Synthesis, Stability, andProtonation Studies of a Self-Complementary DodecamerContaining the ModifiedNucleoside 2�-Deoxyzebularine

M. Vives1

R. Eritja2

R. Tauler1

V. E. Marquez3

R. Gargallo1

1 Department of AnalyticalChemistry,

University of Barcelona,Martı i Franques, 1-11,

E-08028 Barcelona, Spain

2 Institut de BiologıaMolecular de Barcelona-CSIC,

Jordi Girona 18-26,E-08034 Barcelona, Spain

3 Laboratory of MedicinalChemistry,

Center for Cancer Research,National Cancer Institute,

Frederick, MD 21702

Received 7 April 2003;accepted 24 July 2003

Abstract: The nucleoside 2�-deoxyzebularine (K) was incorporated into the self-complementarydodecamer 5�-CGTACGKGTACG-3� by solid-phase 2-cyanoethylphosphoramidite chemistry usingdimethoxytrityl (DMT) as the 5�-hydroxyl protecting group. Standard synthesis cycles using tri-chloroacetic acid and short ammonia treatment (50°C for 30 min) were found to be the optimalconditions to obtain the desired dodecamer with minimum acid and basic degradation of the acid-and base-sensitive 2-pyrimidinone residue. The protonation equilibria of the K nucleoside and of thedodecamer at 37°C were studied by means of spectroscopically monitored titrations. For the Knucleoside, a pKa value of 3.13 � 0.09 was obtained. For the dodecamer, four acid-base specieswere found in the pH range 2–12, with pKa values of 9.60 � 0.07, 4.46 � 0.16, and 2.87 � 0.19.Melting experiments were carried out to confirm the proposed acid-base concentration profiles.Finally, kinetic experiments were also carried out at several pH values to evaluate the stability ofthe K nucleoside and of the dodecamer. An increased stability was shown by the K nucleoside when

Correspondence to: R. Gargallo; email: raimon@apolo.qui.ub.es

Contract grant sponsor: Ministerio de Educacion y Cultura;Contract grant number: MCYT BQ42000-0788

Contract grant sponsor: Generalitat de Catalunya; Contractgrant number: 2001SGR00056.

Contract grant Sponsor: University of Barcelona (M.V.)Biopolymers, Vol. 73, 27–43 (2004)© 2003 Wiley Periodicals, Inc.

27

incorporated into the dodecamer. Multivariate methods based on both hard- and soft-modeling wereapplied for the analysis of spectroscopic data, allowing the estimation of concentration profiles andpure spectra. © 2003 Wiley Periodicals, Inc. Biopolymers 73: 27–43, 2004

Keywords: 2�-deoxyzebularine; spectroscopy; DNA structure; Multivariate Curve Resolution;factor analysis

INTRODUCTION

The incorporation of modified bases into oligonucle-otides is a useful approach for the study of recognitionprocesses between proteins and nucleic acids. Theseprocesses take place between functional groups of theprotein amino acid chains and the nucleobases, andthe specificity of these interactions depends on theprotein’s ability to recognize a characteristic nucleicacid sequence. The present article reports the substi-tution of 2�-deoxycytidine for 2�-deoxyzebularine(K), which chemically corresponds to the deletion ofthe exocyclic amino group from cytosine. Both 2�-deoxyzebularine (1-�-D-deoxyribofuranosyl-1,2-di-hydropyrimidin-2-one) and the corresponding ribo-side were identified as comparable inhibitors of cyti-dine deaminase, but only the riboside (zebularine)showed cellular cytotoxicity against L1210 leukemiacells in vitro and important antitumor activity inL1210-bearing mice.1,2 Because of the fluorescentnature of the aglycon, modified oligonucleotideswhere key cytosines have been replaced with 1,2-dihydropyrimidin-2-one have been synthesized to ex-plore the role of ribozyme function.3 Additionally,based on the reduced stability of the glycosyl bondunder mildly acidic conditions (pH 3), this substitu-tion provides a useful approach to the generation ofapurinic/apyrimidinic (abasic) sites.4 The generationof abasic sites is a common form of DNA damage.Contrary to its behavior under acidic conditions,based-catalyzed hydrolysis causes degradation of the1,2-dihydropyrimidin-2-one ring leading to the gen-eration of the known mutagen, 1,3-propanedialdehyde(malondialdehyde, MDA).5,6 MDA has been shown toreact rapidly at the amino group of amino acids andnucleobases, such as adenine and cytosine, causingimportant mutations.1,7 This chemical behavior is aconcern because both acidic and basic conditions areencountered during the dodecamer synthesis based onthe solid-phase method by 2-cyanoethylphosphora-midite chemistry.8,9 Several works have describedimproved conditions for oligonucleotide synthesis.10

Some of them substitute the more common 5�-hy-droxyl protector group employed in the dodecamersynthesis, 4,4�-dimethoxytrityl (DMT), with a moreacid labile group such as 9-phenylxanthene or pixyl

group,11 and/or modify the basic treatment for thecleavage of the dodecamers from the supports.12 Thework described here is divided into two parts. First,the synthesis of a self-complementary dodecamercontaining the K nucleoside, 5�-CGTACGKGTACG-3�, is described. For this, and due to the 2-pyrimidi-none basic catalyzed hydrolysis previously described,two different treatments for the cleavage of the do-decamer from the supports were evaluated. The ob-jective of the second part is to characterize the pro-tonation equilibria of the K nucleoside and of thedodecamer at 37°C and 150 mM ionic strength. Forthis purpose, acid-base titrations were carried out andmonitored by molecular absorption, fluorescence, andcircular dichroism (CD) spectroscopies. Melting ex-periments at different pH values were also carried outto check the validity of the concentration profilesproposed for the dodecamer. Kinetic studies of the Knucleoside and of the dodecamer at several pH valueswere also carried out to evaluate the acid and basecatalyzed hydrolysis of 2-pyrimidinone residues.

EXPERIMENTAL

Reagents and Solutions

Synthesis. HPLC grade solvents were from Merck(Germany). Standard 2-cyanoethyl phosphoramiditeswere obtained from Cruachem, Ltd. (United King-dom), and dry solvents employed in dodecamer syn-thesis were from SDS (France). Analytical TLC wasrun on aluminum sheets coated with silica gel 60 F254from Merck. Column chromatography was performedon silica gel 60 A from SDS. The rest of the reagentswere purchased from Aldrich (USA) and Fluka (Swit-zerland). 2�-Deoxyzebularine was prepared as de-scribed in Barchi et al.2

Analytical Studies. Samples were prepared in Ultra-pure water (Millipore, USA) with the appropriatebuffer compounds: sodium monohydrogenphosphate,potassium dihydrogenphosphate, NaOH, acetic acid,sodium acetate, hydrochloric acid. NaCl was added toadjust the ionic strength to 150 mM. Reagents wereanalytical reagent grade and were from Panreac(Spain), Probus (Spain), and Merck.

28 Vives et al.

Synthesis

Synthesis of 5�-O-DMT-2�-Deoxyzebularine. 2�-De-oxyzebularine2 (294 mg, 1.38 mmol) was dried afterthree successive evaporations of anhydrous pyridine.The residue was dissolved in 3 mL of pyridine andtreated with 610 mg (1.79 mmol) of 4,4�-dimethoxy-trityl chloride. After 3 h of magnetic stirring at roomtemperature 1 mL of methanol was added to stop thereaction. The yellow solution obtained was evapo-rated to eliminate the solvents. The oily residue wasdissolved in 20 mL of DCM and the solution washedtwice in 5% NaHCO3 and saturated NaCl. The or-ganic phase was dried (Na2SO4) and concentrated todryness. The residue was purified by column chroma-tography on silica gel packed with 1% pyridine/DCMand eluted with a 0–8% methanol gradient in dichlo-romethane (DCM). The collected fractions containingthe pure product were concentrated to dryness toobtain a white foam.

Yield 51%, TLC (7% methanol/DCM) Rf � 0.34.1H-NMR (CDCl3) �(ppm): 8.5 (m, 2H, H6,H4), 7.2–7.4 (9H, phenyl DMT), 6.8 (4H, phenyl DMT), 6.2 (t,1H, H1�), 6.0 (m, 1H, H5), 4.5 (m, 1H, H3�), 4.1 (m,1H, H4�), 3.8 [s, 6H, -OCH3(DMT)], 3.5(m, 2H, H5�),2.8–2.35 (m, 2H, H2�). 13C-NMR (CDCl3) �(ppm):41.9 (C2�), 55.2 (CH3-O DMT), 61.8 (C5�), 72.3(C3�), 86.5 (C q DMT), 87.0 (C4�), 87.5 (C1�), 113.3–144.0 (C DMT), 156.2 (C6), 158.6 (C2), 165.5 (C4).MS (FAB�, nitrobenzyl alcohol): 515.2 (M � H),538.3 (M � Na), 303.4 (100%, DMT�), expected forC30H30O6N2: 514.5.

Synthesis of 5�-O-DMT-2�-Deoxyzebularine-3�-O-2-Cyanoethyl-N,N-Diisopropylphosphoramidite. 5�-O-DMT-2�-deoxyzebularine (360 mg, 0.7 mmol) wasdissolved in 5 mL of anhydrous DCM and 0.50 mL(2.8 mmol) of N-ethyldiisopropylamine was added.The reaction mixture was cooled by ice, and 0.20 mL(0.91 mmol) of 2-cyanoethyl N,N-diisopropyl-chloro-phosphoramidite was added dropwise with a syringe.After the addition the mixture was allowed to reachroom temperature. After 30 min of magnetic stirring,the progress of the reaction was checked by TLC[AcOEt/DCM/Et3N (45:45:10)]. The reaction wasjudged completed, solvents were evaporated, and theoily residue was dissolved in 30 mL of ethyl acetate(AcOEt) and 1.5 mL of triethylamine (Et3N). Thesolution was washed twice with 5% NaHCO3 andsaturated NaCl. The organic phase was dried(Na2SO4) and concentrated to dryness. The residuewas purified by column chromatography on silica gelpacked with an AcOEt/DCM/Et3N (45:45:10) solu-tion and eluted with AcOEt/DCM (1:1). The collected

fractions containing the pure product were concen-trated to dryness to obtain a white foam.

Yield 71%, TLC [AcOEt/DCM/Et3N (45:45:10)]Rf � 0.60. 1H-NMR (CDCl3) �(ppm): 8.5 (m, 2H,H6,H4), 8.46–8.39 (m,4H, 2-cyanoethyl), 7.2–7.4(9H, phenyl DMT), 6.8 (4H, phenyl DMT), 6.22 and6.24 (two t, 1H, Hl�), 6.0 (m, 1H, H5), 4.5 (m, 1H,H3�), 4.1 (m, 1H, H4�), 3.8 [s, 6H, -OCH3(DMT)],3.4(m, 2H, H5�), 2.65–2.43 (m, 2H, H2�), 1.0–1.3(14H, diisopropyl). 31P-NMR (CDCl3, and 85%H3PO4 as external reference). �(ppm): 149.5 and150.0 (two diastereoisomers).

Synthesis of 5�-CGTACGKGTACG-3� Dodecamer.The dodecamer was prepared in 1 �mol scale on anApplied Biosystems 392 automatic DNA synthesizerusing commercially available 2-cyanoethyl phos-phoramidites of the natural bases (Abzl, Cbzl, T, Gdmf)and the 5�-O-DMT-2�-deoxyzebularine-3�-O-2-cya-noethyl-N,N-diisopropylphosphoramidite previouslysynthesized. The last DMT group at the 5� end wasnot removed to facilitate purification. Removal ofbase [dimethylaminomethylidene (dmf) on N2-dGand benzoyl (bzl) on N6-dA and N4-dC] and phos-phate protecting groups and cleavage from the longchain amino alkyl-controlled pore glass (LCAA-CPG)support were evaluated with two treatments using con-centrated ammonia: at room temperature overnight, andat 55°C for 30 min. Ammonia solutions were filtrated,centrifuged, concentrated to dryness, and dissolved inUltrapure water (Millipore). Dodecamer was purified byreverse phase HPLC. The major peak was collected andanalyzed by MS. The collected fractions of pure productwere concentrated by liophilization.

Yield: 83 OD260 (from five syntheses of 1 �mol),MS (MALDI-TOF): m/z 3645.42 [M � Na�], (ex-pected m/z 3646 [M � Na�]).

Procedure

The experimental set-up used for spectroscopicallymonitored acid-base titrations has already been de-scribed elsewhere.13,14 Melting experiments were car-ried out at 1°C increments with a temperature ramp of0.5°C/min and monitored either simultaneously bymolecular absorption and CD, or individually by flu-orescence. At each temperature, a complete spectrumwas recorded. Each sample was heated at 65°C for 5min and allowed to equilibrate at the starting temper-ature for 30 min before the melting experiment wasstarted. After each experiment, the sample was cooledto starting temperature and the final spectrum wascompared to the initial one in order to confirm thereversibility of the process. Kinetic experiments were

Dodecamer Containing 2�-Deoxyzebularine 29

carried out at 37°C and monitored by molecular ab-sorption and fluorescence. The processes were con-sidered finished when no more spectral changes wereobserved.

Apparatus

Synthesis. HPLC and mass spectrometry were usedfor the dodecamer purification and characterization.HPLC was performed on a Shimadzu instrument asfollows: solvent A was 0.1 M triethylammonium ac-etate/acetonitrile (TEAA/ACN) (95:5) and solvent Bwas 0.1 M TEAA/ACN (3:7); column: a polystyrenecolumn, Hamilton PRP-1, 250 � 10 mm; flow rate: 3mL/min. A 0–80% B linear gradient in 20 min anddetection at 260 nm was applied for DMT-oligonu-cleotides, and a 0–50% B linear gradient in 20 minand detection at 260 nm was applied for oligonucle-otides after the removal of the DMT group. MALDI-TOF mass spectrometry measurements were performedin a Voyager DE-RP Applied Biosystems instrumentusing 3-hydroxypicolinic acid as the matrix.

Analytical Studies. Molecular absorption spectra wererecorded on a Perkin-Elmer lambda-19 spectrophotom-eter. Fluorescence spectra were recorded on an Aminco-Bowman series 2 spectrofluorimeter (�exc: 303 nm, �em:325–500 nm). These instruments were located in thesame laboratory, which enabled the simultaneous re-cording of molecular absorption and fluorescence spec-tra. CD spectra were recorded on a Jasco J-720 spec-tropolarimeter equipped with Neslab RET-110 temper-ature control unit. This spectropolarimeter allowed thecalculation of molecular absorption spectra by means ofthe mathematical treatment of the output signal includedin the software, allowing the simultaneous recording ofmolecular absorption and CD spectra. Hellma quartzcuvettes (path length of 1.0 cm) were used. pH measure-ments were performed with an Orion model 701A pHmeter (precision of �0.1 mV) and a combined Ross pHelectrode (Orion 9103).

Data Treatment

Information about the thermodynamic behavior ofevery system studied in this work was obtained frommultivariate analysis of experimental data. Hence, inacid-base titrations, the information to be recoveredconsisted of the number of acid-base species presentalong the titration, their concentration profiles, and thepure spectrum for each one of those species. Simi-larly, in melting experiments, the information con-sisted of the number of conformations, their concen-tration profiles, and the pure spectrum for each one of

them. Finally, for kinetic experiments, the informa-tion consisted of the number of species, their kineticprofiles, and their pure spectra.

Two different data treatments were applied in thiswork depending on the kind of spectroscopy data. Forthe spectroscopy data recorded along the titrations thehard-modeling mass-action law based procedure EQ-UISPEC15 was used. The spectroscopy data recordedalong the melting and kinetic experiments were ana-lyzed with the soft-modeling MCR-ALS16 procedure.Both procedures were applied to calculate the concen-tration profiles and the pure spectra of the spectro-scopically active species present in the system fromthe decomposition of the experimental data matrix Daccording to the equation

D � CST � E (1)

where C and ST are, respectively, the data matricescontaining the concentration profiles and the purespectra for each one of the species or conformationspresent in the experiment. E contains the residualnoise not explained by the proposed species or con-formations in C and ST (Scheme 1a).

Hard-Modeling: EQUISPEC Procedure

EQUISPEC15 procedure is a hard-modeling proce-dure widely applied for equilibrium constants deter-minations from spectroscopic data. EQUISPEC isbased on the postulation of an initial chemical modeldefined by: the stoichiometries of the proposed spe-cies, the approximate values of their equilibrium con-stants, and the compliance of the mass-action law.This procedure is especially useful when applied tothe study of chemical equilibria of monomers or oli-gomers, that is, when mass-action law is fulfilled andno secondary effects related to large polymeric struc-tures, like polyelectrolytic effects, polyfunctional ef-fects, or conformational changes, are present. On theother hand, for large polymeric systems or when an-alyzing data from melting and kinetic experiments,application of hard-modeling procedures is rather dif-ficult or even impossible because it is difficult topostulate a physical or chemical model describing theobserved data variance. In these cases, however, dataanalysis is still possible by applying soft-modelingfactor analysis-based methods because they do notrequire the postulation or fulfillment of a particularphysical or chemical model. Multivariate Curve Res-olution using Alternating Least Squares (MCR-ALS)16 is a soft-modeling method that has beenwidely applied for the study of acid-base and confor-mational transitions of polynucleotides.17–19

30 Vives et al.

SCHEME 1 Different arrangements and decompositions of experimental data matrices. (a)Analysis of a single matrix containing either molecular absorption, CD, or fluorescence datarecorded along a single acid-base, thermal, or kinetic experiment. (b) Analysis of two data matricescontaining molecular absorption and CD (or fluorescence) spectra simultaneously recorded along anacid-base titration [Dabs, DCD (fl)]. (c) Analysis of several data matrices containing molecularabsorption spectra recorded along kinetic experiments at several pH values. In order to improve theresolution, the pure molecular absorption spectra resolved by EQUISPEC for the acid-base speciesin the protonation studies are also included at the top. The scheme can be written as [Sabs

Tacid-base;Dabs

kinetic 1; . . . ; Dabskinetic n]. (d) Analysis of several data matrices containing molecular absorption

and CD data simultaneously recorded along melting experiments at several pH values. In order toimprove the resolution, the pure molecular absorption and CD spectra resolved by EQUISPEC forthe acid-base species in the protonation studies are also included at the top. The scheme can bewritten as {[Sabs

T, SCDT]acid-base; [Dabs, DCD]pH 1; . . . ; [Dabs, DCD]pH n}. D, data matrix; C,

concentration profiles matrix; ST, pure spectra matrix; E, residual matrix; N, active spectroscopicspecies, conformations, or degradation products.

Soft-Modeling: MCR-ALS

The MCR-ALS16 procedure applied in this work con-sisted of the following steps:

1. Data arrangement: For an experiment moni-tored by molecular absorption, the recordedspectra were collected in a table or matrix Dabs.The dimensions of this matrix were Nr rows x�m columns, where Nr were the spectra re-corded at successive pH, temperature, or timevalues, and �m was the number of wavelengthsmeasured (Scheme 1a). For an experiment mon-itored by fluorescence or CD spectra, spectrawere collected in a table or matrix Dfl or DCD,accordingly.

2. Determination of the number of acid-base spe-cies or conformations, n: The number of spec-troscopically active chemical species or confor-mations n was estimated by applying severalmethods, such as singular value decomposition(SVD),20 evolving factor analysis (EFA),21 orSIMPLISMA.22

3. ALS optimization: The ALS optimization pro-cedure is an iterative method used to solve Eq.(1) for the proposed number of species or con-formations n. This iterative process is startedeither with an initial estimation of the concen-tration profiles C or with an initial estimation ofthe pure spectra ST (Sabs

T, SflT, or SCD

T de-pending on the spectra measured: molecularabsorption, fluorescence, or CD) for each one ofthe n species or conformations proposed. Aninitial estimation of C can be obtained fromEFA.21 Alternatively, it is possible to obtain aninitial estimation of ST from SIMPLISMA.22

The iterative optimization advances in two steps,the order of which depends on the initial estimation(C or ST) used. Hence, when the ALS optimizationstarts with an initial estimation of the C data matrixthe calculation is carried out as follows:(a) In the first step, ST is calculated from the solutionof the least squares optimization problem:

min �D* � CST� (2)

where D* is the SVD reproduced data matrix for then species considered, and C is the initial estimationfrom EFA. The constraint of non-negativity in thespectral responses is applied along this optimizationstep, that is, the calculated spectra in ST are con-strained to be positive or zero for molecular absorp-

tion and fluorescence data. This constraint is not ap-plied for CD data.

(b) In the second step, C is recalculated by solvingthe least squares optimization problem:

min �D* � CST� (3)

where ST are the pure spectra calculated in the pre-vious step. In this case the concentration values in Care constrained not only to be positive but also to giveunimodal concentration profiles, that is, only onemaximum is allowed in each of the n concentrationprofiles. Because the total concentration of oligonu-cleotide is constant throughout the experiment, a clo-sure constraint is applied for the concentration profilesin this step, that is, the sum of the different speciesconcentrations at each pH, temperature, or time valueis forced to be a known constant value.

(c) Steps (a) and (b) are repeated alternatively untilthe data matrix D* is well explained within experi-mental error. Steps (a) and (b) are swapped when theALS optimization starts with an initial estimation ofthe pure spectra ST.

The results obtained by factor analysis-basedmethods, such as MCR-ALS, can show some mathe-matical ambiguities (rotational of intensity ambigu-ities)16 and rank-deficiency problems,23–25 which canproduce results different from the true ones. The ro-tational ambiguities take place when the pure spectraor concentration profiles for two or more species orconformations overlap. The intensity ambiguities pro-duces that the concentration profiles and pure spectra arescaled by an unknown factor. The rank-deficiency23–25

is a difficulty commonly present in chemical-reaction-based systems when the number of independent reac-tions plus one is lower than the number of chemicalspecies involved in these reactions. This fact meansthat concentration profiles obtained will not be lin-early independent and the number of detected com-ponents will be less than the number of chemicalspecies really present in the system. A way to over-come these difficulties is the simultaneous analysis ofseveral data matrices obtained under different exper-imental conditions.

In this work, the simultaneous analysis was appliedseveral times (Scheme 1b–d). To carry out such ananalysis, an augmented data matrix is first built up byjoining several data matrices before solving Eq. (1).The augmented data matrix can be built in threedifferent ways. First, matrices containing spectra re-corded using different instrumental techniques (likemolecular absorption, Dabs; fluorescence, Dfl; and/orCD, DCD) along a single experiment can be grouped

32 Vives et al.

in a new augmented row-wise data matrix, which canbe written as [Dabs, DCD, or Dfl] according to MAT-LAB notation.26 As an example, the augmented datamatrix obtained when recording an acid-base titrationsimultaneously by molecular absorption and CD (orfluorescence) is shown in Scheme 1(b). This analysisimproves the resolution of the system due to theachievement of the selectivity conditions of the spe-cies in some of the techniques applied. In this case,CD spectra show a degree of selectivity higher thanmolecular absorption spectra, that is, the shape andintensity of CD spectra are more characteristic of aparticular species than absorption spectra.

A second way to perform the simultaneous analy-sis is by building up a column-wise augmented datamatrix. This is done by grouping the single matricescontaining the spectra recorded at different experi-mental conditions but using the same instrumentaltechnique. As an example, Scheme 1(c) shows the si-multaneous analysis of several matrices containing mo-lecular absorption spectra recorded along different ki-netic experiments (for details, see Results and Discus-sion). The grouping of these individual data matricesresults in an augmented data matrix [Dexperiment1; . . . ;Dexperimentn]. This analysis is adequate to solve systemswhere a deficiency rang problem is present, that is,where a single mathematical component actually corre-sponds to a mixture of two or more conformations.

The third way is building up a row- and column-wise augmented matrix, which contains experimentaldata from different experiments and different tech-niques.24,25 This approach is always desirable becausethe best resolution of the system studied is achievedthrough the solution of the lack of selectivity and thedeficiency rank problems. As an example, the analysisof molecular absorption and CD data recorded simul-taneously along different melting experiments isshown in Scheme 1(d) (for details, see Results andDiscussion).

All MCR calculations were performed using in-house MATLAB (version 6, The Mathworks Inc.,Natick, MA) routines (codes and tutorials are freelyavailable at the URL: www.ub.es/gesq/mcr/mcr.htm).

RESULTS AND DISCUSSION

Dodecamer Synthesis

Dodecamer was prepared using solid-phase 2-cyano-ethylphosphoramidite chemistry.10 The required syn-thon to incorporate 2�-deoxyzebularine into oligonu-cleotides was prepared starting from the correspond-ing nucleoside that was synthesized as previously

described.2 The DMT group was selected for protec-tion of the 5�-end. Reaction of 2�-deoxyzebularinewith DMT chloride in pyridine gave the desiredDMT-nucleoside that was reacted with 2-cyanoethyl-N,N-diisopropylchlorophosphine to yield the desiredphosphoramidite. The 2�-deoxyzebularine phosphora-midite was purified by silica gel chromatography andcharacterized by proton and phosphorous NMR. Solidphase synthesis study using this phosphoramidite re-vealed a step-coupling yield similar to standard phos-phoramidites (�98%).

On the other hand, to investigate the dodecamerstability under the basic treatment of cleavage fromthe support, LCAA-CPG, two syntheses at small scalewere carried out. The last DMT was not removed tofacilitate purification using reversed phase HPLC.One of the oligonucleotides was treated with concen-trated ammonia at 55°C for 30 min and another withconcentrated ammonia at room temperature over-night. In both cases (Figure 1a) and in the area cor-responding to DMT-containing oligonucleotides amajor eluting peak (at 12.18 min) was found with asmaller peak near the major peak (at 11.62 min). Fromboth syntheses these two eluting peaks were collectedand, after the DMT group was removed, analyzed bymeans of HPLC. Chromatograms showed a singleeluting peak from the collected previously eluted peakat 12.18 min (Figure 1b) and a still mixture of themajor and minor components from the collected pre-viously eluted peak at 11.62 min (Figure 1c). Molec-ular absorption analyses of major component showedan absorption band maximum at 260 nm characteristicof the four natural nucleosides absorption with ashoulder at 303 nm characteristic of the 2�-deoxyze-bularine absorption. Mass-spectrometry analysis ofthis material confirmed the mass expected for thedodecamer (m/z 3645.42 [M � Na�]). Mass-spec-trometry analysis of the minor component showed am/z 3614.28 [M � Na�] peak not in agreement withthe dodecamer structure without the 2-pyrimidinonering (m/z expected for dodecamer without 2-pyrimidi-none base 3552 [M � Na�]), indicating no abasic siteformation6 under the experimental conditions of theacidic 5�-DMT group deprotection steps. This mate-rial could be an impurity or a degradation productfrom the basic catalyzed hydrolysis of 2�-deoxyzebu-larine produced by basic cleavage from the LCAA-CPG support. It was noticeable from the chromato-grams that the area of the minor component washigher after treatment with concentrated ammonia atroom temperature overnight than after 30 min at50°C. Moreover, HPLC analysis showed some smalleluting peaks at higher retention times, and these wereassigned to sequences still carrying some protecting

Dodecamer Containing 2�-Deoxyzebularine 33

groups. These small peaks did not appear in the sam-ple treated with a 30 min treatment at 50°C. Fromthese results, all syntheses were finished with a con-centrated ammonia treatment at 50°C for 30 min.Figure 1(d) shows the HPLC analysis of the finalpurified compound (from the major collected previ-ously eluted peaks in Figure 1b and 1c).

Protonation Studies

Acid-base titrations of K nucleoside monitored usingmolecular absorption (Figure 2a, corresponding todata matrix Dabs) and fluorescence (Figure 2b, matrixDfl) were carried out at 37°C and 0.15 M ionicstrength to determine the pKa value of the nucleosideand to characterize the fluorescence signal before its

incorporation to the oligonucleotide. The main trendsobserved in molecular absorption spectra upon proto-nation were a red shift of the maximum from 303 to308 nm and a slight decrease of intensity. Fluores-cence spectra obtained at neutral pH values showed adouble band with maxima at 360 and 382 nm. Uponprotonation the intensity of fluorescence increaseddramatically and finally a maximum at 357 nm and ashoulder over 380 nm were measured.

Matrices Dabs and Dfl were grouped in a row-wiseaugmented data matrix with dimensions number ofrows � 18 pH values and number of columns � 297wavelengths (i.e., 121 wavelengths measured in mo-lecular absorption from 240 to 360 nm and 176 wave-lengths measured in fluorescence measurements from325 to 500 nm). This augmented matrix is given in

FIGURE 1 (a) Reversed-phase HPLC analysis from the 5�-CGTACGKGTACG-3� synthesisended with the last 5�-DMT protector group, with a 0–80% B linear gradient in 20 min. (b)Reversed-phase HPLC analysis after extraction of the 5�-DMT protector group from the collectedpeak eluted in (a) at 12.18 min, with a 0–50% B linear gradient in 20 min. (c) Reversed-phase HPLCanalysis after extraction of the 5�-DMT protector group from the collected peak eluted in (a) at 11.62min, with a 0–50% B linear gradient in 20 min. (d) Reversed-phase HPLC analysis of the finalproduct. Analysis was carried out with a polystyrene column Hamilton PRP-1, 250 � 10 mm; flowrate: 3 mL/min; solvent A was 0.1 M triethylammonium acetate/acetonitrile (95:5) and solvent Bwas 0.1 M triethylammonium acetate/acetonitrile (3:7).

34 Vives et al.

Scheme 1(b) and written in a shorter way as [Dabs,Dfl]. The analysis of this augmented matrix by EQ-UISPEC15 allowed the calculation of the pKa valuefor the N(3) position at the experimental conditions(3.13 � 0.09). This value is rather close to the pKa

value determined for 2-pyrimidinone base at differentexperimental conditions of 20°C (pKa � 2.527). Fig-ure 2c, d shows the calculated pure molecular absorp-tion spectra, Sabs

T, and the pure fluorescence spectra,Sfl

T, which agree with the experimental observations,as commented on above. Figure 2e shows the resolvedconcentration profiles C, which shows that at neutralpH the K nucleoside is fully deprotonated.

Once the protonation equilibrium of the K nucle-oside was studied, a similar procedure was applied tostudy the acid-base behavior of the dodecamer. Frombasic to neutral pH the maximum of the molecularabsorption spectra shifted from 261 to 256 nm (Figure3a, matrix Dabs). From neutral to acidic pH, a slightshift from 256 to 261 nm and a hyperchromic effectwere also observed, a fact that could reflect the basestacking disruption when protonation takes place. CDspectra recorded at basic pH values showed a negativeband at 240 nm (Figure 3b, matrix DCD). Upon pro-tonation the intensity increased and the positionshifted to 250 nm, which would reflect a B-helicalform at neutral pH values. At more acidic pH values,the CD negative band shifted to 244 nm, indicatingthe partial disruption of the B-helical form due to theprotonation of cytosine and adenine bases. At pHvalues below 2, the complete breaking of the double-stranded structure was reflected in the low signal-to-noise ratio observed. Finally, fluorescence spectra forthe pH range 7.7–2.2 (Figure 3c, matrix Dfl) reflect adramatic change in the fluorescence of the fully pro-tonated dodecamer species due to the protonation ofthe K nucleoside. No precipitate was observed alongthe titrations either at acidic or basic pH values.

As molecular absorption and CD data were simul-taneously recorded, an augmented [Dabs, DCD] matrixwas built up (Scheme 1b). On the other hand, asfluorescence data were obtained in an independentexperiment, Dfl matrix had to be analyzed individually(Scheme 1a). In all the cases, data were analyzed byEQUISPEC15 considering the presence of three, four,or five acid-base species. The reliability on thosemodels was analyzed according to the final resultsobtained with each one of them, that is, according tothe chemical sense of the calculated pKa values, con-centration profiles, the pure spectra, and the residualmagnitude of the variance not explained by eachmodel. The system including four species, that is,three pKa values, was finally considered to be themost chemically meaningful. For this system, the

calculated pKa values were: 9.60 � 0.07, 4.46 � 0.16,and 2.87 � 0.19. Figure 3d–f show, respectively, thepure molecular absorption spectra Sabs

T, the pure CDspectra SCD

T, and the pure fluorescence spectra SflT

for each one of the four proposed species. Figure 3gshows the corresponding calculated concentrationprofiles C. The explanation for these species is asfollows: at pH values around 10, the dodecamershows all nitrogenated bases deprotonated, with neu-tral cytosine, adenine, and 2-pyrimidinone, but withanionic thymine and guanine at N(3) (N(3)-T) andN(1) (N(1)-G) positions, respectively. At neutral pH,the protonation of N(3)-T and N(1)-G shows the do-decamer with neutral bases. At pH around 3.5, proto-nation of adenine at N(1) and cytosine at N(3) isobserved. Finally, at lower pH values, the dodecamershows protonation of 2-pyrimidinone at N(3) and gua-nine at N(7). Concentration profiles C show that dou-ble-stranded B-dodecamer is only present as a majorspecies in a very narrow region around pH 7. Thecalculated pKa values for the four natural bases intothe dodecamer are quite similar to the pKa valuesfound in literature for the free bases.28 As expected,the calculated pKa value for 2-pyrimidinone whenincorporated into the dodecamer is slightly lower thanthat calculated for the K nucleoside, that is, protona-tion is a little more difficult when the base formshydrogen bonds with guanine bases in the dodecamer.

The calculated pure spectra (Figure 3d–f) agreewell with the observed experimental trends. The pro-tonation of cytosine and adenine bases is more clearlyreflected in the calculated molecular absorption andCD pure spectra (Sabs

T, SCDT) than in the calculated

fluorescence pure spectra (SflT). Sfl

T for the neutral(labeled as 3) and intermediate (labeled as 2) speciesare very similar. However, the protonation of the2-pyrimidinone base at the K nucleoside (see above)is clearly reflected in the dramatic increase of fluo-rescence for the pure spectrum of the fully protonatedspecies (labeled as 1).

Cooperative effects related to the polymeric structureof polynucleotides can usually be denoted in protonationequilibria by the presence of pronounced slopes in spe-cies distribution concentration profiles.13,29 When theseeffects are present, the acid-base equilibria for eachprotonation site in the polynucleotide cannot be charac-terized by a single constant pKa value. However, in thepresent case, from the slope of the acid-base concentra-tion profiles C (Figure 3g), no cooperative behaviorrelated to the polymeric structure could be deduced, thatis, the pKa value for each one of the nitrogenated baseswas not affected by the protonation/deprotonation ofneighboring bases. This conclusion is reinforced by theresults obtained when applying EQUISPEC,15 which

Dodecamer Containing 2�-Deoxyzebularine 35

FIG

UR

E2

36 Vives et al.

uses a hard-modeling approach for the study of proto-nation equilibria. From the results obtained in previousresearch, hard-modeling methods can be used withoutproblems for the study of protonation equilibria for oli-gonucleotides like dodecamers.14

Melting Studies

Melting experiments at pH values 3.1, 3.9, and 6.7were carried out to confirm the conclusions derivedfrom the previous protonation studies. The reversibil-ity of the unfolding processes at pH 3.1 and 3.9 waschecked by recording one spectrum after a long sta-bilization time at the initial temperature. Reversibilityof the melting process at pH 6.7 was studied in detailwith an annealing experiment monitored by molecularabsorption and CD. While the melting experiments atpH 3.9 and 6.7 were reversible, melting at pH 3.1 wasirreversible, probably because of the labile behaviorof 2-pyrimidinone residues in acidic medium and hightemperature. Kinetic experiments were carried out toconfirm this point (see below). Due to the irrevers-ibility of the melting process at pH 3.1, experimentaldata were not analyzed in this case. At pH 6.7, the lossof the initial structure is reflected in: the increase ofabsorbance with a shoulder formation over 270 nm;the blue shift and decrease of the intensity of the CDband at 250 nm, typical of B-DNA; and the fluores-cence intensity decrease with the differentiation of theband over 350 nm, characteristic of the K nucleoside.

Spectroscopic data were analyzed by means of thesoft-modeling procedure MCR-ALS. For this purposemolecular absorption and CD spectra recorded alongmelting experiments at pH 3.9 ([Dabs and DCD]melting pH 3.9)and 6.7 ([Dabs and DCD]melting pH 6.7) and annealingexperiment at pH 6.7 were ([Dabs and DCD]annealing pH 6.7)arranged in a row-wise augmented data matrix, to-gether with the molecular absorption and the CD purespectra resolved by EQUISPEC from the previousprotonation studies of the dodecamer (Figure 3d,e)([Sabs

T and SCDT]acid-base). The augmented matrix

used in these analyses is given in Scheme 1(d) and iswritten in a shorter way as ([Sabs

T, SCDT]acid-base;

[Dabs, DCD]melting pH 3.9; [Dabs, DCD]melting pH 6.7;[Dabs, DCD]annealing pH 6.7). The dimensions of thisaugmented matrix are: number of rows � 95 (three

estimated EQUISPEC spectra plus 28 spectra mea-sured in the melting experiment at pH 3.9 plus 31spectra measured in the melting experiment at pH 6.7plus 33 spectra measured in the annealing experimentat pH 6.7) and number of columns � 244 (122 wave-lengths recorded in every technique from 235 to 356nm). Simultaneous MCR-ALS analysis of all thesedata sets resolved four components in the temperaturerange under study with the molecular absorption andCD pure spectra, Sabs

T, SCDT (Figure 4a,b). Two of

the calculated pure spectra of these four componentswere coincident with the pure spectra previously ob-tained for the dodecamer with neutral nucleobases andfor the dodecamer with protonated cytosine and ade-nine bases. The other two components were assignedto unfolded random-coil conformations. Calculatedconcentration profiles for the melting experiment atpH 3.9 and pH 6.7 are shown in Figure 4(d,e), respec-tively. Calculated concentration profiles for the an-nealing experiment at pH 6.7 are shown in Figure 4(f).Figure 4(d) shows the presence at initial temperaturevalues of a mixture of the dodecamer with deproto-nated nucleobases, and, as a main component, of thedodecamer with protonated adenine and cytosine.This agrees with the protonation species previouslyobtained at this pH (Figure 3g). When temperature isincreased, the transition of these two species to ran-dom coil-conformations is observed. From Figure4(d), a Tm value � 42°C was estimated from thecrossing point of the two resolved concentration pro-files (curves 2 and 5) for the dodecamer with proton-ated adenine and cytosine. A Tm value � 44°C wasdetermined from the crossing point of resolved con-centration profiles (curves 3 and 6) for the deproto-nated dodecamer at pH 3.9. Calculated concentrationprofiles obtained for the melting experiment at pH 6.7(Figure 4e) showed the transition between the initialdouble-stranded conformation with all nucleobasesdeprotonated (B-form) and the final random-coil con-formation. For this melting process, a value for Tm �50.5°C from the resolved concentration profile C wasdetermined. For the annealing process at pH 6.7, avalue for Tm � 50.9°C was determined. From the lackof significant differences in the calculated Tm valuesfor the melting and the annealing experiments, it was

FIGURE 2 Protonation studies of K nucleoside. (a) Experimental molecular absorption spectrarecorded along the pH range 2.1–7.4, Dabs. (b) Experimental fluorescence spectra recorded in thesame pH range, Dfl. (c) Pure molecular absorption spectra for the acid-base species calculated byEQUISPEC, Sabs

T. (d) Pure fluorescence spectra. SflT. (e) Concentration profile, C. Dashed line

(1�): protonated base. Continuous line (2�): deprotonated base. The arrows indicate the pH valueswhere K nucleoside kinetic experiments were carried out.

Dodecamer Containing 2�-Deoxyzebularine 37

FIG

UR

E3

38 Vives et al.

concluded that the reversibility of the process is at pH6.7.

Fluorescence data obtained in the melting experi-ment at pH 6.7 were analyzed individually (Scheme1a). In Figure 4(c) the resolved fluorescence speciesnormalized spectra Sfl

T are given. The value of Tm

� 50.3°C was determined from the resolved concen-tration profiles C. The concentration profiles obtainedfrom the fluorescence experiment at pH 6.7 were notplotted because they were very similar to Figure 4(d–f). To study the possible existence of nonhyperchro-mic transitions assigned to intermediate hairpin-shaped structures,30 MCR-ALS16 analysis of experi-mental CD data was also performed. No intermediatestructures were detected in this case.

From the results obtained, it was deduced that Tm

values increase with the pH. This fact was related to therelatively more stable double-stranded structure at neu-tral pH values. Upon protonation, a disruption of thestructure was observed, and, consequently, the corre-sponding Tm value decreased. Tm values estimated fordodecamer at pH 3.9 and 6.7 were quite different, whichconfirmed that one species was predominant in eachregion, as depicted in Figure 3(g). Simultaneous analysisof melting experiments augmented with estimated purespectra obtained in protonation studies allowed the res-olution of the melting process of the mixture of speciesat pH 3.9 and confirmed that at pH 6.7 only one specieswas predominant. Tm values calculated from molecularabsorption-CD or fluorescence monitored melting ex-periments at pH 6.7 showed good agreement, indicatingthat both approaches gave a similar description of themelting process for the dodecamer sequence. Whereasthe melting experiment monitored by molecular absorp-tion showed the global perturbation of the duplex, thesame experiment monitored by fluorescence showed lo-cal perturbations in the neighborhood of the fluorophore(in this case, the 2-pyrimidinone base). Because Tm

values obtained from molecular absorption and fluores-cence experiments were quite similar, it was concludedthat the global duplex disruption is coincident with the

local perturbation of the helix around the K � G base pair.This would be related to a similar strength of the K � Gbase pair in relation to the global duplex stability.

Kinetic Studies

Due to the acidic and basic labile behavior of theresidues containing the 2-pyrimidinone base, it wasdecided to study the stability of the K nucleoside andof the dodecamer as a function of time during theacid-base studies. Hence, kinetic experiments of theK-nucleoside at 37°C and at pH values 2.1, 3.1, 4.1,9.2, and 11.9 were carried out, monitored by molec-ular absorption and fluorescence. Finally, the labilebehavior of the K nucleoside when incorporated intothe dodecamer was studied at pH 3.1 at 37°C.

The molecular absorption spectra of the K nucle-oside at pH 2.1, 3.1, and 4.1 showed a progressivedecrease of absorbance with a blue shift of the max-imum (from 310 to 307 nm for experiment at pH 2.1,from 304 to 301 nm for experiment at pH 3.1, andfrom 303 to 300 nm for experiment at pH 4.1) alongtime. Individual analysis of these experiments bymeans of MCR-ALS revealed the presence of onlytwo components at each pH value, which were as-signed to the K nucleoside and to a mixture of theribose ring and the 2-pyrimidinone base, obtainedafter a N-glycosyl bond cleavage without base struc-ture breakage.5 However, from the concentration pro-file obtained along acid-base experiments (Figure 2e),it was clear that at the beginning of the kinetic exper-iment at pH 3.1 there should be a mixture (1:1) ofprotonated and deprotonated K nucleoside, both hy-drolyzed. This fact produces a rank deficiency prob-lem, as has been previously shown in melting exper-iments (see above) and in previous works.23,24 Thisproblem can be solved by MCR-ALS simultaneousanalysis of the matrix obtained in different kineticexperiments at pH 2.1 [Dabs

kinetic pH 2.1]; pH 3.1[Dabs

kinetic pH 3.1], and pH 4.1 [Dabskinetic pH 4.1] together

with the molecular absorption pure spectra resolved by

FIGURE 3 Protonation studies of the 5�-CGTACGKGTACG-3� dodecamer. (a) Experimentalmolecular absorption spectra recorded along the pH range 1.9–10.4, Dabs. (b) Experimental CDspectra recorded along the same pH range, DCD. (c) Experimental fluorescence spectra recordedalong the pH range 2.2–7.7, Dfl. (d) Pure molecular absorption spectra for the acid-base speciescalculated by EQUISPEC, Sabs

T. (e) Pure CD spectra, SCDT. (f) Pure fluorescence spectra, Sfl

T. (g)Concentration profile, C. Symbol (“X”) indicates the pH values where melting experiments werecarried out. The arrow indicates the pH value where the dodecamer kinetic experiment was carriedout. (1) Dodecamer with all the bases protonated; (2) dodecamer with protonated cytosine andadenine and deprotonated 2-pyrimidinone, thymine, and guanine; (3) dodecamer with neutral bases;(4) dodecamer with neutral cytosine, adenine, and 2-pyrimidinone, but with anionic thymine andguanine.

Dodecamer Containing 2�-Deoxyzebularine 39

FIG

UR

E4

MC

R-A

LS

resu

ltsof

the

anal

ysis

ofm

eltin

gex

peri

men

tsof

5�-C

GT

AC

GK

GT

AC

G-3

�do

deca

mer

.(a

)Pu

rem

olec

ular

abso

rptio

nsp

ectr

aS a

bsT

and

(b)

pure

CD

spec

tra

S CD

Tfo

rth

esp

ecie

sca

lcul

ated

bysi

mul

tane

ous

MC

R-A

LS

anal

ysis

ofm

eltin

gat

pH3.

9,m

eltin

gat

pH6.

7,an

dan

neal

ing

atpH

6.7.

(c)

Pure

norm

aliz

edflu

ores

cenc

esp

ectr

a,S fl

T,f

orth

em

eltin

gex

peri

men

tatp

H6.

7.(d

)C

once

ntra

tion

profi

les

for

the

mel

ting

atpH

3.9.

(e)

Con

cent

ratio

npr

ofile

for

the

mel

ting

atpH

6.7.

(f)

Con

cent

ratio

npr

ofile

for

the

anne

alin

gat

pH6.

7.(2

)an

d(3

)as

inFi

gure

3;(5

)ra

ndom

coil

conf

orm

atio

nfr

omth

edo

deca

mer

with

prot

onat

edad

enin

ean

dcy

tosi

nean

dde

prot

onat

edth

ymin

e,gu

anin

e,an

d2-

pyri

mid

inon

e;(6

)ra

ndom

coil

conf

orm

atio

nob

tain

edfr

omth

edo

deca

mer

with

neut

raln

ucle

obas

es.P

repa

ratio

nof

solu

tions

:8.2

mM

acet

icac

id,1

.93

mM

sodi

umac

etat

e,15

0m

MN

aCl,

pH3.

9;10

mM

KH

2PO

4,

10m

MN

a 2H

PO4,

149

mM

NaC

l,pH

6.7.

40 Vives et al.

EQUISPEC in the previous protonation studies for pro-tonated and deprotonated K nucleoside [Sabs

Tacid-base](Figure 2c). The augmented matrix was built up ac-cording to Scheme 1(c) and is written in a shorter wayas [Sabs

Tacid-base; Dabskinetic pH 2.1; Dabs

kinetic pH 3.1;

Dabskinetic pH 4.1]. The dimensions of this augmented

matrix are number of rows � 363 (2 estimated EQ-UISPEC spectra, plus 67 spectra measured in thekinetic experiment at pH 2.1, plus 110 spectra mea-sured in the kinetic experiment at pH 3.1, plus 184

FIGURE 5 Kinetic experiments of K nucleoside. (a) Pure molecular absorption spectra SabsT. (b)

Concentration profile C at pH 2.1. (c) Concentration profile C at pH 3.1. (d) Concentration profileC at pH 4.1. (1�) Protonated K nucleoside; (2�) deprotonated K nucleoside; (3�) mixture of riboseand protonated nucleobase after glycosyl bond cleavage; (4�) mixture of ribose and deprotonated Knucleobase after glycosyl bond cleavage. Preparation of solutions: 17.7 mM HCl, 147.8 mM NaCl,pH 2.1; 4.7 mM HCl, 149.9 mM NaCl, pH 3.1; 8.2 mM acetic acid, 1.93 mM sodium acetate, 150mM NaCl, pH 4.1.

Dodecamer Containing 2�-Deoxyzebularine 41

spectra measured in the kinetic experiment at pH 4.1)and number of columns � 102 (102 wavelengthsrecorded from 259 to 360 nm). The analysis revealedthe presence of four components with the molecularabsorption pure spectra shown in Figure 5(a). Calcu-lated molecular absorption pure spectra of two ofthese four components were practically identical tothe pure spectra obtained for protonated and deproto-nated K nucleoside (Figure 2c). The other two com-ponents were assigned to the species resulting fromthe hydrolysis process: a mixture of a ribose ring witha protonated nucleobase and a mixture of ribose ringand deprotonated nucleobase. Calculated concentra-tion profiles obtained for the kinetic experiments atpH 2.1 and 4.1 (Figure 5b,d) showed the N-glycosylbond cleavage with time for the protonated and de-protonated K nucleoside, respectively. From thecrossing points of curves in these concentration pro-files, it was observed that 50% of the glycosyl bondhydrolysis took place after 0.23 and 6.5 h, respec-tively. Calculated concentration profiles obtained forthe kinetic experiment at pH 3.1 (Figure 5c) showedthe glycosyl bond hydrolysis for a mixture of theprotonated and the deprotonated K nucleoside givingthe corresponding mixture of ribose ring and proton-ated and deprotonated nucleobase. The resultingnucleobases obtained in these hydrolysis processeswere also in acid-base equilibrium, which at this pH

value is shifted to the deprotonated nucleobase (pKa

2.527). This equilibrium could explain the apparentabsence, at the beginning of this experiment, of theexpected 1:1 ratio for the mixture of deprotonated andprotonated K nucleoside. From the crossing point ofcurves in the concentration profile given in Figure5(c), it was observed that 50% of the depyrimidina-tion process took place after 0.4 (curves 1� and 3�) and0.68 (curves 2� and 4�) h for protonated and deproto-nated K nucleoside, respectively. Concentration pro-files (curve 1�) given in Figure 5(b) and (c) for thehydrolysis of protonated K nucleoside at pH 2.1 and3.1, respectively, were fitted to exponential curvesand the obtained rate constant values were 1.35� 10�3 and 4.82 � 10�4 s�1. In the same way,concentration profiles (curve 2�) given in Figure 5(c)and (d) for the hydrolysis of deprotonated K nucleo-side at pH 3.1 and 4.1, respectively, were fitted toexponential curves and the obtained rate constantvalues 3.41 � 10�4 and 3.13 � 10�5 s�1 were ob-tained. A rate value of 1.05 � 10�4 s�1 was previ-ously reported5 for the hydrolysis of K nucleoside atpH 3.0 and room temperature. This value agrees wellwith the values given in this work at pH 3.1, althoughin this previous work the acid-base equilibrium of Knucleoside was not considered. As expected, the gly-cosyl bond hydrolysis was pH-dependent and fol-lowed pseudo-first-order kinetics, which can be ex-

FIGURE 6 Kinetic experiment of K nucleoside at pH 11.9 (18.4 mM NaOH, 135.1 mM NaCl).(a) Pure spectra resolved by MCR-ALS, Sabs

T. (b) Concentration profiles resolved by MCR-ALS,C. Inset: amplification of the interval 0–2 h. Continuous line: K nucleoside. Dashed line: interme-diate from the nucleophilic attack by hydroxyl ions over the 2-pyrimidinone positions C5¢C6.Dotted line: breakage of the 2-pyrimidinone ring.

42 Vives et al.

pressed for the hydrolysis of protonated and deproto-nated 2-pyrimidinone, respectively, as

d�dKH]/dt � �k1��dKH]n and d�dK]/dt

� �k2��dK]n (4)

where k1� and k2� are pseudoconstants that depend onthe concentration of proton (i.e., depending on pH).

Molecular absorption spectra recorded in a kinetic ex-periment at pH 9.2 (72 h) showed few changes, a findingthat was related to the higher stability of the glycosyl bondat this pH value. Only a small absorbance change wasobserved at 270 nm. Whereas these changes at pH 9.2 wererather small, noticeable changes were, however, observedat pH 11.9. MCR-ALS analysis determined the presence ofthree main components along the kinetics at pH 11.9.Figure 6 shows pure spectra Sabs

T and concentration pro-files C. It was observed that the third species is the onlyspecies after 75 h of reaction. This species was assigned tothe breakage of the 2-pyrimidinone ring. The second spe-cies was related to the intermediate species formed duringthe nucleophilic attack by hydroxyl anions over the 2�-pyrimidinone positions C5¢C6. Finally, the first species isthe initial K nucleoside.1,9

Kinetics of the dodecamer sample was also studiedat pH 3.1 for 24 h. During this time, no spectralchanges were observed that could be related to thehydrolytic deavage of the glycosyl bond in the dodec-amer. From the comparison of these results with thoseobtained for the K nucleoside in the same experimen-tal conditions, it was concluded that the stability ofthe K nucleoside increased dramatically upon incor-poration into the dodecamer. Moreover, the observedstability was still enough to carry out the protonationstudies satisfactorily without considering the degrada-tion of the dodecamer. Finally, the nonreversibilityobserved for the dodecamer unfolding experiment atpH 3.1 probably showed a temperature-dependence ofdepyrimidination process of the K nucleoside.

M.V. thanks the University of Barcelona for a Ph.D grant.Maqbool A. Siddiqui is thanked for his help.

REFERENCES

1. Barchi, J. J.; Haces, A.; Marquez, V. E.; McCormack,J. J. Nucleosides Nucleotides 1992, 11, 1781–1793.

2. Barchi, J. J.; Musser, S.; Marquez, V. E. J Org Chem1992, 57, 536–541.

3. Driscoll, J. S.; Marquez, V. E.; Plowman, J.; Liu, P. S.;Kelley, J. A.; Barchi, J. J. J Med Chem 1991, 34, 3280–3284.

4. Adams, C. J.; Murray, J. B.; Arnold, J. R. P.; Stockley,P. G. Tetrahedron Lett 1994, 35, 1597–1600.

5. Iocono, J. A.; Gildea, G.; McLaughlin, L. W. Tetrahe-dron Lett 1990, 31, 145–178.

6. Berthet, N.; Constant, J. F.; Demeunynck, M.; Michon,P.; Lhomme, J. J Med Chem 1997, 40, 3346–3352.

7. Baik, M. H.; Friesner, R. A.; Lippard, S. J Am ChemSoc 2002, 124, 4495–4503.

8. Basu, A. K.; O’Hara, S. M.; Valladier, P.; Stone, K.; Mols,O.; Marnett, L. J. Chem Res Toxicol 1988, 1, 53–59.

9. Nair, V.; Turner, G. A.; Offerman, R. J. J Am ChemSoc 1984, 106, 3370–3371.

10. Brown, T.; Brown, D. J. S. In Oligonucleotides andanalogues, a practical approach; Eckstein, F., Ed.; IRLPress: Oxford, 1991, pp. 1–25.

11. Gildea, B.; McLaughlin, L. W. Nucleic Acids Res1989, 17, 2261–2281.

12. Zhou, Y.; Ts’o, P. O. Nucleic Acids Res 1996, 24,2652–2659.

13. Gargallo, R.; Tauler, R.; Izquierdo-Ridorsa, A.Biopolymers 1997, 42, 271–283.

14. Vives, M.; Gargallo, R.; Tauler, R. Biopolymers 2001,59, 477–488.

15. Dyson, R.; Kaderli, S.; Lawrence, G. A.; Maeder, M.;Zuberbuhler, A. D. Anal Chim Acta 1997, 353, 381–393.

16. Tauler, R.; Smilde, A.; Kowalski, B. R. J Chemom1995, 9, 31–58.

17. Jaumot, J.; Escaja, N.; Gargallo, R.; Gonzalez, C.;Pedroso, E.; Tauler, R. Nucleic Acids Res 2002, 30:e92.

18. Gargallo, R.; Vives, M.; Tauler, R.; Eritja, R. BiophysJ 2001, 81, 2886–2896.

19. Vives, M.; Gargallo, R.; Tauler, R. Anal Chem 1999,71, 4328–4337.

20. Maeder, M.; Zilian, A. Chemom Intel Lab Syst 1988, 3,205–213.

21. Windig, W.; Stephenson, D. A. Anal Chem 1992, 64,2735–2742.

22. Golub, G. H.; Van Loan, C. F. Matrix Computations,Johns Hopkins University Press: Baltimore, 1989.

23. Izquierdo-Ridorsa, A.; Saurina, J.; Hernandez-Cassou, S.;Tauler, R. Chemom Intell Lab Syst 1997, 38, 183–196.

24. Saurina, J.; Hernandez-Casou, S.; Tauler, R.; Izquierdo-Ridorsa, A. J Chemom 1998, 12, 183–203.

25. Amhrein, M.; Sirnivaman, B.; Bonvin, D.; Schumacher,M. Chemom Intell Lab Syst 1996, 33, 17–33.

26. Matlab® routines available at http://www.ub.es/gesq/mcr/mcr.htm

27. Fox, J. J.; Van Praag, D.; Wempen, I.; Doerr, I. L.;Cheong, L.; Knoll, J. E.; Eidinoff, M. L.; Bendich, A.B.; George, B. J Am Chem Soc 1959, 81, 178–187.

28. Izatt, R. M.; Christensend, J. J.; Rytting, J. H. ChemRev 1971, 71, 439–480.

29. Gargallo, R.; Tauler, R.; Izquierdo-Ridorsa, A. AnalChim Acta 1996, 331, 195–205.

30. Davis, T. M.; McFail-Isom, L.; Keane, E.; Williams,L. D. Biochem 1998, 37, 6795–6798.

Reviewing Editor: Dr. George J. Thomas, Jr.

Dodecamer Containing 2�-Deoxyzebularine 43