Investigation of double-stranded DNA/drug interaction by ESI/FT ICR: orientation of dissociations relates to stabilizing salt bridges

JOURNAL OF MASS SPECTROMETRYJ. Mass Spectrom. 2008; 43: 1531–1544Published online 02 June 2008 in Wiley InterScience(www.interscience.wiley.com) DOI: 10.1002/jms.1430

Investigation of double-stranded DNA/drug interactionby ESI/FT ICR: orientation of dissociations relatesto stabilizing salt bridges

Ying Xu,1,2 Carlos Afonso,1 Ren Wen2 and Jean-Claude Tabet1∗

1 Universite Pierre et Marie Curie-Paris 6, UMR 7613 Synthese, Structure et Fonction de Molecules Bioactives, Paris, F-75005, France2 School of Pharmacy, Fudan University, 138 Yi Xue Yuan Road, Shanghai 200032, China

Received 27 November 2007; Accepted 21 April 2008

Noncovalent complexes of DNA and Hoechst 33258 were investigated by ESI–FT/ICR MS in variousactivation modes (collision-induced dissociation (CID), sustained off-resonance irradiation collision-induced dissociation (SORI–CID), infrared multiphoton dissociation (IRMPD) and electron detachmentdissociation (EDD)). The binding selectivity of Hoechst 33258 was confirmed by the comparative studyof its noncovalent association with different DNA sequences. The CID spectra of [ds + HO − 5H]5−

obtained with a linear hexapole ion trap resulted in unzipping of the strands. This outcome is a clue to thedrug-binding mode, shading light on the localization of the binding sites of Hoechst 33258 to the DNAsequence. The IRMPD and SORI–CID experiments mainly gave DNA backbone cleavages and internalfragment ions. From this result, information on the localization of the binding sites of Hoechst 33258in the DNA sequence was obtained. No sodium cationization was observed on the DNA sequence ionsalthough they were present on fragmentation of the duplex, indicating that the backbone cleavages weregenerated from the single strand associated with the Hoechst 33258 where the number of alkali cation isrestricted. Under electron detachment (ED) conditions, multiple EDs were achieved for the [ds + HO −5H]5− ion without any significant dissociation. The presence of drug appears to enhance the stability of themultiply charged system. It was proposed that the studied noncovalent complex involved the formationof zwitterions and consequently strong salt-bridge interactions between DNA and drug. Copyright 2008John Wiley & Sons, Ltd.

KEYWORDS: FT/ICR; DNA/drug complex; zwitterion; salt bridge; IRMPD; EDD

INTRODUCTION

Mass spectrometry (MS) has become a powerful toolfor studying macromolecules, thanks to the developmentof ionization/desorption techniques. Electrospray ioniza-tion (ESI)1 and matrix-assisted laser desorption ionization(MALDI)2 allow ionization into the gas phase of large intactbiomolecules (such as peptides, proteins, oligonucleotides,and carbohydrates) for mass spectrometric analysis. In par-ticular, ESI–MS is adaptable for structural characterizationof intact macromolecules.3 Unlike other ionization meth-ods, ESI–MS can analyze solutions under gentle conditionsof pH, concentration, and temperature.4 ESI mass spectraof biopolymers, such as proteins and nucleic acids, exhibitseries of multiply charged macromolecules without yieldingextensive dissociations because of their low internal energytransfer during the desolvation step.5

In recent years, the use of MS as an analytical tool hasbeen extended to all stages of drug discovery,5 – 7 including

ŁCorrespondence to: Jean-Claude Tabet, Universite Pierre et MarieCurie-Paris6, UMR 7613 Synthese, Structure et Fonction deMolecules Bioactives, Paris, F-75005, France.E-mail: [email protected]

screening,8 target identification and characterization,9 earlydrug metabolisms,10 pharmacokinetics,11 noncovalent asso-ciation of drugs and their targets.8,12 – 14 Electrospray, asone of the soft ionization/desorption methods, is knownto preserve noncovalent associations during transferal fromsolution to gas phase. However, these complexes may bemultiply charged, which could result in coulomb repul-sions and conformation distortion. For instance, specificintermolecular interactions between DNA and DNA-bindingdrug possessing antitumor, antiviral, or antibacterial activ-ities have been thoroughly studied in the gas phase.8,15

The gentle nature of the ESI process allows maintenance,at least partially, the initial conformation of preformedcomplexes in solution. In addition, ESI–MS being a sen-sitive and specific technique requires a modicum of sampleto give accurate binding stoichiometry,16 selectivity, andsometimes binding affinity constant,17 and to elucidate thecomplex structure.18 – 20 Noncovalent drug/DNA complexeshas been investigated in the negative-ion mode as it gener-ates more intense signals due to the presence of numerousgas-phase acidic sites (i.e. phosphodiester groups) in theDNA backbone.21,22 Most studies have been carried out ona conventional instrument such as quadrupole ion trap that

Copyright 2008 John Wiley & Sons, Ltd.

1532 Y. Xu et al.

presents relatively limited mass resolving power and massaccuracy. For such investigations, Fourier transform massspectrometers (e.g. ion cyclotron resonance mode (FT/ICRMS)) combined with ESI are expanding. It constitutes anefficient analytical platform to investigate large biomolec-ular systems,23 such as oligonucleotides,24,25 duplexes,26 aswell as their noncovalent complexes with drugs.27,28 Sucha platform provides the benefits of ultra-high resolvingpower and accurate mass measurements as well as differention excitation modes. These techniques include collision-induced dissociation (CID) occurring either in the linearion trap by ‘on axis’ CID processes29 or in the ICR cell bysustained off-resonance irradiation collision-induced disso-ciation (SORI–CID).30 Ion activation can also be providedunder high vacuum conditions by infrared multiphoton dis-sociation (IRMPD)31 allowing further product ion excitationfor consecutive dissociations. More recently, electron cap-ture dissociation (ECD) for positive ion excitation,32 andelectron detachment dissociation (EDD)33 for negative-ionexcitation (achieved with the use of high kinetic energyelectron beam) have been combined with FT–ICR instru-ments. EDD generates odd-electron anion species after oneor multiple electron releases34 as also observed with elec-tron photodetachment.35 These odd-electron species couldbe unstable and thus, should be very useful for de novoDNA strand sequencing.36 However, the EDD processesapplied for studying gas-phase behavior of double-strandedDNA (ds) and ds/drug complexes have not been extensivelyinvestigated. These ion activation techniques should providecharacteristic series of product ions,37 – 40 possibly within low-abundance relatives to the precursor ion for EDD. On thisbasis, these ion activation methods have been employed tocharacterize the noncovalent complexes of nucleic acids anddrugs.41 – 43

Among the most studied DNA-binding drugs, Hoechst33 25822,44 – 47 has been investigated by various techniques.In the current work, this drug was selected to explorethe potential of ESI–FT/ICR as an investigative tool ofHoechst 33 258 noncovalent associations with different 12-mer oligonucleotide duplexes. In addition, the investigationof the mechanisms of dissociation of such complexes couldlead to better understanding of how the drug interacts witholigonucleotide duplexes. As IRMPD and SORI–CID arepresently the most used activation modes with FT/ICRinstrument, and both approaches follow the lowest energydissociation pathways, they are employed to evaluate thechemical properties of the noncovalent ds/drug complex(i.e. in terms of stability, dissociation orientation, andpossible binding sites). Finally, electron detachment (ED)(see Section ‘Experimental’) has been explored to obtainmore information on the stability of the ternary complexes(double-stranded DNA plus drug). This study highlightsthe interaction properties of DNA-binding drug towardvarious self-complementary DNA oligomers containingvarious numbers of G–C base pairs. On the basis of theseresults a model is proposed for binding sites of the drug ondsDNA.

EXPERIMENTAL

ChemicalsSingle-stranded oligodeoxynucleotides (ss) dodecamerswere purchased from DNA-technology (Denmark). Theyconsisted of three self-complementary oligonucleotides:d(GCGCGATCGCGC), d(GCGCAATTGCGC) and d(GCGAAATTTCGC), noted as žATž, žA2T2ž and žA3T3ž, respec-tively. Hoechst 33 258 (noted as HO) was purchased fromSigma-Aldrich (St. Louis, MO, USA).

Sample preparationSolutions of self-complementary oligonucleotide strandswere prepared in 1 M NH4OAc without further purification,and annealed at 80 °C and slowly cooled to room temperatureto allow duplex formation. Equimolar solutions of eacholigonucleotide and Hoechst 33 258 (stock solution 1 mg/mlin CHCl3/CH3OH 1 : 9) were prepared, and then diluted toobtain a final concentration of 5 pmol/µl in 50/50 (v/v)CH3OH/100 mM NH4OAc for MS analysis.

Mass spectrometryThe experiments were performed on a hybrid Qh-FT/ICR(ApexQe, Bruker Daltonics, Billerica, MA, USA) equippedwith an off-axis Apollo II ESI source operated in the negative-ion mode. The instrument is equipped with a 7 T activelyshielded superconducting magnet. Sample solutions wereinfused directly into the mass spectrometer by a syringepump at a flow rate of 2 µl/min. The capillary voltage wasset to �3.7 kV and the drying gas (N2) temperature to 150 °C.IRMPD was performed with a 25 W CO2 laser (Synrad,Mukilteo, WA, USA) operating at 10.6 µm. Photon irradiationenergy was set at 80% of full laser power. The irradiation timewas varied from 50 to 80 ms to evaluate the fragmentationorientation. Ions before being transferred to the ICR cellare accumulated in the hexapole collision cell operated asa linear ion trap during 0.5–1 s. The precursor ions wereselected in the quadrupole with a relatively large isolationm/z window (50 Th), and in some cases these ions of interestcan be interrogated further in the ICR cell with a narrowerwindow. For CID in the collision cell, precursor ions (withz charge(s)) were activated with a collision energy range of13 ð z eV to 20 ð z eV (Elab D z ð V). For SORI/CID, argonwas introduced into the ICR cell to raise the pressure up to10�6 mbar using a pulsed valve; the ions were then activatedusing a 600-Hz frequency offset for 250 ms with an excitationamplitude of 3.3 Vp�p. A pumping delay of 3–5 s was appliedprior to the ion analysis in order to reestablish a relativelysuitable vacuum in ICR cell (from 10�9 to 10�10 mbar range)insuring high-resolution conditions. For ED experiments, theED heater was set to 1.9 A, the bias voltage was 18.5 V, andthe lens electrode was kept at �19 V. The emitted electronsfrom the cathode were injected into the ICR cell for a durationof 1 s, as necessary to perform an efficient ED from selectedmultiply charged ions.34

All mass spectra and dissociation spectra were acquiredwith ApexControl (version 1.0, Bruker Daltonics) in thebroadband mode from m/z 200 to 3000. The transient wasamplified and digitized using 1 M data points (for full massspectra) or 512 k data points (for product ion spectra). The

Copyright 2008 John Wiley & Sons, Ltd. J. Mass Spectrom. 2008; 43: 1531–1544DOI: 10.1002/jms

DNA/drug interaction by ESI/FT ICR 1533

transient was zero filled once, apodized using the sine-squared function, and Fourier transformed. For SORI–CIDand IRMPD experiments, the presented spectra are obtainedfrom the average of 50 transient signals, whereas in the caseof EDD, 100 transient signals were averaged.

RESULTS AND DISCUSSION

As a well-known DNA-binding drug, Hoechst 33 258 pos-sesses both binding modes: minor groove binding and majorgroove intercalation.44 However, its primary mode is the minorgroove binding, with an affinity constant 50 times stronger.44

This type of drug can associate with A–T base pairs byhydrogen bonding in the minor groove of oligonucleotideduplexes, rather than the intercalation in major groove by�-stacking interactions.47,48 To explore the specificity of nonco-valent interaction of Hoechst 33 258 for DNA duplexes, threedifferent 12-mer oligonucleotides were selected with variousratios of A–T base pairs in the middle of nucleotide sequencessuch as d(GCGCGATCGCGC), d(GCGCAATTGCGC), andd(GCGAAATTTCGC).

ESI mass spectra of noncovalent ternary complexesUsing the FT/ICR instrument, mass resolving power andmass measurement accuracy are sufficiently high for unam-biguous ion identification as well as distinction of the[ds � 2nH]2n� and [ss � nH]n� isobaric ions. The negativeESI mass spectra of the HO/duplex complexes preformedin solution showed that the formation of 5-charged complexions was predominant under our experimental conditions(Fig. 1). For the mixture of žA3T3ž and Hoechst 33 258,the noncovalent [ds C HO � 5H]5� (m/z 1541.331) complexwas definitively the most favored ion (Fig. 1(a)). In addi-tion, low-abundance [ds C HO � 4H]4� (m/z 1926.928) and[ds C HO � 6H]6� (m/z 1284.274) complexes as well asthe free [ds � 4H]4� (m/z 1821.108) and [ds � 6H]6� (m/z1213.428) duplexes were detected. This trend is consistentwith the strong affinity of Hoechst 33 258 for the žA3T3žduplex. In the case of žA2T2ž/HO (result not reported),a similar mass spectrum was obtained. The žA2T2ž/HOnoncovalent complex was the base peak as observed withžA3T3ž, but the relative abundance of the ds/drug com-plex when compared to the free duplex was lower. In thecase of žATž/HO, the duplex [ds � 5H]5� (m/z 1457.286)was the base peak and the ternary [ds C HO � 5H]5� (m/z1542.129) complex was observed with a relatively low abun-dance (Fig. 1(b)). A similar trend was observed for the othernegative complex ions with different charge states. As aresult, the increase of noncovalent complex formation wasobserved with the rise of the A–T base pair number in theDNA sequence. It is consistent with the proposed bindingmode of Hoechst 33 258 mentioned previously, this drug israther a minor groove binder, which associates preferentiallywith A–T base pairs by hydrogen bonding.

It is an additional example, which shows that ESI–MSis an appropriate tool for the analysis of noncovalentcomplexes. It preserves, at least partially, the complexconformation during the desorption step from solution togas phase under soft desolvatation conditions. The binding

affinity and the stoichiometry determined by this techniquecan be used for DNA-binding drug selectivity screening.Specifically, it has been shown that Hoechst 33 258 iscapable of forming 1 : 2 complexes with duplex žA2T2ž.44

In the present experiments, with the high sensitivity andhigh dynamic range of the FT/ICR mass spectrometer,the noncovalent associations of two units as well as threeunits of Hoechst 33 258 with the duplex were detected. Interms of relative abundance, the tendency of the double-strand oligonucleotides to form the 1 : 2 complexes and1 : 3 complexes with the drug is in the order: žATž >žA2T2ž ³ žA3T3ž. This result is consistent with the sequence-dependent binding stoichiometry as known.44 The detectedcontribution of multibinding mode could indicate eitheran intercalation of Hoechst 33 258 in the G–C-rich regionof oligonucleotide sequence44 or a minor groove bindingprocess involving hydrogen and Van der Waals bonding.49

Interestingly, sodiated species are observed in a lower extentin the case of the triply charged ss/HO complex than forthe 5-charged ds/HO complex, independently of the DNAsequence. This trend could suggest that cationization takesplace in the desorbed free single strand rather than in thedesorbed ss/HO complex ions. Consequently, in the caseof the 5-charged ds/HO complex, the alkali cations are,very likely, mainly localized on the drug-free DNA strand.This relatively simple ESI–MS approach highlights thestoichiometry and the specificity of the interaction. Hence,it is indicative of the real interaction mode. Additionalinformation may be obtained from MS/MS experiments.In the following discussion, the reported MS/MS resultswere restricted to the study of one oligonucleotide sequence,žA3T3ž, which shows the best specificity resulting in thehighest ion abundance.

IRMPD of žAnTnž/HO complexesIRMPD, becoming one of the most useful tools for the struc-ture elucidation of biomolecules,50,51 has been employedto investigate the specific interactions involved in non-covalent complexes and the binding sites of high-affinityRNA ligands.52 The IR laser irradiation of the noncovalentžA3T3ž/HO complex was investigated within various irra-diation time but with a constant laser power (Fig. 2). Beingconsidered a ‘slow heating process’,53 the irradiation time isan important parameter, which influences the dissociationdegree of the precursor ion as well as the production ofconsecutive decompositions. Thus, the relative contributionof low m/z product ions increases with the rise of the irradi-ation time,54 which has been optimized and set to 60–70 ms.The ion excitation being nonselective to precursor ions, theinitially formed product ions are also submitted to IRMPDprocesses leading to extensive consecutive fragmentations.For these reasons, the excitation duration does not exceed70 ms in order to maintain the precursor ions as base peakin the IRMPD spectra, and also to avoid the formation ofconsecutive product ions with too many low m/z ions.

Using 60 ms of irradiation time (Fig. 2(a)), the IRMPDspectrum of the [ds C HO � 5H]5� complex (m/z 1541.270)constituted by žA3T3ž displays very low abundance productions. Two product ions are observed at m/z 1820.820 and


1534 Y. Xu et al.

Figure 1. Negative-ion ESI–FT/ICR mass spectra of equimolar mixtures of (a) žA3T3ž/HO and (b) žATž/HO at a concentration of5 µM in 50 : 50 H2O/MeOH (50 mM AcONH4). ds represents the double strand and ss the single strand.

m/z 1354.945 corresponding to the complementary single-stranded [ss � 2H]2� ion and [ss C HO � 3H]3� species.The [ss � 3H]3� ion is also observed at m/z 1213.542;however, its complementary [ss C HO � 2H]2� ion is notdetected, which is likely due to its prompt consecutivedissociations. Furthermore, other oligonucleotide sequenceproduct ions are observed. They are generated through theelimination of a nucleobase as a first step and cleavageof the neighbored phosphate ester bond generating thecharacteristic wn�

j (n D 1 and 2) and �ai � Bi�� typeions.55 Using a 70 ms irradiation time, extensive consecutivefragmentation occurs. The IRMPD spectrum (Fig. 2(b))displays several product ions corresponding to conventionalDNA backbone cleavages, which are reported in Table 1. Inparticular, singly charged �ai � Bi�� species55 (with i D 2,3 and 4, respectively at m/z 426.080, m/z 715.122, m/z1044.168) as well as the complementary wn�

j ions with

n D 1 (i.e. j D 1, 2) and n D 2 (i.e. j D 6, 7, 8, and 9)(Table 1) are observed. This suggests that these cleavages takeplace by consecutive activation of multiply charged singlestrands released by strand separation from the ds/drugcomplex. Similar behavior has been described by Wilsonand Brodbelt54 as well as by De Pauw et al.56 In addition,these sequence ions are further dissociated into a third iongeneration, in particular, to give rise to the formation ofsingly charged internal product ions (listed in Table 1), suchas �C2 : C2�� or �C10 : C10�� (m/z 466.039), �A5 : A6�� or(A4 : A5�� (m/z 803.103), �T7 : T9�� (m/z 1089.120), formedafter consecutive decompositions of �aiC1 � BiC1�� or of wj�1

�

(Scheme 1) not usually observed in such extent. They mustbe, for instance, from the �aiC1 � BiC1�� ion involving the baseelimination from the (i C 1) position, the internal product ionis produced through the release of a second nucleobase fromthe [i � �n C 1�] position associated with the corresponding



Figure 2. IRMPD spectra of [ds C HO � 5H]5� of žA3T3ž/HO mixture (5 µM in 50 : 50 100 mM AcONH4/MeOH) using 80% IR laserpower recorded with irradiation time of (a) 60 ms and (b) 70 ms.

phosphodiester bond cleavage, giving rise to the formationof a wjCn

�-like ion. Such internal product ions are called(Bi�n : Bi).

Finally, it should be noted that the fragmentationpathways observed from the [ds C HO � 5H]5� ion activatedby IRMPD are similar to those of the [ds � 5H]5� freeduplexes (results not reported herein). This indicates thatthe product ions related to the DNA backbone cleavagesare produced after consecutive dissociations of chargedsingle strands obtained from the dissociation of the [ds CHO � 5H]5� ternary complex. However, no indication allowsspecifying their origin by consecutive cleavages either from[ds � 5H]5� or [ss C HO � nH]n� (n D 2 or 3).

It is noteworthy that the free [ds � 5H]5� species pro-duced through the loss of neutral drug is not detected in theIRMPD spectra. Although abundant backbone cleavages ofoligonucleotide are observed, no product ion that remainedassociated with Hoechst 33 258 such as [ds C HO � wj]n� and

[ds C HO � �ai � Bi�]n� have been detected. Such a trend iscontrary to that observed from low-charge double-strandedDNA,37 PNA/DNA/PNA triplexes,16 and DNA/peptidecomplexes.57,58 For instance, in the case of negatively chargedPNA/DNA/PNA triplexes, product ions corresponding toPNA noncovalently linked to both the (ai � Bi) and wj DNAproduct ions have been detected.16 Therefore, noncovalentDNA/PNA binding has been preserved whereas cleavageof DNA backbone occurred, consistent with the existenceof strong noncovalent bonds between the strands. In thesame way, in the case of DNA duplexes with low-chargedstate covalent bond dissociation is observed rather thandouble-strand separation.37

Then, it can be hypothesized that the noncovalentassociations between Hoechst 33 258 and DNA duplexare very fragile in the case of the [ds C HO � 5H]5�

species. Unfortunately, the abundance of lower charge statecomplex ion is too weak to explore the charge effect


1536 Y. Xu et al.

Table 1. Ions list of IRMPD spectrum of [ds C OH � 5H]5� (m/z 1541) recorded with 70-ms irradiation time

m/z exp.aIi/I[dsCHO�5H]5�

(%) m/z theo. Assignment ChargeError(ppm)

306.04965 25.7 306.04966 w1 �1 0.03426.08206 1.5 426.08202 a2 � C �1 0.09466.04224 13.2 466.04221 C2 : C2/C10 : C10 �1 0.06481.04188 1.5 481.04188 T9 : T9 �1 0.00490.05338 12.3 490.05344 A4 : A4/A5 : A5 �1 0.12506.04838 9.6 506.04836 G3 : G3/G11 : G11 �1 0.04635.10227 17.0 635.10218 w2 �1 0.14715.12817 24.6 715.12840 a3 � G �1 0.32795.09469 4.7 795.09473 C2 : G3/C10 : G11 �1 0.05803.11091 7.6 803.11105 A4 : A5 �1 0.17819.10613 2.6 819.10597 G3 : A4 �1 0.20845.11553 1.8 845.11532 A6 : C10 �2 0.25917.64000 9.4 917.63969 w6 �2 0.341044.18067 4.7 1044.18091 a4 � A �1 0.231074.16911 5.3 1074.16850 w7 �2 0.571089.13342 11.1 1089.13396 T7 : T9 �1 0.501108.15228 2.3 1108.15234 C2 : A4 �1 0.051130.51491 2.9 1130.51351 ss � GH � C5H6O2

b �3 1.241163.18801 2.0 1163.19353 ss � GHb �3 4.751213.54222 9.9 1213.54334 ssb �3 0.921230.69758 3.8 1230.69730 w8 �2 0.231354.94514 1.8 1354.94372 ss C HOb �3 1.051378.17937 6.4 1378.18033 T7 : C10 �1 0.701387.22671 3.5 1387.22610 w9 �2 0.441402.19179 3.2 1402.18800 A6 : T9 �1 2.701541.29251 100.0 1541.29369 ds + HOb �5 0.771691.23961 1.8 1691.23794 A6 : C10 �1 0.991820.82049 2.3 1820.81864 ssb �2 1.02

a The listed m/z values consist of monoisotopic peaks.b The cationized forms were not taken into account.

on the covalent/noncovalent bond cleavage orientation.Indeed, species with lower charge state should yield moredirect covalent bond cleavages rather than noncovalentbond separations. As a result, the dissociation spectra of[ds C HO � 5H]5� and [ds � 5H]5� gave a similar trend inproduct ions (results not reported herein).

So, it is difficult to distinguish the binding site of theHoechst 33 258 in the DNA sequence using only IRMPD.However, the absence of drug loss leading to the DNAdouble strand as the product ion must indicate that the minorgroove binding is preferred to the intercalation. However,noncovalent systems involving intercalators, loss of the drugin neutral or charge form is generally observed.54,56 It must beenlightened that the wn�

j (n D 1 and 2) sequence is interruptedfrom w2

� to w62�. Similar behavior has been considered as

an indication of the binding site with the studied drugby Wilson and Brodbelt.54 Such hypothesis must be takencarefully because the absence of the Thy base release fromthe precursor ion could be due to drug binding or due to thelow gas-phase basicity of thymine59,60 as compared to Gua,Ade, and Cyt, which is unfavorable to its protonation andconsequently its elimination.

CID and SORI/CID of the žAnTnž/HO complexesFigure 3 shows the CID spectrum of the [ds C HO �5H]5� ion (m/z 1541.306) using the hexapole collision cellwith argon as target gas used as a linear ion trap forion accumulation. A very different behavior is observedin CID as compared to that obtained in the IRMPDexperiments. Indeed, conventional CID performed in thehexapole collision cell mainly gives the first generation ofproduct ions. They correspond to the complementary [ss �3H]3�/[ss C HO � 2H]2� and [ss � 2H]2�/[ss C HO � 3H]3�

pairs of product ions related to the noncovalent dissociationsof the ternary complex by strand separation according toScheme 2.

It should be noted that the abundance of single strandwith 3 or 2 negative charges is significantly larger than thatof the binary [ss C HO � nH]n� complexes (n D 2 and 3) asshown for each complementary pair of product ions (i.e.[ss � 3H]3�/[ss C HO � 2H]2� and [ss � 2H]2�/[ss C HO �3H]3�). The origin of these dissymmetrical signal intensitydistribution merits closer scrutiny. This particular behaviorcould be explained by consecutive dissociations involvingloss of the neutral HO drug from the [ss C HO � nH]n� (withn D 2 and 3) complexes contributing to the relatively high



Scheme 1. Formation of internal product �C2 : C2�� ion (m/z 466).

Figure 3. CID spectrum on the hexapole collision cell of [ds C HO � 5H]5� of žA3T3ž/HO mixture (5 µM in 50 : 50 100 mM

AcONH4/MeOH) using collision energy of 17 V ( : loss of guanine, : consecutive loss of 98).

abundance of the doubly charged and triply charged singlestrands. Furthermore, singly charged HO drug release couldalso occur consecutively from the [ss C HO � 3H]3� anionand thus enhance the relative abundance of the doublycharged single strand. The [HO � H]� ion (m/z 423.194) wasnot detected, but this could be due to poor transfer of thelow m/z ions under our experimental conditions.

Other very minor competitive dissociations can beobserved such as the direct neutral base release (closed starsin Fig. 3) that is specifically observed for guanine withinabout 10% of the precursor ion abundance. This guanine

neutral loss is also observed with each first generation ofproduct ion either in competition with the loss of the [�a1 �G� � H] neutral or followed by the consecutive neutral lossof 98.032 u, (noted as opened star in Fig. 3) i.e. correspondingto hydroxy methylene-2 furan (C5H6O2). The latter elementalstructure was already proposed by Marshall et al. fromECD/IRMPD experiments,38 who additionally proposedanother isobaric structure such as C3H4N3O (98.035 u) that isnot considered in our case. Indeed, such elemental structurecorresponds to a neutral radical species, which should bedifficult to produce under our CID conditions. In spite of the


1538 Y. Xu et al.

relatively weak accuracy reached in the 1100–2100 Th range,the more realistic interpretation is that the 50-end deoxyriboseis released concomitantly with the GH nucleobase either inone step or by a stepwise process. This double elimination isobserved from the single-strand [ss � nH]n� anion, the binary[ss C HO � nH]n� complexes as well as the ternary [ds CHO � 5H]5� complex. However, it is difficult to rule out thepossibility that these additional product ions are producedthrough consecutive decompositions of the [�ds C HO �5H� � GH]5� and [�ds C HO � 5H� � �GH C 98�]5� ions bysingle-strand anion release. Similar behavior concerning thedirect GH release has been observed from the CID spectra ofDNA duplexes presenting a lower charge state,37 as wellas from PNA/DNA and PNA/DNA/PNA,16 which areconsistent with our proposed interpretation.

Finally, the [ss � nH]n� product ions present a largeramount of satellite peaks (due to multiple HC/NaC

exchanges) than the binary [ss C HO � nH]n� complexes.Similar comment has been proposed for the interpretation ofthe negative-ion ESI mass spectra. Thus, confirming that thesodium ions are mainly retained on the ‘free of drug’ sin-gle strand, after dissociation of [ds C HO � 5H]5� complex,rather than at the single strand retaining the Hoechst 33 258drug.

Scheme 2. Competitive charge distribution on product ions ofthe [ds C HO � 5H]5� precursor ion generated into hexapolelinear ion trap. The solid lines are related to the major processand the dotted lines correspond to consecutive dissociationsas minor pathways.

However, a contradiction seems to arise when consider-ing the lower m/z range where very low-abundance productions can be detected, thanks to the high dynamic range ofthe FT–ICR mass spectrometer (i.e. the 500–1100 Th range).Indeed, the detected product ion series: w2

�, �a3 � G��, w62�,

�a4 � A��, and w72�, are not accompanied by satellite peak

corresponding to HC/NaC exchanges. This behavior is incon-sistent with the assumption that they are generated throughconsecutive dissociations of the single-strand product ions,which are actually multisodiated. Alternatively, the absenceof cationization in these �ai � Bi�� and wj

� product ions sug-gests that they are generated from consecutive dissociationof [ss C HO � nH]n� which present less alkali adducts thanthe direct free [ss � nH]n� single strand. Note that in this m/zrange, internal product ions are not detected in contrast towhat was observed in the IRMPD experiments. This confirmsthat these internal product ions are related to a third gener-ation. Consecutive dissociations are enhanced by raising thecollision energy. However, no additional information on thecomplete oligonucleotide sequence and on the drug-bindinglocalization is obtained.

The introduction of SORI/CID mode, consisting in off-resonance excitation, leads to several excitation–relaxationsequences allowing a step-by-step increase of the precursorion internal energy. This mode is more suitable for disso-ciation of multiply charged dsDNA/drug precursor ionsthan the CID processes occurring in the hexapole linearion trap. In order to perform these experiments, argon waspulsed into the ICR cell to induce multiple collisions. SinceSORI–CID is a resonant-like technique, the formed productions do not undergo further activation and will not giveextensive consecutive dissociations unless they accumulatesufficient internal energy by activation prior to the initialdissociation. Thus, the SORI–CID spectrum (Fig. 4) of thenoncovalent [ds C HO � 5H]5� complex, presents less prod-uct ions when compared to the IRMPD spectrum (Fig. 2(b)),but more than those observed from CID on the hexapolecollision cell (Fig. 4). Under these SORI–CID conditions, the

Figure 4. SORI–CID spectrum of [ds C HO � 5H]5� of the žA3T3ž/HO mixture (5 µM in 50 : 50 100 mM AcONH4/MeOH) using a600-Hz frequency offset and an excitation amplitude of 3.3 Vp�p.



[ds C HO � 5H]5� ions dissociated into [ss C HO � nH]n�

and [ss � nH]�5�n�� (n D 2 and 3) ions through strand separa-tion as observed in the collision cell (Fig. 3). With sufficientexcitation energy, consecutive single-strand backbone cleav-ages take place.

The majority of the structurally informative product ionsgenerated by this experiment consisted of wj

� and �ai � Bi��

type ions, as with IRMPD and CID in collision cell, withoutthe presence of sodium cation. As noted previously, theHoechst 33 258 separated readily from the [ss C HO � nH]n�

ions to generate consecutively the free single-strand anions,thus unfortunately losing binding site information. It isdifficult to get additional structural information, with theused methods, due to the weak strength of the noncovalentinteraction between Hoechst 33 258 and the DNA duplexesin the gas phase. However, the sequence disruption betweenthe w2

� and w62� product ion series confirms the previous

assertion that the drug-binding location is close to the[AATTTC] sequence as proposed by Wilson and Brodbelt.54

To conclude the investigation performed, under low-energy collision conditions, it should be pointed out thaton the hexapole collision cell, the ternary [ds C HO � 5H]5�

complexes led mainly to strand separation processes. Thisinvolves collisional activation and cooling with the argontarget gas under moderate pressure conditions. On theother hand, dissociation experiments carried out inside theICR cell led to extensive consecutive dissociations with theproduction of DNA backbone cleavages. Such consecutivedissociations are expected under IRMPD conditions asthe product ions can be activated by the laser. However,under SORI–CID condition, only the precursor ion isactivated slightly off resonance. Thus, a question shouldbe asked concerning the cause of the observed consecutivedissociation that could be related to the internal energytransferred as well as the target gas pressure during collisionactivation process.

ED of žAnTnž/HO complexesIn order to explore the noncovalent interaction propertiesof this particular ternary complex, ED dissociation hasbeen attempted using relatively high electron energy34

compared to the ECD mode. Unfortunately, decompositionrate constants were too weak to observe the resulting productions from both the [ss � nH]n� and [ds � nH]n� precursorions. As shown in the initial experiments described byZubarev33 and Hakansson,34 [ss � nH]n� are characterizedby a relative strong stability in EDD experiments. However,thanks to the large dynamic range of the FT/ICR analyzer,which allows strong signal amplification, it is possible todetect very low abundance product ions that can be relatedto the consecutive base loss or double-stranded productions.34 In our experiments, no fragmentation was observedfrom the [ds C HO � 5H]5� (Fig. 5(a)). Indeed, only one andtwo EDs take place as displayed in the ED spectrum, incontrast with single-strand anions in the previous worksof Hakansson34 and De Pauw (with photodetachment).35

Thus, the emerging product ions are [ds C HO � 5H]4�ž

(m/z 1926.678), [ds C HO � 5H]3�žž (m/z 2568.930) withoutany consecutive dissociation processes related to the odd-electron species dissociations or strand separations. The two

unpaired electrons must be dispersed in the oligonucleotidebackbone.

In order to enhance rate constant ion dissociations,consecutive ion activation by using the IRMPD mode afterthe application ED process34 has been used. The recordedED/IRMPD product ion spectrum of the ternary complex[ds C HO � 5H]4�ž displays an ion fingerprint that is almostsuperimposable (Fig. 6) to the IRMPD spectrum of theeven-electron [ds C HO � 4H]4� ion. Interestingly, the EDspectrum of ‘free of drug’ [ds � 5H]5� ion displays more EDsthan that of the ternary complex [ds C HO � 5H]5�. In thiscase, [ds � 5H]4�ž, [ds � 5H]3�žž, and [ds � 5H]2�žžž wereobserved in the same experimental conditions (Fig. 5(b)).The observation of third ED was not detected from the EDspectrum of [ds C HO � 5H]5�, indicating that the presenceof drug likely increased the local electron affinity of thestudied complex. In other words, this means that the presenceof drug stabilizes the multiply charged system.

PROPOSED ASSOCIATION MODEL

These experiments show that the ternary žA3T3ž/HOcomplex with five negative charges is relatively stable underlow-energy collision conditions. From both experimentalcollision conditions (i.e. hexapole collision cell and ICRcell), the generation of the [ds � nH]n� duplex ions hasalso been observed. This behavior, according to Wilson andBrodbelt,54 indicates that the Hoechst 33 258 interacts as aminor groove binder with the [50-GCGAAATTTCGC-30/50-CGCTTTAAAGCG-30] duplex in a 1 : 1 stoichiometry.

On the one hand, Wilson and Brodbelt54 clarified the roleof G–C base pair character on DNA backbone cleavage andon direct base release, because it enhances DNA interstrandhydrogen bonding. This behavior involved dipole–dipoleinteractions, and favored a proton transfer that is requiredfor nucleobase release. On the other hand, De Pauw et al.56

have shown that the gas-phase basicity of the drug must beconsidered in addition to the sequence dependence. Indeed,they showed that the increase in hydrogen bonding withthe drug raised the stability of the complex, especially thenumber of positive charges on the drug although the netcharge was negative. Wilson and Brodbelt54 emphasizedthat duplex/drug complexes containing a high proportionof nucleobase pairs with relatively low basicity, such asA–T61 (adenine 942.8 and thymine 880.9 kJ mol�1), mainlyresulted in strand separation for drugs binding to the minorgroove. In this case, the drug remains attached to one of thegenerated single-strand anions. This behavior was attributednot only to a higher oligonucleotide/drug association butalso to a weak duplex stability (low Tm value) as it presents arelatively high number of A–T base pairs. Thus, the stabilityof the drug-binding interaction depends on the A–T basepair number, and also on the number of positive chargesconsidered to be localized on the drug.

Conversely, as shown by a lot of studies, the increase ofG–C base pair number in the duplex (or triplex) sequencesresults in a larger amount of G base release (i.e. B or [B � H]�)depending upon the charge state of ds/drug complexes.Such guanine elimination initiates the sequential single-strand backbone dissociations leading to the formation of the


1540 Y. Xu et al.

Figure 5. ED spectra of (a) [ds C HO � 5H]5� of žA3T3ž/HO mixture and (b) [ds � 5H]5� of žA3T3ž solution (5 µM in 50 : 50 100 mM

AcONH4/MeOH).

complementary �ai � Bi�� and wj� product ions. This trend

is attributed to the relatively high basicity59 and acidity ofguanine nucleobase62 presenting, therefore, an amphotericcharacter.

Several studies considered that the base release canbe explained by proton migration from the neighboringdeoxyribose rather than from the very acid phosphategroup, which is more distant. Zwitterionic DNA formswere assumed to be generated during ion activation63 and,more recently, such a structure was considered to be formedfrom the multiply charged aggregate ions generated whendesolvation takes place.16 In latter studies, the presence ofzwitterions may explain the higher stability of duplex ortriplex anions presenting a low-charge number, such speciesleading favorably to covalent bond cleavages (base release).Indeed, protons located at DNA interstrand allow, in thesame time, double-strand stabilization as well as inducingnucleobase release under ion activation.

In the present investigation on the žA3T3ž/HO complex

presenting a relatively high charge state, which decomposescompetitively by either minor guanine loss or single-strandelimination, could be rationalized by considering double-strand stabilization by one or several protons located at theDNA interstrand. In addition, ss/drug stabilization can takeplace through the formation of one or several salt bridgesbetween the different partners, constituting the complex (i.e.phosphate group and protonated sites of drug).

The latter assertion implies that one or several protonsmust be present on the basic sites of the drug as demonstratedby De Pauw et al.56 Furthermore, to maintain the interactionwith the multiprotonated drug, several phosphodiestersites must be deprotonated for the net charge state toremain constant. Consequently, the following model canbe proposed herein (Scheme 3).

In this model, several protons stabilize the double-strandassociation through hydrogen bonding mainly betweenseveral G/C base pairs depending on the total chargeof the ternary complex. In addition, even if the bonding



Figure 6. (a) ED/IRMPD spectrum of [ds C HO � 5H]4�ž of žA3T3ž/HO mixture; using 80% IR laser power recorded with irradiationtime of 80 ms. (b) IRMPD spectrum of [ds C HO � 4H]4� using 80% IR laser power recorded with irradiation time of 60 ms. (Asteriskdenotes harmonic peaks and double asterisk denotes noise peaks).

Scheme 3. Model proposed to rationalize the associationin the ternary complexes constituted by [50-(GCG AAA TTTCGC)-30]2C Hoechst 33 258, based on hypothetical chargenumber and location. This involves existence of a zwitterionicform yielding salt-bridge formation.

positions are uncertain, stabilizing the drug association,several deprotonated diester phosphates interact with theprotonated drug, via the formation of salt-bridge systemsas it was shown in other studied systems.18 The charge

and alkali distributions on each strand are considered tobe produced either in solution or in droplets. During thedesolvation of the solvated aggregates, the charges aremaintained depending on the relative basicity/acidity of thesites that stay protonated or deprotonated. Consequently,the number of nucleotide residues presenting a zwitterionicstructure depends upon the total charge of the complex.Thus, a larger total charge number will reduce the formationof zwitterions.16

On the other hand, this model is supported by theexistence of one or more HC/NaC exchanges on the phos-phodiester groups of the DNA backbone. Indeed, the extentof this exchange limits the number of zwitterion residues,especially on the single strand that is not noncovalentlybound to Hoechst 33 258. This may explain why the binary[ss C HO � nH]n� (n D 2,3) presents a lower number ofsodium cations than the [ss � nH]n� product ion (inset of


1542 Y. Xu et al.

Fig. 4). In addition, this is consistent with our assump-tion that the decomposition of [ss C HO � nH]n� resultsmainly in an unstable [ss � nH]n� anion. This instabilityis attributed to the presence of protons on the nucleobasesof the [ss C HO � nH]n� product ion. This behavior resultsfrom the distribution of the alkali cations in the cation-ized [ds C HO � 5H]5� precursor ion. It is likely that thealkali cations are mainly retained on the single strand towhich no drug is attached and thus, after strands separa-tion, the excess of protons (balanced by negative chargesat the phosphates) are localized at the nucleobases of the[ss C HO � nH]n� product ion. These protons promote theconsecutive dissociation of the single strand and explain itsinstability.

CONCLUSION

In this study, the complexation of Hoechst 33 258 with DNAduplexes has been reinvestigated using FT/ICR instrumen-tation in order to explain the gas-phase dissociation ofsuch complexes. The binding mode is confirmed, as firstdemonstrated in the literature, by the comparative study ofits noncovalent association with different DNA sequences.In the FT/ICR mass spectra, it has been shown that thenoncovalent complex relative abundance increases with therise of the A–T base pair number in the DNA sequence.The high affinity of Hoechst 33 258 to the žA3T3ž duplexis consistent with its minor groove binder main attach-ment mode. Among different ion activation modes, theCID experiments of the noncovalent complex performedin the linear hexapole ion trap gives a relatively simple acti-vation spectrum. The decomposition of [ds C HO � 5H]5�

leads mainly to two dissociation pathways, giving drug-free single strand and their complementary complex ions(doubly or triply charged). So this result is actually a clueof the drug-binding mode. In addition, another fragmen-tation pathway was observed that involves the loss of theguanine, which is followed by a loss of 98.032 u. This lossof 98 u is attributed to hydroxyl methylene-2 furan neu-tral (C5H6O2) produced through consecutive dissociations.The IRMPD and SORI–CID experiments should make itpossible to obtain complementary information on the local-ization of the binding sites of Hoechst 33 258 in the DNAsequence. When the internal energy is sufficient, extensiveconsecutive fragmentations of the single strand can occur,leading to DNA backbone cleavages and internal productions. It should be noted that the sodium cationization wasnot observed for the DNA sequence ions, although it ispresent in the duplex as well as in the single strand. Thisindicates that the backbone cleavages are generated from thesingle strand associated with the Hoechst 33 258. It is sup-posed that the [ss C HO � nH]n� (n D 2, 3) could dissociatepromptly due to the presence of protonated bases, whichinduces the base loss, and consequently, the DNA backbonecleavage. The cationized single strand is relatively stable,which leads to the absence of sodium containing sequencefragments. Finally, with ED experiments, multiple EDs wereobtained for the 5-charged complex ion without any signifi-cant dissociation. The presence of drug appears to enhance

the stability of the multiply charged system. The formationof zwitterions was considered to explain the oligonucleotidebackbone cleavage interruption from ds/drug complex. Theexistence of such species may allow to explain, on one hand,the stabilization of the double strand by the presence ofprotons on the DNA interstrand and, on the other hand, theexistence of salt bridges that strongly stabilize the Hoechst33 258 complexation to a single strand.

AcknowledgementsThe authors gratefully acknowledge funding provided by the CNRSand the University Pierre & Marie Curie which supported this workand also the SM3P platform.

REFERENCES

1. Fenn JB, Mann M, Meng CK, Wong SF, Whitehouse CM.Electrospray ionization for mass spectrometry of largebiomolecules. Science 1989; 246: 64.

2. Karas M, Bachmann D, Bahr U, Hillenkamp F. Matrix- assistedultraviolet laser desorption of non-volatile compounds.International Journal of Mass Spectrometry and Ion Processes 1987;78: 53.

3. Cristoni S, Bernardi LR. Development of new methodologiesfor the mass spectrometry study of bioorganic macromolecules.Mass Spectrometry Reviews 2003; 22: 369.

4. Loo JA. Studying noncovalent protein complexes byelectrospray ionization mass spectrometry. Mass SpectrometryReviews 1997; 16: 1.

5. Woods AS. The mighty arginine, the stable quaternary amines,the powerful aromatics and the aggressive phosphate: their rolein the noncovalent minuet. Journal of Proteome Research 2004; 3:478.

6. Woods AS, Kaminski R, Oz M, Wang Y, Hauser K, Goody R,Wang HYJ, Jackson SN, Zeitz P, Zeitz KP, Zolkowska D,Schepers R, Nold M, Danielson J, Graeslund A, Vukojevic V,Bakalkin G, Basbaum A, Shippenberg T. Decoy peptides thatbind dynorphin noncovalently prevent NMDA receptor-mediated neurotoxicity. Journal of Proteome Research 2006; 5:1017.

7. Lee MS, Kerns EH. LC/MS applications in drug development.Mass Spectrometry Reviews 1999; 18: 187.

8. Hofstadler SA, Griffey RH. Analysis of noncovalent complexesof DNA and RNA by mass spectrometry. Chemical Reviews 2001;101: 377.

9. Prakash C, Shaffer CL, Nedderman A. Analytical strategies foridentifying drug metabolites. Mass Spectrometry Reviews 2007;26: 340.

10. Tozuka Z, Kaneko H, Shiraga T, Mitani Y, Beppu M, Terashita S,Kawamura A, Kagayama A. Strategy for structural elucidationof drugs and drug metabolites using (MS)n fragmentation in anelectrospray ion trap. Journal of Mass Spectrometry 2003; 38: 793.

11. Marzo A, Bo LD. Tandem mass spectrometry (LC-MS-MS): apredominant role in bioassays for pharmacokinetic studies.Arzneimittelforschung 2007; 57: 122.

12. Woods AS, Moyer SC, Wang HY, Wise RA. Interaction ofchlorisondamine with the neuronal nicotinic acetylcholinereceptor. Journal of Proteome Research 2003; 2: 207.

13. Woods AS, Fuhrer K, Gonin M, Egan T, Ugarov M, Gillig KJ,Schultz JA. AngiotensinII/acetylcholine non-covalent com-plexes analyzed with MALDI-ion mobility-TOFMS. Journal ofBiomolecular Techniques 2003; 14: 1.

14. Hofstadler SA, Sannes-Lowery KA. Applications of ESI-MS indrug discovery: interrogation of noncovalent complexes. NatureReviews Drug Discovery 2006; 5: 585.

15. Woods AS, Koomen J, Ruotolo B, Gillig KJ, Russell DH,Fuhrer K, Gonin M, Egan T, Schultz JA. A Study of peptide-peptide interactions using MALDI ion mobility o-TOF and



ESI-TOF mass spectrometry. Journal of the American Society forMass Spectrometry 2002; 13: 166.

16. Delvolve A, Tabet J-C, Bregant S, Afonso C, Burlina F,Fournier F. Charge dependent behavior of PNA/DNA/PNAtriplexes in the gas phase. Journal of Mass Spectrometry 2006; 41:1498.

17. Powell KD, Ghaemmaghami S, Wang MZ, Ma L, Oas TG,Fitzgerald MC. General mass spectrometry-based assay for thequantitation of protein-ligand binding interactions in solution.Journal of the American Chemical Society 2002; 124: 10 256.

18. Woods AS, Ferre S. The amazing stability of the arginine-phosphate electrostatic interaction. Journal of Proteome Research2005; 4: 1397.

19. Jackson SN, Wang HYJ, Yergey A, Woods AS. Phosphatestabilization of intermolecular interactions. Journal of ProteomeResearch 2006; 5: 122.

20. Jackson SN, Wang HYJ, Woods AS. A study of the fragmentationpatterns of the phosphate-arginine noncovalent bond. Journal ofProteome Research 2005; 4: 2360.

21. Oehlers L, Mazzitelli CL, Brodbelt JS, Rodriguez M, Kerwin S.Evaluation of complexes of DNA duplexes and novelbenzoxazoles or benzimidazoles by electrospray ionization massspectrometry. Journal of the American Society for Mass Spectrometry2004; 15: 1593.

22. Gupta R, Beck JL, Ralph SF, Sheil MM, Aldrich-Wright JR.Comparison of the binding stoichiometries of positivelycharged DNA-Binding drugs using positive and negativeion electrospray ionization mass spectrometry. Journal of theAmerican Society for Mass Spectrometry 2004; 15: 1382.

23. Hendrickson CL, Emmett MR, Marshall AG. Electrosprayionization Fourier transform ion cyclotron resonance massspectrometry. Annual Review of Physical Chemistry 1999; 50: 517.

24. Hofstadler SA, Sannes-Lowery KA, Hannis JC. Analysis ofnucleic acids by FTICR MS. Mass Spectrometry Reviews 2005;24: 265.

25. Muddiman DC, Smith RD. Sequencing and characterizationof larger oligonucleotides by electrospray ionization Fouriertransform ion cyclotron resonance mass spectrometry. Reviewsin Analytical Chemistry 1998; 17: 1.

26. Wunschel DS, Muddiman DC, Smith RD. Mass spectrometriccharacterization of DNA for molecular biological applications:advances using MALDI and ESI. Advances in Mass Spectrometry1998; 14: 377.

27. Sannes-Lowery KA, Cummins LL, Chen S, Drader JJ, Hofs-tadler SA. High throughput drug discovery with ESI – FTICR.International Journal of Mass Spectrometry 2004; 238: 197.

28. Dong XC, Xu Y, Afonso C, Jiang WQ, Laronze JY, Wen R,Tabet J-C. Non-covalent complexes between bis-ˇ-carbolinesand double-stranded DNA: a study by electrospray ionizationFT-ICR mass spectrometry (I). Bioorganic and Medicinal ChemistryLetters 2007; 17: 2549.

29. Cody RB, Burnier RC, Freiser BS. Collision-induced dissociationwith Fourier transform mass spectrometry. Analytical Chemistry1982; 54: 96.

30. Gauthier JW, Trautman TR, Jacobson DB. Sustained off-resonance irradiation for CAD involving CAD technique thatemulates infrared multiphoton dissociation. Analytica ChimicaActa 1991; 246: 211.

31. Woodin RL, Bomse DS, Beauchamp JL. Multiphoton dissocia-tion of molecules with low power continuous wave infraredlaser radiation. Journal of the American Society for Mass Spectrome-try 1978; 100: 248.

32. Zubarev RA, Kelleher NL, McLafferty FW. Electron capturedissociation of multiply charged protein cations. A nonergodicprocess. Journal of the American Society for Mass Spectrometry 1998;120: 3265.

33. Budnik BA, Haselmann KF, Zubarev RA. Electron detachmentdissociation of peptide di-anions: an electron-holerecombination phenomenon. Chemical Physics Letters 2001; 342:299.

34. Mo JJ, Hakansson K. Characterization of nucleic acid higherorder structure by high-resolution tandem mass spectrometry.Analytical and Bioanalytical Chemistry 2006; 386: 675.

35. Gabelica V, Tabarin T, Antoine R, Rosu F, Compagnon I,Broyer M, De Pauw E, Dugourd P. Electron photodetachmentdissociation of DNA polyanions in a quadrupole ion trap massspectrometer. Analytical Chemistry 2006; 78: 6564.

36. Gabelica V, Rosu F, Tabarin T, Kinet C, Antoine R, Broyer M, DePauw E, Dugourd P. Base-dependent electron photodetachmentfrom negatively charged DNA strands upon 260-nm laserirradiation. Journal of the American Chemical Society 2007; 129:4706.

37. Gabelica V, De Pauw E. Comparison of the collision-induceddissociation of duplex DNA at different collision regimes:evidence for a multistep dissociation mechanism. Journal of theAmerican Society for Mass Spectrometry 2002; 13: 91.

38. Hakansson K, Hudgins RR, Marshall AG, O’Hair RAJ. Electroncapture dissociation and infrared multiphoton dissociation ofoligodeoxynucleotide dications. Journal of the American Societyfor Mass Spectrometry 2003; 14: 23.

39. Sannes-Lowery KA, Hofstadler SA. Sequence confirmation ofmodified oligonucleotides using IRMPD in the external ionreservoir of an electrospray ionization fourier transform ioncyclotron mass spectrometer. Journal of the American Society forMass Spectrometry 2003; 14: 825.

40. Kellersberger KA, Yu E, Kruppa GH, Young MM, Fabris D. Top-down characterization of nucleic acids modified by structuralprobes using high-resolution tandem mass spectrometry andautomated data interpretation. Analytical Chemistry 2004; 76:2438.

41. Iannitti-Tito P, Weimann A, Wickham G, Sheil MM. Structuralanalysis of drug-DNA adducts by tandem mass spectrometry.Analyst (Cambridge, United Kingdom) 2000; 125: 627.

42. Iannitti P, Sheil MM, Wickham G. High sensitivity andfragmentation specificity in the analysis of drug-DNA adductsby electrospray tandem mass spectrometry. Journal of theAmerican Chemical Society 1997; 119: 1490.

43. Cummins LL, Chen S, Blyn LB, Sannes-Lowery KA, Drader JJ,Griffey RH, Hofstadler SA. Multitarget affinity/specificityscreening of natural products: finding and characterizinghigh-affinity ligands from complex mixtures by using high-performance mass spectrometry. Journal of Natural Products 2003;66: 1186.

44. Bailly C, Colson P, Henichart JP, Houssier C. The differentbinding modes of Hoechst 33 258 to DNA studied by electriclinear dichroism. Nucleic Acids Research 1993; 21: 3705.

45. Schlosser G, Vekey K, Malorni A, Pocsfalvi G. Combination ofsolid-phase affinity capture on magnetic beads and massspectrometry to study non-covalent interactions: example ofminor groove binding drugs. Rapid Communications in MassSpectrometry 2005; 19: 3307.

46. Rosu F, Gabelica V, Houssier C, De Pauw E. Determination ofaffinity, stoichiometry and sequence selectivity of minor groovebinder complexes with double-stranded oligodeoxynucleotidesby electrospray ionization mass spectrometry. Nucleic AcidsResearch 2002; 30: e82/1.

47. Wan KX, Shibue T, Gross ML. Non-covalent complexes betweenDNA-binding drugs and double-stranded oligodeoxynu-cleotides: a study by ESI ion-trap mass spectrometry. Journalof the American Chemical Society 2000; 122: 300.

48. Gabelica V, De Pauw E, Rosu F. Interaction between antitumordrugs and a double-stranded oligonucleotide studied byelectrospray ionization mass spectrometry. Journal of MassSpectrometry 1999; 34: 1328.

49. Kopka ML, Yoon C, Goodsell D, Pjura P, Dickerson RE. Themolecular origin of DNA-drug specificity in netropsin anddistamycin. Proceedings of the National Academy of Sciences ofthe United States of America 1985; 82: 1376.

50. Little DP, Speir JP, Senko MW, O’Connor PB, McLafferty FW.Infrared multiphoton dissociation of large multiply charged


1544 Y. Xu et al.

ions for biomolecule sequencing. Analytical Chemistry 1994; 66:2809.

51. Little DP, Aaserud DJ, Valaskovic GA, McLafferty FW. Sequenceinformation from 42-108-mer DNAs (Complete for a 50-mer) bytandem mass spectrometry. Journal of the American ChemicalSociety 1996; 118: 9352.

52. Griffey RH, Hofstadler SA, Sannes-Lowery KA, Ecker DJ,Crooke ST. Determinants of aminoglycoside-binding specificityfor rRNA by using mass spectrometry. Proceedings of the NationalAcademy of Sciences of the United States of America 1999; 96: 10 129.

53. Marshall AG, Hendrickson CL, Jackson GS. Fourier transformion cyclotron resonance mass spectrometry: a primer. MassSpectrometry Reviews 1998; 17: 1.

54. Wilson JJ, Brodbelt JS. Infrared multiphoton dissociation ofduplex DNA/Drug complexes in a quadrupole ion trap.Analytical Chemistry 2007; 79: 2067.

55. McLuckey SA, Van Berkel GJ, Glish GL. Tandem massspectrometry of small, multiply charged oligonucleotides.Journal of the American Society for Mass Spectrometry 1992; 3:60.

56. Rosu F, Pirotte S, De Pauw E, Gabelica V. Positive and negativeion mode ESI-MS and MS/MS for studying drug-DNAcomplexes. International Journal of Mass Spectrometry 2006; 253:156.

57. Alves S, Woods A, Tabet JC. Charge state effect on the zwitterioninfluence on stability of non-covalent interaction of single-stranded DNA with peptides. Journal of Mass Spectrometry 2007;42: 1613.

58. Terrier P, Tortajada J, Buchmann W. A study of noncovalentcomplexes involving single-stranded DNA and polybasiccompounds using nanospray mass spectrometry. Journal of theAmerican Society for Mass Spectrometry 2007; 18: 346.

59. Greco F, Liguori A, Sindona G, Uccella N. Gas-phase protonaffinity of deoxyribonucleosides and related nucleobases byfast atom bombardment tandem mass spectrometry. Journal ofthe American Chemical Society 1990; 112: 9092.

60. Rodgers MT, Campbell S, Marzluff EM, Beauchamp JL. Site-specific protonation directs low-energy dissociation pathwaysof dinucleotides in the gas phase. International Journal of MassSpectrometry and Ion Processes 1995; 148: 1.

61. Hunter EP, Lias SG. Evaluated gas phase basicities and protonaffinities of molecules: an update. Journal of Physical and ChemicalReference Data 1998; 27: 413.

62. Gross J, Leisner A, Hillenkamp F, Hahner S, Karas M, Schafer J,Lutzenkirchen F, Nordhoff E. Investigations of the metastabledecay of DNA under ultraviolet matrix-assisted laserdesorption/ionization conditions with post-source-decayanalysis and hydrogen/deuterium exchange. Journal of theAmerican Society for Mass Spectrometry 1998; 9: 866.

63. Wan KX, Gross J, Hillenkamp F, Gross ML. Fragmentationmechanisms of oligodeoxynucleotides studied by H/D exchangeand electrospray ionization tandem mass spectrometry. Journalof the American Society for Mass Spectrometry 2001; 12: 193.


Documents

Investigation of double-stranded DNA/drug interaction by ESI/FT ICR: orientation of dissociations relates to stabilizing salt bridges