Peptide de novo sequencing facilitated by a dual-labeling strategy

Article abstract of DOI:10.1021/ac050540k

A novel peptide derivatization strategy based on guanidination and amidination is presented. Mass-coded labels help distinguish N- and C-terminal fragment ions produced by collision-induced dissociation and are of general utility since peptide N-termini a

Full text of DOI:10.1021/ac050540k

Anal. Chem. 2005, 77, 6300-6309
Peptide de Novo Sequencing Facilitated by a
Dual-Labeling Strategy
Richard L. Beardsley, Laura A. Sharon, and James P. Reilly*
Department of Chemistry, Indiana University, Bloomington, Indiana 47405
or Mascot.12 These algorithms compare MS/MS spectra to the
A novel peptide derivatization strategy based on guanidi-
nation and amidination is presented. Mass-coded labels
help distinguish N- and C-terminal fragment ions pro-
duced by collision-induced dissociation and are of general
utility since peptide N-termini are coded. The amidine
labels also promote specific fragmentation pathways that
elucidate N-terminal residues and provide valuable in-
ternal calibrants. This strategy is demonstrated with the
tryptic peptides of several model proteins, including two
that are phosphorylated. Additionally, interpreted peptide
sequences are matched against a database of over 80 000
proteins to assess the selectivity of this sequencing
approach.
hypothetical fragment ion masses of database sequences and
calculate scores that quantify the validity of assignments. Although
this general approach to protein identification has been success-
fully utilized in numerous experiments, database matching does
possess limitations. Since candidate sequences for assignments
are selected based on precursor ion masses, these algorithms will
be confused by database errors, genetic mutations, and modifica-
tions that occur either post-translationally or during sample
handling. Incorporating peptide modifications into the database
matching approach is often impractical since this dramatically
increases the size of the database leading to more false-positive
assignments and increased search times. Since the mapping of
post-translational modification (PTM) sites is critically important
for deciphering the functions of proteins, this represents a serious
drawback. Furthermore, organisms without sequenced genomes
cannot be studied using database-matching techniques. In light
of these limitations, there is a need for methods that extract
protein-identifying information directly from spectra without
comparison to databases. A number of different de novo sequenc-
ing approaches have been developed in recent years to help
achieve this goal.
De novo sequencing involves interpreting mass differentials
between consecutive peaks in a spectrum. Unfortunately, this
seemingly simple task represents a significant challenge for a
number of reasons. Peptides do not typically yield a complete,
contiguous series of ions. Sequencing ambiguity also arises
because leucine and isoleucine residues are isobaric (113.0841 u
each). Furthermore, the similar masses of lysine (128.0950 u) and
glutamine (128.0586 u) residues are indistinguishable in instru-
ments such as quadrupole ion traps that have low mass accuracy.
Adding to these obstacles, both N- and C-terminal fragment ions
are commonly formed by most activation methods and distin-
guishing them is not straightforward. If mass spacings between
peaks from different ion series (e.g., b- and y-ions) are assigned
to amino acids, incorrect sequences will be derived. Several
derivatization strategies have been developed to help overcome
some of these challenges. Some lead to more predictable
fragmentation patterns while others allow N- and C-terminal
fragment ions to be distinguished. Keough and co-workers
developed a strategy in which peptide N-termini are derivatized
with sulfonic acid groups.13,14 Due to the acidity of this group,
The investigation of biological systems by mass spectrometry
has rapidly evolved in recent years due to advancements in both
instrumentation and bioinformatics. While the development of this
field is ongoing, a number of technologies now exist that greatly
facilitate the characterization of complex biological mixtures.1,2
A
necessary component of this type of work is the ability to
confidently identify proteins in an expeditious manner. The two
most common approaches used to achieve this goal involve
tandem mass spectrometry3,4 and MALDI mass mapping.5-10 In
either experiment, peptides generated by proteolysis are analyzed
by following some form of chromatographic or electrophoretic
separation. Subsequently, proteins are assigned by comparing
mass spectrometric data with predictions based on sequences in
databases. In a typical proteomic experiment, thousands of spectra
may be submitted to automated search routines such as Sequest11
* To whom correspondence should be addressed. Fax: (812) 855-8300.
E-mail: reilly@indiana.edu.
(1) Aebersold, R.; Mann, M. Nature 2003, 422, 198-207.
(2) Wolters, D.; Washburn, M.; Yates, J. Anal. Chem. 2001, 73, 5683-5690.
(3) Hunt, D. F.; Yates, J. R.; Shabanowitz, J.; Winston, S.; Hauer, C. R. Proc.
Natl. Acad. Sci. U.S.A. 1986, 83, 6233-6237.
(4) Biemann, K.; Scoble, H. Science 1987, 237, 992-998.
(5) Mann, M.; Hendrickson, R. C.; Pandey, A. Annu. Rev. Biochem. 2001, 70,
437-473.
(6) Henzel, W.; Billeci, T.; Stults, J.; Wong, S.; Grimley, C.; Watanabe, C. Proc.
Natl. Acad. Sci. U.S.A. 1993, 90, 5011-5015.
(7) Pappin, D.; Hojrup, P.; Bleasby, A. Curr. Biol. 1993, 3, 327-332.
(8) Yates, J.; Speicher, S.; Griffin, P.; Hunkapiller, T. Anal. Biochem. 1993,
214, 397-408.
(9) James, P.; Quadroni, M.; Carafoli, E.; Gonnet, G. Protein Sci. 1994, 3, 1347-
1350.
(10) Mortz, E.; Vorm, O.; Mann, M.; Roepstorff, P. Biol. Mass Spectrom. 1994,
23, 249-261.
(12) Perkins, D. N.; Pappin, D. J. C.; Creasy, D. M.; Cottrell, J. S. Electrophoresis
1999, 20, 3551-3567.
(11) Eng, J. K.; Mccormack, A. L.; Yates, J. R. J. Am. Soc. Mass Spectrom. 1994,
5, 976-989.
(13) Keough, T.; Youngquist, R. S.; Lacey, M. P. Proc. Natl. Acad. Sci. U.S.A.
1999, 96, 7131-7136.
6300 Analytical Chemistry, Vol. 77, No. 19, October 1, 2005
10.1021/ac050540k CCC: $30.25 © 2005 American Chemical Society
Published on Web 08/24/2005

postsource decay13 and collision-induced dissociation14 (CID)
produce only y-type ions. The absence of other fragment ion types
facilitates de novo sequence interpretation. Although the presence
of proline residues sometimes limits sequence coverage, this
technique generally leads to highly informative fragmentation
spectra. A drawback of this approach is decreased sensitivity in
positive ion mode experiments due to the presence of a permanent
negative charge. Gaskell and co-workers demonstrated that
labeling peptide N-termini with phenyl isothiocyanate enables gas-
phase Edman degradation.15 They observed enhanced cleavage
of the N-terminal peptide bond leading to the predominant
formation of complementary b1 and yn-1 fragment ion pairs in CID
experiments. Although this enables identification of N-terminal
residues, no other sequence ions are typically formed due to the
high efficiency of this fragmentation pathway. Despite this
limitation, they demonstrated that, with high mass accuracy FT-
ICR-MS, the N-terminal residue identity and precursor ion mass
were sufficient constraints to enable database matching.16 Of
course, this would not be sufficient to identify PTM sites, genetic
mutations or account for database errors.
A number of groups have taken advantage of isotopic labeling
that, in addition to serving as mass-coded tags for relative
quantitation, have also provided mass signatures for N- and
C-terminal fragment ions and facilitated de novo sequencing. For
example, proteolytic 18O/16O labeling has been utilized to code
peptide C-termini during the course of tryptic digestion.17-19 As a
result, y-type fragment ions were formed as mass-separated pairs
that could easily be distinguished from b-ions. Alternatively,
several researchers have implemented deuterium labeling strat-
egies.20-22 These methods typically involve postdigestion chemical
derivatizations. For example, Goodlett and co-workers esterified
acidic groups with either D0- or D3-methanol to facilitate the
differentiation of C- and N-terminal ion-types.21 Similarly, James
and co-workers modified peptide N-termini with D0-/D4-nicotinyl-
N-hydroxysuccininmide to identify N-terminal fragment ions in
complementary MS/MS spectra.20 While this approach to de novo
sequencing is globally applicable, the derivatization chemistry
leads to unwanted side reactions with lysine and tyrosine residues
that complicate data. As an alternative to most deuterium-based
labeling techniques, Chen and co-workers devised an in vivo D4-/
D0-lysine incorporation strategy that takes advantage of the
specificity of tryptic proteolysis to generate C-terminally mass-
coded peptides.22 This approach provides a means of avoiding wet
chemistry steps following tryptic digestion. However, it is not of
general utility since the introduction of an isotopic label in vivo is
not always feasible. Also, tryptic peptides terminated with arginine
do not contain the isotopic label. Although it does not involve
isotopic labeling, the mass-coded abundance tagging technique
developed by Cagney and Emili also employs a mass signature
at the C-termini of tryptic peptides.23 In this approach, lysine
residues are converted to homoarginines using O-methylisourea
and are thus mass separated from their unlabeled counterparts.
Therefore, the C-termini of lysine-containing tryptic peptides are
shifted by 42 Da and the C-terminal fragment ions can be
distinguished from N-terminal ones. Much like the approach
involving in vivo incorporation of D4-/D0-lysine, this technique is
not global, as arginine-containing peptides are not labeled.
Similarly, Brancia and co-workers recently presented a strategy
that employed 15N- and 13C-coded guanidine tags to impart a 3-Da
mass shift on lysine residues and facilitate sequence interpreta-
tion.24
We now present a derivatization strategy that utilizes both
guanidination25-27 and amidination28,29 to assist peptide sequencing.
This approach facilitates identification of N- and C-terminal
fragment ions by labeling N-termini with amidine moieties that
differ by a methylene group (i.e., 14 u). Lysine residues are
converted to homoarginines to prevent amidination of the side-
chain ꢀ-amino groups. The simple and efficient reactions are
inexpensive and can be completed rapidly with minimal side
reactions.28,30 This is a global approach to protein identifications
since peptide N-termini are mass-coded. The only peptides
excluded from this labeling are those whose N-terminus is blocked
(e.g., by acetylation of the protein N-terminus). Furthermore, the
amidine groups promote specific fragmentation pathways29 that
facilitate de novo sequencing by providing sequence information
that is often absent and internal calibrants that can dramatically
improve mass accuracy. We demonstrate this sequencing ap-
proach using the tryptic peptides of several standard proteins.
EXPERIMENTAL SECTION
Materials. Hemoglobin (human), R-casein (bovine), serum
albumin (bovine, BSA), and TPCK-treated trypsin (bovine) were
obtained from Sigma (St. Louis, MO). Tris(hydroxymethyl)-
aminomethane (Trizma base), S-methylisothiourea hemisulfate,
and ammonium hydroxide were also supplied by Sigma. Aceto-
nitrile and trifluoroacetic acid (TFA) were purchased from EM
Science (Gibbstown, NJ). Thiopropionamide was supplied by TCI
America (Portland, OR). Anhydrous diethyl ether, thioacetamide,
and ammonium bicarbonate were purchased from Fisher (Fair
Lawn, NJ). Iodomethane, iodoacetamide, dithiothreitol (DTT),
poly(propylene glycol), and formic acid were obtained from
(14) Keough, T.; Lacey, M. P.; Fieno, A. M.; Grant, R. A.; Sun, Y. P.; Bauer, M.
D.; Begley, K. B. Electrophoresis 2000, 21, 2252-2265.
(15) Summerfield, S. G.; Steen, H.; O’malley, M.; Gaskell, S. J. Int. J. Mass
Spectrom. 1999, 188, 95-103.
(16) Van Der Rest, G.; He, F.; Emmett, M. R.; Marshall, A. G.; Gaskell, S. J. J.
Am. Soc. Mass Spectrom. 2001, 12, 288-295.
(17) Shevchenko, A.; Chernushevich, I.; Ens, W.; Standing, K. G.; Thomson, B.;
Wilm, M.; Mann, M. Rapid Commun. Mass Spectrom. 1997, 11, 1015-
1024.
(18) Qin, J.; Herring, C. J.; Zhang, X. L. Rapid Commun. Mass Spectrom. 1998,
12, 209-216.
(23) Cagney, G.; Emili, A. Nat. Biotechnol. 2002, 20, 163-170.
(24) Brancia, F. L.; Montgomery, H.; Tanaka, K.; Kumashiro, S. Anal. Chem.
2004, 76, 2748-2755.
(25) Beardsley, R. L.; Karty, J. A.; Reilly, J. P. Rapid Commun. Mass Spectrom.
(19) Uttenweiler-Joseph, S.; Neubauer, G.; Christoforidis, A.; Zerial, M.; Wilm,
2000, 14, 2147-2153.
M. Proteomics 2001, 1, 668-682.
(20) Munchbach, M.; Quadroni, M.; Miotto, G.; James, P. Anal. Chem. 2000,
72, 4047-4057.
(26) Brancia, F. L.; Oliver, S. G.; Gaskell, S. J. Rapid Commun. Mass Spectrom.
2000, 14, 2070-2073.
(27) Hale, J. E.; Butler, J. P.; Knierman, M. D.; Becker, G. W. Anal. Biochem.
(21) Goodlett, D. R.; Keller, A.; Watts, J. D.; Newitt, R.; Yi, E. C.; Purvine, S.;
Eng, J. K.; Von Haller, P.; Aebersold, R.; Kolker, E. Rapid Commun. Mass
Spectrom. 2001, 15, 1214-1221.
(22) Gu, S.; Pan, S. Q.; Bradbury, E. M.; Chen, X. Anal. Chem. 2002, 74, 5774-
5785.
2000, 287, 110-117.
(28) Beardsley, R. L.; Reilly, J. P. J. Proteome Res. 2003, 2, 15-21.
(29) Beardsley, R. L.; Reilly, J. P. J. Am. Soc. Mass Spectrom. 2004, 15, 158-
167.
(30) Beardsley, R. L.; Reilly, J. P. Anal. Chem. 2002, 74, 1884-1890.
Analytical Chemistry, Vol. 77, No. 19, October 1, 2005 6301

Aldrich (Milwaukee, WI). Octadecyl-derivatized silica gel (BioBa-
sic 18) was supplied by Thermo Electron (San Jose, CA).
Synthesis of S-Methyl Thioacetimidate and S-Methyl
Thiopropionimidate.28,31Thioacetamide (11 g) was dissolved in
1 L of anhydrous diethyl ether. Subsequently, 8.8 mL of io-
domethane was added to this solution and the mixture was allowed
to stand at room temperature for 14 h. The precipitate was
collected by vacuum filtration and stored over desiccant at ambient
temperature without further purification.
Scheme 1. Guanidination of Lysine Residues
with S-Methylisothiourea (A) and Amidination of
N-Termini with S-Methylthioacetimidate or
S-Methylthiopropionimidate (B)
Thiopropionamide (1.8 g) was dissolved in 100 mL of 99.5%
pure acetone. This solution was warmed to 60 °C in a water bath
before adding 3.8 mL of iodomethane. The reaction mixture was
incubated for 1 h, without stirring, at the bath temperature. The
product was collected after evaporation of the solvent in a vacuum
chamber. The crystals were stored at ambient temperature over
desiccant and were not further purified. It was not necessary to
synthesize new batches of amidination reagents for each labeling
experiment as these could be stored for several months without
significant degradation.
Tryptic Digestions. Tryptic peptides from R-casein, hemo-
globin, and albumin were generated using TPCK-treated trypsin.
Stock solutions of R-casein and hemoglobin (100 µM) were
prepared in 25 mM ammonium bicarbonate. Prior to tryptic
digestion, the disulfide bonds of BSA were disrupted by reduction
with DTT and the free sulfhydryl groups carbamidomethylated
using iodoacetamide. An aliquot of 20 mM DTT buffered in 100
mM ammonium bicarbonate was added to a 400 µM solution of
aqueous BSA in a 1:1 (v/v) ratio. This mixture was incubated at
70 °C for 45 min. Equal-volume aliquots of the reduced sample
and 110 mM aqueous iodoacetamide were combined and stored
in the dark at ambient temperature for 30 min to complete the
carbamidomethylation of cysteine residues. Tryptic digestions
were performed by adding 100 µL of protein stock or reduced/
carbamidomethylated solutions to 5 µg of lyophilized trypsin. Each
digestion was allowed to incubate at 37 °C for 12 h before being
stored at -20 °C.
Labeling Reactions. Peptides were derivatized using both
guanidination and amidination reactions. First, lysine residues
were converted to homoarginines using S-methylisothiourea
hemisulfate (Scheme 1A). A 1 M mixture of this reagent was
prepared in 6% NH4OH (v/v) and combined with 20 µL of digest
solution (100 µM protein) in a 1:1 ratio (v/v). The reaction mixture
was incubated for 1 h at 65 °C. Prior to performing the amidination
reactions, NH4OH was removed using a SpeedVac (Jouan,
Winchester, VA). Next, the guanidinated peptide mixture was
reconstituted in H2O. Acetamidination and propionamidination
derivatizations were performed as previously described by Beard-
sley and Reilly (Scheme 1B).28 However, in the present experi-
ments, only N-termini were labeled since lysine residues were
already guanidinated. A 43.4 g/L solution of S-methyl thioace-
timidate was prepared in 250 mM Trizma base and mixed 1:1 (v/
v) with guanidinated peptides. Similarly, propionamidination
reactions were prepared by making 1:1 mixtures of guanidinated
peptides and 46.2 g/L S-methyl thiopropionimidate in 250 mM
Trizma base. Each reaction was incubated for 1 h at ambient
temperature before acidifying the mixtures by adding TFA to a
concentration of 2% (v/v). The reaction mixtures were combined
and cleanup was achieved by solid-phase extraction using a 20 ×
1 mm BioBasic C18 Javelin guard column (Thermo Electron Co.,
San Jose, CA). The reaction mixture was loaded on to the column
using a 50 µL/min flow of 95% (v/v) water and 5% (v/v) acetonitrile
that was buffered in 0.1% formic acid. Purified peptides were
recovered by changing the mobile phase to 50% (v/v) acetonitrile
and collecting the effluent.
Liquid Chromatography-Tandem MS of Labeled Pep-
tides. Reversed-phase liquid chromatography was performed
using a column that was constructed by packing BioBasic C18
(Thermo Electron Co.) media into a 50-mm length of 254-µm-i.d.
polyetheretherketone (PEEK) tubing (Upchurch Scientific, Oak
Harbor, WA). Ten-microliter aliquots of the labeled and purified
tryptic digests were injected onto this column in each experiment.
The concentrations of hemoglobin, R-casein, and albumin were
2, 2, and 35 µM, respectively. A linear gradient of increasing
acetonitrile was used in all LC-MS/MS experiments. A flow rate
of 5 µL/min was established by precolumn splitting of eluent
delivered by a Waters 2795 Separations Module (Waters, Milford,
MA). No effort to optimize sensitivity by using nanospray with
<100-µm-i.d. columns was made in this work. Buffer A consisted
of 0.1% aqueous formic acid, and buffer B was 0.1% formic acid in
acetonitrile. All separations were carried out by increasing the
concentration of buffer B from 5 to 40% (v/v) over 120 min. The
effluent was directed to the electrospray ionization source (Z-
spray) of a quadrupole time-of-flight (Q-TOF) mass spectrometer
(Q-Tof Micro, Micromass, Manchester, U.K.). A potential of +3.0
kV was applied to the electrospray needle in all experiments. MS
and tandem MS spectra were acquired using the survey scan
option provided in the manufacturer’s software (MassLynx). The
three most intense peaks were selected from each MS scan in
real time and subsequently fragmented by low-energy CID. Argon
was employed as the target gas in all experiments. After five MS/
MS scans, a precursor ion mass was added to an exclusion list.
The collision energy (16-50 eV) applied to each precursor ion
was varied depending on both charge state and m/z.
Sequence Interpretation and Protein Identification. MS/
MS spectra were interpreted manually with help from the mass
signatures provided by the different N-terminal amidine groups.
(31) Thumm, M.; Hoenes, J.; Pfleiderer, G. Biochim. Biophys. Acta 1987, 923,
263-267.
6302 Analytical Chemistry, Vol. 77, No. 19, October 1, 2005

First, tandem mass spectra of differentially labeled peptides were
paired by searching the total ion chromatograms for precursor
masses that were separated by 14 u (a difference of 7 m/z for [M
+ 2H]2+ ions). Since propionamidinated peptides eluted ∼1 min
later than their acetamidinated counterparts, it was not necessary
to search the entire chromatogram when pairing precursor ions.
In our previous work involving differential amidination of both
N-termini and lysine residues, larger differences in retention time
were sometimes observed.28 Next, product ions were assigned
as N- or C-terminal fragments based on the mass differentials
observed between paired spectra. Singly charged fragment ions
spaced by 14 or 0 u were assigned as N- or C-terminal fragments,
respectively. The mass spacings between adjacent ions of the same
series were used to deduce peptide sequence information.
Although model proteins were used in this work, interpretations
were made directly from the paired spectra without using
knowledge of the sequences. Protein sequences and potential
modifications were obtained from the Swiss-Prot knowledge base
after peptide sequence interpretations. Sequencing was terminated
when no more peaks of a series could be unambiguously assigned
from their mass differentials. Peaks with poor signal-to-noise ratios
were not included in interpretations. Interpreted sequences were
submitted to a search of the NCBI reference sequence database
using the Blast algorithm.32 For all searches, this database was
filtered for mammalian proteomes only and contained 81 351
protein sequences. The sequences submitted for comparison to
this database represented the longest single contiguous segment
that was interpretable for each peptide. Although Blast searching
assigns scores based on similarity or homology, we required that
the interpreted sequence align exactly with a sequence in the
database. Since leucine and isoleucine are isobaric, they were
treated equivalently in all assignments. Protein identifications were
also constrained by limiting matches to those in which the
precursor ion mass was consistent with the mass of the matching
peptide. No enzyme specificity was required, and a 400 ppm error
tolerance was accepted when matching precursor and database
peptide masses.
Figure 1. Q-TOF tandem mass spectra of the [M + 2H]2+ precursor
ions of (A) unmodified and (B) guanidinated/acetamidinated EFTP-
PVQAAYQK.
Despite the predominance of cleavage between T and P, a series
of y-type ions from y3 to y10 were also observed. The spectrum of
the labeled peptide (Figure 1B) also displays these sequence ions.
However, the most striking feature of this spectrum is the
predominance of complementary b1 and y11 fragment ions resulting
from cleavage of the N-terminal peptide bond between E and F.
These products were not observed from the unmodified peptide
whereas they are among the most abundant in the latter example.
Despite the high efficiency of N-terminal peptide bond dissociation,
the relative intensity distribution of the other sequence ions
remains remarkably similar to the unmodified example. Therefore,
the enhancement of this dissociation pathway has increased the
overall information content in the spectrum, allowing for facile
identification of the N-terminal residue using the newly formed
b1 and y11 product ions. These data are typical of the fragmentation
observed from dozens of peptides that we have studied. While
the yn-1 and b1 ions are usually not observed from unmodified
peptides, they are often the most abundant peaks when the
N-terminus is amidinated. In previous work involving CID of
amidinated peptides in an ion trap, we discussed the enhancement
of yn-1 fragment ions but not that of b1 ions.29 These ions were
likely formed in those experiments, but the low-mass cutoff that
is inherent to resonance excitation in an ion trap prevented the
analysis of small product ions.38
RESULTS AND DISCUSSION
Fragmentation Properties of Guanidinated/Amidinated
Peptides. We recently reported that amidination enhances the
formation of yn-1 fragment ions.29 We proposed that this cleavage
pathway is promoted via a hydrogen bond-stabilized cyclic
intermediate involving the N-terminal amidine group and back-
bone carbonyl oxygen. Based on this model, we expected that
doubly labeled (guanidinated/amidinated) peptides would frag-
ment similarly. LC-MS/MS experiments were done using un-
modified and doubly labeled tryptic digests of standard proteins
to verify this. The Q-TOF tandem mass spectra of the [M + 2H]2+
EFTPPVQAAYQK precursor ion displayed in Figure 1 are typical
examples of this study. Cleavage of the TP peptide bond associated
with formation of y92+, y9+, and a series of internal ions was clearly
the most efficient fragmentation pathway of the unmodified peptide
(Figure 1A). It is well known that peptide bonds adjacent and
N-terminal to proline residues are highly labile in CID.3,33-37
Peptide Sequence Interpretation Using Mass-Coded N-
Termini. In the first step of this peptide sequencing strategy,
(35) Yu, W.; Vath, J. E.; Huberty, M. C.; Martin, S. A. Anal. Chem. 1993, 65,
(32) Altschul, S.; Madden, T.; Schaffer, A.; Zhang, J.; Zhang, Z.; Miller, W.;
Lipman, D. Nucleic Acids Res. 1997, 25, 3389-3402.
(33) Loo, J. A.; Edmonds, C. G.; Smith, R. D. Science 1990, 248, 201-204.
(34) Loo, J. A.; Edmonds, C. G.; Smith, R. D. Anal. Chem. 1993, 65, 425-438.
3015-3023.
(36) Vaisar, T.; Urban, J. J. Mass Spectrom. 1996, 31, 1185-1187.
(37) Breci, L.; Tabb, D.; Yates, J.; Wysocki, V. Anal. Chem. 2003, 75, 1963-
1971.
Analytical Chemistry, Vol. 77, No. 19, October 1, 2005 6303

ions (e.g., b-ions) are separated by 14 u. These derivatized spectra
are remarkably easy to compare since they are nearly identical
with regard to the types and intensities of fragment ions formed.
This similarity facilitates the pairing of complementary acetamidi-
nated and propionamidinated tandem mass spectra. As in Figure
1, the complementary b1/yn-1 ions are abundant features of labeled
peptides (Figure 2B and C). CID of the unmodified peptide did
not yield either of these fragment ions (Figure 2A). These peaks
allow the N-terminal residue to be easily identified thereby
providing a valuable starting point for elucidating sequences.
Furthermore, the first two N-terminal residues can easily be
identified when both yn-1 and yn-2 are formed. By comparison,
the unmodified spectrum contains only the yn-2 fragment ion,
which by itself does not allow direct interpretation of the first two
N-terminal residues. In total, a contiguous y-ion series including
y11-y4 was observed, and by using the mass differences between
adjacent peaks, 8 of the 12 residues of this peptide (FFVAPFPE)
were identified. In addition to b1, the b2 - NH3 and b4 - NH3
fragment ions were also observed in both spectra as 14-Da
separated pairs. In our experience, following N-terminal amidi-
nation, b-type ions other than b1 are always accompanied by
neutral loss of NH3.29 As is common in instruments that employ
a collision cell for ion activation, some immonium and internal
fragment ions were also formed. These ions include the PEV, PE,
and PF products as well as the immonium ions of phenylalanine
and proline. Since these internal fragment ions are not labeled,
they can be misinterpreted as y-type ions. Fortunately, they are
normally in the low-mass range and do not interfere with the
majority of interpretations.
Q-TOF tandem mass spectra of the unmodified and amidinated
[M + 2H]2+ ions of YLGYLEQLLR were acquired during the
analysis of R-casein and are displayed in Figure 3. Since this is
not a lysine-containing peptide, it was not guanidinated during
the labeling reactions. However, the amidinated N-terminal amino
groups provided the usual mass signatures. As in the previous
example, the yn-1 (y9) and b1 ions are very abundant products
formed by the derivatized precursor ions and the spectra appear
strikingly similar (Figure 3B and C). Also, the unmodified
precursor ion yielded only a weak signal for y9 and no signal for
the b1 ion (Figure 3A). By matching isobaric peaks in these
spectra, it was possible to identify the complete y-ion series and
therefore infer the entire sequence of this peptide. This case also
demonstrates how the derivatizations facilitate distinguishing
glutamine and lysine residues. The observed mass difference of
128.0558 u between the y3 and y4 ions in Figure 3B is consistent
with glutamine whose mass is 128.0586. However, the monoiso-
topic mass of underivatized lysine (128.0950 u) is only 0.0364 u
higher. In the present example, there is no ambiguity in the
assignment since guanidinated lysine residues have a mass of
170.1168 u. The use of guanidination to distinguish these residues
would be even more critical if instruments with low mass accuracy
were used (e.g., ion trap). A potential complication of shifting the
mass of lysine to 170.1168 is that the diamino acid combinations
of AV, LG, and IG differ by only 0.0113 u (170.1055 u each). If
these sequences are present within a peptide and do not fragment
upon activation, they could be misinterpreted as homoarginines.
CID mass spectra of the [M + 2H]2+ precursor ions of
acetamidinated and propionamidinated LLVVYPW are displayed
Figure 2. Q-TOF tandem mass spectra of the doubly labeled [M +
2H]2+ precursor ions of FFVAPFPEVFGK from R-casein. Spectra
A-C display the fragmentation spectra of the [M + 2H]2+ unmodified,
acetamidinated, and propionamidinated precursor ions, respectively.
lysine residues are converted to homoarginines. This has two
consequences. First, it prevents the amidination of lysines. Second,
it shifts the mass of lysine by 42 u, making it much easier to
distinguish from glutamine. To investigate this approach, tryptic
peptides of hemoglobin, albumin, and R-casein were used. The
Q-TOF tandem mass spectra of unmodified, acetamidinated, and
propionamidinated FFVAPFPEVFGK displayed in Figure 2A-C,
respectively, provide a typical example of how the labeling
facilitates interpretations. These spectra were labeled after the
peptide was identified using the interpreted sequence. Therefore,
not all of the labeled peaks (e.g., internal ions) were deduced
manually. C-Terminal fragment ions (e.g., y-ions) of labeled
peptides (Figure 2B and C) appear as isobaric pairs in separate
MS/MS spectra regardless of the N-terminal label. Since the
amidine groups differ by a methylene unit, N-terminal fragment
(38) Louris, J.; Cooks, R.; Syka, J.; Kelley, P.; Stafford, G.; Todd, J. Anal. Chem.
1987, 59, 1677-1685.
6304 Analytical Chemistry, Vol. 77, No. 19, October 1, 2005

Figure 4. Q-TOF tandem mass spectra of (A) acetamidinated and
(B) propionamidinated [M + 2H]2+ precursor ions of LLVVYPW from
hemoglobin.
a b-ion series from b1 to b5 was inferred leading to the sequence
LLVVY. The C-terminal residues, PW, were not identifiable from
the b-ion series since b6 was not observed. The suppression of
cleavages C-terminal to proline residues is a common attribute of
peptide fragmentation and often contributes to incomplete se-
quence coverage as shown here. However, with slightly more
sophisticated data interpretation methods, it may be possible to
completely sequence peptides from data such as these. In the
present case, the b5 and y2 ions can be interpreted as a
complementary fragment ion pair that is representative of the
entire peptide sequence since the sum of their masses is equal to
the doubly protonated monoisotopic mass of the precursor ion.
LLVVYPW was produced during tryptic digestion of hemoglo-
bin, but is terminated by tryptophan rather than lysine or arginine.
Enzymatic cleavage of the peptide bond C-terminal to aromatic
residues is a common side reaction resulting from the chymot-
ryptic specificity of pseudotrypsin that is formed upon tryptic
autoproteolysis.39,40 Due to this and the presence of other pro-
teolytic enzymes found in biological samples, it is often necessary
to allow for no cleavage specificity when predicting candidate
peptides from databases using matching algorithms. The conse-
quences of doing this are that the databases being searched
become much larger, searches are slower, and the likelihood of
false positive assignments increases. De novo sequencing is
unaffected by the presence of nonspecific peptides since it involves
data interpretation without prior knowledge of database sequences.
Sequencing of Phosphorylated Peptides. Characterizing
post-translational modifications is an important goal in protein
Figure 3. Q-TOF MS/MS spectra of the [M + 2H]2+ (A) unmodified,
(B) acetamidinated, and (C) propionamidinated precursor ions of
YLGYLEQLLR from R-casein.
in Figure 4A and B, respectively. It is noteworthy that the
unmodified form of this peptide was not observed. The lack of
basic residues in this sequence must limit its ionization yield.
Furthermore, the protonation of amidinated peptides was likely
enhanced due to the high basicity of the labels. Although we have
only observed a limited number of nonbasic peptides, amidination
may be generally useful for such molecules. Unlike the examples
shown above, this peptide primarily yields b-ions upon CID
presumably due to the absence of a C-terminal basic residue.
Much like the previous examples, the b1 ion is a prominent feature
in this spectrum. However, the complementary yn-1 ion that is
typically observed is absent. The yn-1 ion may be formed in the
same reaction that produces b1 but undergo further dissociation
to yield the intense y2 peak resulting from cleavage between
tyrosine and proline. The lack of a basic residue such as lysine,
arginine, or histidine makes it more likely that one of the ionizing
protons is located on the peptide backbone, thus facilitating charge
site-directed fragmentation of the Y-P peptide bond. By using
the 14 u mass differentials between peaks in this pair of spectra,
(39) Keil-Dlouha, V., Zylber, N., Imhoff, J. M., Tong, N. T., Keil, B. FEBS Lett.
1971, 16, 291-295.
(40) Karty, J.; Ireland, M.; Brun, Y.; Reilly, J. J. Chromatogr., B 2002, 782, 363-
383.
Analytical Chemistry, Vol. 77, No. 19, October 1, 2005 6305

Figure 6. Q-TOF tandem mass spectra of dual-labeled TVD-
MESTEVFTK. The MS/MS spectrum of the acetamidinated/guanidi-
nated derivative is displayed in (A), whereas (B) shows the propion-
amidinated/guanidinated form.
from yn fragment ions is evidently a favored process. This effect
is a well-known characteristic of phosphopeptides analyzed by
CID. Consistent with previous examples, the unmodified peptide
did not yield yn-1 or b1 fragment ions. Despite the prevalence of
H3PO4 losses, the data are amenable to peptide sequencing
because amino acids are identified using the mass spacings in
the y-ion series. The observation of a y-ion series from y13 - 98
to y7 - 98 allowed the first seven residues, beginning at the
N-terminus, to be sequenced. Unlike most database-matching
techniques, this sequence interpretation was possible without first
using the precursor ion mass to generate a list of candidate
peptides from a database. Since this peptide is phosphorylated,
the assigned peptide would not have been included as a candidate
without considering potential modifications. Evidence for the site
of phosphorylation was provided by the appearance of y5, y5 -
98, and y4 ions. The observation of y4, but not y4 - 98, strongly
suggests that the N-terminal residue of y5 was the site of
phosphorylation. This interpretation is consistent with the phos-
phorylation sites that have previously been reported for this
protein. These spectra also demonstrate the deleterious effect that
proline residues can have in peptide sequencing. It is well known
that dissociation of the C-terminal peptide bond of proline is often
suppressed in CID. The low abundance of y12 - 98 and absence
of y6 - 98 illustrate this effect. Without observing the y6 - 98
ion, it was not possible to directly confirm the presence of the
proline or asparagine residues in the middle of this sequence.
The data displayed in Figure 6 also demonstrate the analysis
of another phosphorylated peptide. The acetamidinated and
Figure 5. Q-TOF tandem mass spectra of the unmodified and
amidinated/guanidinated phosphopeptide VPQLEIVPN(pS)AEER from
R-casein. Spectra A-C display the fragmentation spectra of the [M
+ 2H]2+ unmodified, acetamidinated, and propionamidinated precur-
sor ions, respectively.
research. The identification and mapping of these modifications
can provide valuable insight into protein function. Since modifica-
tion sites are not predicted from genomic sequences, they are
not identified by database-matching algorithms unless all potential
sites are considered as both modified and unmodified. Analogous
to the case of nontryptic peptides, this approach rapidly increases
the database size and thereby leads to increased false-positive
assignments and longer search times. Therefore, it would be
advantageous if alternative approaches, such as de novo sequenc-
ing, could elucidate post-translational modifications.
Tryptic peptides from R-casein were analyzed to test the
compatibility of guanidination/amidination labeling with the
analysis of phosphorylated peptides. ESI Q-TOF tandem mass
spectra were acquired during an LC separation and Figure 5
displays MS/MS spectra of the [M + 2H]2+ VPQLEIVPN(pS)-
AEER phosphopeptide. Results for unmodified, acetamidinated,
and propionamidinated peptides appear in parts A, B, and C of
this figure. In all of these examples, the loss of H3PO4 (-98 u)
6306 Analytical Chemistry, Vol. 77, No. 19, October 1, 2005

propionamidinated peptide spectra appear in Figure 6A and B,
respectively. Using the mass differentials in a series of C-terminal
fragment ions, it was possible to derive the sequence DME(pS)-
TEVF. This is part of TVDMESTEVFTK from the R-S2 casein
precursor. Although phosphorylation of the serine residue has
been previously observed,41 it was an unexpected result since this
modification is not predicted in the Swiss-Prot entry for this
protein.42 Evidence of this phosphorylation site is derived from
both the mass of the precursor, which was 80 u heavier than the
mass predicted of dual-labeled TVDMESTEVFTK, and the 69 u
mass differential between y7 and y6 ions. This is consistent with
the addition of a single phosphate. The 69 u differential does not
match an unmodified amino acid residue but can be explained by
the loss of H3PO4 from a phosphoserine to form dihydroalanine.
Since only a couple of examples of post-translationally modified
peptides have thus far been studied, it would be inappropriate to
claim that the method is universally applicable for analyzing
modifications. However, we do not know of any side reactions
that affect other PTMs and, therefore, expect this sequencing
approach to be generally applicable to these types of analyses.
Side Reactions and Artifacts of Labeling. During the study
of model tryptic peptides, we observed two side reactions that
must be explained. As described in our previous work with
amidinated peptides, N-terminal amidine groups are susceptible
to hydrolysis when the N-terminal residue is Ser or Thr, leading
to formation of acetyl or propionyl groups.28 The consequences
of this reaction are that the mass of the N-terminal label increases
by 1 u, and the fragmentation properties change. Specifically, in
the few cases where N-terminal hydrolysis was observed, yn-1 and
b1 ions were not formed by CID. The phosphopeptide in Figure
6 (TVDMESTEVFTK) is such an example. As shown in these
spectra, neither b1 nor yn-1 ions were formed. Peptides that were
terminated by either Ser or Thr were the only ones that did not
produce these fragment ions. We previously described how
amidine groups promote cleavage of the N-terminal residue, and
it appears that removal of this basic site through hydrolysis
prevents this dissociation pathway.29 Although the presence of yn-1
and b1 ions facilitates identification of the N-terminal residue, it
should also be possible to use the absence of these ions to deduce
the presence of Ser or Thr at that position.
gated the possibility that b1 ions may serve as internal calibrants,
significantly reducing mass errors in MS/MS spectra. The well-
known benefits of high mass accuracy in proteomic research apply
to all types of protein identification experiments (e.g., database
matching and de novo sequencing) since accurate masses allow
for tighter constraints, leading to fewer errors.43 However, during
the course of a typical proteomic experiment, it is difficult to
implement internal calibrations since they require mixing a
calibrant on-line with LC effluent prior to mass analysis. Not only
does this method dilute the effluent but it is very difficult to match
the calibrant and analyte concentrations. Furthermore, including
calibrant masses in MS/MS spectra is not feasible since precursor
ion isolation necessarily excludes other masses. A quasi-internal
calibration alternative to on-line mixing has been to use a dual
ESI source in which one source sprays the LC effluent while the
other contains a reference compound of known mass (i.e., Lock
Spray, Micromass). In this case, the reference channel is intermit-
tently sampled and mass corrections are made in real time based
on the errors observed for this ion. A drawback of this approach
is that the use of a separate reference channel reduces the analyte
duty cycle, which is often critical in proteomic investigations
involving complex mixtures. The use of b1 ions for calibration
overcomes these disadvantages since it is available without on-
line mixing or the introduction of a second ionization source. b1
ions are limited to the 19 unique masses representing the common
amino acids. They are easy to identify because they are ubiqui-
tously observed and typically appear as one of the most intense
CID fragment ions of amidinated peptides.
To examine the effectiveness of this internal calibration
approach, LC-MS/MS experiments with guanidinated/amidi-
nated tryptic peptides of simple model proteins were performed
using a Q-TOF mass spectrometer. The TOF analyzer was initially
externally calibrated using poly(propylene glycol) prior to the
experiment. Following data acquisition, MS/MS spectra were
internally calibrated using the lock mass utility of the instrument’s
software. This feature calculates the difference between the
observed and expected m/z for each given ion. The relative error
calculated for this peak is then used to compensate the entire
mass spectrum. Therefore, all masses are shifted by an equal
percentage of a peak’s nominal mass. An example of the effect of
this internal calibration method is displayed in Table 1. In this
table, the mass accuracies of externally and internally calibrated
peaks from the CID spectrum of the [M + 2H]2+ precursor ion
of propionamidinated VNVDEVGGEALGR are compared. Mass
errors of ∼40 ppm were observed for the peaks of this spectrum
prior to internal calibration. These errors were largely dependent
on the quality of the external calibration, as well as how recently
it was performed. Mass errors commonly drift with increasing
time between calibration and analysis due to temperature fluctua-
tions and the instability of power supplies. Regardless of this
instability, the errors observed following the correction using b1
ions were typically less than 10 ppm. The mass accuracies shown
in Table 1 demonstrate this improvement. Without the use of FT-
ICR MS, mass accuracies less than 10 ppm are difficult to routinely
achieve unless some form of internal calibration is applied. This
method should be generally applicable in the analysis of complex
An additional sample handling artifact that we observed is the
deamidation of Asn residues. This is a well-known effect that is
enhanced under basic conditions, such as those used during
guanidination.40 This reaction converts Asn to Asp residues, thus
increasing its mass by 1 u. This will obviously affect peptide
sequencing, causing Asn to be misinterpreted as Asp. Of the
several dozen peptides analyzed in this work, only three were
found to undergo this side reaction. This reaction likely occurs
during guanidination, since the pH is elevated to 10.5 during that
step. Although deamidation led to a few misinterpretations, we
believe that software that utilizes interpreted sequences to identify
proteins could account for such occurrences by considering the
possibility of Asn in all cases where Asp is detected.
Internal Calibration Using b1 Fragment Ions. As shown
above, amidine groups promote fragmentation of the N-terminal
peptide bond to produce abundant b1 and yn-1 ions. We investi-
(41) Brignon, G.; Ribadeaudumas, B.; Mercier, J.; Pelissier, J.; Das, B. FEBS
Lett. 1977, 76, 274-279.
(43) Clauser, K. R.; Baker, P.; Burlingame, A. L. Anal. Chem. 1999, 71, 2871-
(42) Swiss-Prot, CAS2_BOVIN, Accession No. P02663, June 14, 2005.
2882.
Analytical Chemistry, Vol. 77, No. 19, October 1, 2005 6307

since the organism under study would be known. However,
homology searches using these data could also be performed in
experiments involving either incomplete genomic sequencing or
an unknown organism. Of the 29 peptides studied here, only one
did not provide a unique match after consideration of both
sequence and precursor mass. The tryptic peptides YLYEIAR of
albumin and YIYEIAR of R-fetoprotein would yield identical
fragment ion masses and cannot be distinguished. Furthermore,
database-matching techniques such as Sequest11 or Mascot12
would also be unable to uniquely match this peptide using low-
energy CID spectra. Regardless of the protein identification
strategy, differentiating between leucine and isoleucine requires
some form of high-energy fragmentation that generates side-chain
cleavages. The production of such side-chain reaction products
was recently demonstrated by 157-nm photodissociation.44
Although the precursor masses and interpreted sequences
were sufficient in this work, it may be necessary to provide further
constraints in future searches. This would be especially important
if only a short segment of a peptide (i.e., <5 residues) were
interpretable. As demonstrated in Table 2, most interpreted
sequences begin with the N-terminal residue. It is clear that these
sequences contain the N-terminus since the analysis begins with
the b1 and yn-1 fragment ions that are produced by amidinated
peptides. In cases such as these, it is possible to further limit
random matches by restricting candidate peptides to those that
contain the interpreted sequence at their N-terminus. Another
strategy to further refine assignments would be to use smaller
fragments of interpretable sequences in addition to the contiguous
sequences shown here. In all of the interpretations shown in Table
2, the longest contiguous sequence that was interpretable was
matched against a database. However, it is often possible to
identify shorter portions of a peptide sequence as well. Incorpora-
tion of this additional sequence information will be useful,
especially in cases where a long contiguous sequence is not
interpretable.
Table 1. Effect of Mass Correction Using the b1 Ion of
VNVDEVGGEALGR
external calibration only
b1 lock mass
mass
error
mass
error
fragment
i.d.
calculated
measured
[M + H]
[M + H]
(ppm)
(ppm)
y12
y11
y10
y9
1215.5969
1101.5540
1002.4855
887.4586
758.4160
659.3476
602.3261
595.2728
466.2301
416.2621
155.1184
1215.5438
1101.5155
1002.4476
887.4226
758.3840
659.3187
602.3038
595.2478
466.2104
416.2477
155.1117
39.98
34.95
37.81
40.57
42.19
43.83
37.02
42.00
42.25
34.59
43.19
0.82
-7.90
-5.09
-2.37
-0.79
1.06
-5.98
-0.84
-0.64
-8.17
0.00
y8
y7
y6
b5*
b4*
y4
b1
peptide mixtures since all N-terminally amidinated peptides
produce b1 ions. Also, the high intensity of b1 ion peaks reduces
the effects of isobaric chemical noise that could otherwise distort
peak shapes and lead to errors in calculating the centroided
masses of calibrants.
Protein Identification. The ability to uniquely identify pro-
teins was considered by matching the tryptic peptides of R-casein,
hemoglobin, and serum albumin against a large database. These
proteins were employed in this work because their sequences and
post-translational modifications are well characterized. For each
peptide, the interpreted sequence (i.e., the portion of sequence
interpretable from comparison of the complementary tandem mass
spectra of acetamidinated and propionamidinated spectra) was
submitted to a Blast32search of the NCBI reference sequence
database constrained to 81 351 mammalian proteins. Since leucine
and isoleucine are isobaric, these residues could not be distin-
guished in these experiments or in the sequence matching
process. Table 2 displays the number of exactly matching
sequences for each peptide. In many cases (16 of 29), there was
sufficient sequence coverage to uniquely match a single protein
in the database. However, there were other examples in which
only short segments of a peptide were sequenced, leading to
multiple matches. To resolve this problem, the precursor mass
was also employed as a constraint. Therefore, a sequence match
was only accepted as an assignment if the interpreted sequence
was contained within a database peptide (no cleavage specificity)
that was consistent with the observed precursor ion mass. As
shown in Table 2, the use of this simple constraint eliminated
almost all false-positive matches and uniquely identified the model
proteins. A conservative error tolerance of 400 ppm was employed
for these matching experiments, although, as indicated by the
errors in Table 1, a more strict tolerance could have been applied.
Nevertheless, the success observed here despite the conservative
error tolerances only emphasizes how powerful this constraint
is. Since many, nearly identical variants of the â-chain of
hemoglobin were present in the database, multiple matches were
observed even after consideration of both precursor mass and
interpreted sequence. Furthermore, the matches to hemoglobin
and albumin of other organisms are reported here as well. In many
cases, it was impossible to distinguish between the proteins of
different organisms since they contain identical peptides. In a
typical experiment, it is usually possible to eliminate these matches
CONCLUSION AND FUTURE DIRECTIONS
A new labeling strategy involving the derivatization of lysine
residues and peptide N-termini has been presented. These
modifications simplify peptide sequence interpretation since the
mass-coded N-terminal labels distinguish N- and C-terminal
fragments. This approach was demonstrated using the tryptic
peptides of several model proteins. An advantage of this strategy
is that labeled lysine residues are no longer isobaric with
glutamine. A unique characteristic of the derivatizations is that
cleavage of the N-terminal peptide bond is promoted by amidine
groups. Therefore, abundant yn-1 and b1 ions are typically
observed in MS/MS spectra. These fragment ions provide
sequence information that is typically unavailable from unmodified
peptides and allow reliable interpretation of N-terminal residues.
Furthermore, we have found that b1 ions can serve as internal
calibrants to achieve mass accuracies of better than 10 ppm. This
attribute facilitates de novo sequencing and could also be used to
further constrain database-searching strategies.
This work has demonstrated a labeling strategy that facilitates
the interpretation of peptide sequences from tandem mass spectra.
(44) Cui, W., Thompson, Matthew S., Reilly, James P. J. Am. Soc. Mass Spectrom.
2005, 16, 1384-1398 (available on-line).
6308 Analytical Chemistry, Vol. 77, No. 19, October 1, 2005

Table 2. Protein Identification Using Tryptic Digests of Model Proteins
interpreted
sequence
matches
sequence and
actual sequence
sequence
precursor mass
R-Casein (Bos taurus)
FFVAPFPEVFGK
FFVAPFPE
1
1
R-casein, S1 precursor
R-casein, S1 precursor
R-casein, S1 precursor
R-casein, S2 precursor
R-casein, S2 precursor
R-casein, S1 precursor
â-casein
YLGYLEQLLR
YLGYLEQLLR
FVAPF
GPLVLNP
FALPQ
VPQLELV
LLYQEPVL
HQGLPQEVLNEN
EMPFPK
FVAPFPEVFGK
LYQGPIVLNPWDQVK
FALPQYLK
VPQLEIVPN(pS)AEER
LLYQEPVLGPVR
HQGLPQEVLNENLLR
EMPFPK
13
1
4
1
1
1
R-casein, S1 precursor
â-casein
1
Hemoglobin (Homo sapiens)
EFTPPVQAAYQK
VNVDEVGGEALGR
MFLSFPTTK
EFTPPVQAA
VNVDEVGGE
MFLSF
FLASVSTVL
GDLSTPDAVM
AHLDNLK
LLVVYPW
LLVVY
1
1
hemoglobin â-chain
hemoglobin â-chain
hemoglobin R-chain
hemoglobin R-chain
hemoglobin â-chain
hemoglobin â-chaina
hemoglobin â-chaina
hemoglobin â-chaina
15
1
FLASVSTVLTSK
FFESFGDLSTPDAVMGNPK
VLGAFSDGLAHLDNLK
LLVVYPWTQR
1
3
12
29
LLVVYPW
Albumin (Bos taurus)
EYEATLEECCAK
HLVDEPQNLIK
YNGVFQECCQAEDK
DDPHACYSTVFDK
VPQVSTPTLVEVSR
FSALTPDETYVPK
YLYEIAR
EYEATLEE
HLVDE
1
30
1
albumin
albumin
FQECCQ
albumin
PHACY
3
albumin
VPQVSTPTLVE
FSALTPDETYV
YLYELAR
3
albuminb
albumin
1
7
albuminb and
R-fetoproteinb
albuminb
albuminb
albumin
QTALVELLK
QTALVE
LGEYGFQNAL
FYAPELLY
DAFLGSFLYE
CTEDYLSLLLNR
13
4
LGEYGFQNALIVR
FYAPELLYYANK
DAFLGSFLYEYSR
MPCTEDYLSLILNR
5
1
albumin
albumin
1
a Denotes presence of matches corresponding to different organisms and variants of the Hb â-chain that contain the identical peptide sequence.
b Denotes matches to multiple organisms that also contained the identical peptide sequence.
Although the initial results are promising, automation of this
method is required before it will be suitable for de novo
sequencing of the thousands of MS/MS spectra typically gener-
ated from complex biological samples. The development of such
an approach is currently being pursued and will be compared to
other de novo sequencing strategies while a mixture of bacterial
proteins is analyzed. This work will also provide the opportunity
to test this method with low-abundance proteins. Although the
reaction conditions for guanidination are somewhat harsh, we have
previously applied this derivatization with 50 fmol of a digest30
and do not expect significant complications. Nevertheless, in this
paper, multiple examples of spectra are presented in which more
information (b1 and yn-1 ions) is available after derivatization, and
it is likely that the added peaks will improve de novo sequencing.
An additional goal of future work will be to evaluate this method
using low-mass accuracy instruments such as ion traps. However,
because low-mass accuracy is a detriment to any protein identifica-
tion strategy, we anticipate that this will make peptide sequencing
more difficult.
ACKNOWLEDGMENT
This work has been supported by the National Science
Foundation Grant CHE0094579 and the National Institute of
Health Grant GM61336. This work was supported in part by a
fellowship from Merck Research Laboratories.
Received for review March 30, 2005. Accepted July 20,
2005.
AC050540K
Analytical Chemistry, Vol. 77, No. 19, October 1, 2005 6309