] were the first toexperimentally demonstrate that structural parame-

ters of the DNA helix change over extended regionsupon ligand binding. Distortions induced by actino-mycin D (AMD) binding were found to involveextended regions of up to 10 bp, and such dynamicchanges in DNA structure were assumed to play animportant role in regulating gene activity. To date, var-ious methods, including X-ray crystallography,NMR, and Raman spectroscopy, have been used tostudy in detail the structures of many short double-stranded oligonucleotides of 15–25 bp and their com-plexes with proteins and low-molecular-weightligands [5–10]. However, it is still a promising and dif-ficult problem to study the conformation and dynam-ics as dependent on the nucleotide sequence, environ-mental conditions, and protein binding for DNAregions of more than 30 bp [1]. Developing new meth-ods for probing the local conformational dynamicproperties of extended DNA regions is therefore animportant task.

Several methods are currently available to study thevariations in f lexibility and widths of the grooves inextended DNA fragments; their gist is analyzing theproducts of fragment cleavage with various chemicalor physical agents. For example, fragments are possi-ble to cleave using pancreatic DNase I [11, 12],hydroxyl radicals [13, 14], laser radiation [15, 16], andX rays [17]. The same methods are used to identify thebinding sites for proteins and other ligands in DNA.

Galas and Schmitz [18] proposed DNase I foot-printing to detect the ligand binding sites in DNA inAbbreviations: AMD, actinomycin D.

STRUCTURAL FUNCTIONAL ANALYSISOF BIOPOLYMERS AND THEIR COMPLEXES

UDC 577.113.7, 577.323.35; 577.323.7

610

MOLECULAR BIOLOGY Vol. 52 No. 4 2018

GROKHOVSKY

1978. The binding of ligands to certain DNA sitesappreciated by the degree of protection from DNase Icleavage. More recent modifications of the methodutilize other DNases and chemical agents, such ashydroxyl radicals, to cleave the DNA sugar-phosphatebackbone [19]. Both natural nucleases and syntheticDNA-cleaving agents have their own specificities, i.e.,some recognize particular DNA sequences, while oth-ers are sensitive to conformational dynamic propertiesof double-stranded DNA [20]. For instance, pancre-atic DNase I recognizes DNA from the minor grooveside and poorly cleaves the homopolymeric regionswhere the minor groove is narrowed (poly-d(A) poly-d(T)) or widened (poly-d(G) poly-d(C)), while effi-ciently cleaving heterogeneous sequences and espe-cially sites where DNA easily bends towards the majorgroove [11, 21, 22]. Hydroxyl radicals better cleaveregions with a broader minor groove [23]. Sequencespecificities of DNA-cleaving agents hinder the local-ization of ligand binding sites. On the other hand, thesensitivity of a cleaving agent to the DNA structure ordynamics make it possible to study how ligand bindingdistorts the DNA double helix in the vicinity of a bind-ing site, provided that the site has exactly been deter-mined with other DNA-cleaving agents.

Ligands change the structural dynamic DNA prop-erties in both binding sites and adjacent regions [11,24, 25]. Bis-netropsin has previously been observed toalter the ultrasonic DNA cleavage pattern [26]. Theintensity of ultrasonic cleavage has been assumed toreflect the local variations in dynamics of the DNAsugar-phosphate backbone [27]. The objective of thiswork was to study the ultrasonic cleavage of AMD–DNA complexes.

EXPERIMENTAL

Isolation and further treatment of DNA fragmentshave been described in detail previously [26, 27].Briefly, DNA fragments were obtained by digestingthe modified plasmids pGEM7Z(f+) (Promega,United States) and pUC9 [28], which contained syn-thetic oligonucleotide inserts in the polylinkers, andphage λ DNA with a set of restriction enzymes. Fragmentswere 3'-end-labeled using [α-32P]dATP, [α-33P]dATP,[α-33P]dCTP, [α-33P]dGTP, or [α-33P]dTTP (Insti-tute of Reactor Materials, Zarechnyi, Sverdlovskregion); additional nonlabeled dNTPs; and the Kle-now fragment of Escherichia coli DNA polymerase I(Boehringer Mannheim, Germany). DNA fragmentswere separated in 5% nondenaturing PAGE , eluted,and precipitated.

To obtain a sample, 10 μL aqueous solution of aradiolabeled DNA fragment (approximately 104 Bq)were combined with 10 μL of 0.2 M NaOAc (рН 6.0)or an AMD solution (2.5–10 μM) in the same bufferin 0.2-mL thin-walled polypropylene microcentrifugetubes (Perkin Elmer, United States). The final con-

centration of the fragment was 5‒10 μg/mL, or ~10 μMbase pairs.

The tubes were placed in a round tef lon plate withholes so that their bottoms with test samples were 1 cmbelow the horn sonicator edge [28]. The plate wasplaced in a water bath with crushed ice. Sonication wascarried out using a UZDN-2T ultrasonic disperser(Ukraine) at 22000 Hz. After sonication, the samplewas combined with 90 μL of 0.15 M NaCl, 50 mMTris-HCl (pH 7.5), 10 mM EDTA, 10 μg/mL tRNA.The mixture was extracted with phenol, and DNA wasprecipitated with ethanol; washed with 70% ethanol;dried; dissolved in 1 μL of 95% formamide supple-mented with 15 mM EDTA (pH 8.0), 0.05%Bromphenol Blue, and 0.05% Xylene Cyanol FF;heated at 90°C for 1 min; quickly chilled to 0°C; andapplied onto denaturing polyacrylamide gel with a lengthof 40 cm and a thickness gradient of 0.15‒0.45 mm.Electrophoresis was carried out at 100 W (2.3 kV) at60‒70°C for 55 min. Prior to exposition, the gel wasfixed with 10% acetic acid and dried on a glass plate.

To quantitatively process the data on DNA frag-ment cleavage with ultrasound or DNase I, the inten-sities of all bands present in gel were measured usingSAFA software [29, 30]. After correcting the lanes ofan autoradiograph manually, the program automati-cally computed the intensities of all bands in all lanesand correlated them with the preset nucleotidesequence of the DNA fragment under study. Absoluteintensities of bands varied among experiments andwere consequently normalized. Changes in bandintensities in gel may involve a total lane so that differ-ent lanes of one gel have different total intensities.Band intensity may also decrease or increase along thelane length. Such effects may arise in the course ofmanipulations with DNA fragments and during sepa-ration in gel. The effects were eliminated using a mov-ing average method with several window sizes. A 31-ntwindow was chosen as optimal by varying the windowsize [31]. Smaller window led to a greater scattering ofresults, while a further increase in window size had vir-tually no effect on the proportion of intensities, butdecreased the number of values to be analyzed by thewindow size for each gel. Scans of autoradiographsand the results of their processing with the SAFA pro-gram are available at http://groh.ru/imb/amd.zip.

RESULTS AND DISCUSSION

Specificity of Actinomycin D Bindingwith Double-Stranded DNA

The natural antibiotic AMD contains a phenoxaz-one chromophore and two identical cyclic depsipep-tides (Fig. 1). AMD noncovalently binds to DNA andthus inhibits gene transcription. In addition, AMDforms complexes with quadruplex [32], hairpin [33, 34],and single-stranded [35, 36] DNAs. The specificity ofAMD binding to DNA has been studied in hundreds

MOLECULAR BIOLOGY Vol. 52 No. 4 2018

ULTRASONIC FOOTPRINTING 611

of works with various biophysical methods. AMD hasbeen examined as a model ligand that displays speci-ficity to certain DNA sequences. Structural, thermo-dynamic, and kinetic aspects of its complexation havebeen analyzed [37]. An AMD analog with an azidogroup in position 7 of the chromophore has been usedin several studies. The sequence specificity of the ana-log has been found to be virtually the same as that ofAMD [38–42].

Based on X-ray crystallography [43] and NMRspectroscopy [45], the planar phenoxazone chromo-phore intercalates between two base pairs, while thetwo pentapeptide lactone rings are positioned in theminor groove and span two base pairs on each side ofthe chromophore. The intercalation of the AMDchromophore requires that DNA be unwound and acavity of approximately 4 Å in width form between

neighbor base pairs. Figure 1 shows the angles thatoccur between base pair axes in the crystal structure ofthe AMD complex with an oligonucleotide containingthe 5'-d(GpC)-3' sequence (di- and tetranucleotidesare designated as NpN and NpNpNpN below). DNAgenerally preserves a B-like conformation in complexwith AMD [46].

Presumably, the main factor that determines spe-cific AMD binding to GpC sequences in double-stranded DNA is that two hydrogen bonds arisebetween 2-amino groups of guanines and carbonyloxygens of threonine residues [44, 47]. Stacking withbases plays an important role in binding site choice[48, 49]. The quinoid moiety of the phenoxazonechromophore binds with guanine more tightly thanwith adenine, while the benzenoid moiety does notdisplay such specificity [50]. The kinetics of specific

Fig. 1. Chemical structure of AMD and a diagram of its complex with DNA as viewed from the minor groove side. Nucleotidesare numbered relative to the AMD chromophore. Angles between base pairs are shown as observed in an AMD–oligonucleotidecomplex in crystal [43, 44]. Arrows indicate the sites where the sugar-phosphate backbone is cleaved upon sonication. Arrow sizesapproximately reflect the DNA cleavage intensities.

–3+1

+3

A

T

49°–2

+2

A Thr

Thr

D-Val D-Val

Pro

Pro

Sar

SarN · MeVal

N · MeValT

26° 45°–1

–1+1

G

C

O

C

G

+2

42°

–2

T

A

+3

–3

T

A

13°NH2

NO

O

H3C

O

N N

HN

OH3C

CH

HN O

O

H3C

O

O

N

CH3

O

NH2

NO

O

CH3

O

NN

NH

O CH3

HC

NHO

O

CH3

O

α β

CH3

18 17

1

3210

5

98

7

O

O

P

O

OHO

O

3' O

P

O

OHO

O

O

O

O OP

O

OHO

O

5'O

P

O

OHO

O

POHO

O

O

P

O

O OH

O

3'O

P

O

O OH

O

O

O

OOP

O

O OH

O

5'O

P

O

O OH

O

PO OH

612

MOLECULAR BIOLOGY Vol. 52 No. 4 2018

GROKHOVSKY

binding is determined by two processes, fast intercala-tion of the chromophore in many DNA sites and slowcomplex adjustment in sites of tight binding [1, 51].The AMD molecule experiences minor conformationalchanges when bound in tight binding sites, and thechanges are similar is sites with different sequences,while substantial changes arise in the DNA structure,differing among sites with different sequences [37].

The sites where AMD binding is the tightest aredouble-stranded DNA sites with self-complementaryGpC dinucleotides, the amino groups of which arehydrogen bonded with two carbonyl groups of AMDthreonines. Less tight AMD complexes are formedwith sequences containing GpG (CpC) [52] or GpT(ApC) [53]. These sites allow only one hydrogen bondto form, and two orientations relative to the helix axisare possible for the AMD molecule. Complexes withopposite orientations of the ligand are additionallypossible for symmetric GpC sequences f lanked bynonsymmetrical nucleotides.

To summarize the literature data, dinucleotidesequences are possible to arrange as follows in theorder of decreasing binding affinity, which correlateswith dissociation time of the AMD–DNA complex:GpC > GpG (CpC) > GpT (ApC) ApT, CpG,ApA (TpT), TpA [40, 53]. A similar order of tetranu-cleotides with central GpC is as follows: TpGpCpA >CpGpCpG > [TpGpGpG(pT)] ~ CpGpCpA (TpGp-CpG) > ApGpCpA (TpGpCpT) > ApGpCpG (CpG-pCpT) > ApGpCpT > TpGpCpC (GpGpCpA) >CpGpCpC (GpGpCpG) > ApGpCpC (GpGpCpT) >GpGpCpC [37, 39, 52, 54, 55].

Generally, pyrimidines are preferential at the 5' endof the GpC dinucleotides, and cytosine occurring atthe 3' end decreases the binding. More distantsequences also affect the AMD binding constant, andtheir effects may change the above binding orders. Forexample, AMD dissociates from GpC sites f lanked byalternating AT pairs at a lower rate than from GpCsites f lanked by homo-AT pairs. Within a region ofalternating AT pairs, the tetranucleotide TpGpCpAprovides a better binding site than ApGpCpT, andCpGpCpA is better than GpGpCpA [56]. DNase Ifootprinting has shown that AMD binding inducesconformational changes in binding site-flankingnucleotide sequences, which are more efficientlycleaved as a result; in particular, this has been observedfor extended AT clusters, which are relatively resistantto DNase I digestion in the absence of the antibiotic[57]. AMD binding is thought to widen the minorgroove in the adjacent DNA regions to a width thatfacilitates a better accommodation of the amino acidresidues that form the active center of the enzyme [21].

Figure 2 shows the cleavage profiles obtained for afree 314-bp DNA fragment and in the presence ofAMD with DNase I and with ultrasound treatment.The fragment regions that are protected from DNase Icleavage are indicated with bars. The left panel dis-

plays two regions of the fragment nucleotide sequenceand diagrams characterizing the relative cleavage ofphosphodiester bonds by ultrasound or DNase I in thepresence of 10 μM AMD as compared with controlcleavage reactions, which were carried out in theabsence of AMD.

To evaluate the changes in band intensities betweenthe lanes of samples sonicated or digested with DNaseI in the presence or absence of AMD, a ratio of differ-ences in their integral densities was calculated. Theordinate in the diagrams shown in Figs. 2 and 3 is theratio of a difference in band density between lanes ofAMD-containing and control samples to their sumafter normalization to the ratio of the total densities ofall bands in the same lanes. This data presentationmethod eliminates the differences in the effect of acleaving agent on different DNA sequences; such dif-ferences may exceed two orders of magnitude in thecase of DNase I.

Negative values in the diagrams suggest a less effi-cient cleavage in the presence of the ligand; positivevalues, a more efficient cleavage. As is seen from theDNase I cleavage profile, AMD-binding sites insequences with central GpC or GpG dinucleotides(shown with a larger font size) were protected fromcleavage, while the adjacent regions usually showed anincrease in cleavage efficiency as compared with thatin the free fragment. As mentioned above, variationsin DNase I cleavage of the sugar-phosphate backbonereflect the geometric heterogeneity of the DNA minorgroove and the rigidity to helix bending towards themajor groove [58, 59]. Ultrasonic cleavage pattersobtained in the presence of AMD also substantiallydiffered from control ones.

As is seen from Fig. 2, AMD greatly changed theultrasonic cleavage pattern of DNA at sites of AMDbinding. Several general regularities were observed forsequences containing a GpC dinucleotide upon ultra-sonic cleavage of their complexes with AMD. Thecleavage efficiency strongly decreased in the case ofthe phosphodiester bond between nucleotides –1 and+1 (nucleotides are numbered relative to the positionof the AMD chromophore, see Fig. 1), in the interca-lation site of the AMD chromophore, and moderatelydecreased in the case of the bond between nucleotides+1 and +2. A higher cleavage efficiency was detectedfor the bonds between nucleotides –2 and –1 and nucle-otides +2 and +3, extending to the adjacent one or twobase pairs. A decrease in cleavage efficiency was detect-able in more distant positions and involved 5‒7 bp insome cases, the effect being greater 5' of the site. Sim-ilar changes in cleavage efficiency were observed forthe sequences GpTpT (and, accordingly, ApApC) andGpG (СpС), but decreases in positions +1 and +2were weaker.

The total ultrasonic cleavage efficiency was fargreater in the presence of 50% glycerol, which greatlyincreases the solution viscosity, but relative intensities

MOLECULAR BIOLOGY Vol. 52 No. 4 2018

ULTRASONIC FOOTPRINTING 613

Fig. 2. Right panel: autoradiograph of gel. A free DNA fragment and its complex with AMD were digested with DNase I or son-icated at 22000 Hz; the products were electrophoretically resolved; gel was autoradiotraphed. Lanes: 1, the fragment before treat-ment; 2, chemical cleavage at purines; 3‒5, sonication of the fragment in в 0.1 M NaOAc (pH 6.0) for 16 min in the presence of0, 10, or 5 μM AMD, respectively; 6‒8, digestion of the fragment with DNase I in the presence of 0, 10, or 5 μM AMD, respec-tively; 9‒11, sonication in the presence of 50% glycerol and the same AMD concentrations. Putative AMD binding sites areshown with bars. Left panel: cleavage diagrams. Ordinate, ratio of a difference in band density between lanes 3 and 4 (at the top)or 6 and 7 (at the bottom), which showed the cleavage of AMD complexes and the control fragment. Positive values suggest amore efficient cleavage of the complex compared with the control; negative values, a less efficient cleavage.

0.2

–0.2

0.1

–0.1

00

5' 3'

Sonication 10 μM

0.2

–0.2

0.1

–0.1

00

5' 3'

DNase I 10 μM

1 2 3 4 5 6 7 8 9 10 11

0.2

–0.2

0.1

–0.1

0

05' 3'

Sonication 5 μM

0.2

–0.2

0.1

–0.1

005' 3'

DNase I 5 μM

Cle

ava

ge e

xte

nt

614

MOLECULAR BIOLOGY Vol. 52 No. 4 2018

GROKHOVSKY

Fig. 3. Ultrasonic and DNase I cleavage patterns of a free DNA fragment and its complex with AMD. Lanes: 1–4, sonication ofthe fragment in 0.1 M NaOAc (pH 6.0) for 16 min in the presence of 0, 10, 5, or 2.5 μM AMD, respectively; 5, 7, chemical cleav-age at purines; 6, the fragment before treatment; 8–11, digestion of the fragment with DNase I in the presence of 0, 10, 5, or2.5 μM AMD, respectively. The right panel shows partial nucleotide sequences of the fragment and band density ratio diagramsobtained for two regions of lanes 2, 3, and 4 (ultrasonic cleavage) compared with lane 1 (cleavage in the absence of AMD).

1 2 3 4 5 6 7 8 9 10 11

0.2

–0.2

0.1

–0.1

0

05' 3'

–0.2

0.1

–0.1

0

05' 3'

0.1

–0.1

0

05' 3'

20 µM

5 µM

10 µM

0.2

–0.2

0.1

–0.1

0

05' 3'

–0.2

0.1

0.2

–0.1

0

05' 3'

0.1

–0.1

0

05' 3'

20 µM

5 µM

10 µM

Cle

ava

ge e

xte

nt

MOLECULAR BIOLOGY Vol. 52 No. 4 2018

ULTRASONIC FOOTPRINTING 615

(the cleavage profile) remained the same (Fig. 2). Thisviscosity dependence of ultrasonic cleavage is a featureof mechanochemical reactions [60].

Cooperative Binding of Several Actinomycin D Molecules

The experimental results are difficult to interpretbecause of the reciprocal effects of AMD moleculesbound in close vicinity of each other. For example, twoAMD molecules can bind to the GpCpGpC sequenceat a higher saturation [61]. The binding is stronglyanticooperative because the peptide lactone rings ofneighbor molecules collide in the DNA minor groove.According to NMR findings, the DNA helix greatlybends towards the major groove in this case, and thecentral CpG sequence is fully unwound because theminor groove is widened to accommodate two bulkypentapeptide cycles [62].

The fragment sequence shown in the upper dia-gram (Fig. 3) harbors two GpCpGpC sites separatedby 13 bp. When AMD was used at 10 μM, the ultra-sonic cleavage pattern of the complex substantially dif-fered from the control one. A dramatically differentcleavage pattern was observed for the region at a half ashigh AMD concentration. Two AMD moleculesapparently bound to DNA at a higher AMD concen-tration, and their chromophores, which were only 2 bpapart, greatly distorted the structure of the doublehelix. A redistribution of AMD molecules occurred ata lower concentration, being reflected in the cleavagepattern. However, appreciable changes in DNase Icleavage pattern were not observed in this situation(Fig. 3). Different cleavage patterns were similarlyobtained with the two agents when other AMD bindingsites were examined at different AMD concentrations.

To quantitatively evaluate the effect of particularbinding sites on ultrasonic cleavage, it is necessary toconstruct the DNA fragments wherein a spacer betweenindividual AMD binding sites is more than 15 bp andlacks sequences suitable for AMD binding (e.g.,homo-(AT) sequences can be used as a spacer).

Character of Changes in Ultrasonic Cleavage of the DNA Sugar-Phosphate Backbone

at Actinomycin D Binding SitesThe extent to which chemical bonds of the sugar-

phosphate backbone are cleaved in a mechanochemi-cal reaction depends on the nucleotide sequence,which determines the conformational dynamic prop-erties of the backbone. The cleavage intensity dependson the fragment length, pH, ionic strength, and tem-perature [26]. A comparison of ultrasonic and DNase Icleavage patterns of AMD complexes with radiola-beled DNA fragments provides more detail on thelocal conformation and dynamics of the DNA doublehelix in ligand binding sites. A large set of dataobtained with different sequences was examined and

statistically analyzed to eliminate the effect that otherAMD molecules bound in the vicinity of the targetbinding site exert on its cleavage pattern and to observethe features that characterize the AMD binding-inducedchanges in ultrasonic cleavage pattern and are commonfor all di- and tetranucleotide sequences. The analysisincluded 90 gel lanes obtained via sonication of freeDNA fragments and their complexes with AMD at highlevels of saturation with the ligand. The total sequencelength was more than 13000 nucleotides. Gel scans, theresults of their analysis with the SAFA program, andmean differences established for all di- and tetranucle-otides are available at http://groh.ru/imb/amd.zip.

Figure 4 shows diagrams of median ratios of thedifferences in cleavage intensity of internucleotidebonds between free and AMD-bound fragments for allpossible dinucleotides. The 5'-3' dinucleotide positionon the abscissa axis corresponds to positions –1 to +1on the diagram. The ordinate shows the median ratiosof differences in cleavage intensity between cleavage inthe presence of AMD and control cleavage for posi-tions around the dinucleotide center. Negative valuessuggest a decrease in cleavage intensity in the presenceof AMD as compared with the control; positive values,an increase in intensity.

Lack of a ligand effect was observed for all dinucle-otides that contained only AT pairs (Fig. 4). The great-est decrease in cleavage was established for the GpCdinucleotide, which is known to be a site where AMDbinds with the highest affinity. In the case of CpG, adistinct decrease in internucleotide bond cleavage ispossible to explain by the facts that CpG dinucleotidesare more efficiently cleaved by ultrasound than allother dinucleotides and that ligand binding at anyneighbor site decreases its cleavage. The effect is welldetectable in the sequences shown in Figs. 2 and 3.

A dramatic difference in the character of cleavagewas observed for the complementary dinucleotidesGpG and CpC, suggesting an asymmetric AMD bind-ing to the respective sequence. The effect was weakerin the case of the GpT‒ApC pair. As for the otherdinucleotides, the character of cleavage was probablydetermined by the presence of high-affinity bindingsites for AMD in position –2 to –1 or +1 to +2,depending on the position of the GC base pair; thisrequires an analysis to be performed at the tetranucle-otide level.

Figure 5 shows diagrams of median ratios of thedifferences in cleavage intensity of internucleotidebonds between free and AMD-bound fragments for allpossible tetranucleotides with a central GpC dinucle-otide. The pseudosymmetric complementary tetranu-cleotides ApGpCpT and TpGpCpA showed similarcleavage patterns in positions –1 and +1 and differ-ences in positions more distant from the intercalationsite; the finding most likely indicates that one of thetwo possible orientations of the phenoxazone chromo-phore predominates in complexes with each of the

616

MOLECULAR BIOLOGY Vol. 52 No. 4 2018

GROKHOVSKY

sequences. Patterns of differences in cleavage betweennonsymmetrical complementary nucleotides, such asApGpCpA and TpGpCpT, substantially varied, indi-cating that ultrasonic cleavage at the ligand bindingsite may differ between the two DNA strands. In thecase of tetranucleotides containing GC pairs in vari-ous terminal positions, the character of cleavage wasdifficult to analyze because the tetranucleotides dif-fered in the likelihood of having a binding site foranother AMD molecule. This analysis should be per-formed at the hexanucleotide level, but the availableexperimental data are not ample enough for the pur-pose.

The extent of chemical bond cleavage in the DNAsugar-phosphate backbone via a mechanochemicalreaction has previously been found to depend on thenucleotide sequence, which determines the conforma-tional dynamic properties of the backbone. The char-acter of ultrasonic cleavage of double-stranded DNAfragments has been carried out at 22 and 44 kHz, anddouble-strand breaks have been found to occur moreoften in the sites that contain the CpG sequence. Thestrand is broken between cytosine and guanosine inthe sequence, leaving the phosphate group at the5' end of guanosine. Therefore, the bond directly adja-cent to the 3' carbon of the 2'-deoxyribose moiety is

Fig. 4. Diagrams of ratios of the differences in internucleotide bond cleavage intensity (Х) between free and AMD-bound frag-ments. The 5'-3' dinucleotide position corresponds to positions –1 to +1 on the abscissa axis. Ordinate, median ratios of differ-ences in cleavage intensity between cleavage in the presence of AMD and control cleavage for positions around the dinucleotidecenter. Negative values suggest a decrease in cleavage intensity in the presence of AMD as compared with the control; positivevalues, an increase in intensity.

Position relative to the AMD binding center

AA

TT

Cle

ava

ge e

xte

nt

0

0.05

−0.05

−0.10

−0.15

−0.20

AT

TA

0

0.05

−0.05

−0.10

−0.15

−0.20

AC

GT

0

0.05

−0.05

−0.10

−0.15

−0.20

AG

CT

0

0.05

−0.05

−0.10

−0.15

−0.20

CC

GG

0

0.05

−0.05

−0.10

−0.15

−0.20

CA

TG

0

0.05

−0.05

−0.10

−0.15

−0.20

−4 −3 −2 −1 +1 +2 +3

GA

GC

+4

0

0.05

−0.05

−0.10

−0.15

−0.20−4 −3 −2 −1 +1 +2 +3

CG

TG

+4

0

0.05

−0.05

−0.10

−0.15

−0.20

MOLECULAR BIOLOGY Vol. 52 No. 4 2018

ULTRASONIC FOOTPRINTING 617

broken. A comparison of the ultrasonic cleavage rateswith literature data on S ↔ N interconversion in theB-DNA sugar-phosphate backbone has shown thatthe rates correlate with the pseudorotation intensity ofthe deoxyribose moiety at the 5' end of the respectivedinucleotide [31]. A correlation has been observedbetween dinucleotide cleavage intensity and the rate ofS ↔ N interconversion (or pseudorotation) of thedeoxyribose moiety in the 5'-terminal nucleotide [63].Deoxyribose S ↔ N interconversion is associated witha reorientation of the C3'–O3' bond, which is brokenin the sugar-phosphate backbone upon hydrodynamicexposure. Molecular dynamics computations and spinrelaxation data from 13C-NMR spectroscopy indicate

that all C–H bonds of the deoxyribose moiety in cyto-sine are far more mobile than in the other bases inB-DNA. The finding indicates that the S ↔ N inter-conversion rate in cytosine is higher than in the furanoserings of the other nucleotides and is close to A-DNAconformations [64].

Predominant cleavage of cytosine compared withthe other bases upon hydrodynamic exposure is possi-ble to explain by the fact that a smaller angle occursbetween the target chemical bond and the helix axis inthe case of cytosine, and bond deformation energy isconsequently greater. In turn, this leads to a lowerenergy barrier to hydrolysis and a higher rate of themechanochemical reaction of cleaving the DNA

Fig. 5. Diagrams of ratios of the differences in ultrasonic cleavage intensity of internucleotide bonds in tetranucleotides betweenfree and AMD-bound fragments. The 5'-3' tetranucleotide position corresponds to positions –2 to +2 on the abscissa axis. Ordi-nate, median ratios of differences in cleavage intensity between cleavage in the presence of AMD and control cleavage for posi-tions around the tetranucleotide center. Negative values suggest a decrease in cleavage intensity in the presence of AMD as com-pared with the control; positive values, an increase in intensity.

Position relative to the AMD binding center

AGCT

TGCA

Cle

ava

ge e

xte

nt

0

0.10

0.20

−0.10

−0.20

−0.30

AGCA

TGCT

0

0.10

0.20

−0.10

−0.20

−0.30

CGCG

GGCC

0

0.10

0.20

−0.10

−0.20

−0.30

AGCC

GGCT

0

0.10

0.20

−0.10

−0.20

−0.30

CGCA

TGCG

0

0.10

0.20

−0.10

−0.20

−0.30

AGCG

CGCT

0

0.10

0.20

−0.10

−0.20

−0.30

CGCC

GGCG

0

0.10

0.20

−0.10

−0.20

−0.30

−4 −3 −2 −1 +1 +2 +3 +4

GGCA

TGCC

0

0.10

0.20

−0.10

−0.20

−0.30

−4 −3 −2 −1 +1 +2 +3 +4

618

MOLECULAR BIOLOGY Vol. 52 No. 4 2018

GROKHOVSKY

sugar-phosphate backbone. To check this hypothesis,ultrasonic cleavage was studied with DNA fragmentscontaining covalently closed GpG dinucleotides,which were crosslinked using the bifunctional reagentcis-diaquadiamine platinum(II) [27]. High-resolutionX-ray crystallography has shown that the 5' and 3'deoxyribose moieties of the platinum adduct in theGpG dinucleotide occur, respectively, in the C3'-endoand C2'-endo conformations, which are typical of A-and B-DNA, respectively [65]. When adduct-con-taining DNA fragments were sonicated, a substantialincrease in cleavage was observed for the 5'-terminal gua-nine of the GpG dinucleotide. The finding indicates thatphosphodiester bond cleavage occurred predominantlybetween the crosslinked bases; the 5' phosphate groupwas left on the second guanine residue [27]. Predomi-nant cleavage at crosslinks supports the above hypoth-esis that the probability for a break to arise in the DNAstrand on exposure to hydrodynamic forces is associ-ated with the conformation of the deoxyribose moiety,which occurs in the C3'-endo conformation 5' of theadduct.

An AMD complex with the self-complementarydouble-stranded oligonucleotide ApApApGpCpTpTpThas been examined by NMR, and all sugar moieties ofthe duplex have been found to occur in the C2-endo con-formation [66]. The finding contradicts the earlier modelof the complex wherein G(C3'-endo)–C(C2'-endo)has been assumed to occur in the intercalation site[47]. A decrease in ultrasonic cleavage of the sugar-phosphate backbone at the guanine residue in theAMD intercalation site also provides indirect evidencefor lack of such a transition and supports the mecha-nism that my colleagues and I proposed for mechano-chemical DNA cleavage [31]. However, to evaluate theeffect of AMD binding on the structural dynamicproperties of DNA, it is necessary to compare theresults with the X-ray, NMR, and molecular dynamicsdata available in the literature. Such a study of ultra-sonic cleavage for all di- and tetranucleotides requiresa thorough statistical analysis of the available experi-mental data and is beyond the scope of this work.

ACKNOWLEDGMENTSThis work was supported by the Program of Basic

Research in the State Academies of Sciences from2013 to 2020 (project no. 01201363818) and the pro-gram “Molecular and Cell Biology” of the Presidiumof the Russian Academy of Sciences.

REFERENCES1. Waring M.J. 2006. Sequence-specific DNA Binding

Agents. Ed. Waring M.J. Royal Soc. Chem. Biomol. Sci.Series.

2. Sarai A., Kono H. 2005. Protein–DNA recognitionpatterns and predictions. Ann. Rev. Biophys. Biomol.Struct. 34, 379–398.

3. Burd J.F., Larson J.E., Wells R.D. 1975. Further stud-ies on telestability in DNA. The synthesis and charac-terization of the duplex block polymers d(C20A10) ·d(T10G20) and d(C20A15)-d(T15G20). J. Biol. Chem.250, 6002–6007.

4. Burd J.F., Wartell R.M., Dodgson J.B., Wells R.D.1975. Transmission of stability (telestability) in deoxy-ribonucleic acid. Physical and enzymatic studies on theduplex block polymer d(C15A15)•d(T15G15). J. Biol.Chem. 250, 5109–5113.

5. Olson W.K., Gorin A.A., Lu X.J., Hock L.M., Zhurkin V.B.1998. DNA sequence-dependent deformability deducedfrom protein–DNA crystal complexes. Proc. Natl Acad.Sci. U. S. A. 95, 11163–11168.

6. Sims G.E., Kim S.H. 2003. Global mapping of nucleicacid conformational space: Dinucleoside monophos-phate conformations and transition pathways amongconformational classes. Nucleic Acids Res. 31, 5607–5616.

7. Svozil D., Kalina J., Omelka M., Schneider B. 2008.DNA conformations and their sequence preferences.Nucleic Acids Res. 36, 3690–3706.

8. Heddi B., Oguey C., Lavelle C., Foloppe N., Hart-mann B. 2009. Intrinsic f lexibility of B-DNA: Theexperimental TRX scale. Nucleic Acids Res. 38 (3),1034–1047. doi doi 10.1093/nar/gkp962

9. Abi-Ghanem J., Heddi B., Foloppe N., Hartmann B.2010. DNA structures from phosphate chemical shifts.Nucleic Acids Res. 38, e18.

10. Fox K.R., Waring M.J. 1984. DNA structural variationsproduced by actinomycin and distamycin as revealed byDNAase I footpnatiag. Nucleic Acids Res. 12, 9271–9285.

11. Neidle S., Pearl L.H. Skelly J.V. 1987. DNA structureand perturbation by drug binding. Biochem. J. 243, 1–13.

12. Bullwinkle T.J., Koudelka G.B. 2011. The lysis-lysog-eny decision of bacteriophage 933W: A 933W repressor-mediated long-distance loop has no role in regulating933W PRM activity. J. Bacteriol. 193, 3313–3323.

13. Tullius T.D. 1988. DNA footprinting with hydroxylradical. Nature. 332, 663–664.

14. Van Dyke M.W., Dervan P.B. 1983. Footprinting withMPE·Fe(II). Complementary-strand analyses of dista-mycin-and actinomycin-binding sites on heteroge-neous DNA. Cold Spring Harbor Symp. Quant Biol.47 (1), 347–353.

15. Spassky A., Angelov D. 1997. Influence of the localhelical conformation on the guanine modificationsgenerated from one-electron DNA oxidation. Biochem-istry. 36, 6571–6576.

16. Vtyurina N.N., Grohovsky S.L., Vasiliev A.B., Titov I.I.,Ponomarenko P.M., Ponomarenko M.P., Peltek S.E.,Nechipurenko Yu.D., Kolchanov N.A. 2012. Contex-tual DNA features significant for the DNA damage bythe 193-nm ultraviolet laser beam. Dokl. Biochem. Bio-phys. 447, 267–272.

17. Grokhovsky S.L., Zubarev V.E. 1991. Sequence-spe-cific cleavage of double-stranded DNA caused by X-rayionization of the platinum atom in the Pt-bis-netrop-sin–DNA complex. Nucleic Acids Res. 19, 257–264.

MOLECULAR BIOLOGY Vol. 52 No. 4 2018

ULTRASONIC FOOTPRINTING 619

18. Galas D.J., Schmitz A. 1978. DNase footprinting: Asimple method for the detection of protein–DNAbinding specificity. Nucleic Acids Res. 5, 3157–3170.

19. Dervan P.B. 1986. Design of sequence-specific DNA-binding molecules. Science. 232, 464–471.

20. Drew H.R. 1984. Structural specificities of five com-monly used DNA nucleases. J. Mol. Biol. 176, 535–557.

21. Suck D., Oefner C. 1986. Structure of DNase I at 2.0 Åresolution suggests a mechanism for binding to and cut-ting DNA. Nature. 321, 620–625.

22. Skamrov A.V., Rybalkin I.N., Bibilashvili R.Sh., Got-tikh B.P., Grokhovskii S.L., Gursky G.V., Zhuze A.L.,Zasedatelev A.S., Nechipurenko D.Yu., Khorlin A.A.1985. Specific protection of DNA by distamycin A,netropsin and bis-netropsins against the action ofDNAse I. Mol. Biol. (Moscow). 19, 177–195.

23. Balasubramanian B., Pogozelski W.K., Tullius T.D.1998. DNA strand breaking by the hydroxyl radical is gov-erned by the accessible surface areas of the hydrogen atomsof the DNA backbone. Proc. Nat. Acad. Sci. U. S. A. 95,9738–9743.

24. Belikov S.V., Grokhovsky S.L., Isaguliants M.G., Sur-ovaya A.N., Gursky G.V. 2005. Sequence-specificminor groove binding ligands as potential regulators ofgene expression in Xenopus laevis oocytes. J. Biomol.Struct. Dynam. 23, 193–202.

25. Wells R.D. 2009. Discovery of the role of non-B DNAstructures in mutagenesis and human genomic disor-ders. J. Biol. Chem. 284, 8997–900.

26. Grokhovsky S. L. 2006. Specificity of DNA cleavage byultrasound. Mol. Biol. (Moscow). 40, 276–283.

27. Grokhovsky S.L., Il’icheva I.A., Nechipurenko D.Yu.,Panchenko L.A., Polozov R.V., Nechipurenko Yu.D.2008. Ultrasonic cleavage of DNA: Quantitative analy-sis of sequence specificity. Biophysics (Moscow). 53,250–251.

28. Grokhovsky S., Il’icheva I., Nechipurenko D.,Golovkin M., Taranov G., Panchenko L., Polozov R.,Nechipurenko Y. 2012. Quantitative analysis of electro-phoresis data-application to sequence-specific ultra-sonic cleavage of DNA. In: Gel Electrophoresis-Princi-ples and Basics. London: InTech. doi 10.5772/2205

29. Das R., Laederach A., Pearlman S.M., Herschlag D.,Altman R.B. 2005. SAFA: semi-automated footprint-ing analysis software for high-throughput quantifica-tion of nucleic acid footprinting experiments. RNA. 11,344–354.

30. Nechipurenko Yu.D., Golovkin M.V., Nechipu-renko D.Yu., Il’icheva I.N., Panchenko L.A., Polo-zov R.V., Grokhovsky S.L. 2008. Quantitative methodsfor analysis of DNA cleavage by ultrasound. Mathematics,Computer, Education: Proc. 15th Int. Conf. Izhevsk, vol.3, pp. 26–35.

31. Grokhovsky S.L., Ilicheva I.A., Nechipurenko D.Yu.,Golovkin M.V., Panchenko L.A., R.V. Polozov,Nechipurenko Y.D. 2011. Sequence-specific ultrasoniccleavage of DNA. Biophys. J. 100, 117–125.

32. Kang H.J., Park H.J. 2009. Novel molecular mecha-nism for actinomycin D activity as an oncogenic pro-moter G-quadruplex binder. Biochemistry. 48, 7392–7398.

33. Brown D.R., Kurz M., Kearns D.R., Hsu V.L. 1994.Formation of multiple complexes between actinomycin Dand a DNA hairpin: Structural characterization bymultinuclear NMR. Biochemistry. 33, 651–664.

34. Chen F.-M., Sha F., Chin K.-H., Chou S.-H. 2004.The nature of actinomycin D binding to d(AACCAXYG)sequence motifs. Nucleic Acids Res. 32, 271–277.

35. Rill R.L., Hecker K.H. 1996. Sequence-specific acti-nomycin D binding to single stranded DNA inhibitsHIV reverse transcriptase and other polymerases. Bio-chemistry. 35, 3525–3533.

36. Wadkins R.M., Vladu B. Tung C.S. 1998. AMD bindsto metastable hairpins in single-stranded DNA. Bio-chemistry. 37, 11915–11923.

37. Gallego J., Ortiz A.R., Pascual-Teresa B. Gago F. 1997.Structure–affinity relationships for the binding of acti-nomycin D to DNA. J. Comp.-Aid. Mol. Design. 11,114–128.

38. Nikitin S.M., Grokhovskii S.L., Zhuze A.L., Mikhai-lov M.V., Zasedatelev A.S., Gurskii G.V., Gottikh B.P1981. Ligands possessing affinity for definite bases orbase sequences of DNA: 5. Analogs of actinomycin Dwith substituents in position 7 of the phenoxazonechromophore. Russ. J. Bioorg. Chem. 7, 289–297.

39. Hampshirea A.J., Fox K.R. 2008. The effects of localDNA sequence on the interaction of ligands with theirpreferred binding sites. Biochimie. 90, 988–998.

40. Bailly C., Marchand C., Waring M.J. 1993. New bind-ing sites for antitumor antibiotics created by relocatingthe purine 2-amino group in DNA. J. Am. Chem. Soc.115, 3784–3785.

41. Bailly C., Graves D.E., Ridge G., Waring M.J. 1994.Use of a photoactive derivative of actinomycin to inves-tigate shuff ling between binding sites on DNA. Bio-chemistry. 33, 8736–8745.

42. Qu X., Ren J., Riccelli P.V., Benight A.S., Chaires J.B.2003. Enthalpy/entropy compensation: Influence ofDNA flanking sequence on the binding of 7-aminoactinomycin D to its primary binding site in short DNAduplexes. Biochemistry. 42, 11960–11967.

43. Lian C.Y. Robinson H., Wang A.H.J. 1996. Structureof actinomycin D bound with (GAAGCTTC)2 and(GATGCTTC)2 and its binding to the (CAG)n:(CTG)n triplet sequence as determined by NMR anal-ysis, J. Am. Chem. Soc. 118, 8791–8801.

44. Kamitori S., Takusagawa F. 1992. Crystal structure ofthe 2:1 complex between d (GAAGCTTC) and the anti-cancer drug actinomycin D. J. Mol. Biol. 225, 445–456.

45. Zhou N., James T. L., Shafer R. H. 1989. Binding ofactinomycin D to [d (ATCGAT)]2: NMR evidence ofmultiple complexes. Biochemistry. 28, 5231–5239.

46. Vekshin N.L. 2009. Biophysics of DNA–ActinomycinNanocomplexes. Pushchino: Foton-Vek.

47. Sobell H.M., Jain S.C., Sakore T.D., Nordman C.E.1971. Stereochemistry of actinomycin–DNA binding.Nature. 231, 200–205.

48. Crothers D.M., Müller W. 1973. Origins of base speci-ficity in actinomycin and other DNA ligands. CancerChemother. Rep. 58, 97–100.

49. Rescifina A., Zagni C., Varrica M.G., Pistarà V., Cor-saro A. 2014. Recent advances in small organic mole-

620

MOLECULAR BIOLOGY Vol. 52 No. 4 2018

GROKHOVSKY

cules as DNA intercalating agents: Synthesis, activity,and modeling. Eur. J. Med. Chem. 74, 95–115.

50. Krugh T.R., Mooberry E.S., Chiao Y.C.C. 1977. Pro-ton magnetic resonance studies of actinomycin D com-plexes with mixtures of nucleotides as models for thebinding of the drug to DNA. Biochemistry. 16, 740–747.

51. Paramanathan T., Vladescu I., McCauley M.J., Rou-zina I., Williams M.C. 2012. Force spectroscopy revealsthe DNA structural dynamics that govern the slowbinding of actinomycin D. Nucleic Acids Res. 40, 4925–4932.

52. Scamrov A.V., Beabealashvilli R.S. 1983. Binding ofactinomycin D to DNA revealed by DNase I footprint-ing. FEBS Lett. 164, 97–101.

53. Waterloh K., Fox K.R. 1992. Secondary (non-GpC)binding sites for actinomycin on DNA. Biochim. Bio-phys. Acta. 1131, 300–306.

54. Chen F.-M. 1992. Binding specificities of actinomycinD to non-self-complementary -XGCY- tetranucleotidesequences. Biochemistry. 31, 6223–6228.

55. Bailey S.A., Graves D.E., Rill R., Marsch G. 1993.Influence of DNA base sequence on the binding ener-getics of actinomycin D. Biochemistry. 32, 5881–5887.

56. Fletcher M.C., Fox K.R. 1996. Dissociation kinetics ofactinomycin D from individual GpC sites in DNA. Eur.J. Biochem. 237, 164–170.

57. Fox K.R., Waring M.J. 1984. DNA structural variationsproduced by actinomycin and distamycin as revealed byDNase I footprinting. Nucleic Acids Res. 12, 9271–9285.

58. Hogan M.E., Roberson M.W., Austin R.H. 1989. DNAflexibility variation may dominate DNase I cleavage.Proc. Nat. Acad. Sci. U. S. A. 86, 9273–9277.

59. Brukner I., Sanchez R., Suck D., Pongor S. 1995.Sequence-dependent bending propensity of DNA asrevealed by DNase I: Parameters for trinucleotides.EMBO J. 14, 1812.

60. Basedow A.M., Ebert E.B. 1977. Ultrasonic degrada-tion of polymers in solution. In: Advances in PolymersScience. Eds. Abe A., Albertsson A.-C., Genzer J. Ber-lin: Springer, vol. 22, pp. 83–148.

61. Scott E.V., Zon G., Marzilli L.G., Wilson W.D. 1988.2D NMR investigation of the binding of the anticancerdrug actinomycin D to duplexed dATGCGCAT: Con-formational features of the unique 2:1 adduct. Biochem-istry. 27, 7940–7951.

62. Chen H., Liu X., Patel D.J. 1996. DNA bending andunwinding associated with actinomycin D antibioticsbound to partially overlapping sites on DNA. J. Mol.Biol. 258, 457–479.

63. Il’icheva I.A., Nechipurenko D.Yu., Grokhovsky S.L.2009. Ultrasonic cleavage of nicked DNA. J. Biomol.Struct. Dynam. 27, 391–398.

64. Huang M., Giese T.J., Lee T.S., York D.M. 2014).Improvement of DNA and RNA sugar pucker profilesfrom semiempirical quantum methods. J. Chem. The-ory Comput. 10, 1538–1545.

65. Todd R.C., Lippard S.J. 2010. Structure of duplexDNA containing the cisplatin 1, 2-{Pt (NH 3. 2} 2+-d(GpG) cross-link at 1.77 Å resolution. J. Inorg. Bio-chem. 104, 902–908.

66. Liu X., Chen H., Patel D.J. 1991. Solution structure ofactinomycin-DNA complexes: Drug intercalation atisolated GC sites. J. Biomol. NMR. 1, 323–347.

Translated by T. Tkacheva

SPELL: 1. OK