Identification of Modification Sites in Large Biomolecules by StableIsotope Labeling and Tandem High Resolution Mass SpectrometryTHE ACTIVE SITE NUCLEOPHILE OF THIAMINASE I*

(Received for publication, May 27, 1997, and in revised form, September 22, 1997)

Neil L. Kelleher, Robb B. Nicewonger, Tadhg P. Begley, and Fred W. McLafferty‡

From the Department of Chemistry, Baker Laboratory, Cornell University, Ithaca, New York 14853-1301

A widely used procedure for site localization of cova-lent protein modifications involves proteolysis, partialchromatographic separation of the resulting complexmixture, and tandem mass spectrometry (MS/MS) toidentify peptides whose molecular weight (Mr) has beenincreased appropriately by the modification. As foundpreviously for MS of small molecules, this study showsthat protein fragment identification can be greatly sim-plified by labeling the modification with stable isotopes.Further, the high resolution capabilities of Fouriertransform MS make possible the direct identification ofCH3/CD3-labeled peptides without chromatographicseparation. Although separate Asp-N, Lys-C, and a-chy-motrypsin digests of thiaminase I (42 kDa) yielded asmany as 70 peptides, FTMS identification of the labeledpeptide localized the modification site of a mechanism-based inhibitor to Arg101-Lys121, Asp90-Gly122, and Gly107-Tyr119, respectively. The measured mass difference val-ues of the two labels agreed with that expected for CH3/CD3, 3.019 Da, with a standard deviation of 0.005 Da,providing persuasive identity verification. MS/MS frag-mentation narrowed the site to Pro109-Phe118 and alsocaused loss of the derivative with a sulfur atom,uniquely identifying Cys113 as the thiaminase I active-site nucleophile among the 379 amino acids.

Biological activity can be dramatically altered by a smallcovalent modification in molecular structure (1, 2). To charac-terize such modifications in small molecules, the sensitivity,speed, and specificity of tandem mass spectrometry (MS/MS)1

(3–5) provide well recognized advantages over other methods(e.g. Edman, NMR, IR, crystallography) (2, 6, 7). For example,identifying drug metabolites in a complex extract is especiallystraightforward if a methyl group of the drug molecule hasbeen partially replaced by CD3; even if the mass spectrumcontains many molecular ions, any showing isotopic peaks sep-arated by 3 Da are probably those of a metabolite (3, 8, 9).Further, MS/MS isolation and fragmentation of each isotopi-cally unique ion group can provide information on the label’slocation in the molecule. Characterizing post-translational or

other covalent modifications in large biomolecules is even morechallenging (6); the most effective MS methods (4, 10, 11) forproteins of known sequence (amounts as low as 3 3 10214 mol)(10) employ extensive proteolysis, chromatographic fraction-ation, and MS/MS of the peptide fractions. The masses of thepeptides originating from the protein without and with thelabel can be compared to identify the labeled peptide, followedby its characterization by MS/MS, or all of the labeled protein’speptides can be examined by MS/MS to identify and character-ize the labeled peptide. Here this examination is simplifiedwith mixed, preferably isotopically mixed, labels. Further, thechromatographic steps are eliminated by utilizing the .105

MS/MS resolving power of Fourier transform (FT) MS (5, 12,40). This is illustrated with identification of the active-sitenucleophile of thiaminase I, a 379-amino acid enzyme. Illus-trating the potential unique applicability of this methodology,MS/MS data sufficient for identification in the Protein DataBank (13) have been obtained by ESI/FTMS recently from10217 mol of a 29-kDa protein (14).

Thiaminase I from Bacillus thiaminolyticus catalyzes thedegradation of thiamin (vitamin B1; Scheme I, 1) by nucleo-philic displacement of its thiazole moiety (15, 17, 41, 42). Treat-ment with the suicide substrate 4-amino-6-chloro-2-meth-ylpyrimidine (2) caused the molecular ion mass to increase bythe 107 Da expected for a single adduct (16). Fragmentation ofthese molecular ions gave fragment ions both without and withthe 107-Da adduct; the smallest adducted fragment corre-sponded to the 99-amino acid region Pro79 to Thr177 (18).2

MS/MS was unsuccessful in producing the smaller protein frag-ments required for further localization (16). Proteolysis shouldachieve this (4, 6, 7, 10, 11), but these much smaller fragmentsmust result in a much more complex mixture that would makemuch more difficult the direct identification of the derivatizedpeptide without preliminary separation. We show here thatthis is still possible with high efficiency and accuracy using amixed covalent label (e.g. substituents H versus CH3, CH3

versus CD3 on 2) that produces a characteristic signature ofneighboring isotopic peaks resolved by FTMS. Other recent MStechniques illustrate related advantages of stable isotope in-corporation (19, 20) or depletion (21) in large biomoleculeswhen measured at isotopic resolution. As an unexpected bonus,fragment ion dissociation results in the loss of the covalentlabel with a sulfur atom, specifically identifying attachment atCys113.

EXPERIMENTAL PROCEDURES

Materials—Proteases and high performance liquid chromatographygrade solvents were used as obtained (Sigma). Recombinant thiaminaseI was isolated from an Escherichia coli overexpression system (18). Thisconstruct was shown earlier to possess a heterogeneous N terminus

* This work was supported by the National Institutes of Health Celland Molecular Biology Training Grant 08-T2GM07273 (to N. L. K.),Grants GM16609 (to F. W. M.) and DK44083 (to T. P. B), and AmericanChemical Society Division of Analytical Chemistry full fellowship spon-sored by Perkin-Elmer Corporation (to N. L. K.). The costs of publica-tion of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby marked “advertisement”in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

‡ To whom correspondence should be addressed.1 The abbreviations used are: MS, mass spectrometry; MS/MS, tan-

dem mass spectrometry; ESI, electrospray ionization; FT, Fouriertransform; Mn1, molecular ion of n charges; NS, nozzle-skimmer disso-ciation; SORI, sustained off-resonance irradiation; RP, resolving power.

2 These MS data suggested the preparation of a Cys113 3 S mutantthat was found to be inactive.

THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 272, No. 51, Issue of December 19, pp. 32215–32220, 1997© 1997 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.

This paper is available on line at http://www.jbc.org 32215

by guest on February 9, 2018http://w

ww

.jbc.org/D

ownloaded from

putatively due to semispecific cleavage of its signal peptide (16).Synthesis—4-Amino-6-chloro-2,5-dimethylpyrimidine (2-CH3) was

synthesized from diethyl methyl malonate in a three-step sequenceusing the route previously described for the synthesis of 2 (18, 22). Theproduct was purified by two stages of silica gel chromatography (90:10chloroform/methanol, 80:20 ethyl acetate/hexanes). 2-CD3 and 4,5,6-13C3-2 were synthesized in the same manner from diethyl trideuteri-omethylmalonate and 13C3-diethylmalonate, respectively (23).

Inactivation and Proteolysis—Thiaminase I (20 ml of 2 mg/ml, 40 mg)in 50 mM TRIS, 2 mM EDTA, 5% glycerol, pH 7.5, was treated with 1 mlof inhibitor (1.0 6 0.1 mg in 10 ml of dimethylformamide) for 16 h atroom temperature, dialyzed against 50 mM TRIS, 2 mM EDTA, pH 8.3,overnight, and heated at 80 °C for 12 min. Heat-denatured enzyme wasincubated with ;0.5 mg of Asp-N, Lys-C, or a-chymotrypsin for 16 and16 h and 10 and 36 min, respectively, and quenched by the addition of5 ml of acetic acid and freezing.

MS Analysis—Intact protein and crude digests were desalted byloading onto reversed-phase peptide traps (Michrom Bioresources;0.1 3 1 cm), washed with 1 ml of 1:98:1 CH3OH/H2O/CH3COOH, andstep-eluted with ;20 ml of 70:28:2 CH3OH/H2O/CH3COOH. This solu-tion (;5 ml) was loaded into either a Nanospray (24) or Picospray (14,25) ESI emitter with a 1–3 mm tip; a voltage of 0.6–1.2 kV applied to thesolution supported ESI and determined the flow rate (;1–50 nl/min).Resulting ions were guided through a heated metal capillary, skimmer,and three radio frequency-only quadrupoles into the ion cell (1029 Torr)of a modified 6-Tesla Finnigan FTMS (26). Transients were stored as128–512 K data sets and analyzed using Odyssey Version 4.0 software.Theoretical isotopic distributions were generated using Isopro Version2.0 (27) and fit to experimental data by least squares to assign the mostabundant isotopic peak and relative ratios of overlapping isotopic dis-tributions. Spectra were calibrated either externally using bovine ubiq-uitin (Mr 5 8,564.64-5) or for peptide mixtures, internally, using iden-tified peptides. The mass difference (in units of 1.0034 Da) between themost abundant isotopic peak and the monoisotopic peak is denoted initalics after each Mr value. For isotopic labels, the Dm values arecorrected for isotopic overlaps such as the contribution of 13C3

1H3 to12C3

2H3. Ion fragmentation utilized nozzle-skimmer (NS) collisionalactivation during ion introduction (28) or sustained off-resonance irra-diation (SORI) for fragmentation in the ion cell (29, 35).

RESULTS AND DISCUSSION

Recombinant thiaminase I yielded three molecular ions(42,127-26, 42,198-26, 42,255-26 Da) due to N-terminal heter-ogeneity (extra Ala, Ala-Gly) (16). Incubation with the suicidesubstrate 2 (18, 22) increased their masses by 108 6 1 Da,consistent with the 107-Da mass difference (Dm) expected forCl2 release during inactivation of one active site in each of thethree enzyme molecules. This spectrum showed the isotopiccluster around 42,198 Da to be ,5% as abundant as thataround 42,305 Da (16), indicating that derivatization was.95% complete, as confirmed by 2-nitro-5-thiol sulfobenzoatethiol titration (18). NS dissociation of the multiply chargedprotein ions from the underivatized enzyme gave 61 differentfragment ion masses whose sequence location could be as-signed from the corrected DNA-derived sequence (16). Frag-ments were identified that represented all parts of the protein,but loss of the covalent label from the molecular ion was notdetected.

Data Interpretation Complexity—For active-site location, theNS spectra of the enzyme without (Fig. 1, a and c) and with(Fig. 1, b and d) its covalently bound label are compared; allthree of the molecular ions show that the label causes a 107-Daincrease. The identification of fragment ions that do (Fig. 1, aand b) and do not (Fig. 1, c and d) show this increase isillustrated by the y301 and y193 ions (30),3 respectively. How-ever, this comprehensive search for labeled fragments requiredcareful cross-comparison of all isotopic clusters in both spectra;the shifted y301

231 peaks (Figs. 1, a versus b) represent ,1% of them/z range recorded. As an example of possible problems, in aspectrum of the original study (Ref. 16 and Fig. 2), the derivat-ized y301 peak at m/z 1,456 is overlapped by a b77

61 fragment ion.

3 Amide bond cleavage yields b and y ions containing the N or Cterminus, respectively.

FIG. 1. Partial ESI/FT mass spec-trum from NS fragmentation of (aand c) wild-type thiaminase I and (band d) this enzyme inactivated with2. Circles, best fit of calculated (calc.) iso-topic abundance distribution; open-headed arrow, mass of underivatized y301

231

ion.

SCHEME I.

Stable Isotope Localization of Biomolecule Modification32216

by guest on February 9, 2018http://w

ww

.jbc.org/D

ownloaded from

In the derivatized sample spectrum (Fig. 1b), the abundance ofthe underivatized y301 ions (33,355 Da) is ,10% that of the33,462-Da ions, consistent with the previous evidence for .95%derivatization (16, 18).

Unlabeled/Labeled Mixtures—Thiaminase I labeled with2-CH3 was mixed 1:1 with the free enzyme to give NS data (notshown) similar to that of the combined Fig. 1, a and b. Now onlya single spectrum is needed to identify the modified y301 by itssatellite peak at 1121 Da, and the relative abundances of thesepeaks will be less affected by any NS fragmentation variance ofseparate measurements. Because the active site is in a regionof the molecular ion stable to NS fragmentation, this onlynarrowed the label location to the 99-amino acid region Pro79-Thr177 (16) (although a modified site in other regions would bewell localized by these data, e.g. the region yielding the NSfragment ions b46, b49, b50, b52, b53). Thus proteolysis was triednext, as it can cleave protein bonds at specific sites to producesmaller fragments.4

Mixture of Two Derivatives—The accuracy of measuring a121-Da mass difference (6 1 Da) is relatively poor because themasses as well as their mass difference are so large. To makethese smaller, thiaminase I was reacted with a 1:1 mixture of 2and 2-CH3 and digested with Asp-N. Inspection of the resultingcomplex spectrum for a pair of peaks of expected Dm 5 14.016was ambiguous, although the 3,816.05-2-Da peak (Fig. 2a)corresponds to that calculated for the Asp-N peptide Asp90-Gly122 with a 121.06-Da label (3,816.06-2). The isotopic clustercorresponding to the peptide with the 107-Da label (3,802-2) is

at the noise level, indicating at least a 20:1 overall reactivitydifference for these suicide substrates.

To avoid such reactivity differences, thiaminase I was la-beled with 2 and 2-CH3 in separate tubes, mixed 1:1, anddigested. The resulting mixture gave a complex spectrum con-taining .80 isotopic distributions of ;40 peptides examined byESI/FTMS (Fig. 2b). Now these peaks (3,802.05-2 and3,816.06-2) are both of sufficient abundance (Fig. 2c) for anaccurate measurement of their mass difference. The Dm valuefound, 14.01 6 0.02 Da (calculated, 14.02 Da), is of far higheraccuracy than that of Fig. 1 and strongly supports the duallabel assignment.

MS/MS—The Asp-N data further restrict the active sitefrom the 79–177 to the 90–122 region of this protein. Isolationand dissociation of the Fig. 2c ions gave the Fig. 3 spectrum,with 11 assigned fragment ions, confirming that the peptide isAsp90-Gly122. However, recognition of the Dm 5 14.02-Da label(indicated as “114”) was difficult because the off-resonanceSORI excitation (27) activated the lower mass precursor ions toa far greater extent; infrared multiphoton dissociation (31)would be superior in this regard. The correct identification ofthe internal ion Pro109-Phe118, also made ambiguous by the lowfragment ion abundance from the 3,816.06-2 precursor (Fig. 3,inset), would have further restricted the active site.

Isotopic Mixture—To minimize differences in substrate reac-tivity as well as SORI excitation, a 1:2 mixture of 2-CH3 and2-CD3 was reacted with thiaminase I, and the products weredigested with Asp-N. The spectrum of this unfractionated di-gest (Fig. 4a) contained ;40 components of sizes 1.2–4.5 kDa.Now even a casual inspection identifies the labeled peptide; thepeak grouping at m/z 955 is highly distinctive with twice the

4 T. D. Wood, Z. Guan, M. Borders, L. H. Chen, G. L. Kenyon, andF. W. McLafferty, submitted for publication.

FIG. 2. a, thiaminase I incubated with a1:1 mixture of 2 and 2-CH3 and Asp-N-digested, with total products yielding thispartial spectrum (m/z 760–766). b, twothiaminase I samples incubated sepa-rately with 2 and 2-CH3, mixed, and Asp-N-digested, with the products yieldingthis spectrum. c, expansion of the circledregion in b. calc, calculated.

FIG. 3. MS/MS spectrum from SORIfragmentation of the Fig. 2c ions.These ions also contained some from thepeptide of Mr 5 3,749.97-0 (designated *).Unidentified ions are designated Un1.

Stable Isotope Localization of Biomolecule Modification 32217

by guest on February 9, 2018http://w

ww

.jbc.org/D

ownloaded from

normal number of isotopic peaks for a peptide of this size (Fig.4b). The measured monoisotopic mass of 3,814.06-0 (i.e. 3,816.07-2) corresponds to Asp90-Gly122 with a 121.06-Da label(3,814.05-0 Da) with a measured 32:68 ratio for the fittedabundances (circles and squares, Fig. 4) of the two components;their measured Dm values of 3.021 and 3.026 Da for the twomost abundant isotopes compare well with 3.019 Da expectedfor m(CD3) 2 m(CH3).

To localize further the modification site within Asp90-Gly122,a region with two internal Lys residues, thiaminase I wastreated with a 1:1 mixture of 2-CH3 and 2-CD3 and digestedwith Lys-C. The product spectrum indicated 22 peptides ofsizes 1 to 8.2 kDa (Fig. 5a); of these, the abundance distribu-tions of the isotopic peak groups beginning at 2,600.37 and2,617.40 Da (Fig. 5b) were highly distinctive with nearly twicethe normal number of isotopic peaks. Fitting the abundanceratios to those expected for the CH3 and CD3 adducts indicatedthese to be in 60:40 and 56:44 ratios for the 2,600 and 2,617peak groups, respectively; any isotope effects are within the 610% error in sample preparation. For the two most abundant

isotopic peaks of the two clusters, Dm 5 3.020 and 3.021 andDm 5 3.015 and 3.015 Da. The 2,617.40-0-Da value matchesthat expected for Arg101-Lys121 with the 121.06-Da label(2,617.39-0), a site localization improved over the Asp-N pep-tide Asp90-Gly122. The labeled 2,600.37 peak at m/z 652 repre-sents the loss of 17.02 6 0.02 Da from M41; loss of ammonia(17.03 Da) has been reported for other peptides with an N-terminal arginine (32–34).

MS/MS of the 2,617.40-0-Da ions gave eight fragment ions,consistent with the Arg101-Lys121 assignment (Fig. 6a). Of thefragments that exhibited broadened isotopic distributions, the1,728.96-0 ions (Fig. 6c) corresponding to y13 localized themodification to the 13-residue region Pro109-Lys121; the com-plementary b8 ion was not labeled. The measured Dm values forthe two most abundant isotopic peaks of the y13 are 3.022 and3.014, and the isotopic abundances correspond to a label ratioof 67:33; differential SORI excitation of the CH3- versus theCD3-labeled molecular ions could have altered their 56:44 ra-tio. NH3 loss occurs from the N-terminal b7 and b8 ions but notfrom y13.

FIG. 4. a, spectrum of peptides from anAsp-N digest of thiaminase I that hadbeen labeled with a 1:2 mixture of 2-CH3and 2-CD3. b, expansion of the circled re-gion in a, with best fits of calculatedabundance distributions for labeling with2-CH3 (squares) and 2-CD3 (circles) in a32:68 ratio.

FIG. 5. a, spectrum of peptides from aLys-C digest of thiaminase I that hadbeen labeled with a 1:1 mixture of 2-CH3and 2-CD3. b, expansion of m/z 650–660,squares and circles as in Fig. 4 corre-sponding to 60:40 and 56:44 ratios for2,600- and 2,617-Da peak groups, respec-tively. calc, calculated.

Stable Isotope Localization of Biomolecule Modification32218

by guest on February 9, 2018http://w

ww

.jbc.org/D

ownloaded from

Active Site Localization with a Third Protease—Within theregion Pro109-Phe118, the less specific enzyme a-chymotrypsincould cleave on the C-terminal side of Leu112, Leu116, Leu117,Phe118, or Tyr119. Its partial proteolysis of the 1:1 2-CH3:2-CD3

inactivated thiaminase I with quenching after 10 and 36 minyielded spectra with 65 and 70 components, respectively. De-spite this complexity, peak groups beginning at 1,614.87-0 inthe 36-min spectrum (Fig. 7) and at 3,619.94-0 Da in the10-min spectrum (not shown) were recognizable by their broad-ened isotopic distributions. These represented CH3/CD3 ratiosof 58:42 and 45:55, with Dm 5 3.027 and 3.026 (two mostabundant isotopic peaks) and Dm 5 3.024 and 3.015, respec-tively. These Mr values correspond (expected, 1,614.86-0 and3,619.92-0) to the 121.06-Da-labeled peptides Gly107-Tyr119

and Gly107-Tyr136 from proteolysis at Tyr106, Tyr119, andTyr136. This simple identification of one out of 70 peptideslimits the labeled site to 13 amino acids without assignment ofany other peptides to those possible from common chymotrypticcleavages (i.e. C-terminal to Phe, Tyr, Trp, and Leu). Unfortu-nately, MS/MS of the Gly107-Tyr119 ions gave only one fragmention cluster displaying the isotopic signature (Fig. 7, inset), witha measured CH3/CD3 65:35 ratio and Dm 5 3.018 and 3.014.Assignment of this as b12 localizes the active site to Gly107-Phe118. These data combined with the Lys-C data localize thelabeled site to Pro109-Phe118, confirming the Pro109-Phe118

poorly indicated (Fig. 3, inset) by Asp-N digestion and MS/MS.The only nucleophilic residues in this region are Cys113 andThr114.

Identification of Cys113 as the Modification Site—The MS/MSspectra of Figs. 3 and 6 contained anomalous peaks that cor-responded to the loss of a singly charged ion or neutral ofunusual mass. Dissociation of the (Asp90-Gly122)51 peptide ions(Fig. 3) gave a (3,660.99 Da)41 ion that corresponds to the loss

of a singly protonated fragment ion of 141.05 Da (neutralmass), whereas losses of 141.03, 141.03, and 141.02 Da wereobserved from y30

41, y3141, and y14

21, respectively. Similarly, in Fig.6 the (Arg101-Lys121)41 peptide ions of 2,619.39-2 and2,622.41-2) produce 2,464.35-2 ions by the losses of 155.04 and158.06 Da; such neutral losses from the y13

21 ion are consistentwith the formation of the 1,573.93-0 Da ion. These reactions allinvolve loss of the isotopic label (Fig. 6, c versus b); the onlyapparent explanation is that collisional dissociation is causingthe Asp90-Gly122 ions to lose (Scheme II) 3a-SH and the Arg101-Lys121 ions to lose 3b-SH. Thus the suicide substrate must bebound to the enzyme through a sulfur linkage, namely a Cysresidue; the only one in these peptides is Cys113, identifying themodification site exactly. This ion fragmentation in which thesuicide substrate is lost with the active-site nucleophile makespossible the identification of a single labeled residue in a 379-amino acid protein using only the spectrum of one enzymaticdigest (Fig. 2 or 5) and one MS/MS spectrum of its labeledpeaks (Fig. 3 or 6) and without assigning the many other Mr

values observed to other peptide products. Another isotopiclabel, 4,5,6-13C3-2, gave similar results.

Lower Performance MS Instrumentation—Without the 105

resolving power (RP) capabilities of FTMS, labeled site locationin large proteins would be more difficult but still possible inmany cases. Quadrupoles can achieve ;103 RP for MS and

SCHEME II.

FIG. 6. a, MS/MS spectrum from disso-ciation of the peaks centered at m/z 656 inFig. 5b. b, expansion of data for the y13

21,155.03-Da fragment ion. c, expansion ofdata for the y13

21-fragment ion; squaresand circles as in Fig. 4 corresponding to a67:33 ratio. calc, calculated.

FIG. 7. Thiaminase I labeled with1:1 2-CH3 and 2-CD3 and digestedwith a-chymotrypsin for 36 min, withthe total products yielding this par-tial spectrum (m/z 806.5–815). Inset,partial spectrum from MS/MS of m/z 808–811.5. Squares and circles as in Fig. 4 cor-responding to a 58:42 and 65:35 ratios, re-spectively. G, Gly; Y, Tyr; calc, calculated.

Stable Isotope Localization of Biomolecule Modification 32219

by guest on February 9, 2018http://w

ww

.jbc.org/D

ownloaded from

MS/MS spectra, whereas time-of-flight instruments canachieve substantially higher RP for MS spectra but much lowerfor MS/MS spectra. The charge states of isotopically unresolvedpeptide ions from ESI would have to be determined by decon-volution (5), although not those of the singly charged ions frommatrix-assisted laser desorption ionization (37). For RP 5 103,the Asp90-Gly122-labeled (Dm 5 3 Da) ions of the Fig. 4 spec-trum would form a single peak although broadened versus thatof an unlabeled peptide (at RP 5 103, each of 4-kDa isotopicpeaks will be 4-Da wide at half-height). Even for a label oflarger Dm, RP 5 103 would require prior chromatographicseparation to identify labeled peaks in more complex proteoly-sis mixtures (e.g. 70 peptides, Fig. 7).

CONCLUSIONS

Successful application of this approach should be possible formany proteins. The covalent label must not be removed com-pletely by the proteolysis or ion fragmentation used, but partialremoval that includes a heteroatom of the labeled amino acid(S, N, O) can substantially restrict its location. High resolutionand accuracy are valuable to detect multiple site derivatizationfrom the Mr shift caused by labeling and to identify with highconfidence the Dm indicative of the labeled peaks. Isotopiclabels minimize reaction rate differences, whereas ion dissoci-ation methods other than SORI minimize m/z discrimination inMS/MS spectra.

ESI/FTMS can obtain MS/MS information from 10217-molsamples (14), so that isotopic labeling could make possibleactive-site localization with orders of magnitude lower samplerequirements than current methodology. Application of thistechnique to nucleotides could be far more specific with onlyMS/MS dissociation, as ion fragmentation is much more exten-sive; complete sequence information has been obtained fromMS/MS spectra of a 50-mer DNA (38). Further, for some studiesthe mixed isotopic labeling could be carried out in the biologicalsystem, such as protein oxidation (39) or post-translation phos-phorylation or glycosylation in an 18O-enriched environment.

Acknowledgments—The authors thank Nino Compabasso, ColleenCostello, Einar Fridriksson, Yuan Gao, Ulrich Haupts, David Horn, andSean Taylor for experimental assistance and helpful discussions.

REFERENCES

1. Lundblad, R. L. (1991) Chemical Reagents for Protein Modification, 2nd Ed.,CRC Press, Inc., Boca Raton, FL

2. Silverman, R. B. (1992) The Organic Chemistry of Drug Design and DrugAction, pp. 178–218 and 244–270, Academic Press, Inc., New York

3. Gaskell, S. J. (ed) (1986) Mass Spectrometry in Biomedical Research, JohnWiley & Sons, Inc., New York

4. Biemann, K., and Papayannopoulos, I. A. (1994) Acc. Chem. Res.27, 370–3785. McLafferty, F. W. (1994) Acc. Chem. Res. 27, 379–3866. Kellner, R., Lottspeich, F., and Meyer, H. E. (1994) Microcharacterization of

Proteins, VCH, Weinheim, Germany7. Creighton, T. E. (1993) Proteins, 2nd Ed., W. H. Freeman and Co., New York8. Caprioli, R. M. (1972) in Biochemical Applications of Mass Spectrometry

(Waller, G. R., ed) pp. 735–771, John Wiley & Sons, Inc., New York9. McCloskey, J. A. (1990) Methods Enzymol. (McCloskey, J. A., ed) pp. 329–338,

Academic Press, San Diego, CA10. Blackburn, R. K., and Anderegg, R. J. (1997) J. Am. Soc. Mass Spectrom. 8,

483–49411. Kurian, E., Prendergast, F. G., Tomlinson, A. J., Holmes, M. W., and Naylor,

S. (1997) J. Am. Soc. Mass Spectrom. 8, 8–1412. Marshall, A. G., and Comisarow, M. B. (1974) Chem. Phys. Lett. 25, 282–28313. Mørtz, E., O’Connor, P. B., Roepstorff, P., Kelleher, N. L., Wood, T. D.,

McLafferty, F. W., and Mann, M. (1996) Proc. Natl. Acad. Sci. U. S. A. 93,8264–8267

14. Valaskovic, G. A., Kelleher, N. L., and McLafferty, F. W. (1996) Science 273,1199–1202

15. Fujita, A. (1954) Adv. Enzymol. Relat. Subj. Biochem. 15, 389–42116. Kelleher, N. L., Costello, C. A., Begley, T. P., and McLafferty, F. W. (1995)

J. Am. Soc. Mass Spectrom. 6, 981–98417. Begley, T. P. (1996) Nat. Prod. Rep. 13, 177–18518. Costello, C. A., Kelleher, N. L., Abe, M., McLafferty, F. W., and Begley, T. P.

(1996) J. Biol. Chem. 271, 3445–345219. Schnolzer, M., Jedrzejewski, P., and Lehmann, W. D. (1996) Electrophoresis

17, 945–95320. Miranker, A., Kruppa, G. H., Robinson, C. V., Aplin, R. T., and Dobson, C. M.

(1996) J. Am. Chem. Soc. 118, 7402–740321. Marshall, A. G., Senko, M. W., Li, W., Li, M., Dillon, S., Guan, S., and Logan,

T. M. (1997) J. Am. Chem. Soc. 119, 433–43422. Hutter, J. A., and Slama, J. T. (1987) Biochemistry 261, 1969-197323. Cox, F. B., and McElvain, S. M. (1943) in Organic Synthesis Collections (Blatt,

A. H., ed.) Vol. II, pp. 279–282, John Wiley & Sons, Inc., New York24. Wilm, M., and Mann, M. (1996) Anal. Chem. 68, 1–825. Valaskovic, G. A., Kelleher, N. L., Little, D. P., Aaserud, D. J., and McLafferty,

F. W. (1995) Anal. Chem. 67, 3802–380526. Beu, S. C., Senko, M. W., Quinn, J. P., Wampler, F. M., III, and McLafferty,

F. W. (1993) J. Am. Soc. Mass Spectrom. 4, 557–56527. Senko, M. W., Beu, S. C., and McLafferty, F. W. (1995) J. Am. Soc. Mass

Spectrom. 6, 229–23328. Loo, J. A., Udseth, H. R., and Smith R. D. (1988) Rapid Commun. Mass

Spectrom. 2, 207–21029. Gauthier, J. W., Trautman, T. R., and Jacobsen, D. B. (1991) Anal. Chim. Acta

246, 21130. Roepstorff, P., and Fohlman, J. (1984) Biomed. Mass Spectrom. 11, 60131. Little, D. P., Speir, J. P., Senko, M. W., O’Connor, P. B., and McLafferty, F. W.

(1994) Anal. Chem. 66, 2809–281532. Campbell, S., Rodgers, M. T., Marzluff, E. M., and Beauchamp, J. L. (1995)

J. Am. Chem. Soc. 117, 12840–1285433. Schnier, P. D., Price, W. D., Jockusch, P. A., and Williams, E. R. (1996) J. Am.

Chem. Soc. 118, 7178–718934. Deery, M. J., Summerfield, S. G., Buzy, A., and Jennings, K. R. (1997) J. Am.

Soc. Mass Spectrom. 8, 253–26135. Senko, M. W., Speir, J. P., and McLafferty, F. W. (1994) Anal. Chem. 66,

2801–280836. Deleted in proof37. Karas, M., and Hillenkamp, F. (1988) Anal. Chem. 60, 2299–230138. Little, D. P., Aaserud, D. J., Valaskovic, G. A., and McLafferty, F. W. (1996)

J. Am. Chem. Soc. 118, 9352–935939. Stadtman, E. R. (1993) Annu. Rev. Biochem. 62, 797–82140. Marshall, A. G. (1996) Acc. Chem. Res. 29, 307–31641. Murata, K. (1982) Ann. N.Y. Acad. Sci. 378, 146–15642. Earl, J. W., and McCleary, B. V. (1994) Nature 368, 683–684

Stable Isotope Localization of Biomolecule Modification32220

by guest on February 9, 2018http://w

ww

.jbc.org/D

ownloaded from

Neil L. Kelleher, Robb B. Nicewonger, Tadhg P. Begley and Fred W. McLaffertyNUCLEOPHILE OF THIAMINASE I

and Tandem High Resolution Mass Spectrometry: THE ACTIVE SITE Identification of Modification Sites in Large Biomolecules by Stable Isotope Labeling

doi: 10.1074/jbc.272.51.322151997, 272:32215-32220.J. Biol. Chem. 

  http://www.jbc.org/content/272/51/32215Access the most updated version of this article at

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/272/51/32215.full.html#ref-list-1

This article cites 34 references, 3 of which can be accessed free at

by guest on February 9, 2018http://w

ww

.jbc.org/D

ownloaded from