Anal. Chem. 1990, 62,693-698 Boca Raton, FL, 1984; pp 3, 30-31, 355-365. (2) Slgel. C. W.; Woolley, J. L. I n Densitometry in Thin Layer Chromatcgraphy; Touchstone, J. C., Sherma. J., Eds.; Wlley: New York, 1979. (3) Sophien, L. H.; Piper, D. L.; Schneiler, 0. H. The Suffapyr/m&%ws; Press of A. Collsh: New York, 1952; p 10. (4) Bratton, A. C.; Marshall, E. 1. J . Bid. Chem. 1939, 128, 537. (5) Sato, S.;Higuchl, S.; Tanaka, S. Anal. Chlm. Acta 1980, 720. 209. (6) Jonhson. K. L.; Jeter, D. T.; Clalbone, R. C. J. Pbarm. Sci. 1975, 64, 1657. (7) Horwltz, J. J. Assoc. Off. Anal. Chem. 1981, 64, 104. (8) Horwltz, J. J. Assoc. Off. Anal. Chem. 1981, 64, 814. (9) Vo-Dlnh, T.; Meler, M.; Wokaun, A. Anal. Chbn. Acta 1986. 187, 139. (10) Lasema, J. J.; Camplglla, A. D.; Wlnefordner, J. D. Anal. Chem. 1989, 67, 1697. (1 1) Vo-Dinh, T. I n Chemlcal Ana/ys/s of Po/ycycllc Aromatic Compounds, Vo-Dinh, T., Ed.; Wlley: New York, 1989; p 451. (12) Torres. E. L.; Wlnefordner, J. D. Ana/. Chem. 1987, 59, 1826. (13) Laserna. J. J.; Berthod, A.; Winefordner, J. D. Talanta 1987, 4 1 , 605. (14) Laserna, J. J.; Berthcd. A.; Wlnefordner, J. D. Microchem. J. 1988, 38, 125. (15) Carrabba, M. M.; Edmonds, R. B.; Rauh, R. D. Anal. Chem. 1987, 59, 2559. (16) Moody, R. L.; Vo-Dinh, T.; Fletcher, W. H. Appl. Spectrosc. 1987, 4 1 , 986. (17) Enlow, P. D.; Bunclck, M. C.; Warmack, R. J.; Vo-Dinh, T. Anal. Chem. 1988, 58, 1119. (18) Alak. A. M.; Vo-Dlnh, T. Anal. Chem. 1987, 59, 2149. (19) Laserna, J. J.; Campiglia, A. D.; Wlnefordner, J. D. Anal. Chim. Acta 1988, 208, 21.
693
(20) Creighton, J. A.; Blatchford. C. G.; Albrecht, M. G. J. Chem. Soc., Faraday Trans. 1979, 75, 790. (21) Brock, T. D. Membrane Filtration: A User‘s GUMS and Reference Msnual; Science Tech, Inc.: Madison, WI, 1983; p 185. (22) Ddllsh, F. R.; Fateley. W. 0.; Bentley, F. F. cherecterlstic Raman FreQUenCieS of Organic Compounds; Wlley: New York, 1974 p 53. (23) Baranska, H.; Labudzlnska, A.; Terpinskl, J. Laser Raman Spectrometry. AnalytcalAppllcations; Ellis Horwood: Chlchester, 1987; p 136. (24) Laserna, J. J.; Torres, E. L.; Winefordner, J. D. Anal. Chim. Acta 1987, 200, 469. (25) Berthod, A.; Lasema, J. J.; Winefordner, J. D. Appl. Spectrosc. 1988, 4 7 , 1137. (26) Blatchford, C. G.; Campbell, J. R.; Crelghton, J. A. Surf. Sci. 1983, 720. 435. (27) Kerker, M.; Sllman, 0.; Wang. D.4. J. phvs. Chem. 1984, 88, 3168. (28) Heritage, J. P.; Bergman, J. G.; Pinczuk, A.; Worlock, J. M. Chem. mys. Len. 1979, 67, 229. (29) Cooney, R. P.; Mernagh, T. P.; Mahoney, M. R.; Spink, J. A. J. phvs. Chem. 1983, 87, 5314.
RECEIVED for review August 22, 1989. Accepted December 26,1989. J. J. Laserna gratefully acknowledges the Direccion General de Investigacion Cientifica y Tecnica (DGICYT), Ministerio de Educacion y Ciencia, Spain, for financial support. This research was supported by NIH Grant NIHR01-GM11373-26.
Effect of Reducing Disulfide-Containing Proteins on Electrospray Ionization Mass Spectra Joseph A. Loo, Charles G. Edmonds, Harold R. Udseth, and Richard D. Smith* Chemical Methods and Separations Group, Chemical Sciences Department, Pacific Northwest Laboratory, Richland, Washington 99352
Electrospray lonlzatlon produces multiply charged molecular ions for blomolecules with molecular weights In excess of 100 000. This allows mass spectrometers with llmlted mass-tocharge range to extend thelr molecular weight range by a factor equal to the number of charges. The maxlmum number of observed charges for peptides and smaller protdns correlates well with the number of baslc amino acld residues (Arg, Lys, His), except for disulfklecontalnlngmdecules, such as lysozyme and bovlne albumin. However, reduction of dlsulfide linkages wtth 1,4-dlthbthrettol (Cleland’s reagent) may alkw the protein to be In an extended conformation and make “burled” baslc resldues available for protonatlon to yield hlgher charged molecular Ions by the electrospray Ionization process. For larger proteins reduction of dlsulflde brldges greatly Increases the maxlmum charge state, but charging of basic amlno acid resldues remalns less efflcient than for smaller proteins.
INTRODUCT10N Electrospray ionization coupled with mass spectrometry (ESI-MS) began with the seminal experiments of Dole (1,2), and after a 20-year gestation, reemerged in the laboratory of Fenn ( 3 4 , who first published results of electrospray ionization mass spectra of large molecules with molecular weights of up to 40000 and demonstrated the propensity for multiple charging produced by this soft ionization technique (6,7).The ability to produce multiply charged molecular ions for biomolecules allows mass spectrometersto analyze compounds with molecular weights exceeding the mass-to-charge ( m / z )range
of the instrument by a factor equal to the molecule’s charge state. In fact, proteins with molecular weights in excess of 130000 have been successfully analyzed in our laboratory by ESI-MS with a quadrupole mass spectrometer of limited m / z range (1700) (8). Initial tandem mass spectrometry (MS/MS) results of m / z selected multiply charged molecular ions of relatively large biomolecules appear promising (9-11). The mechanism for desorption of multiply charged molecular ions from highly charged liquid droplets remains uncertain. The ion evaporation model of Iribarne and Thomson (12,13),in which Coulombic explosion of rapidly evaporating liquid droplets of high charge density results in smaller charged droplets and continues this sequence until solute field-assisted desorption occurs, is widely accepted. However, slightly different mechanisms involving expulsion of very small charged droplets, or perhaps multiply charged molecular clusters as the Rayleigh limit is approached, may also operate. In this model, these very small droplets would then form the analyte ions of interest by evaporation of neutral solvent molecules. Regardless of the detailed mechanismb) for the small submicrometer diameter droplets, the efficient transfer to the gas phase of highly charged macromolecules from the surface of the shrinking droplets is now well established. The maximum number of charges a molecule can retain by the ESI process for poly(ethy1ene glycol) oligomers has been estimated by Wong et al. (5)based upon an electrostatic repulsion argument. In their model the maximum charge state consistent with molecular stability was determined. Unsurprisingly, their ESI-MS data indicated multiple charging somewhat less than predicted by this approach. One explanation advanced for this observation was that ion evaporation is determined by
0003-2700/90/0362-0693$02.50/00 1990 American Chemical Society
694
ANALYTICAL CHEMISTRY, VOL. 62, NO. 7, APRIL 1, 1990
the kinetics for transport to the surface and the chargestate-dependent rate of ion evaporation from the droplet surface (7). For peptides and proteins, we had previously observed (14) [as later supported by Covey et al. (15)] an approximate correlation between the maximum positive (multiply protonated) charge state observed in our positive ion electrospray ionization mass spectra and the total number of basic amino acid residues (Le., arginine, lysine, and histidine) in the primary sequence plus the N-terminal amino group. For example, glucagon (average molecular weight, M,= 3483) possesses two arginines, one lysine, one histidine, and the N-terminal amino group; its ESI mass spectrum shows (M + 5HI5+as the highest multiply charged species. However, the correlation for hen egg white lysozyme, a protein with 18 basic amino acids, is poor; (M + 14H)I4+is the most highly charged molecular ion observed. While the solution pH strongly affects protein charge state in solution, the effect upon ESI mass spectra is often quite small. A general assumption is that quaternary structure is lost for electrosprayed proteins, and for multimeric proteins the noncovalently bonded subunits are observed separately in the electrospray ionization mass spectra (8). For example, an ESI mass spectrum of lactate dehydrogenase from rabbit muscle (M, = 140000) yields multiply charged ions indicative of the four identical subunits ( M , = 35000) (8). Clearly, the electrostatic forces leading to gas-phase ion formation are more than sufficient to separate the weakly bonded and similarly charged subunits in the absence of the stabilizing role of the solvent. Little is yet known regarding the effect of higher order structure on the extent of multiple charging. Thus, the factors governing both maximum charge state and the distribution of states remain largely undetermined. In this study we have investigated aspects of the effect of protein structure upon ESI mass spectra by comparison of proteins with and without reduction of native disulfide bridges. By cleavage of cysteinecysteine bonds, and thus by affecting the higher order structure of the protein, a dramatic effect on the number of positive charges is observed. The analytical utility of disulfide reduction reactions with ESI-MS and its relevance to collisionally activated dissociation (CAD) studies are addressed.
EXPERIMENTAL SECTION The instrumentation used in this study and typical operating Samples conditions have been previously described (8,9,16,17). in an aqueous acidic buffer solution, consisting of 1-5% (v/v) glacial acetic acid, are introduced through a 100 rm i.d. fused silica capillary at a rate of 1 gL/min and mixed with a liquid sheath electrode, typically methanol, flowing at 3 pL/min, at the tip of the electrospray ionization source. Analyte and sheath flow are independently controlled by separate syringe pumps. A potential of +5 kV is applied to the sheath electrode, producing highly charged liquid droplets of 1 km diameter (18) at atmospheric pressure in a flow of dry nitrogen to aid the desolvation process. The ESI source is mounted 1.5 cm from the entrance of the quadrupole MS. Highly charged ions are sampled through a 1-mm nozzle orifice and 2-mm skimmer and are efficiently transported through a cryopumped region by the radio frequency (rf) quadrupole lens to a quadrupole mass spectrometer for detection. For positive ions, the typical focusing lens voltage is +1 kV, with the nozzle at +200 V and the skimmer at ground potential. The mass spectrometer (Extrel Co., Pittsburgh, PA) used for these studies has an effective m / z range of 1700. Scan rates varied between 30 and 2 min over the entire m/z range. With the current data system (Teknivent,St. Louis, MO), peak abundancesare collected at integer m / z values. Routine calibration of the m / z scale for ESI-MSwas performed with low molecular weight polymers, such as poly(ethy1ene glycol) (average molecular weight lOOO), monitoring both the singly charged (singly sodiated) and doubly charged (doublysodiated)molecular ion distributions. Calibration was periodically checked by ESI-MS with well-characterized
-
i 1 i
c:
p
/'
1
'
I
wlDTT/
1 /
i
l
/
.i _
/
X
U
i
3
i
3w,
,6P4 , ,
c
si: Nc
l " " I
3c
Eosic A-
'
151: -0 Ac d s
'
I
'
233
Figure 1. Correlation between basic amino acid residues (arginine,
lysine, and histidine) and maximum positive charge observed in ESI mass spectra for a number of peptides and proteins examined in our laboratory. The dashed reference line has a slope of 1. (See Table I for a more complete listing.)
protein samples, such as horse heart cytochrome c (M, = 12360). While 50-100 pmol of material was consumed for each spectrum, we have demonstrated the ability to analyze low femtomole quantities with our present instrumentation (19). The protein samples were commerciallyobtained from Sigma Chemical Co. (St. Louis, MO) with the exception of lactate dehydrogenase (CALBIOCHEM Corp., San Diego, CA) and methionine-human growth hormone (kindly provided by Dr. Ian Jardine, Finnigan MAT) and used without further purification. ESI mass spectra were obtained for the native proteins immediately after glacial acetic acid was added to the aqueous solution. Reduction of disulfide bonds was carried out with Cleland's reagent, 1,4-dithiothreitol (DTT) (20). A small amount of an aqueous DTT solution was added to an aliquot of the undenatured protein solution (-0.1 mM) in distilled water, such that the final DTT concentration was approximately 5-10 mM. After a period of over 12 h at room temperature, glacial acetic acid was added ( 1 4 % )to the protein sample containing DTT and ita mass spectrum obtained. A commercial sample of an S-carboxymethyl derivatized bovine a-lactalbuminwas prepared in similar fashion.
RESULTS AND DISCUSSION As first demonstrated by Fenn and co-workers (6, 7), electrospray ionization produces multiply charged (typically protonated) molecular ions for high molecular weight peptides and proteins. For compounds with molecular weights greater than 10000, a typical ESI mass spectrum shows a bellshaped distribution of charge states, composed of multiply protonated molecular ions, effectively extending the molecular weight range addressable by a factor equal to the number of charges. For a pure compound, any two peaks in the ESI mass spectrum provide sufficient information to determine both the charge state and the ion mass since adjacent peaks differ by only one charge (6-11,14-16). Additional peaks can be utilized for improved molecular weight precision and accuracy. The ability to shift the multiple charge distribution to higher mlz (a decrease of charge state) by changing the liquid sheath composition (14) or by increasing the collision energy in the high-pressure region of the atmospheric pressure inlet (such that the more highly charged species undergo CAD) has been previously demonstrated (16). For most peptides and proteins, with molecular weights ranging from a few hundred to -40000, a relatively good linear correlation is observed between the number of basic amino acid residues (plus one for the N-terminus) and the maximum positive charge state observed in the ESI mass spectrum, as demonstrated in Figure 1and Table I. Additionally, tandem mass spectrometry (9) of the various multiply protonated species for melittin (M, = 2845) (up to the (M +
-
695
ANALYTICAL CHEMISTRY, VOL. 62, NO. 7, APRIL 1, 1990 100
Table I. Peptides and Proteins Analyzed by Electrospray Ionization-Mass Spectrometry amino acid residues disulfide basic total (+N-term) linkages
Hen Egg W h i t e Lysozyme
I
1103
A
IMW 14,3061 12*
i
'
maximum
ESI charge
peptide/protein
M,
Leu-enkeplhalin bradykinin gramicidin S angiotensin I me1ittin glucagon cytochrome c (horse) a-lactalbumin (bovine) lysozyme (hen egg) myoglobin (horse) j3-lactoglobulin
555 1060 1141 1296 2846 3 483 12 360
5 9 10 10 26 29 104
1 3 2 4 6 5 25
0 0 0 0 0 0 0
1 3 2 4 6 5 21
14 175
123
17
4
13
14306
129
19
4
14
16 950
153
33
0
29
18277
162
21
2
19
trypsin inhibitor 20 091 181 (soybean) Met-human 22 255 192 growth hormone carbonic 29 024 259 anhydrase (bovine) lactate 35 500 333 dehydrogenase (rabbit muscle) bovine albumin 66266 582 77 500 684 ovotraneferrin bovine albumin 133OOO 1164 dimer
22
2
23
24
2
21
39
0
42
51
3
55
100 98 200
17 15 34
71 73 120
B
6HI6+species) produced by electrospray ionization support this assumption (i.e., that the positive charges reside on the three lysine residues, the two arginine sites, or the N-terminal amino group). Biemann and co-workers (21) have also speculated that protonation for peptides occurs at basic amino acid sites (when present). This suggestion is also supported by gas-phase proton affinity data for amino acids; arginine, histidine, and lysine were found to have the highest proton affinities (22). However, ESI mass spectra of selected proteins with molecular weights exceeding loo00 (e.g., lysozyme) do not yield the expected high positive charge states that would be indicated by their amino acid sequence. A common feature of such proteins is the presence of more than one disulfide bridge. For example, hen egg lysozyme and bovine a-lactalbumin each have four disulfide bonds. Bovine albumin has as many as 17 cysteinecysteine bridges in its primary sequence (23), and the maximum number of charges observed in its ESI mass spectrum (8) (slightly greater than 60 positive charges) is much less than the 100 basic residues seemingly available for protonation. A number of studies dealing with the mass spectrometric analysis of disulfide-containing peptides with fast atom bombardment (FAB-MS) (24-27) and plasma desorption (PD-MS) (28,29) have been published. Monitoring the reduction reaction of disulfide bonds with reducing agents (such as DTT) by mass spectrometry allows one to assess the rate and extent of reaction and obtain important information regarding the number of disulfide bonds (by measuring the small mass increase associated with the reaction to form sulfhydryl groups) and the nature of the linkages (e.g., whether the peptide is composed of several peptide chains linked by interchain disulfide bonds). For example, it is possible to follow the reduction reaction of insulin, by FAB-MS and PD-MS,
I
1302
I
I
i 600
800
1000
1200 m/z
I '%e3
1400
16'00
1800
Figure 2. (A) E S I mass spectra of hen egg white lysozyme in 5 % glacial acetic acid and (B) upon the addition of DTT.
and monitor the disappearance of the intact molecular ion as the ions indicative of the A-chain and the B-chain, bound by two disulfide bridges in the native state, appear (25,26, 28, 29). ESI-MS has been used (11)to distinguish the oxidized form of somatostatin ( M , = 1637.9), a cyclic peptide due to an intramolecular cysteine-cysteine linkage, from ita reduced state (M, = 1639.9) in much the same manner as reported by PD-MS (28) and FAB-MS (25,30). A 1 and 0.67 m / z unit difference between the doubly and triply charged molecular ions, respectively, of the oxidized and reduced state was measured by ESI-MS instead of measuring a molecular weight difference of 2 for the singly charged molecular ion. Tandem mass spectrometry (11)of the doubly and triply charged reduced state yields much more fragmentation versus the corresponding oxidized state since two bonds must be broken to generate a typical product ion from the cyclic portion of the molecule. Martin and Biemann (30) have previously published similar results for the singly charged molecular ion generated by FAB. A recently published study (31)demonstrated the capability to monitor the reduction reaction of the cyclic peptide, oxytocin, by 0-mercaptoethanol with ion spray MS and MS/MS. Alternatively, oxidation of disulfide-containing peptides by performic acid (oxidizing half-cystinyl groups to cysteic acid) followed by mass spectrometric analysis of peptides can determine the number of disulfide linkages present (27). Since disulfide bond reduction in proteins (M, > 10 000) using DTT has been noted (32), the same type of information should be obtainable with electrospray ionization mass spectrometry for substantially larger proteins. Because it is well established that protein conformation in solution depends in a complex fashion upon the intramolecular forces dictated by its amino acid sequence, the direct reductive cleavage of disulfide bonds should greatly affect the higher order structure of the protein. A protein with disulfide bonds in its native conformation may have several basic amino acid sites that are effectively "buried" in the globular structure, resulting in more hydrophobic environments that are not as readily protonated. Disulfide-cleavage reactions should allow the protein to "relax" into more extended conformations, allowing such buried sites to be more readily protonated. Evidence for this is demonstrated by the following comparisons of mass spectra of ESI generated gas-phase ions from highly charged liquid droplets for disulfide-containingproteins, and its D'IT reduced form. The typical electrospray ionization mass spectrum for lysozyme is shown in Figure 2A with multiple charging clearly evident up to the (M + 14H)14+ species. However, as previously discussed, egg white lysozyme
696
ANALYTICAL CHEMISTRY, VOL. 62, NO. 7, APRIL 1, 1990
100
1237
Met
A
Human G r o w t h H o r m o n e
I M W 22,2551
Table 11. M,Determination for Three Proteins by ESI-MS
118*
Mr
protein
1
13'0
-
B
lysozyme (hen egg)b native reduced a-lactalbumin (bovine)' native reduced carboxymethylated Met-human growth hormoned native reduced
M , (measured)
(expected)a
14306.5 f 4.1 14311.5 & 6.4
14306.2 14 314.1
14 174.7 f 6.1 14 180.8 f 5.7 14645.6 f 5.8 (14 703.0 f 2.7)e
14 175.0 14 183.0 14647.4 (14 705.4P
22254.2 f 4.8 22259.3 f 7.4
22 255.2 22 259.2
nFrom amino acid sequence. bFigure 2. cFigure 4. dFigure 3. text for explanation.
e See
600
800
1000
1200 miz
1400
1600
1800
loo--
Flgure 3. (A) Electrospray ionization mass spectra of Met-human growth hormone in 5 % acetic acid and (B) upon reduction with DTT. 100
___
"*11182
I
1
I M W 66,3001
]
1232
with D T T
7I I1ilZSi
al
a
,977,481
772. 775
1189
1577
-
8661
A
1211
A
1289
Bovine a-Lactalbumin I M W 14,1751
I1255
Bovine Albumin
carboxymethylated
[/
1047 1051
600 1+222 1226
1000
1200 miz
1400
1600
1800
Figure 5. (A) ESI mass spectra of bovine albumin before and (e)after reduction of its disulfide linkages.
1333 1338
0
800
LT_______
600
800
10'00
1200 m/z
1400
1600
1800
Flgure 4. (A) ESI mass spectra of bovine a-lactalbumin, (e) after reduction with DTT, and (C) a mass spectrum from a commercial sample of the reduced and carboxymethylated form.
has a total sum of 18 basic amino acids. Breaking cysteinecysteine linkages with the addition of DTT allows greater multiple protonation to occur, as demonstrated in Figure 2B. The same effect has been observed for a number of other proteins. Met-human growth hormone ( M , = 22 255) has a total sum of 23 basic amino acids and an N-terminal group, yet its ESI mass spectrum (Figure 3A) shows only a trace of the (M + 21H)21+molecular ion, with the (M + 18H)1wspecies dominant. The addition of DTT to reduce its two disulfide bonds (Figure 3B) increases the relative abundance of the more highly charged molecular ions, allowing a maximum charge state where all basic sites can be protonated. Similar results are shown in Figure 4 for bovine a-lactalbumin, a protein with a covalent structure quite similar to chicken lysozyme that also possesses four disulfide linkages. Its typical ESI mass spectrum (Figure 4A) shows the 12+ species as the highest charged ion, compared to the total of 16 basic amino acids. Reduction of the disulfide bonds with DTT allows the molecule to obtain up to 16 positive charges (Figure 4B). Reduction of disulfide bonds with subsequent carboxymethylation (with iodoacetate) yields similar results, but with even greater protonation than anticipated, as shown in Figure 4C. From Bojesen's proton affinity (PA) data (22),glutamine has a similar gas-phase PA as lysine. It may be possible that
three of the five glutamine residues of a-lactalbumin can be protonated under our ESI conditions. Similarly, the data from lysozyme (Figure 2) and Met-human growth hormone (Figure 3) indicate additional charging beyond predicted values. Both proteins contain Gln residues (3 and 14, respectively) as possible protonation sites. More experiments are needed to verify these results. In addition to an increase in multiple charging (up to +20) for a-lactalbumin, an increase in molecular weight due to eight carboxymethyl groups (an approximate molecular weight increase of 472 from the disulfide-bearing molecule), is deduced from the m / z shift observed (Table 11). This shift allows one to readily determine the number of disulfide linkages in large proteins by ESI-MS. A closer inspection of the mass spectrum of S-carboxymethyl-a-lactalbumin also reveals that each multiply charged peak is actually a doublet with a peak separation consistent with an extra carboxymethylated moiety (Table 11). Alkylation of similar-sized proteins with iodoacetate have found contributions due to reaction of the histidine and methionine side chains to form the carboxymethyl modification (33,34). Lactalbumin has one methionine and three histidine residues which may undergo this type of reaction. An examination of the rate of reaction of bovine a-lactalbumin with iodoacetate indicated that the single methionine residue was the most reactive site (34). Although the measured peak positions listed allowed an M, accuracy and precision of generally better than *0.03%, improvements in the data system to allow recording of peak positions in less than one m / z unit intervals should
ANALYTICAL CHEMISTRY. VOL. 62, NO. 7, APRIL 1. 1990 100
to tax the electrospray ionization process by demanding too high a charge concentration on a small region of the droplet surface.
Bovine Albumln Dlmer (MW 133,000) 1256
A
'100
1800
wlDTT
)O +
mu,, ,
0
600
8oo
iuc::
..ino
iioo
1600
is00
ill, 7
Figure 6. (A) ESI mass spectra of bovine albumin dimer before and (6) after the addition of DTT.
markedly enhance M. measurements and should allow for unambiguous determination of the number of disuliide bonds (8).
--
697
Figures 5 and 6 depict similar results obtained for bovine albumin (M, 66 600) and the native bovine albumin dimer species (M, 133000). Again, the dramatic increase in relative abundance of the higher charged species is evident. Over 175positive charges can be resolved in the maSS spectrum for the albumin dimer molecule treated with DTT, an increase of over 50 charges relative to the unreduced form. The bimodal multiple charge distribution obtained after reaction with dithiothreitol is probably due to incomplete reduction or re-formation of the disulfide bonds. Separation of the reaction products by LC/MS (31) or capillary electrophoresis-MS (8,35) should clarify the situation. Reactions are typically performed in buffered solutions with a pH of approximately 8-9 (20, 32). No attempt was made in this report to study the pH dependence of the effect of protein denaturing agents such as urea or guanidinium chloride upon the reduction reaction. However, the reaction in an unbuffered distilled water solution proceeded much slower than that reported for the buffered system (32). For example, Figure 5 compares mass spectra of unreduced bovine albumin (top figure) and the reaction product of albumin with DTT sampled after 12 h. A spectrum obtained after only 4 h shows only a slight increase in abundance of the higher charged molecular ions. Nearly complete reduction of bovine albumin with dithiothreitol is reported to occur in approximately 1h at pH 8.2 (32). It is conceivable that even after 12 h with our system, reduction of disulfide bonds remains incomplete. Its globular structure may he partially retained, preventing protonation to the extent expected from the number of basic amino acid residues. Also, such large molecules may begin
CONCLUSIONS The ability to produce and observe by ESI-MS a dramatic increase in the number of positive charges for disulfide-containing proteins as large as M,> 130000 upon the reduction of disulfide linkages with 1,Cdithiothreitol has been demonstrated. The enhanced protonation is presumably a result of allowing the protein molecule to attain a more extended tertiary structure, thus allowing otherwise inaccessible sites to he protonated. A good linear correlation between the maximum positive charge state observed in the ESI mass spectrum and the number of hasic residues has been oberved in the absence of disulfide bonds. The appearance of more highly charged molecular ions due to the reduction of disuKde bonds of peptides and proteins can he readily monitored by ESI-MS. The potential for quantitative determination of the number of disulfide bonds, particularly for larger proteins remains to he determined. The production of more multiply charged molecular ions extends the mass spectrometer's molecular weight limit for MS and MS/MS experiments of large peptides and proteins. The efficient protonation is especially important for instruments of limited m/z range and the analysis of disulfidecontaining proteins with limited number of basic amino acid residues. Highly charged parent ions allows more highly charged daughter ions to he produced and deteded (9,11,36). For example, with a triple quadrupole instrument (1400 m/z limit), doubly charged sequence ions were detected in the MS/MS spectrum of the (M 3H)3Cion from the B-chain of bovine insulin (oxidized, M , = 3496) (36). Abundant 4+ product ions were formed from CAD of the (M + 6H)6*ion of melittin; however, the highest charged fragment ions from dissociation of the (M + 5H)5Cspecies were 3+ (9). Greater CAD efficiencieshave been obtained for more highly charged species from large peptides (9, 37, 39) and proteins (10, 11, 37-39), facilitated by the higher energy collisions possible and perhaps by the strained (extended)conformation due to charge repulsion. Reduction of disulfide bridges, in addition of 'opening" up the molecule and allowing product ions to be generated from regions formerly enclosed by disulfide linkages, allows ESI-MS/MS to directly access and probe more highly charged protein species.
+
LITERATURE CITED (1) Dole. M.: Mack, L. L.: Hines. R. L.: Mobley, R. C.: Ferguson. L. D.: Alice. M. B. J . wlem. mys. 1988. 49, 22462249. (2) Mack. L. L.: Kralk. P.: Rheude, A.: Dole. M. J . Chem. phys. 1970, 52. 4977-4986. (3) Yamashta, M.: Fenn. J. B. J . mys. chem. 1984, 88. 4451-4459. (4) WhnehOUSe. C. M.: Dreyer. R. N.: Yamashta. M.: Fenn, J. 8. Ansf. chem. 1985, 57, 675-679. (5) Wong, S. F.; Meng. C. K.; Fenn. J. 8. J . phys. chem. 1988, 92.
-F ~A E- ~ri"L.""" . ~
(61 Mann, M.: Meng. C. K.: Fen". J. B. Anal. chem. 1989. 6 1 , 1702-1708. (7) Meng, C. K.: Mann. M.: Fenn, J. 8. 2.mys. D : At., Mol.. C ~ t e m 1988. 10, 361-368. (8) Loo, J. A.; Wselh, H. R.: Smith, R. D. Ansl. Biochem. 1989. 179, 404-412. (9) Barinaga. C. J.: Edmonds, C. G.: Udseth. H. R.: Smnh, R. D. R a w C o n " . Mass Spectrom. 1989, 3, 160-164. (10) Smith, R. D.; Barinaga. C. J.: W&. H. R. J . phya. Chsm. 1989, 93, 5019-5022. (11) Loo, J. A.; Edmonds, C. 0.: Barinaga, C. J.: vdseth. H. R.: smith. R. 0..unpubliihed resuns. (12) Iribarne, J. V.: Thomson. B. A. J . Chem. Phys. 1978. 64. 2287-2294. (13) Thamson, B. A,: Iribarne, J. V. J . C b m . mys. 1979. 71. 4451-4463. (14) Lw. J. A.: vdseth. H. R.: Smith. R. D. Bkmsd. E n w . Mass Spectrom. 1988, 17, 411-414. (15) Covey, T. R.: Bonner. R. F.: Shushan. 8. I.:Henion. J. RBpidcCmmun. Mass Swim" 1988. 2. 249-256. (16) Loo. J. A,; Udwh. H. R.: Smith. R. D. RabcCmmUn. Mas Spec*om. 1988. 2, 207-210.
Anal. Chem. 1990, 62,698-703 Smith, R. D.; Barinaga, C. J.; Udseth, H. R. Anal. Chem. 1988, 60, 1948-1952. Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S.F.; Whitehouse, C. M. Mass Spectrom. Rev. lW0, 9 , 37-70. Edmonds, C. G.; Loo, J. A.; Fields, S. M.; Barinaga. C. J.; Udseth, H. R.; Smith, R. D. In froceedhgs of the Second Internatkmal Symposium on Mass Spectrometry in the Health and Lite Sciences (San Francisco, CA, August 1989): Burlingame, A. L., McCloskey, J. A., Eds.; Elsevier: Amsterdam, in press. Cleland, W. W. Biochemistry 1964,3 , 480-482. Johnson, R. S.;Martin, S.A.; Bemann. K. Int. J. Spectrom. Ion f r o cess 1988,86, 137-154. Bojesen, G. J. Am. Chem. Soc. 1987, 109, 5557-5558. Peters, T.. Jr. I n Advances in Protein Chemistry; Anfinsen, C. B., Edsall. J. T., Richards, F. M., Eds.; Academic Press: Orlando FL. 1985, VOi. 37, pp 161-245. Larsen, B. S.; Yergey, J. A.; Cotter, R. J. Biomed. Mass Spectrom. 1985, 10, 586-587. Buko, A. M.; Fraser, B. A. Biomed. Mass Spectrom. 1985, 12. 577-585. Morris,-H. R.; Pucci, P. Blochem. Bbphys. Res. Commun. 1985, 126, 1122-1 128. Sun, Y.; Smith, 0.L. Anal. Biochem. 1988. 172, 130-138. Chalt, B. T.; Field, F. H. Biochem. Biophys. Res. Commun. 1988, 134, 420-426. Roepstroff, P.; Nielsen, P. F.; Klarskov. K.;H0jrup, P. Biomed. fnviron. Mass Spectrom. 1988, 16, 9-18. Martin, S. A.; Biemann, K. Int. J. Mass Spectrom. Ion frocesses 1987, 78, 213-228.
(31) Lee, E. D.; Miick, W.; Henion, J. D.; Covey, T. R. J. Am. Chem. SOC. 1989, 7 1 7 , 4600-4604. (32) Bewley, T. A.; Li, C. H. Int. J. Protein Res. 1989, 7 , 117-124. (33) Fruchter, R. G.; Crestfield, A. M. J. Biol. Chem. 1987, 242, 5807-5812. (34) Castellino, F. J.; Hill, R. L. J. Bioi. Chem. 1970,245, 417-424. (35) Loo, J. A.; Jones, H. K.; Udseth, H. R.; Smith, R. D. J. Microco/umn Sep. 1988,5 , 223-229. (36) Loo, J. A.; Barinaga, C. J.; Edmonds, C. G.; Udseth, H. R.; Smith, R. D. Collected Abstracts; Proceedings of the 37th ASMS Conference on Mass Spectrometry and Allied Topics, May 21-26, 1989, Mbmi Beach, FL; American Society for Mass Spectrometry: East Lansing, MI, pp 586-587. (37) Smith, R. D.; Loo, J. A.; Barinaga, C. J.; Edmonds, C. G.: Udseth, H. R. J . Am. SOC. Mass Spectrom., in press. (38) Smith, R. D.; Barinaga, C. J. Rapid Commun. Mass Spectrum, in press. (39) Smith, R. D.; Loo, J. A.; Edmonds, C. G.; Barlnaga, C. J.; Udseth, H. R. Anal. Chem., in press.
RECEIVED for review July 18,1989. Accepted January 2,1990. We thank the National Science Foundation (DIR 8908096) and the U.S. Department of Energy, through PNL Internal Exploratory Research, for support of this research under Contract DE-AC06-76RLO 1830. Pacific Northwest Laboratory is operated by Battelle Memorial Institute.
Hadamard Transform Measurement of Tandem Fourier-Transform Mass Spectra Evan R. Williams, Stanton Y. Loh, and Fred W. McLafferty* Chemistry Department, Baker Laboratory, Cornell University, Ithaca, New York 14853-1301 Robert B. Cody Nicolet Analytical Instruments, 6416 Schroeder Road, Madison, Wisconsin 53711 The slmuna~~lus collection of mutlrple spectra uslng tandem (MSIMS) and " h a 1 (IIIIS/MS/MS) mass epectrometry from multiple precursors Is demonstrated to yield correspondlngly enhanced sensltlvtly. This approach utlllzes Hadamard transform deconvdutlon and takes advantage of the muitldrannel dlseoclatlon capablllty of Fowler-transform mass spectrometry. By application of this to an 11-component mlxture, the 11 spectra of the products of dlssoclatlng 11 different comblnetlons of six of the component molecular ions are measured; Hadamard traneformatlon ylekls indivldual spectra of the precursor ions exhlbltlng a signal-to-nolse lmprovement of 1.8X over spectra measured separately, as predicted by theory. Precursor ion selection with hlgh speclflcity and product formatlon wlth hlgh abundance reproduclMRLy are crltlcal; spurlous peaks resutling from Imperfect reproducibility can be mlnlmlred by using simultaneous equatlon coefflcients reflectlng the degree of precursor dlssoclation. Extendon of this technique to MS" spectra is demonstrated with simultaneous MS/MS/MS monitoring of three precursors and three daughters yielding nine spectra repre8enUng the nhe pmslbk cbsoclatlon pathways. For MS" spectra, codlng the product relathshlps for each additlonal step (e.g., precutsor daughter, daughter granddaughter) requlres elhrination of haU of the remaining ions. No Ions are lost for coding in an Improved Hadamard approach In which the combined daughter spectrum of the selected half of the precursors Is subtracted from that of the other half.
-
-
Identification of scores of compounds in complex mixtures and structural elucidation of large molecules presents a dif-
ficult challenge in chemical analysis. Tandem maas spectrometry (MS/MS, MS") (1, 2) greatly increases the information obtainable from these samples and is widely used for solving "needle-in-the-haystack"type problems of trace components in complex mixtures. However, applying MS/MS to many components is hampered by the great sensitivity loss encountered if the MS-I1 spectrum of each precursor must be measured individually, discarding the precursor ions of all other components. When analysis time or sample quantity is limited, e.g., on-line trace analysis for toxic substances in incinerator exhaust, peptide sequencing, etc., the use of this technique is limited to just a few precursors. MS" spectra, in which mass-selected products are dissociated to produce further spectra, present an even greater measurement problem, since the number of possible dissociation or reaction pathways increases exponentially with n. Recently two methods have been developed for simultaneous collection of MS/MS spectra in which only half, instead of all other, of the precursor ions must be discarded. Both methods utilize Fourier-transform (FT) ion cyclotron resonance (ICR) mass spectrometry (3-15). One uses a second FT step (14, 15) analogous to 2D-NMR; after initial radio frequency (rf) excitation of the precursor ions into cyclotron orbits, a deexcitation rf pulse after a specific delay time tl changes the effective abundance of specific precursor ions as a function of the phase difference, which is dependent on the precursor m / z value. The secondary mass spectra are then recorded as a function of tl; the Fourier transform of this function for a specific mass peak then shows the abundance originating from each precursor. In our Hadamard transform (HT) method (3),half of the p precursor ions are selected and dissociated simultaneously,and the spectrum of the combined
0003-2700/90/0362-0698$02.50/00 1990 American Chemical Society
