View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

Noncovalent Association Phenomena of 2,5-Dihydroxybenzoic Acid With Cyclic and Linear . A Matrix-Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometric and X-ray Crystallographic Study

Andrea Mele and Luciana Malpezzi Dipartimento di Chimica del Politecnico and C.N.R.—Centro Studi sulle Sostanze Organiche Naturali, Milano, Italy

D- and 19 glucose derivatives were investigated by positive and negative matrix assisted laser desorption/ionization time-of-flight using 2,5-dihydroxyben- zoic acid (DHB) as the matrix. The set of substrates includes oligomers of amylose and cellulose, native ␣-, ␤-, and ␥-cyclodextrin, and chemically modified ␤- and ␥-cyclodextrins. These analytes were chosen to modulate molecular weight, polarity, and capability of establishing noncovalent interactions with guest . In the negative-ion mode, the DHB Ϫ matrix gave rise to charged multicomponent adducts of type [M ϩ DHB Ϫ H] (M ϭ ) selectively for those analytes matching the following conditions: (i) underi- vatized chemical structure and (ii) number of glucose units Ն4. In the positive-ion polarity, only some amylose and cellulose derivatives and methylated ␤-cyclodextrins provided small ϩ amount of cationized adducts with the matrix of type [M ϩ DHB ϩ X] (X ϭ Na or K), along ϩ with ubiquitous [M ϩ X] . The results are discussed by taking into account analyte– matrix association phenomena, such as hydrogen bond and inclusion phenomena, as a function of the molecular structure of the analyte. The conclusions derived by mass spectro- metric data are compared with the X-ray diffraction data obtained on a single crystal of the 1:1 ␣-cyclodextrin Ϫ DHB noncovalent adduct. (J Am Soc Mass Spectrom 2000, 11, 228–236) © 2000 American Society for Mass Spectrometry

yclomaltooligosacchrides (cyclodextrins, CDs) tion mass spectrometric techniques were successfully are truncated, cone-shaped molecules able to applied to the field of host–guest [7]. Cinclude neutral guests within their molecular The majority of mass spectrometric studies on cavity, giving rise to noncovalent host–guest (h–g) charged gaseous association of CDs with guest mole- inclusion complexes. The driving force for the inclusion cules were carried out by using either fast bom- of organic guests into the cavity of cyclodextrins is the bardment (FAB) [8–14] or electrospray ionization (ESI) attractive interaction between the lipophilic cavity of [15–28]. Matrix-assisted laser desorption/ionization the host and apolar parts of the guest , typi- mass spectrometry (MALDI-MS) is a key analytical and cally aromatic rings. Host–guest inclusion complexes structural tool for chemistry [29], and have attracted the interest of many scientists for the many efforts were devoted to the optimization of ex- uniqueness of their chemical behavior and structural perimental protocols [30–35], structural elucidation by features [1], and led technologists to practical applica- post-source-decay (PDS) fragmentation [36, 37], and to tions, especially in the pharmaceutical [2] and food the understanding of collision-induced dissociation industry [3]. The physical characterization of h–g ad- (CID) [38] and the mechanism of desorption/ionization ducts mainly relies upon high resolution nuclear mag- [39]. Nevertheless, there are only a few examples of netic resonance (NMR) [4], thermal methods [5], and detection of noncovalent adducts of CDs under condi- X-ray diffraction studies [6]. More recently, soft ioniza- tions of matrix-assisted desorption/ionization [40–42]. A few other studies deal with noncovalent complexes of proteins [43–50] and hydrogen-bonded assemblies [51, Address reprint requests to Dr. Andrea Mele, Dipartimento di Chimica del Politecnico, Via Mancinelli, 7 I-20131 Milano, Italy. E-mail: andrea.mele@ 52]. polimi.it In MALDI practice, 2,5-dihydroxybenzoic acid

© 2000 American Society for Mass Spectrometry. Published by Elsevier Science Inc. Received September 7, 1999 1044-0305/00/$20.00 Revised October 21, 1999 PII S1044-0305(99)00143-9 Accepted October 22, 1999 J Am Soc Mass Spectrom 2000, 11, 228–236 DHB-OLIGOSACCHARIDES ASSOCIATIONS BY MALDI 229

(DHB) is a matrix particularly well suited for carbohy- mJ per shot, 3 ns pulse width of a single laser shot). drate determination. From the point of view of su- Negative ion spectra were recorded by reversing the pramolecular chemistry, DHB is potentially capable of polarity of the acceleration potential. The laser irradi- forming inclusion complexes with CDs, because of its ance was selected manually over an arbitrary nonlinear aromatic character, and establishing nonspecific inter- scale ranging from 0 (null power) to 180 (full power). actions with saccharides, mainly dipole–dipole and/or Once it reached the detection threshold for the interest- hydrogen bond, because of the presence of COOH and ing peaks, the irradiance was raised 10% to 20%. The OH functional groups. Observation of matrix-analyte data were acquired with a laser irradiance within the clusters in MALDI-MS is well known and documented range 88–93 of the arbitrary scale of the instrument for [53], but so far no attempt to provide a rationale in the positive polarity, and 92–95 for the negative ion terms of noncovalent associations and specific host– mode. All the spectra were acquired by scanning the guest interactions has been done. We report here a sample spot along the x axis and averaging 100 laser systematic investigation on formation and detection of shots. The data were processed on a Sun Spark 2 matrix-analyte associations of cyclic and linear oligosac- computer using the manufacturer’s software. Mass cal- charides with DHB. The results are interpreted and ibration was achieved using the matrix and ␥-cyclodex- discussed, focusing mainly on the role played by non- trin as internal references, and bovine insulin (Sigma, covalent interactions such as hydrogen bonding and Milan, Italy) as an external reference. The hydrophilic inclusion phenomena as a function of the structural and lipophilic analytes were dissolved in distilled water features of the analytes. The information obtained by and grade cyclohexane, respectively. The mass spectrometric data are also compared with X-ray concentration of the solutions was typi- crystallographic data of the noncovalent adduct cally 100 pmol/␮L, corresponding to 0.1 mM. The ␣-CD Ϫ DHB chosen as a model compound [54]. samples were prepared by mixing 0.5 ␮L of matrix solution with 0.5 ␮L of analyte solution directly on the sample slide and allowing evaporation of the solvent Experimental ␮ under a N2 stream. Sandwich type preparation (0.5 L ␮ Materials of matrix solution and evaporation, 0.5 L of analyte solution and evaporation, and then 0.5 ␮L of matrix ␣ Glucose (C6H12O6) 1, -D-glucose pentaacetate solution and evaporation) was used only for nonwater (C16H22O11) 2, sucrose (C12H22O11) 3, sucrose octaac- soluble 2, 4, 15, and 17. ␣ ␤ etate (C28H38O19) 4, -cyclodextrin (C36H60O30) 10, -cy- ␥ clodextrin (C42H70O35) 11, -cyclodextrin (C48H80O40) ␤ X-ray 12, heptakis(2,6-di-O-methyl)- -cyclodextrin (C56H98O35) ␤ 13, triacetyl -cyclodextrin (C84H112O56) 15 were pur- Crystals, suitable for the X-ray single crystal analysis, chased from Aldrich (Milan, Italy) and used as such. were obtained after many attempts by slowly cooling Oligomers of (i) amylose and (ii) cellulose were pur- two different hot solutions containing ␣-cyclodextrin ϭ chased from Sigma (Milan, Italy): (i) maltose (C12H22O11) and DHB (solvent: CH3CN/H2O 2/3 v/v) with ana- 5, maltotriose (C18H32O16) 6, maltotetraose (C24H42O21) lyte to matrix molar ratios 1:1 and 1:5, respectively. The 7, maltopentaose (C30H52O26) 8, maltohexaose (C36H62O31) best crystal (platelet, 0.3 ϫ 0.3 ϫ 0.05 mm) was isolated ؅ 9 and (ii) cellotriose (C18H32O16) 6 , cellotetraose from the latter solution and used for crystal structure ؅ ؅ (C24H42O21) 7 , cellopentaose (C30H52O26) 8 . Hep- characterization. Crystals coming from the 1:1 solution ␤ takis(2,3,6-tri-O-methyl)- -cyclodextrin (C56H98O35) 14 gave identical unit cell parameters. was obtained from Sigma (Milan, Italy) and used with- The X-ray diffraction data collection of the complex out further purification. A sample of randomly func- were performed with the four-circle automatic Siemens ␤ tionalized hydroxypropyl- -cyclodextrin (C105H196O56 P4 diffractometer, using graphite monochromated Cu-K␣ considering full derivatization) 16 was presented by radiation (␭ ϭ 1.54179 Å) and ␪/2␪ scan technique. Chiesi Farmaceutici (Parma, Italy). Octakis(3-O-butanoyl- Unit cell parameters were determined using 30 reflec- ␥ 2,6-di-O-pentyl)- -cyclodextrin (Lipodex E, C84H112O56) tions in the range 18° Յ 2␪ Յ 50°; a total of 9428 ϭ 17 was synthesized as reported in the literature [55]. 65 reflections (4685 unique, Rint 0.0477) were collected mM solution of the matrixes were prepared by dissolv- up to 136.2° in 2␪ and index range: Ϫ1 Յ h Յ 9, Ϫ17 Յ ing DHB or ␣-cyano-4-hydroxycinnamic acid (CHCA) k Յ 16, Ϫ22 Յ l Յ 22. Empirical adsorption correction (Sigma, Milan, Italy) in a 2:3 (v/v) mixture of acetoni- was applied to the intensities data, using the ␺-scan trile and aqueous 0.1% (v/v) trifluoroacetic acid (TFA). method [56]; no decay correction was deemed neces- sary. The complex crystallizes in the orthorhombic MALDI-TOF-MS space group P 221 21, with a half molecule in the asymmetric unit. The cell parameters are: a ϭ 8.273(2), The spectra were acquired on a Shimadzu-Kratos b ϭ 16.786(2), and c ϭ 21.939(3) Å. Kompact Maldi II instrument operating in the linear The structure was determined by using, as a starting mode with an accelerating potential of 20 kV and model, the coordinates of the of the ␣-CD mole- ␭ ϭ equipped with a UV pulsed laser (N2, 337 nm, 100 cules (apart from those of primary C and O atoms 230 MELE AND MALPEZZI J Am Soc Mass Spectrom 2000, 11, 228–236

which could deviate from the expected position) of isomorphous structures found in the Cambridge Struc- tural Database [57]. Subsequent refinements and differ- ence Fourier maps revealed the missing atoms of the ␣-CD molecule, of the guest molecule and of the water molecules. Most of the H atoms, not located by the ⌬␳ map, were introduced at calculated positions and re- fined in a riding model. Two water molecules were found on the electron density map; minor peaks of electron density suggested the presence of some resid- ual weakly populated water sites, which were retained in the final refinement with a common isotropic tem- perature factor. The H atoms of the water molecules were not introduced. The refinement was carried out on F2 by full-matrix least-squares procedure with SHELXL97 [58] for 428 parameters; final stage con- verged to R ϭ 0.0865 for 3557 observed reflections having I Ն 2␴(I).

NMR Spectroscopy High resolution NMR spectra were acquired on a Bruker ARX 400 at a nominal temperature of 305.0 Ϯ

0.1 K using D2O as solvent. The chemical shifts were referenced to external DSS using a coaxial capillary Scheme 1 tube. Standard water presaturation techniques have been applied to suppress the signal of residual HOD. The NMR samples were prepared by dissolving suit- alter the preferred route for the ion formation, which is able quantities of pure host, pure guest (DHB) and 1:1 constantly the alkali metal attachment rather than pro- ؅ ؅ ؅ mixture of host and guest in 1 mL of D2O in order to tonation. The spectra of compounds 6, 6 , 7 , 8, 8 , 13, obtain 9 mM solutions. The experiments were carried and 14 showed small but significant peaks (see relative out on ␤-cyclodextrin 11 and permethyl ␤-cyclodextrin intensity values on Table 1) assigned to the multicom- ϩ 15. The spectra and a table with the chemical shift ponent ions of the general formula [M ϩ DHB ϩ X] values of selected protons are available as supplemen- (X ϭ K for 6, 6؅, 7؅, 8, 8؅, 13,Xϭ Na or K for 14). These tary material. ions can be regarded as cationized supramolecular adducts of matrix and analyte. Figure 1 shows in detail Results and Discussion the positive-ion MALDI-TOF spectrum of compound 14 with the peaks assignment. Intriguingly, these adducts The molecular structure of the oligosaccharides ana- are not formed by native cyclodextrins 10, 11, and 12, lyzed throughout the present work are displayed in whose ability to form host–guest inclusion complexes Scheme 1. They are representative of native mono- and with aromatic molecules, such as DHB, is well docu- linear oligosaccharides (1, 3, 5–9), chemically modified mented [1]. analogs (2 and 4), cyclic maltooligosaccharides (cyclo- Experiments carried out by using CHCA as a matrix dextrins 10, 11, and 12), chemically modified, water- did not show the presence of any supramolecular soluble ␤-cyclodextrins (13, 14, and 16), and lipophilic matrix-saccharide adduct (data not reported). derivatives of ␤- and ␥-cyclodextrin (15 and 17, respec- The results of negative ion MALDI mass spectra of tively). The experimental results relating to positive-ion compounds 1–17 in DHB are reported in Table 2. These MALDI-TOF spectra using DHB as a matrix are re- data confirm the low capability of the compounds to Ϫ ported in Table 1. The analytes are detected as sodium provide significant amount of [M Ϫ H] ions, but point ϩ or potassium cationized species [M ϩ Na] and [M ϩ out that medium sized linear and cyclic oligosaccha- ϩ K] (only maltose 5 gave rise to detectable amount of rides (compounds 7 to 12) afford peaks consistent with Ϫ ϩ ϩ Ϫ ϩ ϩ ϩ Ϫ Ϫ dehydrated [M H2O Na] and [M H2O K] the general formula [M DHB H] . Once again, species) independently of either the molecular weight experiments performed with CHCA as a matrix showed Ϫ Ϫ of the sugar or the presence of protecting groups on all neither [M Ϫ H] nor [M ϩ CHCA Ϫ H] for all the or some of the hydroxyls of the sugar skeleton. Linear examined compounds (data not reported). Figure 2 and cyclic maltooligosaccharides share this behavior, as shows a detail of the spectrum of ␥-cyclodextrin 12 as Ϫ shown by compounds 6 to 12. Moreover, the hydro- an example. The formation of the [M ϩ DHB Ϫ H] philic or lipophilic nature of the analyte (compare adducts appears to correlate with the number n of compounds 10, 11, and 12 with 14, 15, and 17) does not glucose units (n Ն 4) of the analyte and is inhibited by J Am Soc Mass Spectrom 2000, 11, 228–236 DHB-OLIGOSACCHARIDES ASSOCIATIONS BY MALDI 231

Table 1. Main results of positive-ion MALDI-TOF spectra of compounds 1–17 Comp. Calcd. MWa Observed peaks (m/z) Assignmentb

1 180.17 203.2; 219.3 [M ϩ Na]ϩ;[Mϩ K]ϩ 2 390.33 413.5; 429.7 [M ϩ Na]ϩ;[Mϩ K]ϩ Ϫ ϩ ϩ ϩ ϩ ϩ ϩ 3 342.30 347.2; 365.7; 381.7 [M H2O Na] ;[M Na] ;[M K] 4 678.59 702.2; 718.3 [M ϩ Na]ϩ;[Mϩ K]ϩ Ϫ ϩ ϩ Ϫ ϩ ϩ 5 342.30 347.2; 363.3 [M H2O Na] ;[M H2O K] 365.3; 381.3 [M ϩ Na]ϩ;[Mϩ K]ϩ 6 504.43 526.9; 543.0; 697.1 [M ϩ Na]ϩ;[Mϩ K]ϩ;[Mϩ DHB ϩ K]ϩ (47) (6؅ 504.43 527.1; 543.2; 697.2 [M ϩ Na]ϩ;[Mϩ K]ϩ;[Mϩ DHB ϩ K]ϩ (47 7 666.57 689.7; 705.7 [M ϩ Na]ϩ;[Mϩ K]ϩ (7؅ 666.57 689.7; 705.7; 859.8 [M ϩ Na]ϩ;[Mϩ K]ϩ;[Mϩ DHB ϩ K]ϩ (17 8 828.72 851.8; 867.6; 1022.0 [M ϩ Na]ϩ;[Mϩ K]ϩ;[Mϩ DHB ϩ K]ϩ (90) (8؅ 828.72 852.8; 867.9; 1022.1 [M ϩ Na]ϩ;[Mϩ K]ϩ;[Mϩ DHB ϩ K]ϩ (17 9 990.86 1013.7; 1029.9 [M ϩ Na]ϩ;[Mϩ K]ϩ 10 972.84 995.6; 1011.5 [M ϩ Na]ϩ;[Mϩ K]ϩ 11 1134.98 1158.2; 1174.1 [M ϩ Na]ϩ;[Mϩ K]ϩ 12 1297.12 1320.3; 1336.2 [M ϩ Na]ϩ;[Mϩ K]ϩ 13 1331.36 1354.5; 1524.6 [M ϩ Na]ϩ;[Mϩ DHB ϩ K]ϩ (31) 14 1429.54 1453.0; 1468.9 [M ϩ Na]ϩ;[Mϩ K]ϩ 1607.5; 1623.3 [M ϩ DHB ϩ Na]ϩ;[Mϩ DHB ϩ K]ϩ (27) 15 2017.75 2040.8; 2056.9 [M ϩ Na]ϩ;[Mϩ K]ϩ 16c 1541.51 1565.0; 1581.0 [M ϩ Na]ϩ;[Mϩ K]ϩ 17 2979.99 3005.8; 3021.9 [M ϩ Na]ϩ;[Mϩ K]ϩ aAveraged values. bThe data in parentheses refer to the intensity ratio [M ϩ X]ϩ/[M ϩ DHB ϩ X]ϩ. cThe data of this line are referred to the most abundant fraction, showing, on average, one hydroxypropyl group per glucose unit.

Figure 1. Expansion of the positive-ion MALDI-TOF spectrum (DHB) of permethylated ␤-cyclodex- trin 14. 232 MELE AND MALPEZZI J Am Soc Mass Spectrom 2000, 11, 228–236

Table 2. Main results of negative-ion MALDI-TOF spectra of compounds 1–17 Comp. Calcd. MWa Observed peaks (m/z) Assignment

1 180.17 — — 2 390.33 — — 3 342.30 — — 4 678.59 — — 5 342.30 — — 6 504.43 — — — — 6؅ 504.43 7 666.57 819.0 [M ϩ DHB Ϫ H]Ϫ 7؅ 666.57 818.9 [M ϩ DHB Ϫ H]Ϫ 8 828.72 981.4 [M ϩ DHB Ϫ H]Ϫ 8؅ 828.72 981.3 [M ϩ DHB Ϫ H]Ϫ 9 990.86 1143.7 [M ϩ DHB Ϫ H]Ϫ 10 972.84 1125.1 [M ϩ DHB Ϫ H]Ϫ 11b 1134.98 1133.3; 1287.8 [M Ϫ H]Ϫ;[Mϩ DHB Ϫ H]Ϫ 12c 1297.12 1296.5; 1450.6 [M Ϫ H]Ϫ;[Mϩ DHB Ϫ H]Ϫ 13 1331.36 — — 14 1429.54 — — 15 2017.75 — — 16 1541.51 — — 17 2979.99 — — aAveraged values. bRatio of the integrated intensities: [M ϩ DHB Ϫ H]Ϫ/[M Ϫ H]Ϫ ϭ 15.4. cRatio of the integrated intensities: [M ϩ DHB Ϫ H]Ϫ/[M Ϫ H]Ϫ ϭ 4.9.

Ϫ the presence of substituents on the OH groups of the HSO4 . Our findings can be accounted for by consider- sugars. Similar results have been recently reported by ing the multicenter ionization model for ion formation Wong et al. [59] in the case of noncovalent adducts of in MALDI proposed by Knochenmuss et al. [60, 61]. The miscellaneous oligosaccharides with the anion dopant key feature of their model is the role played by two

Figure 2. Expansion of the negative-ion MALDI-TOF spectrum (DHB) of ␥-cyclodextrin 12. J Am Soc Mass Spectrom 2000, 11, 228–236 DHB-OLIGOSACCHARIDES ASSOCIATIONS BY MALDI 233

Ϫ [M ϩ DHB Ϫ H] species can be grouped into two families: (i) low molecular-weight, underivatized sug- ars (1, 3, and 5), characterized by a limited number of hydrogen-bonding donor groups, and (ii) protected sugars (compounds 2, 4, 13–17) for which the chemical derivatization of the OH groups upsets the hydrogen- bond profile with respect to the native saccharide. Ϫ Scheme 2 Furthermore, fairly abundant [M ϩ DHB Ϫ H] ions ␤ 3 were obtained from either [ -D-Glc-(1 4)]n-D-Glc ,n ϭ 3 and 4, corresponding to compounds 7؅ and 8؅) ␣ 3 ϭ electronically excited matrix molecules (m*H, according respectively), and [ -D-Glc-(1 4)]n-D-Glc (n 3, 4, to the notation used by the authors) in mutual spatial and 5, corresponding to compounds 7, 8, and 9, respec- vicinity as precursor species for both protonated and tively), independently of the different conformational cationized analyte species. The two m*H species “pool preferences of cellulose and amylose oligomers, linear their energy” affording efficient charge separation. The and helical, respectively [62]. Cyclodextrins 10, 11, and proton (or cation) thus generated is captured by the 12 are able to form noncovalent host–guest inclusion analyte, which is supposed to have sufficient proton (or complexes with the anion of the matrix by interaction of cation) affinity. One consequence of this two-step the apolar cavity of the saccharide with the aromatic model is a large production of deprotonated matrix ions ring of the deprotonated matrix. However, the multi- Ϫ Ϫ m corresponding, in our case, to [DHB Ϫ H] ions. center model of Knochenmuss et al. requires two ex- The latter may mediate both the desorption step, by cited matrix molecules in “close proximity” [60]. As- forming strong aggregates with those analytes able to suming that a true inclusion complex cyclodextrin– establish attractive interactions with them (e.g., those DHB is actually preformed in the crystal, the host compounds with a large number of hydrogen bond cyclodextrin would act as a spacer keeping the included donor groups, such as free OHs), and the ionization DHB molecule separated from the neighbor nonin- process. Indeed, in a noncovalent aggregate of type cluded matrix molecules, thus preventing any interac- Ϫ Ϫ [DHB Ϫ H] ...[carbohydrate] (see Scheme 2), the tion among them, mandatory for generating the m negative charge is largely delocalized onto the carbox- species as primary ions. Therefore, the role, if any, ylate oxygens and stabilized by resonance. The forma- played by inclusion phenomena in the laser desorp- Ϫ tion of the above-mentioned [DHB Ϫ H] ... [carbohy- tion/ionization processes leading to the gaseous [M ϩ Ϫ drate] complex allows the ionization of the analyte to DHB Ϫ H] seems to be marginal with respect to other occur by attachment of a preformed and thermodynam- intermolecular associations, like hydrogen bonded as- ically stable carboxylate anion rather than by deproto- semblies. These deductions, inferred from mass spec- nation of any of the OH groups of the carbohydrate. trometrical data only, are substantiated by the experi- This route prevents the localization of a negative charge mental results of the X-ray structure analysis on a single onto the saccharides which, for structural reasons, are crystal of the 1:1 noncovalent adduct ␣-CD–DHB. The unable to provide efficient stabilization. packing diagram of the complex is shown in Figure 3. According to the data of Table 2, the nature of the The ␣-CD molecules are arranged in the typical chan- attractive interaction between the analyte and the ion- nel-type structure in a head-to-tail fashion. In the chan- ized matrix is likely to be mainly the hydrogen bond. nels, adjacent ␣-CD molecules are linked by hydrogen Experimental evidence for the importance of polar bonds between the primary hydroxyl groups and sec- interactions between carboxylic acid anions and neutral ondary hydroxyl groups. The guest molecules (DHB) carbohydrates in the gas phase was recently provided appear disordered, showing rotation about one of the by Selva et al. [24]: gaseous protonated and deproto- twofold axes of the benzene ring. The carboxylic group nated 1:1:1 ␤-cyclodextrin Ϫ ketoconazole Ϫ tartaric is disordered over two adjacent positions that are half acid [␤CD Ϫ KC Ϫ TA] ternary species were obtained occupied also by one of the two hydroxyl groups; the by ESI-MS from the related multicomponent associa- other hydroxyl group is disordered over the two related tion. para positions. Most importantly, DHB molecules are The rank orders of the polar or lipophilic noncova- located outside the CD cavity and are situated among the lent intermolecular interactions were assigned by using channels, forming chains, in the head-to-tail mode, via tandem-MS data. In particular, within the deprotonated hydrogen bonds. This type of noninclusion host–guest ternary species, the strongest supramolecular attractive complex has been referred to as association compound interaction appeared to be that of polar type between [63]. The chains of perfectly coplanar guest molecules the TA anion and ␤CD, only due to dipole and hydro- are elongated parallel to the ␣-CD channels, in the c gen bonding involving the external hydrophilic surface crystallographic axis direction. Fully and partially po- of ␤CD, provided that no evidence of an inclusion sitioned water molecules contribute to a network of complex formation between TA anion and ␤CD was hydrogen bonds inside and outside the ␣-CD macro- found by NMR experiments in D2O solution [25]. cycles. The solid-state arrangement of matrix and ana- In the present study, the compounds not yielding the lyte is nearly a perfect match for the topological require- 234 MELE AND MALPEZZI J Am Soc Mass Spectrom 2000, 11, 228–236

follow the regular trend observed for the negative Ϫ charged counterparts [M ϩ DHB Ϫ H] . Also, the ϩ abundance of [M ϩ DHB ϩ X] is one to two orders of ϩ magnitude less than [M ϩ X] . The distribution of the ϩ cationized multicomponent adducts [M ϩ DHB ϩ X] along the analyte set 1–17 seems to be random or, at least, not straightforwardly imputable to a well defined repertoire of intermolecular interactions, such as that Ϫ discussed above in the case of the [M ϩ DHB Ϫ H] adducts. As an example, the mass spectrometric re- sponse of methyl-substituted ␤-cyclodextrins 13 and 14 is different compared to those shown by either underi- vatized ␤-CD 11 or functionalized analogs 15 and 16. Unfortunately, all the attempts of preparing crystals of putative complexes between methylated cyclodextrins and DHB, suitable for X-ray analysis, have been so far unsuccessful. NMR data, although referred to as solu- tion state, indicate that 11 and 14 share similar capabil- ities of interacting with DHB, either by forming true inclusion compounds, or by nonspecific interactions of the external surface of the cyclic carbohydrates with DHB (experimental data available as supplementary material). Thus, neither presumed different inclusion capabilities, nor the different hydrogen-bond profile of 11 with respect to 14, provide a rationale for the ϩ observed formation of [M ϩ DHB ϩ X] adducts. The positive-ion data rather suggest that the ablation of ϩ [M ϩ DHB ϩ X] is the result of the excited-state ma- trix action on a preformed aggregation of the saccharide and DHB in the crystal. Accordingly, the different behavior of methylated vs. native ␤-cyclodextrin is consistent with a different solid-state preorganization of matrix and analyte that is not predictable a priori on the basis of the known complexing ability of the hosts and their hydrogen-bond profiles.

Conclusions Figure 3. Packing diagram of the ␣-CD–DHB complex. Top: projection along the a axis. Bottom: projection along the c axis. H The present work demonstrates that linear and cyclic atoms and disordered water molecules are omitted, for clarity. The oligosaccharides derived from D-glucose may form bonds involving disordered atoms of the guest molecules are negative charged, noncovalent aggregates of type [M ϩ indicated with dotted lines. Hydrogen bonds are indicated with Ϫ dashed lines. DHB Ϫ H] provided the following conditions are sat- isfied: (i) the analyte is made up of a relatively large number n of units (n Ն 4) and (ii) Ϫ ments for both the generation of [DHB Ϫ H] (spatial there is an absence of protecting groups on the hydroxyl proximity of two matrix molecules) and its attachment groups of the sugar. The result is independent of the to the carbohydrate via dipolar interaction and hydro- linear or cyclic structure of the oligosaccharide. Oli- ␣ ␣ 3 gen bonds. Single crystals of the -CD–DHB 1:1 com- gomers of amylose and cellulose, [ -D-Glc-(1 4)]n-D- ␤ 3 plex were obtained from a solution with matrix to Glc and [ -D-Glc-(1 4)]n-D-Glc, respectively, afford analyte molar ratio ϭ 5, that is two orders of magnitude similar results, despite their helical and linear confor- less than the value generally used for MALDI-TOF mation, respectively. The driving force for the aggrega- spectra (see Experimental section). Nevertheless, the tion is likely to be the establishment of intermolecular negative-ion MALDI-TOF spectrum carried out by us- matrix/analyte interactions, such as dipole–dipole and ing the same solution used for the crystalization of the hydrogen bonds. Indeed the formation of [M ϩ DHB Ϫ Ϫ complex afforded only one signal related to the carbo- H] is inhibited by methylation, acetylation, or other hydrate at m/z 1125.9, once again assigned to the forms of chemical derivatization of the OH groups. The Ϫ gaseous association [␣-CD ϩ DHB Ϫ H] . solid-state structure of the 1:1 association compound Comparison of Tables 1 and 2 points out that the between ␣-CD and DHB, as deduced by X-ray diffrac- ϩ formation of the cationized [M ϩ DHB ϩ X] does not tion experiments, reveals that the matrix molecules are J Am Soc Mass Spectrom 2000, 11, 228–236 DHB-OLIGOSACCHARIDES ASSOCIATIONS BY MALDI 235

Ϫ suitably arranged for generating [DHB Ϫ H] , via mul- 9. Sawada, M.; Shizuma, M.; Takai, Y.; Adachi, H.; Takeda, T.; ticenter ionization mechanism [60], and attaching to the Uchiyama, T. Chem. Commun. 1998, 1453–1454. analyte through hydrogen bonds. Mass spectrometric 10. Mele, A.; Panzeri, W.; Selva, A. Eur. Mass Spectrom. 1997, 3, and diffraction data indicate a negligible role of inclu- 347–354. ϩ Ϫ Ϫ 11. Mele, A.; Panzeri, W.; Selva, A. J. Mass Spectrom. 1997, 32, sion phenomena in generating [M DHB H] ad- 807–812. ducts. 12. Davey, S. N.; Leigh, D. A.; Smart, J. P.; Tetler, L. W.; Truscello, ϩ ϩ ϩ The detection of [M DHB X] adducts in the A. M. Carbohydr. Res. 1996, 290, 117–123. positive-ion mode is not directly related to clear-cut 13. Mele, A.; Selva, A. J. Mass Spectrom. 1995, 30, 645–647. structural features of the compounds examined here, 14. Juo, C.-G.; Shiu, L.-L.; Chen, C. K.-F.; Luh, T.-Y.; Her, G.-R. such as the hydrogen bond properties invoked for the Rapid Commun. Mass Spectrom. 1995, 9, 604–608. negative ions. Nevertheless, the participation of inclu- 15. Selva, A.; Redenti, E.; Zanol, M.; Ventura, P.; Casetta, B. Org. sion phenomena can be ruled out on the basis of the Mass Spectrom. 1993, 28, 983–989. ϩ 16. Cescutti, P.; Garozzo, D.; Rizzo, R. Carbohydr. Res. 1995, 290, following points: (i) [M ϩ matrix ϩ X] adducts are 105–115. only formed by using DHB as a matrix, although both 17. Lamcharfi, E.; Chuillon, S.; Kerbal, A.; Kunesch, G.; Libot, F.; CHCA and DHB are good potential guests for host– Virelizier, H. J. Mass Spectrom. 1996, 31, 982–986. guest complexes of cyclodextrins. (ii) ␤-CD 11 and its 18. Lemesle-Lamache, V.; Wouessidjeve, D.; Cheron, M.; Duch- permetylated derivative 14 show a different response ene, D. Int. J. Pharm. 1996, 141, 117–124. toward the same guest molecule, DHB, although the 19. Prokai, L.; Ramanathan, R.; Nawrocki, J.; Eyler, J. J. Inclusion high resolution NMR results do not show a dramatic Phenom. Mol. Recogn. Chem. 1996, 25, 117–120. difference in the inclusion properties between these two 20. Cunnif, J. B.; Vauros, P. J. Am. Soc. Mass Spectrom. 1995, 6, hosts. (iii) The behavior of the open chain oligosaccha- 437–447. ؅ ؅ 21. Camilleri, P.; Haskins, N. J.; New, A. P.; Saunders, M. R. Rapid rides 6–9 and 6 –8 depends on the molecular weight Commun. Mass Spectrom. 1993, 7, 949–952. rather than on the presumed complexing capabilities 22. Selva, A.; Redenti, E.; Zanol, M.; Ventura, P.; Casetta, B. Eur. related to their molecular conformation. Mass Spectrom. 1995, 1, 105–106 (and errata corrige, Eur. Mass Spectrom. 1995, 1, 330). 23. Selva, A.; Redenti, E.; Zanol, M.; Ventura, P.; Casetta, B. J. Mass Supplementary Material Available Spectrom. 1996, 31, 1364–1370. Figure 1. Expansion of the NMR spectra of (a) top: free 24. Selva, A.; Redenti, E.; Zanol, M.; Ventura, P.; Casetta, B. J. Mass ␤-cyclodextrin and (b) bottom: 1:1 complex ␤-cyclodex- Spectrom. 1998, 33, 729–734. 25. Redenti, E.; Ventura, P.; Fronza, G.; Selva, A.; Rivara, S.; trin-DHB. Only the spectral region related to the carbo- Plazzi, P.; Mor, M. J. Pharm. Sci. 1999, 88, 599–607. hydrate protons is displayed. 26. Bakhtiar, R.; Kaifer, A. E. Rapid Commun. Mass Spectrom. 1998, Figure 2. Expansion of the NMR spectra of (a) top: 12, 111–114. free permethyl-␤-cyclodextrin and (b) bottom: 1:1 com- 27. Sun, W.; Cui, M.; Liu, S.; Song, F.; Nikolaevich, Y. Rapid plex permethyl-␤-cyclodextrin-DHB. Only the spectral Commun. Mass Spectrom. 1998, 12, 2016–2022. region related to the carbohydrate protons is displayed. 28. He, F.; Ramirez, J.; Garcia, B. A.; Lebrilla, C. B. Int. J. Mass Table 1. NMR chemical shift variations of cyclodex- Spectrom. 1999, 182/183, 261–273. trins and DHB protons for the 1:1 complexes of CDs 11 29. Harvey, D. J. J. Chromatogr. A 1996, 720, 429–446. 30. Bartsch, H.; Ko¨nig, W. A.; Straßner, M.; Hintze, U. Carbohydr. and 14 with DHB. Res. 1996, 286, 41–53. 31. Chankvetadze, B.; Endresz, G.; Blaschke, G.; Juza, M.; Jaku- Acknowledgments betz, H.; Schurig, V. Carbohydr. Res. 1996, 287, 139–155. 32. Nonami, H.; Tanaka, K.; Fukuyama, Y.; Erra-Balsells, R. Rapid The authors wish to thank Professor Antonio Selva for discussions Commun. Mass Spectrom. 1998, 12, 285–296. and encouragement, Mr Walter Panzeri (CNR—Milano) and Mr 33. Nonami, H.; Fukui, S.; Erra-Balsells, R. J. Mass Spectrom. 1997, Emanuele Canu (Shimadzu Italia—Milano) for technical assis- 32, 287–296. tance. Dr. Marcus Juza (ETH, Zu¨ rich) provided a sample of 34. Hao, C.; Ma, X.; Fang, S.; Liu, Z.; Liu, S.; Song, F.; Liu, J. Rapid compound 17 as a generous gift. Commun. Mass Spectrom. 1998, 12, 345–348. 35. Fauvelle, F.; Jaquinod, M.; Pe´tillot, Y.; Forest, E. Eur. Mass Spectrom. 1996, 2, 381–384. References 36. Yamagaki, T.; Ishizuka, Y.; Kawabata, S.; Nakanishi, H. Rapid 1. Chem. Rev. (Special Issue) 1998, 98. Commun. Mass Spectrom. 1996, 10, 1887–1890. 2. Fro¨mming, K.-H.; Szejtli, J. Cyclodextrins in Pharmacy; Kluwer 37. Yamagaki, T.; Nakanishi, H. Rapid Commun. Mass Spectrom. Academic: Dordrecht, 1994. 1998, 12, 1069–1074. 3. See, for example, Duby, P.; Huynh-Ba, T. Eur. Pat. Appl. EP 38. Cancilla, M. T.; Penn, S. G.; Lebrilla, C. B. Anal. Chem. 1998, 70, 545,085; Chem. Abstr. 1993, 119, 138068k. 633–672. 4. Inoue, Y. Annu. Rep. NMR Spectrosc. 1993, 27, 59–101. 39. Wong, C. K. L.; Chan, T.-W. D. Rapid Commun. Mass Spectrom. 5. Szejtli, J. Cyclodextrin Technology; Kluwer Academic: Dordre- 1997, 11, 513–519. cht, 1988; pp 92–97 and references therein. 40. Henke, C.; Steinheim, C.; Janshoff, A.; Steffan, G.; Luftmann, 6. Saenger, W. Angew. Chem. Int. Ed. 1980, 19, 344–362. H.; Sieber, M.; Galla, H. Anal. Chem. 1996, 68, 3158–3165. 7. Vincenti, M. J. Mass Spectrom. 1995, 30, 925–939. 41. Gallagher, R. T.; Ball, C. P.; Gatehouse, D. R.; Gates, P. J.; 8. Madhusudanan, K. P.; Katti, S. B. J. Mass Spectrom. 1998, 33, Lobell, M.; Derrick, P. J. Int. J. Mass Spectrom. Ion Processes 1017–1022. 1997, 166, 523–531. 236 MELE AND MALPEZZI J Am Soc Mass Spectrom 2000, 11, 228–236

42. Williams, T. L.; Fenselau, C. Eur. Mass Spectrom. 1998, 4, 53. See, for example, Tables 4 and 5 of [39]. 379–383. 54. The detailed crystal data have been deposited into the Cam- 43. Xiang, F.; Beavis, R. C. Org. Mass Spectrom. 1993, 28, 1424– bridge Crystallographic Data Centre as supplementary publica- 1429. tion No. CCDC 132368. Copies of the data can be obtained, 44. Rosinke, B.; Strupat, K.; Hillenkamp, F.; Rosenbusch, J.; free of charge, on application to the CCDC, 12 Union Road, Dencher, N.; Kru¨ ger, U.; Galla, H. J. Mass Spectrom. 1995, 30, Cambridge CB21EZ, UK (fax: ϩ44-(0)12233336033 or e-mail: 1462–1468. [email protected]). 45. Juhasz, P.; Biemann, K. Carbohydr. Res. 1995, 270, 131–147. 55. Schurig, V.; Grosenick, H. J. Chromatogr. A 1994, 666, 617–625. 46. Glocker, M. O.; Bauer, S. H. J.; Kast, J.; Volz, J.; Przybylski, M. 56. Sheldrick, G. M. SHELXTL; Siemens Analytical X-ray Instru- 1996, J. Mass Spectrom. 31, 1221–1227. ments Inc.; Madison, Wisconsin, 1990. 47. Bekir, S.; Masselon, C.; Zenobi, R. J. Mass Spectrom. 1998, 33, 57. Allen, F. H.; Kennard, O. Chem. Des. Autom. News 1993, 8, 994–1002. 31–37. 48. Jespersen, S.; Niessen, W. M. A.; Tjaden, U. R.; van der Greef, 58. Sheldrick, G. M. SHELXL97. Release 97-2. Program for the J. J. Mass Spectrom. 1998, 33, 1088–1093. Refinement of Crystal Structures; University of Go¨ttingen, 49. Farmer, T. B.; Caprioli, R. M. J. Mass Spectrom. 1998, 33, 697–704. Germany, 1999. 50. Moniatte, M.; Lesieur, C.; Vecsey-Semjen, B.; Buckley, J. T.; 59. Wong, A. W.; Cancilla, M. T.; Voss, L. R.; Lebrilla, C. B. Anal. Pattus, F.; van der Goot, F. G.; Van Dorsselaer, A. Int. J. Mass Chem. 1999, 71, 205–211. Spectrom. Ion Processes 1997, 169, 179–199. 60. Knochenmuss, R.; Dubois, F.; Dale, M. J.; Zenobi, R. Rapid 51. Joliffe, K. A.; Crego Calama, M.; Fokkens, R.; Nibbering, Commun. Mass Spectrom. 1996, 10, 871–877. N. M. M.; Timmermann, P.; Reinhoudt, D. N. Angew. Chem. 61. Zenobi, R.; Knochenmuss, R. Mass Spectrom. Rev. 1998, 17, Int. Ed. 1998, 37, 1247–1251. 337–366. 52. Crego Calama, M.; Hulst, R.; Fokkens, R.; Nibbering, N. M. M.; 62. Lee, S.; Wyttenbach, T.; Bowers, M. T. Int. J. Mass Spectrom. Ion Timmermann, P.; Reinhoudt, D. N. Chem. Commun. 1998, Processes 1997, 167/168, 605–614. 1021–1022. 63. Wenz, G. Angew. Chem. Int. Ed. 1994, 33, 803–822.