Báo cáo khoa học: Binding of berberine to human telomeric quadruplex – spectroscopic, calorimetric and molecular modeling studies pot

The complete thermodynamic profile for berberine bind-ing to the quadruplex, at 25C, shows a small negative enthalpy DH of 1.7 kcalÆmol1, an entropy change with TDS of +6.5 kcalÆmol1, and

quadruplex – spectroscopic, calorimetric and

molecular modeling studies

Amit Arora1, Chandramouli Balasubramanian1, Niti Kumar1, Saurabh Agrawal1, Rajendra P Ojha2 and Souvik Maiti1

1 Proteomics and Structural Biology Unit, Institute of Genomics and Integrative Biology, CSIR, Delhi, India

2 Biophysics Unit, Department of Physics, DDU Gorakhpur University, India

The telomere is a region of highly repetitive DNA at

the end of a linear chromosome that protects the

terminal ends of chromosomes from being recognized

as damaged DNA and allows faithful chromosome

replication during the cell cycle [1,2] Mammalian

telomeres consist of several kilobase pairs of

double-stranded G-rich DNA and a 100–200 base

single-stranded overhang on the 3¢-end [3,4] A host of

telomere-associated proteins, including TRF1, TRF2

and POT1, ensures that the telomeric single-stranded

overhang does not trigger DNA damage response

pathways or lead to abnormal chromosomal

rearrange-ments [5–7] Exposure of the 3¢-end due to uncapping

results in cellular senescence and apoptosis [8,9]

Telo-merase, a ribonucleoprotein reverse transcriptase

enzyme (composed of both RNA and proteins), uses its internal RNA component (complementary to the telomeric single-stranded overhang) as a template for synthesis of telomeric DNA A(GGGTTA)n, directly at the ends of chromosomes Telomerase is present in most fetal tissues, normal adult male germ cells, inflammatory cells, proliferative cells of renewal tissues, and most tumor cells Importantly, telomerase

is active in a majority of human cancer cells but is inactive in most normal somatic cells [10] It has been shown previously that formation of intramolecular G-quadruplexes by the telomeric G-rich strand inhibits the activity of telomerase [10] Therefore, ligand-induced stabilization of intramolecular telomeric G-quadruplexes has become an attractive strategy for

Keywords

berberine; hydration; quadruplex;

quadruplex–ligand interaction;

thermodynamics

Correspondence

S Maiti, Proteomics and Structural Biology

Unit, Institute of Genomics and Integrative

Biology, CSIR, Mall Road, Delhi 110 007,

India

Fax: +91 11 2766 7471

Tel: +91 11 2766 6156

E-mail: souvik@igib.res.in

(Received 11 April 2008, revised 16 May

2008, accepted 9 June 2008)

doi:10.1111/j.1742-4658.2008.06541.x

This study examines the characteristics of binding of berberine to the human telomeric d[AG3(T2AG3)3] quadruplex By employing UV-visible spectroscopy, fluorescence spectroscopy and isothermal titration calorime-try, we found that the binding affinity of berberine to the human telomeric quadruplex is 106 The complete thermodynamic profile for berberine bind-ing to the quadruplex, at 25C, shows a small negative enthalpy (DH) of )1.7 kcalÆmol)1, an entropy change with TDS of +6.5 kcalÆmol)1, and an overall favorable free energy (DG) of)8.2 kcalÆmol)1.Through the temper-ature dependence of DH, we obtained a heat capacity (DCp) of )94 (± 5) calÆmol)1ÆK)1 The osmotic stress method revealed that there is an uptake of 13 water molecules in the complex relative to the free reactants Furthermore, the molecular modeling studies on different quadruplex– berberine complexes show that berberine stacking at the external G-quartet

is mainly aided by the p–p interaction and the stabilization of the high negative charge density of O6 of guanines by the positively charged N7 of berberine The theoretical heat capacity (DCp) values for quadruplex– berberine models are)89 and )156 calÆmol)1ÆK)1

Abbreviations

H 2 TMPyP 4 , 5,10,15,20-tetrakis(1-methyl-4-pyridyl)-21H,23H-porphine; ITC, isothermal titration calorimetry; MMPBSA, molecular mechanics Poisson–Bolzmann surface area; SASA, solvent-accessible surface area.

the development of anticancer ligands A molecule that

has (a) a p-delocalized system, (b) a partial positive

charge in the center of the molecular scaffold and

(c) positively charged substituents to interact with the

grooves, loops and the negatively charged phosphate

backbone is most likely to interact with, and thus

stabilize, G-quadruplexes A number of

G-quadruplex-interacting agents with the above-mentioned features,

typically porphyrins [11–16], anthraquinones [17],

perylenes [18] and carbocyanines [19], have been

developed and shown to promote and⁄ or stabilize

quadruplex structures In past few years, Neidle and

co-workers have reported a number of trisubstituted

acridine analogs with a variety of side-chain

modifi-cations and stereoisomer variations exhibiting strong

G-quadruplex binding, high selectivity for quadruplex

over duplex DNA, and associated telomerase

inhibitory activity in the nanomolar range [20–22]

The rational design of new therapeutic agents that

bind to quadruplexes in a structure-specific manner is

of considerable interest and urgency Many small

molecules that bind to quadruplexes have proven to be

effective therapeutic agents, although the exact mode

of binding and nature of thermodynamic forces that

regulate DNA–ligand interactions are often poorly

understood This limited knowledge hampers many

efforts to rationally modify existing ligands and⁄ or

design new therapeutic agents that bind to target

quadruplex structures with predictable affinity and

specificity Characterization of the forces that govern

quadruplex–ligand interactions traditionally relies on

detailed knowledge of the thermodynamic and

struc-tural properties of the ligand, the DNA, and the

com-plex Berberine, an isoquinoline alkaloid from plants,

is a planar molecule with an extended p-delocalized

system having a partial positive charge on N7 [23]

(Fig 1) It has been shown that berberine and its

analogs bind to telomeric G-quadruplex and inhibit the telomerase activity [24,25] Studies show that these molecules have high selectivity for G-quadruplex over duplex DNA, and the aromatic moieties of the mole-cule play a dominant role in quadruplex binding, implying that this molecule could be an attractive scaf-fold to develop new ligands targeting G-quadruplex selectively To obtain comprehensive knowledge on the interaction of this scaffold, we performed spectro-scopic, calorimetric and molecular modeling studies to obtain thermodynamic and structural details of the quadruplex–berberine interaction

Results

Equilibrium binding studies by UV-visible spectroscopy

To gain insight into the interaction between berberine and the G-quadruplex formed by the telomere, UV spectra of berberine in the absence and presence of quadruplex were obtained The resulting absorption spectra are illustrated in Fig 2 The UV spectra of berberine show two distinct peaks at 341 and 421 nm Addition of increasing amounts of quadruplex results

in hypochromicity (34–40%) and a moderate batho-chromic shift of 7 nm for the high-energy peak from

341 to 348 nm in the UV-visible spectra of berberine, and hyperchromicity and a red shift of 19 nm for the lower-energy peak from 421 to 440 nm, indicating

Fig 1 Chemical structure of berberine.

Fig 2 Absorbance spectra of 1 l M berberine in 50 m M MES buffer (pH 7.4) and 100 m M KCl in the absence and presence of succes-sive additions of quadruplex at 25 C The inset is the Scatchard plot of r ⁄ C versus r, where r is the ratio of bound berberine to the total base pair concentration, and C is the concentration of free ligand Data were fitted to the McGhee–von Hippel neighbor exclu-sion model.

interaction of berberine with quadruplex The

occur-rence of sharp isobestic points at 359, 383 and 445 nm

clearly indicates the existence of equilibrium in the

binding Ligand binding with DNA through

intercala-tion usually results in hypochromicity and

bathochro-mism due to strong stacking interactions between an

aromatic chromophore and the base pairs of DNA

These spectral characteristics suggest a mode of

bind-ing that involves a stackbind-ing interaction between

berber-ine and the quartet of quadruplex The absorbance

change at 341 nm of the berberine absorption spectra

upon successive addition of quadruplex was used to

construct a Scatchard plot Analysis of this Scatchard

plot yielded a binding affinity of (1.2 ± 0.2)· 106m)1

and a binding site density of 0.9 at 25C

Equilibrium binding studies by the fluorescence

method

Fluorescence emission spectra for berberine in the

absence and presence of different amounts of

quadru-plex were recorded in order to study the binding event

Figure 3 shows the effect of successive addition of

quadruplex on the fluorescence emission spectra of

berberine It is seen that increasing the concentration

of quadruplex results in a gradual increase in the

rescence intensity of berberine The ratio of the

fluo-rescence intensity of berberine in the presence and

absence of quadruplex is about 50 The kmaxin the

flu-orescence emission spectra shifts to the blue end by

5 nm The spectral changes arise from the change in

the environment of berberine, which reveals that

berberine is binding with quadruplex The change in

fluorescence intensity at 522 nm due to addition of

quadruplex solution was used to construct the binding

isotherm (inset of Fig 3) Analysis of this isotherm

following 1 : 1 binding stoichiometry using Eqn (7)

(see Experimental procedures) gives a binding affinity

of (1.2 ± 0.1)· 106m)1 at 25C Thermodynamic

parameters calculated for the quadruplex–berberine

binding are presented in Table 1

Equilibrium binding studies by the isothermal

titration calorimetry (ITC) method

With recent advances in the sensitivity and reliability

of the calorimeter, ITC has become an important tool

for the direct measurement of thermodynamic

para-meters in various biological interactions [26,27] ITC

yields thermodynamic parameters such as Gibbs free

energy change (DG), enthalpy change (DH), and

entropy change (DS), along with the number of

bind-ing sites (n) in a sbind-ingle experiment Also, determination

of binding enthalpy as a function of temperature yields changes in heat capacity (DCp) associated with an interaction that provides valuable insights into the type and magnitude of forces involved Therefore, we have utilized ITC to characterize the thermodynamics of binding of berberine to quadruplex Calorimetric titra-tions were performed at different temperatures to directly measure the binding enthalpy Figure 4A shows a typical titration curve obtained at 25C The area under the heat burst curves was determined by integration to yield the heat of injection associated with the reaction These injection heats were corrected

by subtraction of the corresponding dilution heats derived from the injection of identical amounts of ber-berine into the buffer alone The corrected isotherms obtained at five different temperatures are shown in Fig 4B All related thermodynamic parameters are presented in Table 2 The binding affinity measured from ITC is (0.4 ± 0.1)· 106m)1 at 25C At all temperatures studied, the binding enthalpies were found to be negative, with their magnitude increasing

Fig 3 Fluorescence emission spectra of 0.5 l M berberine in

50 m M MES buffer (pH 7.4) and 100 m M KCl in the absence and presence of successive additions of quadruplex at 25 C The inset

is the plot of DF versus quadruplex concentration Data were fitted

to Eqn (7) to extract the binding affinity (Experimental procedures).

Table 1 Thermodynamic parameters obtained for quadruplex–ber-berine binding at 25 C K b is the binding constant determined from spectroscopic titrations in 50 m M MES buffer (pH 7.4), DG is the net binding free energy calculated using DG = )RT ln K b , and DH is the binding enthalpy determined directly by ITC and used to calcu-late the entropy change, using DG = DH )TDS.

K b (·10 6

M )1) DG (kcalÆmol)1) DH (kcalÆmol)1) TDS (kcalÆmol)1)

1.2 ± 0.1 )8.2 ± 0.8 )1.7 ± 0.2 6.5 ± 0.7

with an increase in temperature In all cases, the

stoi-chiometry was found to be one mole of ligand binding

per mole of quadruplex

Apparent discrepancy between spectroscopic and calorimetric binding constants

Examination of Tables 1 and 2 shows that the Kb val-ues determined from the spectroscopic and ITC data differ by one order at 25 C, despite the use of identi-cal salt and buffer conditions The binding constant can be determined accurately when titrant is added to

a fixed and constant concentration [Q0] of DNA, such that [Q0] is in the range of 1⁄ KA In UV and fluores-cence binding experiment, [Q0] = 1 lm and 0.5 lm respectively, which is in the range of 1⁄ KA (1.2 lm) However, the ITC experiment was performed at [Q0] = 10 lm, which is much larger than 1⁄ KA (1.2 lm) ITC could not be used in this case to obtain

an accurate value for Kb, as the low site concentration required would give a heat signal below the sensitivity

of the instrument For a quadruplex concentration of

1 lm in the cell, the heat output was not significantly greater than the heats of ligand dilution If the dilution heats were small and the binding enthalpy was large, then it would be possible to obtain a binding isotherm using a quadruplex concentration of 0.5–1 lm How-ever, these conditions are not met Despite these prob-lems, ITC can still be used to accurately and directly measure the binding enthalpy and stoichiometry for this interaction ITC remains an invaluable technique for determining binding enthalpies, even in cases where the binding constant cannot be determined accurately [28]

Heat capacity measurements Supplementary Fig S1 shows the temperature depen-dence of all thermodynamic parameters No curvature

in the plots of thermodynamic constants versus tem-perature is apparent over this temtem-perature range, and all of the plots are fitted with linear functions The heat capacity change (DCp) for a binding interaction

Fig 4 (A) Sample thermogram for the calorimetric titration of

100 l M berberine into 10 l M quadruplex at 25 C (B) Integrated

heats (after subtraction of heat of dilution for berberine) versus

ber-berine to quadruplex molar ratio plot at 10 C (h), 15 C (O), 20 C

(D), 25 C (,) and 30 C (e) The first data point was eliminated in

the data fit.

Table 2 Thermodynamic parameters obtained from ITC experiments for quadruplex–berberine binding in 50 m M MES buffer (pH 7.4) buffer containing 100 m M KCl Thermodynamic parameters were obtained for berberine binding to the preformed telomeric quadruplex at 25 C The quadruplex concentration in the cell was 10 l M and the berberine concentration in the syringe was 100 l M DG was determined using the relationship DG = )RT ln K b , where R is the universal gas constant, T is temperature, and K b is the binding affinity for the quadruplex– berberine interaction DH and DS correspond to the enthalpy and entropy change for the binding, respectively, and DG corresponds to the free energy change of binding DCpis the heat capacity change associated with the quadruplex–berberine interaction and is calculated using Eqn (1) as described in the text Values determined from microcalorimetric data.

Temperature (C) DH (kcalÆmol)1) K b · 10)6( M)1) n DG (kcalÆmol)1) TDS (kcalÆmol)1) DC p (calÆmol)1ÆK)1)

can be determined from the temperature dependence

of the observed binding enthalpy using the standard

relationship:

The slope of the resulting line of DH versus

tempera-ture (T) in supplementary Fig S1 yields DCp of

)94 ± 5 calÆmol)1ÆK)1 for the binding of berberine to

quadruplex Thus, berberine binding to quadruplex is

associated with a negative heat capacity change that

falls within a range that is frequently observed for

both nucleic acid–ligand and protein–ligand

inter-actions [29,30]

Hydration change due to the binding obtained by

the osmotic stress method

The osmotic stress method has been used extensively

to evaluate the participation of water molecules in a

wide variety of biochemical reactions [30] Any

equilib-rium that involves changes in the water molecules

associated with a biopolymer is sensitive to changes in

the water activity (aW) [31–33] Water activity can in

turn be manipulated by the addition of low molecular

weight cosolutes, which themselves do not interact

with the biopolymer but are assumed to change the

water activity Equilibria that are coupled with

hydra-tion changes are influenced by the osmolyte

concentra-tion, as described by Qu & Chaires [33]:

dlnðKs=K0Þ=d½Osm ¼ Dnw=55:6 ð2Þ

where ln(Ks⁄ K0) is the change in the binding free

energy, [Osm] is the osmolality (moles of solute per kg

of solvent) of the solution, and Dnwis the difference in

the number of bound water molecules between the

complex and the free reactants The change in the

binding affinity upon change in osmolyte concentration

is shown in supplementary Fig S2 As the

concentra-tion of osmolyte increases, the affinity of berberine for

the quadruplex binding site decreases This observation

is consistent with the acquisition of water by the

com-plex relative to the DNA Using Eqn (2), the average

number of exchanged water molecules is found to be

13 ± 2

Molecular modeling studies

Computational methods are widely used to

investi-gate biomolecules and complexes, and have been

shown to be valuable for a deeper understanding of

the structural, dynamic and energetic properties The

mixed hybrid NMR structure of the human telomeric

quadruplex was used for study (supplementary Fig S3) The structure has two external G-quartets that can act individually as binding sites for berber-ine [25] Berberberber-ine was docked against these external G-quartets, and the complexes were simulated in aqueous solution The rmsd values of the heavy atoms of the whole complex (black) and without loop residues (gray) are shown in supplementary Fig S4 The rmsd values of both the MH1 (5¢-end) and MH2 (3¢-end) complexes remain < 3 A˚ The rmsd values for the G-quartet (without loop residues) and berberine are conserved in both cases, and stay at < 1 A˚ during the last 2 ns (inset in supplementary Fig S4) The fluctuations of the loop residues are obvious, as they are not held tightly by hydrogen bonds and hence are free to move during dynamics This observation has also been reported in previous studies on G-quadruplex structures [34,35]

The stacking of berberine over the external G-quartet

is shown in Fig 5 for the models Berberine stacking over the G-quartet plane is aided by the formation of strong p–p aromatic stacking interactions between the berberine scaffold and the G-quartet plane In addition, the positively charged nitrogen atom in berberine positions itself on the axis passing through the center of the G-quartet plane Hence, a strong electrostatic inter-action can be expected between the positively charged nitrogen and the highly electron-rich central area of the G-quartet plane, due to the guanine carbonyl lone pairs The positioning of the nitrogen atom was observed to

be fairly retained in all models during dynamics This suggests that the electrostatic interaction between the negatively charged clouds formed by O6 of guanines and the positively charged nitrogen atom plays an important role in this stabilization

The relative free energy components for the complex formation were estimated by the molecular mechanics Poisson–Bolzmann surface area (MMPBSA) approach The calculations were performed on the basis of the single trajectories of the quadruplex–berberine complexes obtained from the explicit solvent simula-tions The estimates are summarized in supplementary Table S1

Theoretical heat capacity calculation The heat capacity change upon complex formation is

an informative measure that can provide insights into the exchange of water during the process The relations connecting the changes in heat capacity to the burial

of polar and nonpolar solvent-accessible surface area (SASA) during complex formation has been proposed

and applied in a number of previous reports [29,36] In

our study, we used the relation proposed by Ren et al

[37], which is given as follows:

DCp¼ 0:382ð 0:026ÞDAnp 0:121ð 0:077ÞDAp ð3Þ

Here, DAnp and DAp represent the changes in SASA for nonpolar and polar groups, respectively A sum-mary of the solvent-accessible areas and DCp values is shown in Table 3 The SASA is reduced upon complex formation, and the majority of the reduction was due

to the burial of nonpolar surface This is reflected

in the negative values of the calculated heat capacity change The calculated values are )89 and )156 calÆmol)1ÆK)1for MH1 and MH2 respectively

Discussion

In order to understand biomolecule–ligand binding in terms of sequence-specific recognition and affinity, it is necessary to complement high-resolution structural data with accurate thermodynamic measurements By using a combination of spectroscopic and ITC tech-niques, we have elucidated a complete thermodynamic profile (DG, DH, DS, Kb, DCpand Dnw) for the binding

of berberine to the telomeric quadruplex UV-visible absorption titration experiments show that the binding

of berberine to G-quadruplexes results in a red shift (10–12 nm) and substantial hypochromicity (34–40%)

in the k341 nmof berberine The red shift in the absorp-tion maxima and the observed hypochromicity of ber-berine in the presence of quadruplex may be interpreted in terms of stacking interactions between the quartet-forming guanine bases and the aromatic groups of berberine Although the observed red shift is intermediate between what is observed for intercalation (> 15 nm) and for outside binding (< 8 nm) [38], the same extent of hypochromicity that is generally seen in the intercalated binding are observed, revealing that berberine interacts with quadruplex through stacking interactions between quartet and berberine, as happens

in case of intercalative binding events Recently, Wei

Table 3 Summary of the changes in SASA in A˚2 and heat capacity changes (DCp) in calÆmol)1ÆK)1calculated for the complex models DAtot

is the change in total accessible surface area, DA np is the change in nonpolar SASA and DA p is the change in polar SASA DC p is calculated using Eqn (3) as described in the text.

A

B

G 20

G 2

G 8

G 16

G 4

G 10

G 14

G

Fig 5 Stacking of berberine on the G-quartet face A-MH1, B-MH2.

The G-quartet is shown as a stick model Berberine is shown as a

ball and stick model.

et al [15] have studied the interaction of cationic

porphyrin [5,10,15,20-tetrakis(1-methyl-4-pyridyl)-21H,

23H-porphine] (H2TMPyP4) with three distinct

G-quadruplex DNAs, parallel-stranded (TG4T)4,

dimer-hairpin-folded (G4T4G4)2, and monomer-folded

AG3(T2AG3)3, by UV resonance Raman spectroscopy,

UV-visible absorption spectroscopy, fluorescence

spec-troscopy, and surface-enhanced Raman spectroscopy

In their UV-visible absorption titration experiments,

the same extent of red shift (11–13 nm) but larger

hypochromicities (56–62%) were observed, indicating

intercalative (containing the end stacking) binding of

H2TMPyP4to these G-quadruplexes Comparing these

observations with ours, it can be concluded that

berberine binds to quadruplex at external quartets

through stacking interactions between the

quartet-forming guanine bases and the aromatic groups of

berberine The lower red shift and moderate

hypo-chromicity can be accounted for by the partial

inter-calation by end stacking with the quartet The 1 : 1

binding stoichiometry, as estimated from the present

results, limits the interaction to a single binding mode

It was seen that addition of increasing concentrations

of quadruplex results in a gradual increase in the

fluorescence intensity of berberine, indicating the

trans-fer of berberine from an aqueous environment to a

hydrophobic environment This observation rules out

the possibility of outside stacking of berberine, where

quenching of the chromophore fluorescence by solvent

molecules could have been continued In a recent

study, Franceschin et al [25], through molecular

modeling, have shown that piperidino-berberine stacks

on the terminal G-tetrad of the quadruplex

The sigmoidal binding isotherms obtained from ITC

experiments are indicative of the existence of either a

single binding site per quadruplex or a number of

equivalent, but not necessarily independent, binding

sites At all temperatures, the stoichiometry for the

quadruplex–berberine binding was found to be 1 : 1,

confirming the results of our spectroscopic binding

studies The binding enthalpies were found to be

neg-ative with increasing magnitudes upon increase in

temperature Berberine binds to quadruplex at 25C

with a small, negative enthalpy (DH) of )1.7

kcalÆmol)1 and an entropy change with TDS of +6.5

kcalÆmol)1 with an overall favorable free energy (DG)

of)8.2 kcalÆmol)1 The favorable binding of berberine

comes from a combination of enthalpy and entropy

terms that vary with temperature The negative values

of DH and positive values of DS are consistent with

the characteristics of a combination of van der Waals,

hydrophobic and electrostatic interactions in the

binding process The temperature dependence of the

binding affinity was used to calculate the van’t Hoff enthalpy, which did not match the calorimetric enthalpy The obtained van’t Hoff enthalpy was )12 kcalÆmol)1, giving DHvH⁄ DHcal ratios in excess of

1 A large difference between the van’t Hoff and calo-rimetric enthalpies could be due to substantial temper-ature-dependent behavior of associated reactions This might originate from changes in hydrophobic hydra-tion [39], in which release of water molecules from hydrophobic surfaces upon binding results in loss of enthalpy (due to stronger hydrogen bonds of struc-tured water) and gain in entropy, a phenomenon known to be temperature-dependent [40–42] Measure-ment of heat capacity changes (DCp) associated with ligand–macromolecule binding can help to differentiate the nature of hydration changes, i.e hydrophobic ver-sus polar hydration [43] Unlike other thermodynamic parameters, which have contributions from various sources, DCpis believed to arise purely from molecular hydration associated with binding [44] This parameter can thus be utilized to estimate the extent of burial or exposure of polar and nonpolar groups to bulk water upon molecular binding The obtained DCp for the quadruplex–berberine interaction was )94 ± 5 calÆmol)1ÆK)1 In the case of intercalative as well as minor groove binding ligands, it was shown that DCp varies from )100 to )400 calÆmol)1ÆK)1 [29,44] How-ever, the large negative heat capacity change is highly correlated with hydration heat capacity changes that arise from burial of the hydrophobic area As we have observed in our UV binding experiment, berberine binds to quadruplex by stacking on the terminal G-tet-rad of the quadruplex, so burial of the hydrophobic group is not extensive enough to show a sufficiently negative heat capacity change upon interaction The free energy (DGhyd) for the hydrophobic transfer of a ligand from aqueous solution to its macromolecular binding pocket is a function of the DCpfor the binding reaction, and is given by the following relationship [44]:

According to this relationship, negative heat capacity changes result in a large negative (favorable) DGhyd value, which, in turn, shows a significant driving force for complex formation As an illustrative example, the Chaires [37] and Wilson [29,45] groups applied Eqn (3)

to show that the major driving force for the binding of various heterocyclic ligands (e.g Hoechst 33258 and penta-amidine) to the minor groove of duplex DNA stems from the hydrophobic transfer of the ligands from solution to the DNA-binding site

Careful inspection of Table 1 clearly shows that the

binding of berberine to the human telomeric

quadru-plex having a small negative value of DH and a large

positive value of TDS is predominantly entropically

driven Recently, Chaires has analyzed the binding

data for 26 DNA–ligand interactions and discussed

distinctive thermodynamic signatures for groove

bind-ing and intercalation [46] Groove-bindbind-ing interactions

are largely entropically driven, whereas intercalation

reactions are driven by large favorable enthalpy

con-tributions and are opposed by entropy In the present

study, the observed low negative enthalpy and high

positive entropy change could constitute further

evi-dence in support of stacking with the terminal quartet

rather than intercalation Rigidifying the DNA ought

to exert an entropic cost, and this is the most likely

explanation for the unfavorable entropy associated

with intercalation [47] In contrast, terminal stacking

in the quadruplex structures should not rigidify the

quadruplex structure, thus reducing the entropic cost

as compared to intercalation The origin of the

favor-able entropic term in the present study is not also

apparent It has been argued that groove binding

shows the favorable entropy because of the release of

water molecules upon complex formation, by

dis-placement of the ‘spine of hydration’ within the

minor groove [48] However, the osmotic stress study

shows that the quadruplex–berberine complex

acquires 13 molecules of water on average per mole

of complex at 25C Similar kinds of observation

have been reported recently by Kiser et al [49], where

uptake of water molecules as well as a positive

entropy change were observed when Hoechst, a

groove binder, bound to oligomeric DNA This

dis-crepancy may be due to the inefficiency of the

osmo-tic stress method in measuring the overall hydration

change, as mentioned by Chaires [46] as well as by

Kiser et al [49]

The simulated structures show that the planar

ber-berine molecule is stacked onto the G-quartet,

posi-tioning the N7 positive charge above the center of the

G-tetrad in the region of high negative charge density

generated by the carbonyl groups to stabilize the

complex via favorable p-stacked interactions between

aromatic residues without significant disruption of

the guanine tetrads The binding enthalpy (DH) for the

quadruplex–berberine complexes originates from the

combination of polar solvation energy and favorable

solute electrostatic, van der Waals and nonpolar

sol-vation energy The conformational entropy for the

complexes arises from the loss of translational and

rotational degrees of freedom The overall free energy

change (DG) for the quadruplex–berberine binding is

found to be more favorable for the MH2 berberine– quadruplex complex (supplementary Table S1) Fur-thermore, the calculated DCp is highly negative and arises due to the loss of nonpolar SASA, whereas the accompanying uptake of water is associated with gain

of polar SASA This is because berberine is nonpolar

in nature, and hence the more nonpolar accessible sur-face gets buried upon interacting with the quadruplex

On the other hand, the positive change in the polar surface area (DAp) confirms that there is exposure of the polar surface on complex formation, further imply-ing the uptake of water molecules within the hydration layer

Conclusion

The single-stranded G-rich telomeric 3¢-overhang at the ends of chromosomes can form unique secondary DNA structures, such as G-quadruplexes, which are known to inhibit telomerase activity and have thus become attractive targets for new anticancer ligands However, a structure-based approach needs to be developed to design a new generation of binding agents that can selectively target such unique second-ary DNA structures Ultimately, a comprehensive understanding of the thermodynamic and structural parameters of quadruplex–ligand complexes would aid

in the design of new quadruplex-selective molecules and help to rationalize their in vivo performance In this study, we have obtained comprehensive data

on the thermodynamic and structural parameters involved in the quadruplex–ligand interaction, using a well-characterized human telomeric quadruplex and

an alkaloid, berberine It has been observed that bind-ing of berberine to the human telomeric quadruplex

is associated with a small, negative enthalpy (DH)

of )1.7 kcalÆmol)1, a favorable free energy (DG)

of )8.2 kcalÆmol)1 and a favorable entropy with TDS value of +6.5 kcalÆmol)1ÆK)1 at 25C A negative heat capacity change was observed when it was calcu-lated using two independent methods experimentally, from the temperature dependence of DH values, and theoretically based on surface area calculations The theoretical value is more negative than the experimen-tal value There was an uptake of 13 water molecules

on average per complex, which provides an unfavor-able contribution to the free energy of the binding Structural studies of the complex obtained from mole-cular dynamic studies reveal that berberine stacks over the G-tetrad, allowing overlap of the p-system of berberine primarily with two bases of each G-tetrad The partial positive charge on the berberine N7 appears to act as a ‘pseudo’ potassium ion, and is

positioned above the center of the G-tetrad in the

region of high negative charge density generated by

the carbonyl groups Extension of this study to other

known and well-established ligands, such as porphyrin

and telomestatin, will be reported in due course

Experimental procedures

Berberine chloride was obtained from Sigma and was used

without any further purification The 22-mer

oligonucleo-tide from the telomere end, d(AGGGTTAGGGTTAGG

GTTAGGG), was obtained from Sigma Genosys USA

Concentrations of oligonucleotide solutions were

deter-mined from the absorbance at 260 nm, using the molar

extinction coefficient for the G-rich strand, calculated by

extrapolation of tabulated values of the dimers and

[51] All other reagents were of analytical grade Milli Q

water was used throughout all the experiments All

experiments were performed in 50 mm MES buffer (pH 7.4)

specified

UV-visible and fluorescence spectroscopy

QuadruplexỜberberine binding constants were determined

by UV fluorescence, and Cary 400 (Varian) and

Fluoro-max 4 (Spex) instruments were used for UV and

fluores-cence titration experiments respectively A fixed berberine

concentration was titrated by increasing the quadruplex

concentration in 50 mm MES buffer (pH 7.4) containing

100 mm KCl Data were transformed into a Scatchard plot

ligand for those sites [52] Data were fitted to the McGheeỜ

von Hippel neighbor exclusion model

To determine the affinity of binding between berberine

and quadruplex, fluorescence experiments were carried out

and varying the quadruplex concentration (0Ờ5 lm) For

analysis of data, the observed fluorescence intensity was

considered as the sum of the weighted contributions from

free berberine and berberine bound to quadruplex:

where F is the observed fluorescence intensity at each

fluores-cence intensities of the initial and final states of titration,

observed in the Scatchard plot, it can be shown that:

ơL0a2b đơL0ợ ơQ ợ 1=Kbỡabợ ơQ Ử 0 đ6ỡ

concentration, and [Q] is the added quadruplex concentra-tion

From Eqns (4,5), it can be shown that:

DF Ử đDF max =2ơL0ỡ đơL0ợ ơQ ợ 1=K A ỡ 

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi đơL0ợ ơQ ợ 1=K A ỡ 2  4ơL0ơQ q

đ7ỡ

are the initial and subsequent fluorescence intensities of the berberine at 522 nm, upon quadruplex addition

ITC

ITC measurements were carried out on a VP-ITC titration calorimeter (MicroCal, Northampton, MA, USA) Before loading, the solutions were thoroughly degassed The refer-ence cell was filled with the respective degassed buffer The preformed quadruplex concentration (10 lm) was kept in the sample cell, and berberine (100 lm) in the same buffer was placed in a syringe of volume 300 lL The berberine solution was added sequentially in 10 lL aliquots (for a total of 25 injections, 20 s duration each) at 4 min intervals Sequential titrations were performed to ensure full occu-pancy of the binding sites by loading and titrating with the same ligand without removing the samples from the cell until the titration signal was essentially constant The heats

of dilution were determined in parallel experiments by injecting a berberine solution of the same concentration in the same buffer The respective heat of dilution is sub-tracted from the corresponding binding isotherm prior to curve fitting origin 5.0 software was used to fit the thermodynamic parameters to the heat profiles

Molecular modeling

Literature studies based on NMR data reveal that the 22-mer human telo22-meric sequence assumes multiple intercon-vertible conformations, comprising the parallel, antiparallel and hybrid-type G-quadruplexes, in the presence of potas-sium ions [53] However, no solution structure obtained in the presence of potassium ions is available for the 22-mer human telomeric sequence Furthermore, the telomeric quadruplex adopts a mixed hybrid conformation in the pres-ence of berberine (1 : 1 molar ratio of berberine: telomeric quadruplex), as shown in the CD spectra (supplementary Fig S5) We chose the hybrid-type NMR structure (Protein Data Bank ID 2hy9) [54] of the human telomeric quadru-plex, as shown in supplementary Fig S3 The 2hy9 is a 26-mer mixed-hybrid structure of the human telo26-meric sequence

in the presence of potassium ions The four adenine residues, two from each terminal end, were removed for comparison

purposes The initial 3D coordinate for berberine was

extracted from the crystal structure of the transcriptional

receptor QacR from the Protein Data Bank (1jum) [55] The

extracted structure was then optimized, and the partial

electro-static potential calculation in the antechamber module

of amber8 The remaining parameters for berberine were

taken from the GAFF forcefield in amber8 [57] The NMR

structures possess two external G-quartets, shown in

supple-mentary Fig S3, that can independently act as binding sites

for berberine stacking The optimized structure of berberine

was docked against the binding sites by defining the external

G-quartet as the active site using the suflexdock module

in sybyl 7.3 [58] This resulted in two quadruplex–

berberine complex models, referred to as MH1 and MH2

the one with the lowest docking energy was considered as

the starting structure for further simulation experiments

Molecular dynamics

Two potassium ions were manually placed in the central

channel between the G-quartet planes in the complex

mod-els The complex models were simulated using the amber8

[59] suite of programs with the Cornell et al all-atom force

field ff99 [60] The complexes were neutralized with

potas-sium ions, with the two inner ions retained inside The

sys-tems were then immersed in a periodic box of TIP3P water

model, which extended approximately 8 A˚ (in each

direc-tion) from the solute in a truncated octahedron unit cell

[61] Simulations were performed with periodic boundary

conditions, and the particle-mesh Ewald method was used

involving bonds to hydrogen atoms were constrained using

equilibration phase, which was 1 fs The direct-space cutoff

used was 10 A˚ Simulations were performed at a constant

temperature of 300 K The Langevin coupling with a

collision frequency of 1.0 was used for temperature

regula-tion [64] A constant pressure of 1 atm with isotropic

mole-cule-based scaling with a relaxation time of 1 ps was used

The equilibration step involves multiple optimization and

relaxation of the solvent and potassium ions in the bulk

solvent with the solute and the two inner potassium ions

fixed with restraints that include 500, 100, 250, 50, 100, 25,

sys-tem was heated from 0 to 300 K at constant volume, and

this was followed by equilibration for 25 ps at a constant

temperature of 300 K and a pressure of 1 atm The

produc-tion phase was started at this stage and continued for 5 ns

All simulations were performed in an SGI Altix 450 cluster

The conformations in the trajectories were collected at

intervals of 2 ps Trajectory analyses were done using the

SASA calculation

The SASA calculation was done using grasp 1.3 [65] The lowest-energy structure evolved during simulation was used for the SASA calculation of the complexes Surfaces for

defined as nonpolar, and the remaining hydrophilic atoms are defined as polar The grasp radii set was used in the calculation

Thermodynamics calculation

The free energy estimates were performed by the MMPBSA approach, where the total free energy of binding is expressed as the sum of the contributions from the gas phase and solvation energies plus an additional term for the solute entropic contribution This can be expressed in the following equation:

G¼ Egasþ Esolvþ TSsolute ð8Þ

corre-sponds to solute entropic effects The analysis was done for the last 2 ns trajectory of the complexes The snapshots for quadruplex and berberine were extracted from the complex trajectories at intervals of 10 ps This yielded 100 snapshots

in total All counterions (except the two spanning the cen-tral channel of the G-quartet planes) and water molecules were stripped out from the trajectory prior to the thermo-dynamic analysis The gas-phase energies of the solutes were calculated using the Cornell et al force field [60] with

no cutoff Solvation free energies were computed as the sum of polar and nonpolar contributions using a contin-uum solvent representation

The polar contribution was calculated with the pbsa program in amber8 The dielectric constants used for the solute and the surrounding solvent were 1 and 80, respec-tively The Cornell et al radii set was used to define atom-centered spheres for the solute atoms, and a probe radius of 1.4 A˚ was used for the solvent to define the dielectric boundary around the molecular surface A lat-tice spacing of two grid points per A˚ was used, and 1000 finite difference iterations were performed, excluding salt effect The nonpolar solvent contribution was estimated

cal-culation for solute entropic contribution was performed with the nmode module in amber8 The snapshots were minimized in the gas phase using the conjugate gradient method for 5000 steps, using a distance-dependent dielec-tric of 4r (r is the interatomic distance) and with a

gradient The frequencies of the vibrational modes were computed for these minimized structures at 300 K, using normal mode analysis methodology The thermodynamic