Matrix-assisted laser desorption ionization and time-of-¯ightmass spectrometry for the sensitive determination of
arylamide±deoxynucleoside adducts
M. Paul Chiarelli*, Xiaomei Gu, Allison A. Aldridge, HuaPing Wu
Dept. of Chemistry, Loyola University, 6525 N Sheridan Rd., Chicago, IL 60626, USA
Received 19 September 1997; received in revised form 19 December 1997; accepted 8 January 1998
Abstract
Matrix-assisted laser desorption/ionization (MALDI) and time-of-¯ight mass spectrometry (TOFMS) are tested for the
determination of arylamide adducts of deoxyguanosine. Fifteen matrices were tested for the sensitive determination of N-
(guanosin-8-yl)-2-acetylamino¯uorene, (dG-C8-AAF), a compound whose structure is representative of carcinogen±DNA
adducts formed by nitroaromatic and arylamine carcinogens. Eight matrices allow adduct detection when 560 fmole or less is
introduced into the TOFMS. Two matrices, a-cyano-4-hydroxycinnamic acid (CHCA) and mercaptobenzothiazole (MBT),
promoted adduct detection when 56 fmol or less of adduct is introduced into the TOFMS. The other seven matrices did not
permit dG-C8-AAF adduct detection at the low pmol level. In all determinations the limit of detection (LOD) is based on the
nucleic acid±arylamide, BH�2 fragment ion formed by the cleavage of the glycosidic bond. The relation between matrix
structure and the mechanism of analyte ionization is discussed as well. # 1998 Elsevier Science B.V.
1. Introduction
The study of carcinogen-modi®ed deoxynucleo-
sides is motivated by the belief that these compounds
are formed in the genesis of cancer. The covalent
attachment of a metabolized carcinogen to a DNA
molecule can induce misreplication(s) in the DNA
strand that may eventually give rise to the formation of
tumors [1]. The structural analysis of carcinogen-
modi®ed DNA derived from in vivo sources is dif®cult
because these molecules are formed in such small
quantities. Analytical methods possessing the sensi-
tivity to detect such low levels of adducts, e.g., 32P-
postlabeling combined with TLC and immunoassays
[2], provide little or no structural information regard-
ing unknown adducts. As a consequence, many DNA
adducts that have been isolated from human sources
have unknown structures. The analysis of adducts
isolated from human sources with analytical metho-
dology based on mass spectrometry may yield insight
into the identity of otherwise unknown environmental
genotoxins as well as mechanisms of carcinogenesis
(metabolism and DNA adduct formation) implied by
speci®c adduct structures. Several mechanisms of
carcinogen metabolism in route to DNA modi®cation
in humans have been suggested [1], however, identi-
fying those primarily responsible for human DNA
modi®cation may require adduct detection and pro-
Analytica Chimica Acta 368 (1998) 1±9
*Corresponding author.
0003-2670/98/$19.00 # 1998 Elsevier Science B.V. All rights reserved.
P I I S 0 0 0 3 - 2 6 7 0 ( 9 8 ) 0 0 0 7 6 - 2
duct ion analysis at the attomole±fmol level depending
upon the tissue from which the DNA is extracted and
the quantity of DNA isolated. The focus in this study
are those DNA adducts formed by arylamine and
nitroaromatic compounds that are metabolized to an
N-hydroxy intermediate and bind to DNA predomi-
nantly at the C8 position of guanine.
Applications of mass spectrometry to the analysis
of DNA damage in humans have been restricted to the
analysis of targeted analytes [3±5]. Most attempts to
detect unknown DNA adducts with structural infor-
mation have been based on fast atom bombardment
(FAB) and tandem mass spectrometry [6]. However,
the LODs associated with methods based on FAB have
not been low enough to obtain structural information
from DNA adducts isolated from human sources. The
coupling of electrospray ionization (ESI) with tandem
MS methods may allow for the structural character-
ization of unknown DNA adducts isolated from
humans. The coupling of capillary chromatography
with ESI has provided low fmol level LODs for
aromatic DNA adducts [7,8]. It is anticipated that
re®nements in the ESI process (e.g., nanospray) that
have enabled sequencing of large peptides at the fmol
level [9] will enable tandem MS determinations of
unknown DNA adducts isolated from humans. In one
study, nanospray ionization coupled with tandem
quadrupole MS permitted product ion analysis of a
diglycidyl ether adduct of deoxyguanosine monophos-
phate at the fmol level [10].
Matrix-assisted laser desorption ionization
(MALDI) may permit structural analysis of DNA
adducts from human sources. The compatibility of
MALDI with multichannel detectors such as time-of-
¯ight [11,12] and Fourier transform mass spectro-
meters [13,14] may allow the molecular weight and
product ion analysis of DNA adducts isolated from
humans with the same sensitivity as a single-ion or
single-reaction monitoring determination performed
with quadrupole mass analyzers. The choice of the
matrix is crucial for the sensitive determination of a
particular analyte. Nucleic acid adducts of PAH
composed of ®ve and six benzene rings have been
detected with as little as 55 fmol introduced into
the TOFMS using matrices whose utility was antici-
pated based on their structural features [11]. In this
study 15 different matrices were surveyed for the
determination of arylamine adducts of deoxyguano-
sine. Eight matrices were found to provide LODs
with 560 fmol or less and two matrices with 56 fmol
or less introduced into the TOFMS. The sensitivity
of the determinations obtained with different matrices
provides some insight into the mechanisms of
desorption and ionization associated with these
analytes.
2. Experimental
The TOFMS used in these studies is of a modi®ed
Wiley±McClaren design and has been described in
detail previously [15] (R.M. Jordan, Grass Valley,
CA). A nitrogen laser (337 nm, 6 mW peak laser
power, 3 ns pulse width, and a � 400 mm2 spot size)
was used to induce desorption (model VSL-337ND,
Laser Science, Newton, MA). The laser output is
focussed onto the probe at a 458 angle to the probe
surface with a single 125 mm focal length silica lens.
The probe tip is inserted into the vacuum interlock and
into the acceleration region so the probe tip is ¯ush
with the ®rst acceleration plate. The detector is a dual
microchannel plate (MCP) operated at a potential of
ÿ1.9 keV. Data is recorded on a Tektronix TDS 520A
oscilloscope and transferred to a IBM-compatible 486/
66 PC for processing. All spectra were acquired in the
positive ion mode.
N-(deoxyguanosin-8-yl)-2-acetylamino¯uorene
(dG-C8-AAF) was supplied by Dr. Frederick A.
Beland at the National Center for Toxicological
Research in Jefferson, AR. 7-Methylguanosine was
purchased from Sigma Chemical (St. Louis, MO).
Both compounds were used without further puri®ca-
tion. Stock solutions (0.25 mg/ml) were prepared by
dissolving each in methanol.
Fifteen different matrices were tested as vehicles for
arylamine±deoxyguanosine adduct desorption and
ionization. These compounds were chosen because
of their demonstrated utility as matrices for other
compounds. Thirteen were purchased from commer-
cial sources (Sigma and Aldrich Chemical) and used
without further puri®cation. 4-Phenyl-a-cyanocin-
namic acid (PCC) and 4-benzyloxy-a-cyanocinnamic
acid (BCC) were synthesized as described in the
literature [11]. Their NMR spectra and MALDI mass
spectra are consistent with those reported previously
[11]. All matrix stock solutions were prepared in 2:1
2 M.P. Chiarelli et al. / Analytica Chimica Acta 368 (1998) 1±9
mixture of acetonitrile:water, made 0.1% in tri¯uoro-
acetic acid. The concentration of each of the matrices
was varied throughout this study to determine the
matrix:analyte ratio that promoted the best sensitivity
(normally about 250:1). The concentrations of the
matrix solutions used in this study for adduct deter-
minations at 560 fmol level and lower were between
120±180 pmol/ml. Samples were prepared for MALDI
analysis by combining nine parts of the matrix solu-
tion with one part of the analyte solution and deposit-
ing 1 ml of the resulting mixture on the sample probe to
dry under ambient conditions prior to insertion into the
TOFMS.
Samples were prepared for quantitative analysis by
combining 1 ml of the dG-C8-AAF solution of vari-
able concentration with 1 ml of the 7-methylguanosine
internal standard (15 pmol/ml) and 8 ml of 0.82 mM 4-
hydroxy-a-cyanocinnamic acid (CHCA). The same
concentration of CHCA matrix was used throughout
the quanti®cation study. 1 ml was placed on the probe
and allowed to dry under ambient conditions. Four
spectra were acquired from different areas of each
sample probe loading. Twenty to thirty shots were
averaged for each spectra. Two probe loadings were
analyzed and eight spectra were averaged for each
point on the graph.
HPLC analyses were conducted with a Beckman
HPLC system (model 421A controller, model 110B
pumps, and a model 163 variable wavelength detector)
at a wavelength of 280 nm. Solutions of guanine
(Rt�18 min), deoxyguanosine (Rt�21 min), and dG-
C8-AAF (Rt�25.5 min) were analyzed individually
and combined with CHCA (Rt�12.5 min) using a
water/methanol gradient (0±15 min:10%±85% metha-
nol, 15±30 min: 85%±100% methanol) at a ¯ow rate
of 1 ml/min.
3. Results and discussion
3.1. AAF adduct fragmentation
One of the major factors in¯uencing the `̀ in-
source'' and `̀ post-source'' fragmentation of analytes
in MALDI-TOFMS is the matrix used as a vehicle for
desorption and ionization. No MALDI spectra
acquired with the matrices shown in Fig. 1 gave a
protonated molecular ion with less than 5 pmol of
AAF adduct introduced into the mass spectrometer.
The only sample-speci®c ion observed in the MALDI
spectra of the AAF adducts is the protonated guanine±
AAF BH�2 ion at m/z 373 formed by cleavage of the
glycosidic bond (reaction 1).
2,6-Dihydroxyacetophenone (2,6-DHAP) [16] is
one such matrix that allows AAF adduct detection
at subpmol levels. 2,6-DHAP is a matrix that has been
shown to minimize the in-source fragmentation (rela-
tive to CHCA) of many peptides that possess post-
translational modi®cations [17]. In this study
approximately 50% of the MALDI spectra
acquired with 5±6 pmol of the AAF adduct on the
probe in a 2,6-DHAP matrix gave a molecular ion at
m/z 489. These results appear to be consistent with the
MALDI spectra of diol-epoxide adducts of deoxygua-
nosine acquired with FTMS [13] and TOFMS [14]
detection. In these studies sample loadings of 20±
70 pmol yielded predominantly BH�2 ions with a 10±
80% relative abundance of (M�H)� ions. MALDI
M.P. Chiarelli et al. / Analytica Chimica Acta 368 (1998) 1±9 3
studies of a variety of small molecules [18] and results
obtained in this study indicate that normal deoxygua-
nosine undergoes the same extent of glycosidic clea-
vage as the AAF adducts studied here and the aromatic
adducts cited above, suggesting aromatic DNA
adducts of deoxyguanosine may yield only BH�2 ions
when subpmol quantities of adducts are determined.
This is in marked contrast to diol-epoxide adducts of
deoxyadenosine [13,14], where (M�H)� ions are
found to be the most abundant ions in MALDI
determinations. The lack of glycosidic cleavage in
deoxyadenosine adducts may be attributed to the
greater glycosidic bond strength of deoxyadenosine
relative to the other three normal deoxynucleosides
[19]. MALDI analysis of deoxyadenosine in a CHCA
matrix in our laboratory was found to yield pre-
dominantly protonated molecular ions when analyzed
under the same laser power conditions as dG-C8-
AAF.
The observation of a BH�2 and (M�H)� ion 116
mass units apart (the mass of the neutral deoxyribose
lost) could serve as a signature for an unknown DNA
adduct in a MALDI spectrum. The observation of both
BH�2 and (M�H)� ions would make the selection of
the BH�2 for product ion analysis (from the same
sample preparation) by post-source decay straightfor-
ward. The desorption/ionization characteristics of the
AAF adduct obtained with the matrices in Fig. 1
suggest only the BH�2 ion may be observed in MALDI
spectra of DNA adducts acquired with quantities of
adducts likely to be obtained from a human source
(e.g., placenta or lung tissue). For instance, 15 mg of
DNA having a substitution level of one adduct in 108
nucleotides will contain 460 fmol of adduct. There-
fore the selection of a potential DNA adduct BH�2 ion
for post-source decay product ion analysis [20] may be
dif®cult depending upon the complexity of the
MALDI spectra.
Fig. 1. The structures of the matrices that promoted detection of the BH�2 ion from dG-C8-AAF with 560 fmol or less introduced into the mass
spectrometer.
4 M.P. Chiarelli et al. / Analytica Chimica Acta 368 (1998) 1±9
One approach that may minimize in-source frag-
mentation of the glycosidic bond is to introduce a short
(nanosecond) time delay between the laser desorption/
ionization event and the extraction of the ions from the
source [21±23]. During the desorption event, the
region between the repeller and extraction plate is
®eld-free. This time delay will minimize the collision
energy of the desorbing deoxynucleoside ions and the
neutral species (matrix molecules, etc.) and hopefully
reduce fragmentation because the deoxynucleoside
ions are not accelerated during the ®rst few moments
of the desorption event. However, delayed extraction
often yields precursor ions with too little internal
energy to yield signi®cant amounts of sample-speci®c
fragment ions during product ion analysis by post-
source decay [24]. These results suggest that once a
BH�2 ion associated with a carcinogen±deoxyguano-
sine adduct is observed in a MALDI spectrum
acquired using delayed extraction, product ion analy-
sis of the BH�2 ion may have to be carried out under
continuous extraction conditions.
The possibility that the adducts may be undergoing
an acid-catalyzed glycosidic cleavage in solution con-
taining the matrix or as the matrix preparation was
drying on the probe was investigated using HPLC with
UV detection. The HPLC±UV analyses of CHCA
matrix solutions (pH�2) and dried probe preparations
(redissolved in methanol) containing guanine, dG, or
dG-C8-AAF suggest that glycosidic cleavage of the
adduct does not occur prior to laser desorption. Evi-
dence for degradation of dG-C8-AAF in the matrix
solution became apparent in the chromatogram only
after heating the matrix/adduct solution for 24 h at
378C. These results are consistent with earlier studies
of oligonucleotides by MALDI-TOFMS and HPLC±
UV where no degradation of the oligonucleotide in
solution with the matrix was apparent prior to MS
analyses [19].
3.2. LOD determinations
All matrices in Fig. 1 were found to yield repro-
ducible observations of the BH�2 fragment ion with at
least 560 fmol loaded on the sample probe. PCC and
BCC gave reproducible signals with 280 fmol intro-
duced into the TOFMS. CHCA and MBT, matrices
whose utility was initially demonstrated for small
peptides [25,26], gave the most sensitive determina-
tions of any of the matrices tested. Reproducible
signals were obtained with 56 fmol loaded on the
Fig. 2. Positive ion MALDI-TOFMS spectra acquired with 560 fmol of dG-C8-AAF and (inset) 56 fmol of dG-C8-AAF introduced into the
TOFMS in the MBT matrix. The ions at m/z 168 and m/z 335.5 are the protonated matrix monomer and dimer ions, respectively.
M.P. Chiarelli et al. / Analytica Chimica Acta 368 (1998) 1±9 5
probe. Representative spectra obtained with the MBT
matrix are presented in Fig. 2. LOD determinations
were carried out by analyzing probe loadings with
progressively smaller amounts of adduct. The LODs
were de®ned as the smallest amount of analyte neces-
sary for reproducible observation of the BH�2 fragment
ion from at least four out of eight different areas
irradiated by the laser on a given probe for two
consecutive probe loadings (S/N�3). The only matrix
that provided signals with consistent intensity to
attempt a calibration plot was CHCA.
MALDI may be a viable means of quantifying DNA
adducts isolated from in vivo sources given the sensi-
tivity and speed of the determinations, and the inex-
pensive nature of the hardware. Motivation for
attempting this calibration plot stems from previous
success in quantifying small molecules by MALDI
[27] and earlier attempts to generate such calibration
plots for large PAH-nucleic acid adducts were unsuc-
cessful [11].
The calibration plot was constructed by combining
the matrix and internal standard with amounts of dG-
C8-AAF that we expect to isolate from 25 mg of
human placenta DNA [28]. Probe loadings containing
50, 100, 200, 600, 800, and 1200 fmol of dG-C8-AAF
and 1.5 pmol of the 7-methylguanosine internal stan-
dard were analyzed. Other details regarding sample
preparation and spectra acquisition are presented in
the experimental section. The 7-methylguanosine
internal standard undergoes glycosidic cleavage dur-
ing the desorption/ionization process (reaction 2), but
to a much lesser extent than dG-C8-AAF, deoxygua-
nosine, or guanosine. The best correlation is achieved
when the BH�2 peak height is plotted over the sum of
the peak heights of both ions associated with the 7-
methylguanosine internal standard. A plot of the ratio
of the ion abundances (m/z 373/(m/z 166� m/z 298))
versus fmol of dG-C8-AAF loaded on the probe is
described by the equation y�((2.43�0.36)�10ÿ4)x �(0.074�0.014). The ion abundance ratios varied from
0.10 to 0.55 and the standard error for the regression is
0.037. The correlation coef®cient (R2) is 0.92.
Comprehensive studies of the factors that govern
the quanti®cation of peptides by MALDI suggest the
best correlation is obtained when the internal standard
and analyte have similar structural features and the
matrix:analyte ratio is kept constant throughout the
determination. Errors in the determination brought
about by differences in analyte and matrix structure
(caused by different solubilities in the matrix and
evaporating solvent) are minimized by using a quan-
tity of internal standard throughout the determination
equivalent to the largest amount of analyte loaded on
the probe, as is done in this study [29]. These results
suggest that better correlations (R2�0.95) may be
obtained after the structure of a particular adduct is
deduced by comparison to its PSD product ion spectra
to that of in vitro-synthesized adduct standards. Then a
standard with a stable isotope label could be used to
estimate the concentration of that adduct in a tissue
sample. Representative spectra of the AAF adduct
acquired with the CHCA matrix and 7-methylguano-
sine internal standard are presented in Fig. 3.
It should be pointed out here that large abundances
of matrix cluster ions in the m/z range 100±1000 that
are observed in MALDI determinations of large pep-
6 M.P. Chiarelli et al. / Analytica Chimica Acta 368 (1998) 1±9
tides are not observed in this study, consistent with
other MALDI determinations of small molecules
[18,27]. This is due to the fact that less laser energy
and smaller matrix:analyte ratios (100:1, in some
cases) are suf®cient for the desorption and ionization
of small molecules relative to large peptides where
matrix:analyte ratios as large as 105:1 are used rou-
tinely [18,27]). In order for a MALDI determination to
be successful, the matrix must isolate analyte mole-
cules from each other. The absolute quantity of matrix
necessary to isolate small molecules (ca. 400±500 dal-
tons) from each other in the solid state is less than for a
peptide having a mass of several thousand daltons. In
the Fig. 2 spectra, only 130 pmol of MBT matrix are
deposited on the probe with the analyte. The only
matrix ions of signi®cant abundance observed in the
m/z 100±1000 range in these studies are the proto-
nated matrix molecular (M�H)� ion and, in some
cases, the protonated dimer (2M�H)� ion. When
matrix:analyte ratios associated normally employed
for the MALDI analysis of large peptides were used
for the MALDI determination of dG-C8-AAF, the
LOD increased. The LOD for dG-C8-AAF increased
to ca. 1 pmol (introduced into the TOFMS) in a 7�104
molar excess of CHCA.
Seven matrices did not yield any sample-speci®c
adduct ions upon laser desorption with low pmol
quantities of adducts. These are 2,5-dihydroxybenzoic
acid, thiosalicyclic acid, 3-hydroxypicolinic acid, 2,4-
dihydroxyacetophenone, 2,4,6-trihydroxyacetophe-
none, salicylamide, and anthranilic acid. 2,5-Dihy-
droxybenzoic acid and 3-hydroxypicolinic acid
were tested because they permitted detection of
diol-epoxide DNA adducts at the 20±70 pmol level
[13,14]. None of these matrices that are useful for
oligonucleotide analyses were useful for the analysis
of dG-C8-AAF.
3.3. Mechanistic considerations
Central to improving analytical determinations of
DNA adducts based on MALDI is a better under-
standing of the mechanisms of desorption and ioniza-
tion of these analytes. The functional groups of these
a-cyanocinnamic acids utilized in this study suggest
certain aspects of the ionization mechanism of these
DNA adducts in MALDI. The question of whether or
not matrix molecules in excited states can protonate
analyte molecules during the MALDI has been the
subject of much investigation and speculation
Fig. 3. Positive ion MALDI-TOFMS spectra of dG-C8-AAF acquired with 1200 fmol and (inset) 50 fmol introduced into the TOFMS in the
CHCA matrix.
M.P. Chiarelli et al. / Analytica Chimica Acta 368 (1998) 1±9 7
[30±33]. Such a mechanism is believed plausible
because many organic molecules are known to
undergo functional group-speci®c changes in pka upon
electronic excitation in solution [34]. The utility of the
methyl ester of CHCA as a matrix suggested that an
analyte protonation mechanism based on excited pro-
ton transfer was operative in MALDI [30]. Proton
transfer from the para-hydroxy group of the CHCA
methyl ester to the analyte is suggested in this study
because carboxylic acids undergo a modest increase in
pka upon excitation to the ®rst excited state (e.g., from
6 to 7 generally) while hydroxyl groups undergo a
decrease in pka, from 10 to 2 [34]. More recently, an
ionization mechanism based on excited state proton
transfer from a hydroxyl to a carbonyl group was used
to explain why ortho-hydroxy acetophenones were
useful as MALDI matrices and para- and meta-hydro-
xyacetophenones were not [33]. The utility of PCC
and BCC as matrices for the sensitive determinations
of AAF adducts in this study and in studies of PAH-
nucleic acids would indicate that an ionization
mechanism involving excited state proton transfer is
not facile in the determination of these adducts. The
structures of PCC and BCC are different from CHCA
insofar as the para-hydroxy group of CHCA is
replaced with a phenyl and benzyloxy group, respec-
tively. Therefore the proton associated with the ioni-
zation step must come from the carboxylic acid group.
Ionization from other protons (i.e., from the phenyl
ring) is not likely here, given that experiments
designed to distinguish random protonation reactions
from functional-group speci®c protonation reactions
in UV laser mass spectrometry indicate that phenyl
and aliphatic protons did not protonate analytes even
at power densities large enough to generate C� ions
[35].
4. Conclusion
Matrix-assisted laser desorption/ionization
(MALDI) and time-of-¯ight mass spectrometry
(TOFMS) is employed to promote detection of dG-
C8-AAF at the fmol level using eight different
matrices. Two matrices, a-cyano-4-hydroxycinnamic
acid and mercaptobenzothiazole, permitted adduct
detection with less than 60 fmol loaded on the laser
probe. These results suggest that the detection of
unknown arylamine adducts formed in human pla-
centa and lung is possible. Now that these adducts may
be detected at the low fmol level, structural analysis
using MALDI and post-source decay product ion
analysis of arylamine nucleoside adducts isolated
from humans should be feasible.
The structures of the matrices that promote fmol-
level detection of these arylamine adducts suggest that
the analyte is protonated by a matrix molecule in the
ground state rather than in excited state.
References
[1] F.A. Beland, M.C. Poirier, in: A.E. Sirica (Ed.), The
Pathobiology of Neoplasia, Plenum Press, New York, 1989,
pp. 57.
[2] M.C. Poirier, A. Weston, Environ. Health Perspect. 104
(Suppl. 5) (1996) 883.
[3] A.K. Chaudhary, M. Nokubo, G.R. Reddy, S.N. Yeola, J.D.
Morrow, I.A. Blair, L.J. Marnett, Science 265 (1994) 1580.
[4] M. MuÈller, F.J. Belas, I.A. Blair, F.P. Guengerich, Chem. Res.
Toxicol. 10 (1997) 242.
[5] T.-Y. Yen, N.I. Christova-Gueoguieva, N. Scheller, S. Holt,
J.A. Swenberg, M.J. Charles, J. Mass Spectrom. 31 (1996)
1271.
[6] M.P. Chiarelli, J.O. Lay, Jr., Mass Spectrom. Rev. 11 (1992)
447.
[7] J. Ding, P. Vouros, Anal. Chem. 69 (1997) 379.
[8] K. Vanhoutte, P. Joos, F. LemieÁre, W. Van Dongen, E.L.
Esmans, J. Mass Spectrom and Rapid Commun. Mass
Spectrom., (1995) S143.
[9] M. Wilm, A. Shevchenko, T. Houthaeve, S. Breit, L.
Schweigerer, T. Fotsis, M. Mann, Nature 379 (1996) 466.
[10] K. Vanhoutte, W. Van Dongen, I. Hoes, F. LemieÁre, E.L.
Esmans, H. Van Onckelen, E. Van den Eeckhout, R.E.J. van
Soest, A.J. Hudson, Anal. Chem. 69 (1997) 3161.
[11] M. George, J.M.Y. Wellemans, R.L. Cerny, M.L. Gross, K. Li,
E.L. Cavalieri, J. Am. Soc. Mass Spectrom. 5 (1994) 1021.
[12] L. Chen, P.D. Devanesan, J. Byun, J.K. Gooden, M.L. Gross,
E.G. Rogan, E.L. Cavalieri, Chem. Res. Toxicol. 10 (1997)
225.
[13] E.A. Stemmler, M.V. Buchanan, G.B. Hurst, R.L. Hettich,
Anal. Chem. 66 (1994) 1274.
[14] R. Hettich, G. Hurst, M. Buchanan, Polycyclic Aromat.
Compd. 6 (1994) 95.
[15] J. Bai, Y.H. Liu, D. Siemieniak, D.M. Lubman, Rapid
Commun. Mass Spectrom. 8 (1994) 687.
[16] J.J. Gorman, B.L. Ferguson, T.B. Nguyen, Rapid Commun.
Mass Spectrom. 10 (1996) 529.
[17] J.J. Pitt, J.J. Gorman, Rapid Commun. Mass Spectrom. 10
(1996) 1786.
[18] R. Lidgard, M.W. Duncan, Rapid Commun. Mass Spectrom.
9 (1995) 128.
8 M.P. Chiarelli et al. / Analytica Chimica Acta 368 (1998) 1±9
[19] L. Zhu, G.R. Parr, M.C. Fitzgerald, C.M. Nelson, L.M. Smith,
J. Am. Chem. Soc. 117 (1995) 6048.
[20] B. Spengler in: J.R. Chapman (Ed.), Protein and Peptide
Analysis by Mass Spectrometry, Methods in Molecular
Biology, Humana, Totowa, NJ, 61, 1996, pp. 43.
[21] Y. Zhu, L. He, J.R. Srinivasan, D.M. Lubman, Rapid
Commun. Mass Spectrom. 11 (1997) 987.
[22] P. Juhasz, M.T. Roskey, I.P. Smirnov, L.A. Haff, M.L. Vestal,
S.A. Martin, Anal. Chem. 68 (1996) 941.
[23] S.M. Colby, J.P. Reilly, Anal. Chem. 68 (1996) 1419.
[24] R. Kaufman, P. Chaurand, D. Kirsch, B. Spengler, Rapid
Commun. Mass Spectrom. 10 (1996) 1199.
[25] R.C. Beavis, T. Chaudhary, B.T. Chait, Org. Mass Spectrom.
27 (1992) 156.
[26] N. Xu, Z.-H. Huang, J.T. Watson, D.A. Gage, J. Am. Soc.
Mass Spectrom. 8 (1997) 116.
[27] M.W. Duncan, G. Matanovic, A. Cerpa-Poljak, Rapid
Commun. Mass Spectrom. 7 (1993) 1090.
[28] R.B. Everson, E. Randerath, R.M. Santella, R.C. Cefalo, T.A.
Avitts, K. Randerath, Science 231 (1986) 54.
[29] A.IA. Gusev, W.R. Wilkinson, A. Proctor, D.M. Hercules,
Fresenius' J. Anal. Chem. 354 (1996) 455.
[30] M.E. Gimon, L.M. Preston, T. Solouki, M.A. White, D.H.
Russell, Org. Mass Spectrom. 27 (1992) 827.
[31] M.P. Chiarelli, A.G. Sharkey, Jr.D.M. Hercules, Anal. Chem.
65 (1993) 307.
[32] R.D. Burton, C.H. Watson, J.R. Eyler, C.L. Lang, D.H.
Powell, M.Y. Avery, Rapid Commun. Mass Spectrom. 11
(1997) 443.
[33] J. Krause, M. Stoeckli, U.P. Schunegger, Rapid Commun.
Mass Spectrom. 10 (1996) 1927.
[34] J.F. Ireland, P.A.H. Wyatt in: V. Gold, D. Bethell (Eds.),
Advances in Physical Organic Chemistry, Academic Press,
London, 12, 1976, pp. 131.
[35] C.D. Parker, D.M. Hercules, Anal. Chem. 58 (1986) 25.
M.P. Chiarelli et al. / Analytica Chimica Acta 368 (1998) 1±9 9
