Artifacts in Four-sector Tandem Mass Spectrometry A. M. Falick,* Katalin F. Medzihradszky and F. C . Walls Mass spectrometry Facility, Department of Pharmaceutical Chemistry, University of California, San Francisco, CA 94143-0446, USA
Several types of artifacts were shown to be present in 4-sector tandem collision-induced dissociation (CID) mass spectra. In CID spectra of protonated peptides produced by liquid secondary-ion mass spectrometry (LSIMS), peaks corresponding to successive losses of matrix molecules from the precursor ion were observed. In addition, peaks corresponding to MH' ions of smaller peptides that were also present in the sample/matrix mixture in greater abundance than the selected precursor ion were observed. Both of these types of artifact peaks were shown to originate from the 'peak-at-every-mass' chemical noise at the same nominal mass as that selected by the first 2 sectors (MSl). These noise ions are transmitted through to the collision cell and produce fragments that are analysed and detected in the next 2 sectors (MS2). A second, unrelated, kind of artifact was found to be due to decompositions in the second field-free region of MS2 in an EBEB geometry machine. These artifacts, which are detectable over only a very limited mass range when using a conventional single-point detector, can be present over a much greater mass range when an array detector is used and when the collision cell is floated above ground potential. A clear understanding of the origins of all peaks in a CID spectrum is important in order to have a firm foundation for interpretation, manual or computer-aided, of the spectra of unknown compounds.
Four-sector tandem liquid secondary-ion mass spectrometry (LSIMS) of peptides and other biopolymers has proved to be extremely valuable in the elucidation of various structural features of these molecules, despite the relatively short time that this technique has been available.'.2 Virtually all of the ions observed in the CID fragment ion spectrum are formed in the collision cell located between the two mass analysers (MS1 and MS2) by unimolecular or collision-induced decomposition (CID) of the selected precursor ion. Partly Because of this fact, the CID fragment-ion spectra of known compounds are usually readily interpretable and all of the significant peaks in the spectrum can normally be assigned to known or chemically reasonable fragmentations. It is important to have a clear understanding of the origins of all of the peaks in the spectra of known compounds so as to have a firm foundation for of the interpretation, manual or c~mputer-aided,~-' spectra of unknown compounds. With this goal in mind, we have examined the CID spectra of a number of peptides, and have found certain artifact peaks that could cause difficulty in spectral interpretation if their existence was not previously known or suspected. These artifact peaks arise either from CID of another precursor ion of the same mass as the one selected in MS1, or from ions that generate peaks at an apparent mass different from that expected on the basis of their m / z value. EXPERIMENTAL Peptides. The peptide Glu-Thr-Tyr-Ser-Lys (ETYSK) was purchased from Sigma (St Louis, MO, USA) and derivatized with hexanol according to a published procedure.(' The carboxymethylated glycopeptide Cys-Glu-Thr(Man)-Val-Ala-Ala-Ala-Arg (CET(Man)VAAAR) was obtained in a Collected highperformance liquid chromatography (HPLC) fraction Author to whom correspondence should be addressed.
from a tryptic digest of recombinant human platelet-derived growth factor B chain (rhPDGF-B) expressed in yeast .' The aminoethylated peptide Gly -Thr -Phe -Ala -1le -Leu-Ser-Glu-Leu-His-Cys-AspLys (GTFAILSELHCDK) was obtained from a tryptic digest of a hemoglobin variant under study in this Iaboratory.* Sequences of all of the peptides used in this work were verified by tandem mass spectrometry. Muss spectrometry. Tandem mass spectrometry (MS/MS) experiments were performed on a Concept IIHH four-sector EBEB tandem mass spectrometer' (Kratos Analytical, Manchester, UK) fitted with an electro-optical array detector,'" a cesium LSIMS ion sourcell and coolable probe.'* Precursor ions were generated with an 18 keV cesium ion primary beam. The matrix used was either glycerol + thioglycerol (1:l) containing 1% trifluoroacetic acid or 3-nitrobenzyl alcohol (NBA). The instrument accelerating voltage was 8 kV and the collision energy for CID was 4 keV. The collision gas helium was used at a pressure sufficient to suppress the precursor ion beam to about 30% of its initial value. The instrument was controlled and data were acquired with a DS-90 data system. Calibration and data display were carried out with the aid of a Mach 3 data processing system. RESULTS AND DISCUSSION Peaks resulting from matrix background ions isobaric with the selected precursor ion Figure l(a) shows the CID fragment-ion spectrum of a glycopeptide isolated as a collected HPLC fraction. The precursor MH' ion of mass 1040.5 was produced in the ion source of MS1, using a glycerol + thioglycerol matrix. The MH' ion was only moderately abundant, A at giving an ion current of approximately 1x the detector of MS1. The peaks that are consistent with the assigned sequence are indicated with the conventional notation. l 3 . I 4 However, there are several strong peaks in the spectrum that do not correspond to any of
0951-4198/90/0318-0322$05.00 318 RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 4, NO. 9. 1990
0
John Wiley & Sons Limited, 1990
ARTIFACTS IN FOUR-SECTOR TANDEM MASS SPECTROMETRY
7 90
miz (b)
Figure 1. (a), Tandem CID spectrum of an HPLC fraction containing the protonated carboxymethylated glycopeptide CET(Man)VAAAR(MH+ = m i r 1040.5) in a glycerol thioglycerol matrix. Expected fragment ions from this peptide are indicated with l4peaks corresponding to glycerol the conventional n~menclature,'~ losses from the precursor ion are marked '-ngly', and peaks related to a second peptide present in the mixture are marked M,H+ and [M,+Na]+. (b), Tandem CID spectrum of the same HPLC fraction in NBA matrix. The peaks marked '-nNBA' correspond to losses of NBA from the selected precursor ion.
+
the predicted fragment ions from this peptide. Most of these peaks (marked '-ngly') were found to correspond to losses of n molecules of glycerol from the MH' ion. There are also two peaks at mlz598.3 and 620.3 that cannot be explained as either normal fragment-ion or matrix-loss peaks. In order to confirm the suspected origin of the glycerol-loss peaks, the CID spectrum of the MH' ion of rnlz 1040.5 was obtained using NBA as the matrix. Figure l(b) shows the result. In this spectrum, the glycerol-loss ions are replaced by a series of peaks corresponding to successive losses of NBA, whereas the peaks at mlz 598.3 and 620.3 were unchanged. However, there was a second peptide (Pro-Val-GlnVal-Arg) present in the same HPLC fraction, which gave an abundant MH' peak (about a factor of 100 more abundant than the mlz 1040.5 peak) and a somewhat less abundant [M+Na]+ peak in the normal LSIMS spectrum of this HPLC fraction. As the inset in Fig. 2 shows, there is also a peak at m/z 599.3 in the CID spectrum that corresponds to the first 13Cisotope peak for this smaller peptide. We attribute the appearance of these artifact peaks to ions of nominal mass 1040 that are a part of the ..MH+.'
_ _ 120 598 4 M,H+
948.4
-2NEA
10
0
mlz
Figure 2. Tandem CID spectrum of background ions of mass 948.4k 0.5. The inset is an expansion of the region around mass 598.
0
John Wiley & Sons Limited, 1990
mir
Figure 3. Portion of CID spectrum of a protonated hemoglobin tryptic peptide, showing artifact peaks at m l z 1071.5 and 1072.5, due to the presence of another peptide in the sample mixture.
'peak-at-every-mass' background or chemical noise that is characteristic of LSIMS spectra. In a tandem instrument, most of this chemical noise is eliminated by MS1. However, as we have previously noted: the portion of the chemical noise that is isobaric with the precursor ion selected in MS1 is also transmitted to the collision cell, where fragment ions can be generated from these noise ions. We infer from our results that the background ions are clusters that may contain several intact molecules of the matrix compound andlor other materials present in the matrix. This hypothesis suggests that there is nothing unique about the particular combination of masses in this sample. If the chemical noise carries components of the major peaks in the spectrum, one should detect the same or related artifact peaks in a CID spectrum of a 'pure' background peak. Figure 2 shows such a spectrum obtained using the same sample and matrix as in Fig. l(b), but selecting the precursor in MS1 as rnlz 948.4. The resolution of MS1 was set so as to pass ions of rnlz 948.4f0.5. The precursor 'ion' is probably a mixture of various ions within this mass range. In the normal LSIMS spectrum of this sample, there was no discernible peak above background at this mass. The CID spectrum of this ion does in fact contain strong fragment ion peaks at masses corresponding to successive losses of NBA from the selcted precursor ion, as well as a strong peak at rnlz 598.3. Other experiments using background precursor ions of other masses gave similar results in NBA. In a glycerol + thioglycerol matrix, CID of background ions gave analogous results with the NBA-loss ions being replaced by the corresponding glycerol-loss ions. Another example of a matrix-related artifact peak can be seen in Fig. 3. This is a portion of the CID fragment ion spectrum of the protonated peptide GTFAILSELHC(Ae)DK. The peaks at rnlz 1071.5 and 1072.5 do not correspond to any expected CID fragment ions from this peptide. On the other hand, a peptide of mass 1071.5 (hemoglobin aT5, Met-Phe-Leu-Ser-Phe-Pro-Thr-Thr-Lys) was present in the HPLC fraction from which this sample came. In the normal LSIMS mass spectrum of this HPLC fraction, the rnlz 1071.5 peak was roughly 8 times as abundant as the peak at m/z 1476.7. Therefore, an ion of mass
RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 4. NO. 9. 1990 319
ARTIFACTS IN FOUR-SECTOR TANDEM MASS SPECTROMETRY
1476.7f0.4 (the resolution of MS1 determines the range of possible masses) consisting of the protonated aT5 peptide and a moiety of nominal mass 405 (1071 + 405 = 1476), must be generated from the analyte+matrix mixture and form a part of the chemical noise background at this mass. The peak at m l z 1072.5, corresponding to the first 13C isotope peak of the protonated aT5 peptide, cannot be related to the first 13C isotope peak (at mlz1477.7) of the peptide GTFAILSELHC(Ae)DK because the resolution of MS1 was sufficiently high to preclude these ions from reaching the collision cell. Instead, this peak must result from a precursor ion containing the protonated aT5 peptide and a background mass of 405, with one 13Catom. We have observed similar artifact peaks in a number of samples containing mixtures of peptides. In addition to the matrix-loss ions and the ions from other compounds present in the mixture, we have sometimes also found peaks corresponding to glycerol cluster ions, e.g., peaks at masses 93, 185, etc., in the CID spectrum. Artifact peaks of the types described above would be expected to be noticeable when analysing a molecular species that gives only a weak MH+ ion, or if the precursor ion mass happens to coincide with a normal matrix cluster ion. However, we have found that the artifact peaks can sometimes be surprisingly large even when the precursor ion is relatively abundant and not obviously related to the matrix cluster ions, as is observed in Fig. 1. This is presumably due to the fact that the clusters are not covalently bound, but are much more loosely associated. In addition, the relative sizes of both the matrix-related and non-matrix-related artifact peaks are strongly correlated with their abundances in the normal LSIMS spectrum. The results presented above were obtained in a 4sector tandem machine using high-energy collisions. However, such peaks should appear equally in any LSIMS experiment that uses high-energy collisional activation. In view of the fact that the proposed cluster ions that give rise to the phenomena described here are presumably loosely bound, it appears likely that these types of artifact ions will also be found in similar experiments employing low-energy collisional activation. Furthermore, our experiments were carried out with the resolution of MS1 set to pass a mass window of k 0.5 mass units. If the resolution of MS1 were set so as to transmit a range of masses into the collision cell, as is sometimes done in triple quadrupole experiments, a correspondingly larger number of noise ions should be transmitted and the artifact ions should be more abundant. On the other hand, operating MS1 at higher resolution could reduce the number of noise ions transmitted, but probably at an unacceptable cost in decreased sensitivity in most cases. The background-related peaks reported here can be as large as or larger than those from the important sequence ions, even when the precursor ion is sufficiently abundant to give a strong CID spectrum. When studying a less abundant component of a mixture of analytes, the more abundant species may contribute significant peaks to the CID spectrum. Knowledge of and accurate characterization of such ions is therefore very important in interpreting CID spectra of unknowns. When working with less abundant precursor
ions, the ultimate sensitivity of LSIMS followed by CID as an analytical technique may be limited by the presence of these background ions.' Peaks formed as a result of metastable decomposition in the field-free region between the electric sector of MS2 (E2) and the magnet of MS2 (B2) We also report here the observation of a second, quite different kind of artifact peak that can appear in 4sector tandem CID spectra, and is particularly important when an array detector is used. It is well known that metastable decompositions in the second field-free region (between the electric sector and the magnet) of a conventional (EB geometry) double-focusing mass spectrometer result in relatively broad peaks that appear at an apparent mass rn* = rn2/M,where rn is the fragment-ion mass and M is the precursor-ion mass.l5 Under certain circumstances, decompositions occurring in the equivalent field-free region of MS2 in a 4-sector tandem machine can give rise to similar peaks appearing in the CID fragment-ion spectrum.16 Figure 4 shows part of the 4-sector CID spectrum of the hexylated peptide E(hexyl)TYSK(hexyl)(MH+ 795.4). The spectrum contains an unusually broad peak whose centroid is at mass 759.8. The normal fragmention peaks present in the spectrum correspond to the expected loss of various amino acid side chains. In normal 2-sector mass Spectra of peptides, a metastable ion peak is often observed that corresponds to loss of water from MH'. The broad peak in Fig. 4 has the same characteristics as these 2-sector peaks; in addition, its measured centroid agrees closely with the calculated apparent mass for such a peak (759.87). Similar peaks have been observed in high-energy CID spectra of a number of peptides acquired with an array detector in this laboratory.
M H + - 18
M H + - 45
MH+
I
- 15
II . I /
I
m* 759.75
1 ~ " ' l " " 1 ' ~ " I ' ' " I ' " ' I ' ' ' ' I " ~ ' I
750
755
760
765
770
775
780
785
m/z
Figure4. Part of the CID spectrum of the protonated peptide E(hexyl)TYSK(hexyl), MHf m / z 795.4. The measured centroid of the broad peak is 759.75. The higher background level above mass 760 may be due to the 795-777 transition occurring as the precursor ions traverse the last part of the electric sector or the magnet of MS2.I6
320 RAPID COMMUNICA-TIONS IN MASS SPECTROMETRY, VOL. 4, NO. 9, 1990
0 John Wiley & Sons Limited, 1990
ARTlFACTS IN FOUR-SECTOR TANDEM MASS SPECTROMETRY
In a 4-sector tandem instrument. MS2 is not scanned in the same way as a normal 2-sector machine. The fragment ions that are analysed in MS2 have a range of kinetic energies given by
E = (V, - V , ) ( m / M )+ V ,
Center mass
m*
(1)
(;j2=E(
-
EIE,- V J V , 1-v,1v0
j
where B, and E, are the values of the magnetic and electric sector fields required in MS2 to allow any selected precursor ion from MS1 to pass through to the detector of MS2. (This scan law reduces to the more familiar linked-scan law B / E = constant when V, = 0.) A consequence of this scan law is that when a particular mass is focused at the collector slit of MS2, only some of the ions leaving the collision cell can pass through the electric sector of MS2 (E2) and thereby reach the second field-free region (second FFR) of MS2, namely, ions whose kinetic energies lie within the pass band of E2.16 In an instrument using a conventional slit preceding the detector of MS2, the energy bandpass of E2, A E 2 , is typically _+ 1% or so. If the collision cell is at ground potential, the group of transmitted ions in this case consists of those whose masses are within +_ 1 % of the mass for which MS2 is set at that moment. If an ion whose mass lies within this range decomposes in the second FFR of MS2, the decomposition can be detected only if the fragment-ion apparent mass ( m * ) is less than 1% different from the ion mass for which MS2 is tuned. If the collision cell is held at some potential other than ground potential, the range of ion masses AmE that can pass through E 2 is given by A m E = ( V o / ( V n - V c ) ) A E 2For . example, if A E 2 = + _ l % and the collision cell is floated at 50% of the normal accelerating voltage, then the value of A m , is f 2 Y 0 . l ~ If MS2 is set to detect fragment ions at mass 1000, only decompositions of ions in the mass range 1020-1001 that result in a metastable ion peak with an apparent mass of 1000 will be observable. (Ions with masses 5 1000 cannot produce an m* peak at mass 1000.) The largest loss observable will be [(lo20 X 1000) - 1000]”2= 10 mass units. Therefore, only a very few chemically reasonable decompositions can give rise to an m* peak in the normal CID fragment ion spectrum under these conditions. In an instrument with an array detector on MS2, however, the situation is somewhat different. The energy bandpass of E2 must be at least wide enough to pass all of the fragment ion masses that can be detected simultaneously on the array. In our instrument, for example, (Kratos Concept IIHH) a range covering approximately +_ 2% of the mass focused on the center of the array can be detected simultaneously. The electric sector bandpass is also about k2%. The collision cell is typically floated at 50% of the accelerating voltage, allowing ions whose masses lie within 3z 4% of the center mass to pass through the electric sector. As is
@
John Wiley & Sons Limited, 1990
I
ih
mass
!2R
where E is the fragment-ion kinetic energy (in eV). V , is the normal accelerating potential of MSl and MS2, and V , is the potential at which the collision cell is floated. In order to detect these ions, both B and E in MS2 must be scanned in accordance with the linked scan law:”
E
m
Array mass range f 4%
-+
-
Figure 5. Diagram showing the maximum possible mass difference between precursor ion, normal fragment ion, and metastable ion peaks. It has been assumed that the mass range of the array is 2 2”/0 about the center mass, the energy bandpass of E2 is +2%, and the collision cell is floated at 50% of the accelerating voltage. In order for a metastable ion peak to be observed, the precursor ion mass, M , must be transmitted by E2, and at the same time, the apparent fragment ion mass, m*,must lie within the array mass range.
illustrated in Fig. 5 , this means that under these conditions, if an ion decomposes in the second FFR of MS2 the apparent mass, m * , of its fragment ion can differ from it by as much as 6% (neutral fragment mass
Several types of artifacts were shown to be present in 4-sector tandem collision-induced dissociation (CID) mass spectra. In CID spectra of protonated peptides produced by liquid secondary-ion mass spectrometry (LSIMS), peaks corresponding to successive losses of matrix molecules from the precursor ion were observed. In addition, peaks corresponding to MH' ions of smaller peptides that were also present in the sample/matrix mixture in greater abundance than the selected precursor ion were observed. Both of these types of artifact peaks were shown to originate from the 'peak-at-every-mass' chemical noise at the same nominal mass as that selected by the first 2 sectors (MSl). These noise ions are transmitted through to the collision cell and produce fragments that are analysed and detected in the next 2 sectors (MS2). A second, unrelated, kind of artifact was found to be due to decompositions in the second field-free region of MS2 in an EBEB geometry machine. These artifacts, which are detectable over only a very limited mass range when using a conventional single-point detector, can be present over a much greater mass range when an array detector is used and when the collision cell is floated above ground potential. A clear understanding of the origins of all peaks in a CID spectrum is important in order to have a firm foundation for interpretation, manual or computer-aided, of the spectra of unknown compounds.
Four-sector tandem liquid secondary-ion mass spectrometry (LSIMS) of peptides and other biopolymers has proved to be extremely valuable in the elucidation of various structural features of these molecules, despite the relatively short time that this technique has been available.'.2 Virtually all of the ions observed in the CID fragment ion spectrum are formed in the collision cell located between the two mass analysers (MS1 and MS2) by unimolecular or collision-induced decomposition (CID) of the selected precursor ion. Partly Because of this fact, the CID fragment-ion spectra of known compounds are usually readily interpretable and all of the significant peaks in the spectrum can normally be assigned to known or chemically reasonable fragmentations. It is important to have a clear understanding of the origins of all of the peaks in the spectra of known compounds so as to have a firm foundation for of the interpretation, manual or c~mputer-aided,~-' spectra of unknown compounds. With this goal in mind, we have examined the CID spectra of a number of peptides, and have found certain artifact peaks that could cause difficulty in spectral interpretation if their existence was not previously known or suspected. These artifact peaks arise either from CID of another precursor ion of the same mass as the one selected in MS1, or from ions that generate peaks at an apparent mass different from that expected on the basis of their m / z value. EXPERIMENTAL Peptides. The peptide Glu-Thr-Tyr-Ser-Lys (ETYSK) was purchased from Sigma (St Louis, MO, USA) and derivatized with hexanol according to a published procedure.(' The carboxymethylated glycopeptide Cys-Glu-Thr(Man)-Val-Ala-Ala-Ala-Arg (CET(Man)VAAAR) was obtained in a Collected highperformance liquid chromatography (HPLC) fraction Author to whom correspondence should be addressed.
from a tryptic digest of recombinant human platelet-derived growth factor B chain (rhPDGF-B) expressed in yeast .' The aminoethylated peptide Gly -Thr -Phe -Ala -1le -Leu-Ser-Glu-Leu-His-Cys-AspLys (GTFAILSELHCDK) was obtained from a tryptic digest of a hemoglobin variant under study in this Iaboratory.* Sequences of all of the peptides used in this work were verified by tandem mass spectrometry. Muss spectrometry. Tandem mass spectrometry (MS/MS) experiments were performed on a Concept IIHH four-sector EBEB tandem mass spectrometer' (Kratos Analytical, Manchester, UK) fitted with an electro-optical array detector,'" a cesium LSIMS ion sourcell and coolable probe.'* Precursor ions were generated with an 18 keV cesium ion primary beam. The matrix used was either glycerol + thioglycerol (1:l) containing 1% trifluoroacetic acid or 3-nitrobenzyl alcohol (NBA). The instrument accelerating voltage was 8 kV and the collision energy for CID was 4 keV. The collision gas helium was used at a pressure sufficient to suppress the precursor ion beam to about 30% of its initial value. The instrument was controlled and data were acquired with a DS-90 data system. Calibration and data display were carried out with the aid of a Mach 3 data processing system. RESULTS AND DISCUSSION Peaks resulting from matrix background ions isobaric with the selected precursor ion Figure l(a) shows the CID fragment-ion spectrum of a glycopeptide isolated as a collected HPLC fraction. The precursor MH' ion of mass 1040.5 was produced in the ion source of MS1, using a glycerol + thioglycerol matrix. The MH' ion was only moderately abundant, A at giving an ion current of approximately 1x the detector of MS1. The peaks that are consistent with the assigned sequence are indicated with the conventional notation. l 3 . I 4 However, there are several strong peaks in the spectrum that do not correspond to any of
0951-4198/90/0318-0322$05.00 318 RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 4, NO. 9. 1990
0
John Wiley & Sons Limited, 1990
ARTIFACTS IN FOUR-SECTOR TANDEM MASS SPECTROMETRY
7 90
miz (b)
Figure 1. (a), Tandem CID spectrum of an HPLC fraction containing the protonated carboxymethylated glycopeptide CET(Man)VAAAR(MH+ = m i r 1040.5) in a glycerol thioglycerol matrix. Expected fragment ions from this peptide are indicated with l4peaks corresponding to glycerol the conventional n~menclature,'~ losses from the precursor ion are marked '-ngly', and peaks related to a second peptide present in the mixture are marked M,H+ and [M,+Na]+. (b), Tandem CID spectrum of the same HPLC fraction in NBA matrix. The peaks marked '-nNBA' correspond to losses of NBA from the selected precursor ion.
+
the predicted fragment ions from this peptide. Most of these peaks (marked '-ngly') were found to correspond to losses of n molecules of glycerol from the MH' ion. There are also two peaks at mlz598.3 and 620.3 that cannot be explained as either normal fragment-ion or matrix-loss peaks. In order to confirm the suspected origin of the glycerol-loss peaks, the CID spectrum of the MH' ion of rnlz 1040.5 was obtained using NBA as the matrix. Figure l(b) shows the result. In this spectrum, the glycerol-loss ions are replaced by a series of peaks corresponding to successive losses of NBA, whereas the peaks at mlz 598.3 and 620.3 were unchanged. However, there was a second peptide (Pro-Val-GlnVal-Arg) present in the same HPLC fraction, which gave an abundant MH' peak (about a factor of 100 more abundant than the mlz 1040.5 peak) and a somewhat less abundant [M+Na]+ peak in the normal LSIMS spectrum of this HPLC fraction. As the inset in Fig. 2 shows, there is also a peak at m/z 599.3 in the CID spectrum that corresponds to the first 13Cisotope peak for this smaller peptide. We attribute the appearance of these artifact peaks to ions of nominal mass 1040 that are a part of the ..MH+.'
_ _ 120 598 4 M,H+
948.4
-2NEA
10
0
mlz
Figure 2. Tandem CID spectrum of background ions of mass 948.4k 0.5. The inset is an expansion of the region around mass 598.
0
John Wiley & Sons Limited, 1990
mir
Figure 3. Portion of CID spectrum of a protonated hemoglobin tryptic peptide, showing artifact peaks at m l z 1071.5 and 1072.5, due to the presence of another peptide in the sample mixture.
'peak-at-every-mass' background or chemical noise that is characteristic of LSIMS spectra. In a tandem instrument, most of this chemical noise is eliminated by MS1. However, as we have previously noted: the portion of the chemical noise that is isobaric with the precursor ion selected in MS1 is also transmitted to the collision cell, where fragment ions can be generated from these noise ions. We infer from our results that the background ions are clusters that may contain several intact molecules of the matrix compound andlor other materials present in the matrix. This hypothesis suggests that there is nothing unique about the particular combination of masses in this sample. If the chemical noise carries components of the major peaks in the spectrum, one should detect the same or related artifact peaks in a CID spectrum of a 'pure' background peak. Figure 2 shows such a spectrum obtained using the same sample and matrix as in Fig. l(b), but selecting the precursor in MS1 as rnlz 948.4. The resolution of MS1 was set so as to pass ions of rnlz 948.4f0.5. The precursor 'ion' is probably a mixture of various ions within this mass range. In the normal LSIMS spectrum of this sample, there was no discernible peak above background at this mass. The CID spectrum of this ion does in fact contain strong fragment ion peaks at masses corresponding to successive losses of NBA from the selcted precursor ion, as well as a strong peak at rnlz 598.3. Other experiments using background precursor ions of other masses gave similar results in NBA. In a glycerol + thioglycerol matrix, CID of background ions gave analogous results with the NBA-loss ions being replaced by the corresponding glycerol-loss ions. Another example of a matrix-related artifact peak can be seen in Fig. 3. This is a portion of the CID fragment ion spectrum of the protonated peptide GTFAILSELHC(Ae)DK. The peaks at rnlz 1071.5 and 1072.5 do not correspond to any expected CID fragment ions from this peptide. On the other hand, a peptide of mass 1071.5 (hemoglobin aT5, Met-Phe-Leu-Ser-Phe-Pro-Thr-Thr-Lys) was present in the HPLC fraction from which this sample came. In the normal LSIMS mass spectrum of this HPLC fraction, the rnlz 1071.5 peak was roughly 8 times as abundant as the peak at m/z 1476.7. Therefore, an ion of mass
RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 4. NO. 9. 1990 319
ARTIFACTS IN FOUR-SECTOR TANDEM MASS SPECTROMETRY
1476.7f0.4 (the resolution of MS1 determines the range of possible masses) consisting of the protonated aT5 peptide and a moiety of nominal mass 405 (1071 + 405 = 1476), must be generated from the analyte+matrix mixture and form a part of the chemical noise background at this mass. The peak at m l z 1072.5, corresponding to the first 13C isotope peak of the protonated aT5 peptide, cannot be related to the first 13C isotope peak (at mlz1477.7) of the peptide GTFAILSELHC(Ae)DK because the resolution of MS1 was sufficiently high to preclude these ions from reaching the collision cell. Instead, this peak must result from a precursor ion containing the protonated aT5 peptide and a background mass of 405, with one 13Catom. We have observed similar artifact peaks in a number of samples containing mixtures of peptides. In addition to the matrix-loss ions and the ions from other compounds present in the mixture, we have sometimes also found peaks corresponding to glycerol cluster ions, e.g., peaks at masses 93, 185, etc., in the CID spectrum. Artifact peaks of the types described above would be expected to be noticeable when analysing a molecular species that gives only a weak MH+ ion, or if the precursor ion mass happens to coincide with a normal matrix cluster ion. However, we have found that the artifact peaks can sometimes be surprisingly large even when the precursor ion is relatively abundant and not obviously related to the matrix cluster ions, as is observed in Fig. 1. This is presumably due to the fact that the clusters are not covalently bound, but are much more loosely associated. In addition, the relative sizes of both the matrix-related and non-matrix-related artifact peaks are strongly correlated with their abundances in the normal LSIMS spectrum. The results presented above were obtained in a 4sector tandem machine using high-energy collisions. However, such peaks should appear equally in any LSIMS experiment that uses high-energy collisional activation. In view of the fact that the proposed cluster ions that give rise to the phenomena described here are presumably loosely bound, it appears likely that these types of artifact ions will also be found in similar experiments employing low-energy collisional activation. Furthermore, our experiments were carried out with the resolution of MS1 set to pass a mass window of k 0.5 mass units. If the resolution of MS1 were set so as to transmit a range of masses into the collision cell, as is sometimes done in triple quadrupole experiments, a correspondingly larger number of noise ions should be transmitted and the artifact ions should be more abundant. On the other hand, operating MS1 at higher resolution could reduce the number of noise ions transmitted, but probably at an unacceptable cost in decreased sensitivity in most cases. The background-related peaks reported here can be as large as or larger than those from the important sequence ions, even when the precursor ion is sufficiently abundant to give a strong CID spectrum. When studying a less abundant component of a mixture of analytes, the more abundant species may contribute significant peaks to the CID spectrum. Knowledge of and accurate characterization of such ions is therefore very important in interpreting CID spectra of unknowns. When working with less abundant precursor
ions, the ultimate sensitivity of LSIMS followed by CID as an analytical technique may be limited by the presence of these background ions.' Peaks formed as a result of metastable decomposition in the field-free region between the electric sector of MS2 (E2) and the magnet of MS2 (B2) We also report here the observation of a second, quite different kind of artifact peak that can appear in 4sector tandem CID spectra, and is particularly important when an array detector is used. It is well known that metastable decompositions in the second field-free region (between the electric sector and the magnet) of a conventional (EB geometry) double-focusing mass spectrometer result in relatively broad peaks that appear at an apparent mass rn* = rn2/M,where rn is the fragment-ion mass and M is the precursor-ion mass.l5 Under certain circumstances, decompositions occurring in the equivalent field-free region of MS2 in a 4-sector tandem machine can give rise to similar peaks appearing in the CID fragment-ion spectrum.16 Figure 4 shows part of the 4-sector CID spectrum of the hexylated peptide E(hexyl)TYSK(hexyl)(MH+ 795.4). The spectrum contains an unusually broad peak whose centroid is at mass 759.8. The normal fragmention peaks present in the spectrum correspond to the expected loss of various amino acid side chains. In normal 2-sector mass Spectra of peptides, a metastable ion peak is often observed that corresponds to loss of water from MH'. The broad peak in Fig. 4 has the same characteristics as these 2-sector peaks; in addition, its measured centroid agrees closely with the calculated apparent mass for such a peak (759.87). Similar peaks have been observed in high-energy CID spectra of a number of peptides acquired with an array detector in this laboratory.
M H + - 18
M H + - 45
MH+
I
- 15
II . I /
I
m* 759.75
1 ~ " ' l " " 1 ' ~ " I ' ' " I ' " ' I ' ' ' ' I " ~ ' I
750
755
760
765
770
775
780
785
m/z
Figure4. Part of the CID spectrum of the protonated peptide E(hexyl)TYSK(hexyl), MHf m / z 795.4. The measured centroid of the broad peak is 759.75. The higher background level above mass 760 may be due to the 795-777 transition occurring as the precursor ions traverse the last part of the electric sector or the magnet of MS2.I6
320 RAPID COMMUNICA-TIONS IN MASS SPECTROMETRY, VOL. 4, NO. 9, 1990
0 John Wiley & Sons Limited, 1990
ARTlFACTS IN FOUR-SECTOR TANDEM MASS SPECTROMETRY
In a 4-sector tandem instrument. MS2 is not scanned in the same way as a normal 2-sector machine. The fragment ions that are analysed in MS2 have a range of kinetic energies given by
E = (V, - V , ) ( m / M )+ V ,
Center mass
m*
(1)
(;j2=E(
-
EIE,- V J V , 1-v,1v0
j
where B, and E, are the values of the magnetic and electric sector fields required in MS2 to allow any selected precursor ion from MS1 to pass through to the detector of MS2. (This scan law reduces to the more familiar linked-scan law B / E = constant when V, = 0.) A consequence of this scan law is that when a particular mass is focused at the collector slit of MS2, only some of the ions leaving the collision cell can pass through the electric sector of MS2 (E2) and thereby reach the second field-free region (second FFR) of MS2, namely, ions whose kinetic energies lie within the pass band of E2.16 In an instrument using a conventional slit preceding the detector of MS2, the energy bandpass of E2, A E 2 , is typically _+ 1% or so. If the collision cell is at ground potential, the group of transmitted ions in this case consists of those whose masses are within +_ 1 % of the mass for which MS2 is set at that moment. If an ion whose mass lies within this range decomposes in the second FFR of MS2, the decomposition can be detected only if the fragment-ion apparent mass ( m * ) is less than 1% different from the ion mass for which MS2 is tuned. If the collision cell is held at some potential other than ground potential, the range of ion masses AmE that can pass through E 2 is given by A m E = ( V o / ( V n - V c ) ) A E 2For . example, if A E 2 = + _ l % and the collision cell is floated at 50% of the normal accelerating voltage, then the value of A m , is f 2 Y 0 . l ~ If MS2 is set to detect fragment ions at mass 1000, only decompositions of ions in the mass range 1020-1001 that result in a metastable ion peak with an apparent mass of 1000 will be observable. (Ions with masses 5 1000 cannot produce an m* peak at mass 1000.) The largest loss observable will be [(lo20 X 1000) - 1000]”2= 10 mass units. Therefore, only a very few chemically reasonable decompositions can give rise to an m* peak in the normal CID fragment ion spectrum under these conditions. In an instrument with an array detector on MS2, however, the situation is somewhat different. The energy bandpass of E2 must be at least wide enough to pass all of the fragment ion masses that can be detected simultaneously on the array. In our instrument, for example, (Kratos Concept IIHH) a range covering approximately +_ 2% of the mass focused on the center of the array can be detected simultaneously. The electric sector bandpass is also about k2%. The collision cell is typically floated at 50% of the accelerating voltage, allowing ions whose masses lie within 3z 4% of the center mass to pass through the electric sector. As is
@
John Wiley & Sons Limited, 1990
I
ih
mass
!2R
where E is the fragment-ion kinetic energy (in eV). V , is the normal accelerating potential of MSl and MS2, and V , is the potential at which the collision cell is floated. In order to detect these ions, both B and E in MS2 must be scanned in accordance with the linked scan law:”
E
m
Array mass range f 4%
-+
-
Figure 5. Diagram showing the maximum possible mass difference between precursor ion, normal fragment ion, and metastable ion peaks. It has been assumed that the mass range of the array is 2 2”/0 about the center mass, the energy bandpass of E2 is +2%, and the collision cell is floated at 50% of the accelerating voltage. In order for a metastable ion peak to be observed, the precursor ion mass, M , must be transmitted by E2, and at the same time, the apparent fragment ion mass, m*,must lie within the array mass range.
illustrated in Fig. 5 , this means that under these conditions, if an ion decomposes in the second FFR of MS2 the apparent mass, m * , of its fragment ion can differ from it by as much as 6% (neutral fragment mass
