A Thermodynamic Perspective on Potential G‐Quadruplex Structures as Silencer Elements in the MYC Promoter

Abstract Multiple G‐tracts within the promoter region of the c‐myc oncogene may fold into various G‐quadruplexes with the recruitment of different tracts and guanosine residues for the G‐core assembly. Thermodynamic profiles for the folding of wild‐type and representative truncated as well as mutated sequences were extracted by comprehensive DSC experiments. The unique G‐quadruplex involving consecutive G‐tracts II–V with formation of two one‐nucleotide and one central two‐nucleotide propeller loop, previously proposed to be the biologically most relevant species, was found to be the most stable fold in terms of its Gibbs free energy of formation at ambient temperatures. Its stability derives from its short propeller loops but also from the favorable type of loop residues. Whereas quadruplex folds with long propeller loops are significantly disfavored, a snap‐back loop structure formed by incorporating a 3’‐terminal guanosine into the empty position of a tetrad seems highly competitive based on its thermodynamic stability. However, its destabilization by extending the 3’‐terminus questions the significance of such a species under in vivo conditions.


Introduction
Guanine(G)-rich sequences have the propensity to fold into non-canonical tetra-stranded structures called G-quadruplexes (G4s). These are formed by stacking of planar hydrogenbondedg uanineq uartets in the presence of cationsl ike Na + or K + that stabilize the stacked quartet arrangementb yt heir coordination within the central cavity of the G-core.D ue to an overrepresentation of potentialq uadruplex-forming sequences in promoter regions of various proto-oncogenes like c-myc, ckit,o rBcl-2,G 4s tructures within ac ellular environment are considered attractive targets for novel therapeutics. Thus, the c-myc oncogenec odes for ap rotein that maya ct as both transcriptional activator andr epressor,b eing involved in the regulation of variousg enes linked to proliferation and growth arrest. [1] Because overexpression of c-myc was found to be associated with aw ide range of human cancers, [2,3] its complex transcriptional regulation employing multiple promoters has been the subjecto fi ntense research over the past three decades. As ar esult, the so-called nuclease hypersensitivity element III 1 (NHE III 1 ), a2 7b ase pair sequence located upstream of the P1 promoter,w as identified as am ajor control element of c-myc expression. [4,5] Notably,d ouble-stranded NHE III 1 was found to be in equilibrium with an on-canonical quadruplex formed after duplex unwinding by intrastrand folding of its purine-rich noncoding strand. [6] Twod ifferent G-quadruplex structures for the single-stranded NHE III 1 were identifieda nd characterized based on DMS footprinting experiments. [7] Following initial assignments to an antiparallel basket and ac hair topology,c ompelling evidence of their folding into three-layered parallel quadruplexes with three propeller loops came from subsequent NMR studies on truncated and mutated sequences. [8] With its six guanine tracts composed of two to four consecutive Gr esidues, the wild-type 27mer MYC single strand may fold into multiple G-quadruplexs tructures with various loop architectures depending on the set of G-tracts recruited to form the G-quadruplex core (Table 1). Site-directed mutagenesis togetherw ith polymerase stop and luciferase reporter assaysi nt he absence and presence of the G-quadruplexb inding porphyrin TMPyP4 pointed to G-tracts II-V as being involvedi nab iologically relevant G4 repressore lement whose formation decreases c-myc expression at the RNA as well as the protein level. [7,9] Because the first G-tract was suggested to not be engaged in the transcriptional gene regulation, subsequent structurals tudies mostly focusedo nt runcateds equences only involving the four central G-tracts of MYC as models of the physiologically activeG -quadruplex( Figure 1A). [10] As a consequence of two stretches comprising four guanines, their foldingi nto ap arallel three-layeredG -quadruplex allows for the formation of four different loop isomers. [11,12] However,a quadruplex with two one-nucleotide (1-nt) and ac entral 2-nt propeller loop was suggested to predominatei nb uffer solu-tion. [8] Even adding more diversity,e xpansion of the truncated sequence to also include the 3'-terminal G 2 -tract VI of the wildtype MYC resulted in ad ifferent major G-quadruplex species that features as nap-backl oop and allows the 3'-terminal guanine base to fill an empty guaninep osition within the 3'-tetrad ( Figure 1B). [13] Due to the importance of the G-quadruplex structures formed by NHE-III 1 for understanding c-myc expression and for ar ational design of drugs targeting the MYC sequence, various efforts have been directed towards identifying and characterizing the physiologically most relevant quadruplex species. [11,14] However,m ajor quadruplexes formed under in vivo conditions will not only depend on their intrinsic thermodynamic stability but also on their folding kinetics and on the differential binding of multiple transcription factorsl ike NM23-H2 or nucleolin. [15,16] Such ac omplex situation requires an individual assessment of binding processes, G4 foldingk ineticsa sw ell as G4 thermodynamic stability. Previous studies have evaluated G4 thermodynamics within ac ontrolled in vitro environment by the determinationo fm elting temperatures and also by van 't Hoff analyses based on at wo-state transition, yet rigorous and systematic investigations on MYC-derived sequences are large-ly missing. We here report on ad etailed thermodynamic profiling of an extensive set of putative MYC quadruplex folds by microcalorimetric methods to allow for am odel-free extraction of thermodynamic parameters. The present studies not only focus on specific loop isomerso fasingle fold but also on Gquadruplex structuresw ith the recruitment of different sets of G-stretches. Althoughi nsufficient to fully describe G4 populations within the cellulare nvironment, such ac omprehensive thermodynamic profiling will add to our understanding of major G4 structures formed in vivo. On the other hand, knowing the folding thermodynamicso fi ndividual MYC sequences in detail gives valuable informationo nt he impact of loop sequence, loop positioning, and loop length on the G4 stability of parallel quadruplexes in general.

NHE III 1 sequence variants
For ad etailed evaluation of quadruplex stabilities, various Grich sequences derived from the purine-rich 27mer NHE III 1 single strand termed MYC were employed (Table 1). Excluding its GG 3'-terminus, the MYC sequence features five G-tracts with potentialf olding into five different G4 speciesd epending on the G-tractsf orming the G-quadruplex core. In addition, one of the guanines at either the 5'-or 3'-end of aG 4 -tract is excluded from aG -columni nathree-layeredq uadruplex. Given three G 4 -tracts in MYC,t his gives rise to 2 2 or 2 3 loop isomers for the five quadruplexes depending on their folding with participation of two or all three MYC G 4 -tracts, respectively.D isregardingt he terminal G 2 -tract VI and assuming exclusive formation of all-parallel G-quadruplexes it can easily be seen that there are 28 possible specieso ft hree-layeredG 4s.
Introducing mutationsw ithin Gt racts constitutes ac onvenient strategy to block formation of otherwise competing G4 structures. Thus, selectiveG !T/A substitutions may restrict foldingi nto quadruplexes with only one defined set of G-columns or even provide for exclusivef olding into as inglel oop isomer.H owever,t os erve as good mimics of the wild-type fold, trappedm utants are required to not noticeably affect the structure and thermodynamic stability when compared to the corresponding wild-type sequence. In fact, due to their position within loop regionsb ase substitutions are suggested to only have as mall impact on theG 4s tructure, but some blurring in stability should be considered when discussing sequences structurally trapped through mutations and truncations (see below). [17][18][19] The selection of truncated and/or substituted sequences is based on representatives that cover aw ide range of structural diversity to provide for extensive informationo ns tructure-stability relationships. For the sake of simplicity and consistency, sequences are nameda ccordingt ot he G-tractsn ot involved in tetrad formation( i.e.,t ract Ia nd VI in MYC-D1,6) and in keeping with the number of nucleotides within the propeller loops (i.e.,t wo 1-nt and ac entral 2-nt propeller loop in MYC-D1, 6 [13] and MYC-D3,6 [20] previously   Figure 1B).

Probing the G-quadruplextopology by their CD signatures
For assessing quadruplex thermal stabilities and driving forces of G4 folding, differential scanning calorimetric measurements (DSC) were performed. [21] In order to lower high melting temperatures for obtaining well-definedp ost-transitional baselines in the accessible temperature range as required for the extraction of reliable transition enthalpies, samples were dissolved in al ow-salt buffer with 10 mm potassium ions. Therefore, G4 foldingw as initially tested by ac omparison of CD spectra acquiredi nb oth 10 mm and 120 mm K + buffers ( Figure 2). CD spectra of all sequences in potassium buffer feature negative and positive amplitudes at about 243 and 263 nm, typical of parallel quadruplexes with exclusive homopolar tetrad stacking. Whereas ab road low-intensity shoulder between 280 and 300 nm distinguishes the snap-back loop quadruplexes MYC- with the involvement of G-tract VI from the regularp arallel species with uninterrupted G-tracts, an egative dip at 290 nm is noticeable in particular for MYC-D5,6, MYC-D1,6, and MYC TT -D1,6. Importantly,e xcept for minor variations in amplitude, CD signaturesw ere conserved at the differents alt concentrationsw ith no apparent change in G4 topology.
Probing the structural polymorphismo fMYC variants 1 HNMR spectrao fa ll MYC-derived sequences were acquired to evaluatet heir folding and to test for structuralh eterogeneities under the solution conditions employed (Figures 3a nd S1). Focusing on the Hoogsteen imino protons pectral region between 10.5 and 12.0 ppm, 12 resonancesa re generally expected for as ingle quadruplexw ith three G-tetrad layers, corresponding to 3 4h ydrogen-bonded guanine bases. Whereas the parent full-length MYC sequence shows extensive polymorphism with multiple G4 speciesa ss uggested by its crowded imino proton spectral region, the spectra of somem utated sequences are in line with as ingle fold without noticeable additionals pecies. These include not only MYC-D1,6[1.2.1], [10] MYC-D1,6[2.1.2], MYC-D1[1.3.1], [13] and MYC del -D1[1.1.1], [22] but also MYC-D3,6 with the formation of as ingle 1.6.1 loop isomer. [20] Other sequences exhibit coexistence of different folds, however,amajor species populated by ! 80 %p redominates over aminor species in mostcases. As traightforwardc omparison and interpretation of thermodynamic parameters for G4 formation is affected by the coexistence of multiple topologies with equilibriap ossibly changed by outer conditions. However,o ther serious problems for any thermodynamic analysism ay also result from sequences that partially form stable higher-order G4 structuresi n vitro depending on sequence, ion concentration, but also on the particular annealing procedure, as hasb een reported recently. [14,23,24] Thus,t he polymorphic MYC sequence was shown to have as trong tendency for intermolecular aggregation and with multimeric structures only melting at very high temperatures the formed fractiono fh igher-order speciesm ay remain unnoticed in G4 denaturation studies. Although thermal stabilities as measuredb ym elting temperatures remain unaffected, enthalpies and entropies of unfolding mayc onsequently be too low andcompromise an acceptable accuracy.
To assesst he potential formation of multimeric structures we also performedn on-denaturing polyacrylamide gel electrophoresis for all sequences followinga nnealing in a1 0mm potassium phosphate buffer used for NMR and DSC studies. Most of the sequences exhibit either as ingle or strongly predominating monomer band under the low-salt conditions, differing in migration according to their number of residues ( Figure S2). Yet, additional slower migrating bands for MYC-D5,6 and in particularf or MYC-D1,6 and MYC TT -D1,6 comprising about 30 % of the total in-lane intensity hint at putative dimer formation for these mutant sequences. However, with no significant intensity of broadened signals in the 1 HNMR spectra with their flat baseline (Figure 3), larger high-melting G4 associates like G-wire speciesc an mostly be excluded for the latter sequences but may gain in relevance with buffers of higheri onic strength.

Thermodynamic profiling of quadruplex formation by DSC
Representative examples of DSC melting profiles are shown in Figure 4. It should be noted that for some of the samples heating above 100 8Cr esulted in ag radual decrease of signal heights in subsequent heating cycles apparently due to partial heat-inducedd amage of the DNA. Therefore, only those ther-   [22] as well as from as eparate CD-based T m determination for MYC-D1,6[1.2.1] (data not shown). After proper baseline corrections, calorimetric enthalpies DH o cal of G4 folding were obtainedb yi ntegration of the meltingt ransition. Likewise, ensuring equilibrium conditions during thermal denaturation,m olare ntropies DS o at the melting temperature were calculated using standard thermodynamic relationships. These also enabled determination of the Gibbs free energy DG o at any reference temperature T from DH o and DS o when changes in molar heatc apacities DC p o are close to zero. [21,25] In fact, although quadruplex folding is expected to be associated with a DC p o ¼ 6 0, changes are generally too small to warrantt heir reliable extractionf rom the thermograms in case of the G4 transitions. [19] Average values from three independentm easurements on different samples are summarized in Table 2. Here, only model-free calorimetric transition enthalpies DH o cal and entropies DS o cal are included because many of the sequences employed may fold into multiple quadruplex species. Consequently,o bserved ratios between calorimetric and van 't Hoff enthalpies determined based on a two-state approximation will hardly yield direct information on (un)folding intermediates and/orcooperative units.

Dependence on concentration and type of cations
Due to the high thermals tability of some quadruplexes and the need for ad efined post-transitional baselinef or proper baselinec orrections, G4 melting was shifted to lower temperatures by using al ow-saltb uffer with 10 mm potassium phosphate, pH 7. To probe differential effects of cation concentration on thermodynamic profiles of the parallel quadruplexes, two representative G-rich sequences, namely MYC-D1,6[1.2.1] and MYC-D3,6, each forming aw ell-defined quadruplexs pecies were also characterized in ah igh-saltb uffer with 120 mm potassiump hosphate, pH 7( Ta ble 2, Figure S3). The significant rise in their thermals tability with a DT m of about 20 8Cc an be attributed to ah igher ionic strength of the buffer solution but also to the specific uptake of K + ions upon G4 folding, being coordinated within the centralG 4c avity as ap rerequisite for quadruplex stabilization. From at hermodynamic perspective, an elevated potassium ion concentrationresultsinamore exothermicf olding of the parallel quadruplex, yeti sa ccompanied by ah igher entropic penalty.T his pointst os tronger intramolecular interactions and ah igher rigidity similart oo bservations reported for aD NA four-wayj unction lacking more specific cation binding. [26] Because the DH8 contribution to the Gibbs free energy predominates even at highert emperatures, the overall thermodynamic stability of the quadruplexes will always increase in the accessible temperature range at higher K + concentrations. On the other hand, relative stabilities amongq uadruplexes are anticipatedt on ot be affected by changing K + concentrations. It shouldb en oted, however,t hat in addition to slower kinetics of (un)folding, absolute differences in enthalpies and Gibbs free energies between G4 species will generally show ag radual decrease with al owered K + concentration. [17] Thus, DDG8 293 between MYC-D1,6[1.2.1] and MYC-D3,6 amountst oÀ6.7 kcal mol À1 in 120 mm potassium phosphate buffer but only to À4.8 kcal mol À1 in 10 mm phosphate.
Ab etter stabilization of guanine quadruplexes with K + compared to Na + ions was showntoalmostequally depend on desolvation and the size of the alkali metal cation. [27] To also test the impact of substituting potassium for sodium ions on the  6. Apparently,r eplacing potassium by sodium ions destabilizes the parallelq uadruplexesbyl owering exothermicities of folding with only partial compensation through ar educed entropic penaltyi nc lose analogy to decreasing K + concentrations. On the other hand, destabilization of parallel but selective stabilization of antiparallel conformationsh as been reported by Na + . [28] Whereas both MYC-D1, 6 6 to an antiparallel or (3 + 1) hybridtype G4 as apparent from the typical CD signature ( Figure S4). Clearly,t his prohibits ad irect comparison of cation impact on the thermodynamic profile of the latter and sets limits to changes in solution conditions.

G4 variants with different G-tracts for G-core assembly
To characterize the thermodynamics of G4 folding for the MYC sequence when recruiting ad ifferent set of G-tracts,i ndividual tracts were either truncated if located at the termini or blocked by G!T/As ubstitutions, forcing them to loop out between adjacent G-columns. In the present study,G -tracts I, II, III, or V were individually blockedi na ddition to truncating the MYC 3'terminus with its G 2 -tract VI to give MYC-D1,6, MYC-D2,6, MYC-D3,6, and MYC-D5,6. Based on previous observations that addition of the two-nucleotide GG 3'-terminus will result in refolding of a MYC-D1p arallel quadruplex into aq uadruplex with an interrupted G-tract and as nap-back loop with positioningo f the 3'-terminal Gi nto the empty position of the outer G-core ( Figure 1B), folding of as equence MYC-D1(1.3.1)w as also characterized in more detail.B ecause all of these MYC-derived sequences carry G 4 tracts, they are expected to fold into different loop isomerst hroughr egister shifts. [19] It should be noted, however, that as ingle G!A/T substitution within ac entral Gtract of MYC-D2,6 and MYC-D1(1.3.1)w as introduced for better comparison, reducing the number of available folding pathways. Before discussing thermodynamic stabilities,i ti si nstructive to look at the imino protons pectralr egion of the variouss equences mimicking MYC G-tract isomers. Notably,i mino resonances for most folds suggest ar ather homogenous G4 population with only small amountso fc oexisting species ( Figure 3 (Table 2). To gether with the favored fold of MYC-D3,6 and MYC-D1(1.3.1), it can thus be assumed that G 4 tracts at 5'-and3 '-termini strongly favor loop isomers that shift the excessive Gt othe overhang rather than to an internal propeller loop. Accordingly,t he observed predominant G4 for MYC-D2,6 likely constitutes the 6-1-1 loop isomer.Onthe other hand, MYC-D5,6 exhibits am ore significant heterogeneityw ith an imino protons pectral region apparently composed of at least two major quadruplex folds.
Gibbs free energies of the five quadruplexes differing in the identityo ft heir G-columns vary considerably by 4kcal mol À1 at 310 Ki na10 mm K + buffer,a nd differences tend to even increasea t293 K. MYC-D1,6 and MYC-D1[1.3.1]f old into the most stable G4 structures, not only being close in their Gibbs free energies but also in their enthalpic and entropicc ontributions. On the other hand, MYC-D2,6 and especially MYC-D3,6 with their singlel ong loop revealarather low stability ( Figure 5). Clearly,t he latter two folds suffer from al ong propeller loop known to generally compromise quadruplex stability.W hereasl onger laterall oops have previously found to confer more exothermic heat to G4 folding through additional loop interactions and the uptake of counterions andw ater, [29] this does not apply to the present propeller loops.T hus, both MYC-D2,6 and MYC-D3,6 exhibit al ess negative molar transition enthalpy DH o cal but also al owered entropic penalty upon foldingw hen compared to the most stable MYC-D1,6a nd MYC-D1[1.3.1]. Noticeably,q uadruplexes derived from MYC-D5,6s harearather similars equence with MYC-D1,6 but have a significantly more positive Gibbs free energy by about3kcal mol À1 when compared to the latter (see below). Also, both favorable enthalpy and unfavorable entropy of folding are conspicuously smaller for MYC-D5,6 that was suggested to not be involved in thebiologically relevant MYC quadruplex.

Snap-back loop quadruplexes
Based on its thermodynamic stabilitya nd also suggestedb y previous NMR studies, [13] the snap-back loop quadruplexe ncompassing the 3'-terminal GG-tract may effectively compete with ar egular parallel quadruplex in case of the full-length MYC sequence. Consequently,s uch snap-back loop quadruplexes have been targeted by variousl igands with sometimes promising binding affinities. [13,22,30] To getm ore insighti nto the impact of additional modifications on the stability of the snapback loop structure, we shortened its two-tetrad bridging central propeller loop by deletiono ft wo nucleotides to give Apparently,l oop residues in this case may participate in additional hydrogen bond or stacking interactions, resulting in am ore favorable enthalpic but unfavorable entropic contribution. [29,31] Given negligible heat capacity effects (see above), the latterw ill generally result in ad ecreased thermal stability due to the strong temperatured ependence of the TDS o term in the Gibbs-Helmholtz equation.
Because the terminal3 '-G fills an emptyp osition to participate in the outer G-tetrad of the snap-backl oop structure, its compatibility with a3 '-extension seems critical for its formation within the nuclear hypersensitivity element IIII 1 .N MR structurals tudies have already shown that such an extended sequence preserves the same G4 topology with the 3'-added nucleotide pointinga way from the quadruplex G-core. [13] In line with the latter report, MYC-D1[1.3.1]T shares as imilar imino proton spectralr egion with MYC-D1[1.3.1]a nd MYC del -D1[1.1.1]a lbeit with small amountso faminor species, indicating the formation of am ajor snap-back loop quadruplex even within an extended sequence context (Figure 3). However, stabilities seem to be compromised with aG ibbs free energy of G4 folding declining to À2.9 kcal mol À1 at 310 K. Such destabilizing effects associated with an enhanced formation of other competitive speciesa re even more pronounced for MYC-D1[1.3.1]TT with a2 -nucleotide 3'-TT addition (Figure 3a nd Ta ble 2). Therefore, when solely based on thermodynamics these results raise doubts with respect to the relevance of such snap-back loop structures under in vivo conditions.

G-register isomers
The two loop isomers MYC-D1,6[1.2.1] and MYC-D1,6[2.1.2] recruit the same four G-tractso fMYC but incorporate different Gs of the two G 4 tracts into their G-quadruplex core. If blocking Tsubstitutions are removed, ad ynamic exchange between such G-register isomersc an occur through as liding mechanism without complete refolding, entropically stabilizing the folded state. [15,19] Onlyl ookinga tt he individual isomers, MYC-D1,6[1.2.1] has am elting temperature higher by 13 8Ca sc ompared to MYC-D1,6[2.1.2].T hisi se xpected based on the longer loop lengths of the latter with their known destabilizing effects in parallelG 4s. Also, ah igher stability of MYC-D1,6[1.2.1] is associated with am ore favorable enthalpy of foldingb ya bout À11 kcal mol À1 that is only partially counteracted by al arger entropicp enalty to give a DDG8 of %À2.5 kcal mol À1 at 310 K. It should be noted that differencesi nDH8 of À7a nd À16 kcal mol À1 between MYC-D1,6[1.2.1] and MYC-D1,6[2.1.2] in buffer solutionsc ontaining 7.5 and 20 mm K + have independently been determined previously based on at wo-state transition. [17,19] This is in good agreement with the present data considering larger differences upon increasing K + concentrations (see above and ref. [18]).

Impact of overhangand loop sequences
As ar esult of using MYC sequences adopting well-defined NMR-derived quadruplex structures with appropriate G!T/A substitutions combined with efforts to add flanking residues for preventing G4 aggregation, [32] mutual comparison of thermodynamic parameters among native and mutated sequences Clearly,anegligible or only small impact of Gm utationsi n trapped quadruplex structures constitutes ak ey assumption when assessing the thermodynamics of different G4 folds. Whereas moderate changes in quadruplexm eltingt emperatures have been reported for T/A substitutions in loops of MYC sequences, [18] only minor thermodynamic perturbations were suggested for thymidine and 2'-deoxyinosine substitutions based on ag lobal thermodynamic analysis. [19] Indeed, due to MYC-D1,6 predominantly folding into the MYC-D1,6[1.2.1] topology with ac entral 2-nt loop as shown by NMR (see above), the mostly identicalt hermodynamic parameters determined for foldingo ft hese two sequences confirmt hat the dual G!T mutations in MYC-D1,6[1.2.1] have only an egligible impact on the thermodynamicsofG 4f ormation( Ta ble2).
Based on these findings, significant differences in the thermodynamic profiles of MYC-D1,6 and MYC-D5,6 are unexpected. Both sequences feature four consecutive G-tracts G 3 -G 4 -G 3 -G 4 for MYC-D1,6 and G 4 -G 3 -G 4 -G 3 for MYC-D5,6 separated by single To rAnucleotides.C onsequently,f ormation of am ost favored MYC-D5,6[1.2.1] fold of similar stability to MYC-D1,6[1.2.1] may be anticipated. Apparently,t he differento verhang and/or loop residues must exert more significant effects on G4 stabilityf or theset wo G-tract variants. To search for the origin of the different thermodynamic stabilities, we substituted MYC-D1,6 in its overhang to form MYC TT -D1,6 with TT flanking residues precedinga nd followingt he 5' and 3' G-tract and to becomeapseudo-inverted MYC-D5,6 sequence with A$T exchange in the loops.I nterestingly,t hese changes had no effect on the G4 meltingt emperature but lowered the exothermich eat of folding by nearly 9kcal mol À1 counteracted by as maller amount of entropyl osses. Overall,t he amount of Gibbs free energy decreased by 1kcal mol À1 at 310 K. Such effects are easily rationalized by overhang sequences involved in specific interactions to act as stabilizing terminal caps of quadruplexes. [10,18] However,al arge part of the observed DDG8 310 difference of 2.8 kcal mol À1 between MYC-D1,6 and MYC-D5,6 and of the accompanied decrease in melting temperature by 10 8Cm ustb ea ttributed to the different loop residues (T-GA-T versus A-TG-A when only assuming 1.2.1 loop isomers). [18] These results emphasize that depending on the particular structural context overhang sequences and loop nucleotides may have more significant effects on the thermodynamics of folding. Also, only comparing meltingt emperatures may be misleadingi nn ot revealing significant changes in enthalpic and entropic contributionst hat determine relative stabilities at ambient temperatures.

Conclusions
The MYC sequence with its multiple G-tractsmay fold into various quadruplexesw ith potential relevance under physiological conditions. Whereas melting temperatures are convenientp arametersf or probing their thermodynamic stability, T m values are not directly linked to free enthalpies of G4 formation at a given temperature, requiring information on enthalpic and entropic contributions to the (un)folding event. Also, athermody-namic characterizationb ased on av an 't Hoff analysis of melting profiles is strictly based on at wo-state transition. This limits an evaluation of folding processesp roceeding through intermediates or to unresolved transitions of as equence with different coexisting quadruplex folds. To overcome these restrictions, ar igorous thermodynamic analysisu sing microcalorimetry is required to allow for am odel-free extraction of parameters.
With more than four consecutive G-tracts in the MYC sequence, those quadruplexes with al ooped-out internal tract are highly unfavorable unless specifically stabilized by potential interactions with ligands or proteins.O nt he other hand, two quadruplexes with av ery similars equence context, MYC-D1,6 and MYC-D5,6, considerably differ in their thermodynamic stability. MYC-D1,6, proposed to fold into the physiologically most relevant G4 and also being the mosts table G4 species, draws its energetic benefits to al arge extent from its favorable loop composition when compared to MYC-D5,6. Thus, recruiting excessive G-tracts in case of oxidative lesions may circumvent deleterious effects on G-quadruplex formation but is expected to yield al ess stable alternative fold. Also, competition by a snap-backl oop quadruplexi nvolving the 3'-terminal GG tract VI mustb em et with caution.A lthougha ss table as MYC-D1,6 when truncated, an extended 3'-end will allow,y et destabilize this particular topology.
The presented comprehensive thermodynamicp rofiling forms as olid basis for rationalizing stabilities of coexisting G4 structures that together with additional modulating interactions are expected to determine relevant populations within the cellular environment. Studying quadruplex structures and their relative populations in vivo remains ac hallengeb ut may be possible by the use of NMR in conjunction with stable isotope labeling. Also, detailed insights into the thermodynamics of different quadruplex folds will further contribute to our understanding of G4 stabilities in general and aid in the prediction of major folding topologies in sequences with multiple Gruns.

Experimental Section
Materials and sample preparation DNA oligonucleotides were purchased from TIB MOLBIOL (Berlin, Germany) and further purified by ethanol precipitation. Concentrations were determined spectrophotometrically by measuring absorbances in an H 2 Os olution at 80 8Cw ith molar extinction coefficients calculated by an earest neighbor model. [33,34] Prior to measurements, oligonucleotide solutions with concentrations as used for the subsequent experiments were annealed by heating to 85 8C for 5min followed by slow cooling to room temperature and storage in ar efrigerator overnight. For the experiments both al ow-salt buffer (10 mm potassium phosphate, pH 7) and ah igh-salt buffer (20 mm potassium phosphate, 100 mm KCl, pH 7) was used.

Circular dichroism (CD)
Spectra were acquired with aJ asco J-810 spectropolarimeter equipped with at hermoelectrically controlled cell holder.M easurements were performed with 1-cm quartz cuvettes at 293 Ko n 5 mm quadruplex in either al ow-salt or high-salt buffer.S pectra were recorded with ab andwidth of 1nm, ar esponse time of 1-2s,a nd as canning speed of 50 nm min À1 and finally blank-corrected.

NMR spectroscopy
NMR spectra were acquired on aB ruker Avance 600 MHz spectrometer equipped with an inverse 1 H/ 13 C/ 15 N/ 19 Fq uadruple resonance cryoprobehead and z-field gradients. Quadruplexes were dissolved in al ow-salt buffer with 10 mm potassium phosphate, pH 7.0. For solvent suppression on the samples in 90 %H 2 O/10 % D 2 OaWATERGATE with w5 element was employed. Data were processed using To pspin 4.0.6. Proton chemical shifts were referenced relative to TSP.

Differential scanning calorimetry (DSC)
DSC measurements were performed on aV P-DSC (Malvern Instruments, Great Britain) with 50 mm oligonucleotide in a1 0mm potassium phosphate buffer,p H7,u nless otherwise stated. Samples were heated from 20 to 100-110 8Cw ith as can rate of 0.5 8Cmin À1 . Equilibrium conditions were confirmed by as ingle experiment with as can rate of 0.25 8Cmin À1 ,y ielding at hermogram superimposable on the thermogram obtained with twice the scan rate. A buffer versus buffer scan was subtracted from the sample scan and cubic baselines were constructed for each transition. Melting temperatures T m and calorimetric enthalpies DH o cal corresponding to the maximum of the DSC peak and the area under the heat capacity versus temperature curve were obtained from the baseline-corrected profiles. DSC curves were fitted with an on-two-state model as implemented in the DSC analysis software. Here, the temperature dependence of the ratio of unfolded and folded population as given by the shape of the DSC thermogram is described by the van 't Hoff relationship with DH o vH ¼ 6 DH o cal .C hanges in heat capacity DC p 8 are too small for their reliable determination and were set to zero, consistent with negligible heat capacity effects upon quadruplex folding. [21] Reported thermodynamic parameters are average values with corresponding standard deviations from at least three independent experiments.