Affecting Proton Mobility in Activated Peptide and Whole Protein Ions via Lysine Guanidination Sharon J. Pitteri,† Gavin E. Reid,‡ and Scott A. McLuckey*,† Department of Chemistry, Purdue University, West Lafayette, Indiana 47907-2084, and Joint Proteomics Laboratory, The Ludwig Institute for Cancer Research, Royal Melbourne Hospital, Parkville, Victoria, Australia 3050 Received July 10, 2003
We have evaluated the effect of lysine guanidination in peptides and proteins on the dissociation of protonated ions in the gas phase. The dissociation of guanidinated model peptide ions compared to their unmodified forms showed behavior consistent with concepts of proton mobility as a major factor in determining favored fragmentation channels. Reduction of proton mobility associated with lysine guanidination was reflected by a relative increase in cleavages occurring C-terminal to aspartic acid residues as well as increases in small molecule losses. To evaluate the effect of guanidination on the dissociation behavior of whole protein ions, bovine ubiquitin was selected as a model. Essentially, all of the amide bond cleavages associated with the +10 charge state of fully guanidinated ubiquitin were observed to occur C-terminal to aspartic acid residues, unlike the dissociation behavior of the +10 ion of the unmodified protein, where competing cleavage N-terminal to proline and nonspecific amide bond cleavages were also observed. The +8 and lower charge states of the guanidinated protein showed prominent losses of small neutral molecules. This overall fragmentation behavior is consistent with current hypotheses regarding whole protein dissociation that consider proton mobility and intramolecular charge solvation as important factors in determining favored dissociation channels, and are also consistent with the fragmentation behaviors observed for the guanidinated model peptide ions. Further evaluation of the utility of condensed phase guanidination of whole proteins is necessary but the results described here confirm that guanidination can be an effective strategy for enhancing C-terminal aspartic acid cleavages. Gas phase dissociation exclusively at aspartic acid residues, especially for whole protein ions, could be useful in identifying and characterizing proteins via tandem mass spectrometry of whole protein ions. Keywords: tandem mass spectrometry • whole protein ion dissociation • guanidination • protein identification
Introduction Mass spectrometry has emerged as a preferred tool for protein identification1,2 due to an attractive combination of sensitivity, specificity, and speed. Of particular importance is the high specificity that mass spectrometric approaches can afford while simultaneously providing widespread applicability. That is, with an appropriate choice of ionization method, entire classes of species, such as peptides and proteins, can be subjected to the same protocol for identification. High specificity derives from the high precision and accuracy with which molecular masses can be obtained. In the case of peptide mass fingerprinting3-5 or the measurement of accurate mass tags,6 in which peptide masses derived from sequence-specific proteolysis are measured, only the mass measurement characteristics of mass spectrometry are of interest. However, in the case * To whom correspondence should be addressed. Phone: (765) 494-5270. Fax: (765) 494-0239. E-mail:
[email protected]. † Department of Chemistry, Purdue University. ‡ Joint Proteomics Laboratory, The Ludwig Institute for Cancer Research, Royal Melbourne Hospital.
46
Journal of Proteome Research 2004, 3, 46-54
Published on Web 10/16/2003
of tandem mass spectrometry (MS/MS) based techniques, the dissociation of mass-selected ions to produce sequence specific structural information is an additional factor in allowing confident protein identification. The most technologically mature approaches that take advantage of tandem mass spectrometry for protein identification involve the fragmentation of mass selected peptide ions derived from specific proteolysis.7,8 These approaches have thus far proven to be most successful in large-scale identification applications because they can be applied to much more complex mixtures than can peptide mass fingerprinting approaches. An alternative approach to the application of tandem mass spectrometry to proteolytically derived peptides, however, is to subject ions derived from whole proteins directly to dissociation.9 Although early whole protein tandem mass spectrometry investigations where conducted using triple quadrupole instrumentation,10-13 limitations of this instrumentation generally do not allow for protein identification based on the resultant product ion spectra. A series of more recent developments based on the use of high magnetic field strength 10.1021/pr034054u CCC: $27.50
2004 American Chemical Society
research articles
Guanidinated Peptide/Protein Dissociation
Fourier transform ion cyclotron resonance (FTICR), however, have demonstrated the capability for whole protein identification and characterization via tandem MSn.14-16 Other forms of tandem MS instrumentation, such as quadrupole/time-offlight17 and quadrupole ion trap mass spectrometers,18 are also being increasingly applied to whole proteins. Of primary importance to the utility of tandem mass spectrometry based approaches in proteomics is the information content represented by the reaction products produced between stages of mass analysis. In the case of peptides, it is usually desirable to observe products from cleavage of many of the amide linkages. In the case of whole protein ions, however, even if it were possible to cleave every amide bond, doing so can give rise to a significant compromise in detection limits due to the distribution of parent ion charge among a large number of products. It is therefore desirable to be able to derive as much structural information about a protein as possible with a minimum number of dissociation channels. For this reason, it is important to understand how protein ions fragment and, to the extent possible, apply conditions that favor the most informative fragmentation channels. New methodologies for the identification and characterization of proteins via whole protein ions (or large polypeptide pieces) based on electrodynamic ion trapping are being developed. Gas-phase ion/ion reactions19 facilitate the approach and are used for charge state concentration and purification,20 as well as to facilitate interpretation of product ion spectra21 following ion trap collision induced dissociation (CID). Previous work has demonstrated that under ion trap CID conditions for whole protein ions, the charge state of the precursor ion plays a major role in determining the observed fragmentation pathways.22-24 Very low charge state protein ions, classified on the basis of the number of charges being less than the number of arginine residues, tend to show losses of one or more small molecules (i.e., NH3 or H2O). At low charge states, i.e., those close to or often slightly greater than the number of arginine residues, preferential cleavages C-terminal to aspartic acid residues are often observed. A further increase in charge (intermediate charge states) often results in an array of nonspecific amide bond cleavages, whereas at the highest charge states normally produced via electrospray, preferential cleavages N-terminal to proline residues are often the dominant fragmentation channels observed. This overall charge state dissociation behavior, over a wide range of charge states and under the long time frame conditions of ion trap collisional activation, appears to be readily rationalized on the basis of proton mobility,25 in close analogy with protonated peptide ion dissociation. Thus, at very low charge states, solvation of the ionizing protons by several basic sites26 is expected to be maximized, allowing basic site localized fragmentation to dominate. At low charge states, proton mobility remains low such that C-terminal aspartic acid cleavage is facilitated by transfer of the acidic proton of the side chain to the C-terminal carbonyl group, and not therefore catalyzed by an ionizing proton, dominates.27 At intermediate charge states, a condition is reached wherein proton mobility is maximized thus giving rise to the greatest extent of nonspecific cleavage. At the highest charge states, proton mobility is once again reduced, likely due to intramolecular Coulomb repulsion. N-terminal proline cleavage has often been found to be prominent under these conditions, presumably due, at least in part, to a relatively high proton affinity of the carbonyl oxygen of the imide bond of proline.28 As a result, the imide bond might
Scheme 1. Guanidination Reaction of Lysine with o-Methyl Isourea to Form Homoarginine
be more often susceptible to nucleophilic attack. Once the cyclic intermediate has been formed, intramolecular proton transfer can then occur to the imide nitrogen and cleave the bond. If proton mobility is a major factor in determining the fragmentation behavior of protein ions, then measures to affect proton mobility should influence the cleavages observed for a given protein ion charge state. Furthermore, such measures might be used to direct protein dissociation to desired fragmentation channels. Of particular interest are the residuespecific cleavages N-terminal to proline residues and C-terminal to aspartic acid residues. These cleavages are attractive because they can provide enzyme-like specificity in gas-phase fragmentation while giving rise to a relatively small number of product ions formed via dissociation. In this work, we focus on the possibility of directing fragmentation behavior to C-terminal Asp channels by limiting proton mobility via guanidination, i.e., the conversion of lysine side chains to homoarginine (Scheme 1). This chemical reaction was first reported in the late 1960s by Kimmel.29 In recent years, it has gained popularity in peptide mass analysis where guanidination has been used to increase the MALDI response of peptides by making the peptide more basic.30-32 Guanidination can also be used to determine the number of lysine residues present in a peptide due to a 42 Da mass increase per lysine residue.33 Lysine guanidination has also been used previously to elucidate the role of proton mobility in the fragmentation behavior of protonated peptides.34 Here, we have performed a systematic study of the fragmentation behavior of model peptides in a quadrupole ion trap mass spectrometer to demonstrate that lysine guanidination is effective in reducing proton mobility and to mimic the overall trends noted previously for protonated protein charge state dissociation. The peptide ion results were extended to an examination of the collision induced dissociation behavior of guanidinated ubiquitin to rationalize the effect of guanidination on proton mobility within protein ions.
Experimental Materials. Synthetic peptides were purchased from SynPep (Dublin, CA). Perfluoro-1,3-dimethylcyclohexane and o-methyl isourea hydrogen sulfate were purchased from Aldrich (Milwaukee, WI). Bovine ubiquitin was purchased from Sigma (St. Louis, MO). Trifluoroacetic acid (TFA) and guanidine-HCl were purchased from Pierce (Rockford, IL). Guanidination. The guanidination procedure used here is based on that employed by Brancia et al.31 1 mL o-methyl isourea (0.5 M) was added to 1 mg/mL aqueous peptide (8 M guanidine-HCl solution for ubiquitin) solution (1 mL). The mixture was adjusted to pH 10.5 using NaOH (5.0 M). The Journal of Proteome Research • Vol. 3, No. 1, 2004 47
research articles reaction proceeded in the dark for 24 h at room temperature. 2 mL aqueous 1% TFA was added to stop the reaction. The solution was then purified using reversed phase high-performance liquid chromatography on a Hewlett-Packard (Palo Alto, CA) 1090 system using an Aquapore RP-300 (7 µm pore size, 100 × 4.6 mm i.d.) column (Perkin-Elmer, Wellesley, MA) operated at 1 mL/min. A linear 60 min gradient from 0 to 100% buffer B was used, where buffer A contained 0.1% aqueous TFA and buffer B contained 0.09% TFA in acetonitrile/water (40/60). Absorbance was monitored at 215 nm and the column was maintained at a temperature of 40 °C. The collected fractions were lyophilized and reconstituted in water for use as stock solutions. Under the conditions used in this study, 100% of the peptides were guanidinated while a range of 6-7 lysines were converted to homoarginines in the case of ubiquitin. Ubiquitin with differing numbers of modified lysine residues elute at different times during HPLC purification. For these experiments, fully guanidinated ubiquitin was collected for use. Mass Spectrometry. Unmodified peptides and protein were used for MS analysis without further purification. 0.1 mg/mL solutions of modified and unmodified samples were prepared in water/methanol/acetic acid (50/50/1) for MS analysis. Peptides were ionized using nano-electrospray. Borosilicate glass capillaries (1.5 mm o.d., 0.86 mm i.d.) were pulled to form nanospray emitters using a P-87 Flaming/Brown type micropipet puller (Sutter Instruments, Novato, CA). A stainless steel wire was inserted into the capillary and typically about 1.4 kV was applied for ionization. Protein solutions were ionized using a laboratory-built capillary electrospray source with a flow rate of 1 µL/min. Typically about 2.5 kV was applied to the electrospray needle for ionization. All experiments were performed on a Hitachi (San Jose, CA) M-8000 quadrupole ion trap mass spectrometer, modified to allow for ion/ion reactions.35 Ion/ion reactions were used to reduce the charge states of precursor and/or product ions via proton transfer using PDCH (perfluoro-1,3-dimethylcyclohexane) as described by Stephenson et al.36 For peptide ions, a typical ion accumulation time was several hundred milliseconds during which [M + 2H]2+ or [M + 3H]3+ ions were produced directly from the sample. For some experiments, this was followed by a PDCH anion injection time of a few milliseconds and an ion/ion reaction time of several hundred milliseconds to produce [M + H]+ ions. Isolation of ions was accomplished using the Hitachi instrument’s filtered noise fields (FNF) and by raising the amplitude of the radio frequency signal applied to the ring electrode of the ion trap to eject unwanted ions. An external waveform generator (Model 33120A, Agilent, Palo Alto, CA) was used to apply a single frequency resonance excitation voltage to the endcaps for 300 ms to induce CID. Product ion spectra were acquired via resonance ejection.37 The peptide ion spectra shown are typically the average of 300 scans. For protein ions, precursor ion accumulation and ion/ion reactions took place as described for the peptide ions. However, during the ion/ion reaction time a single-frequency resonance excitation voltage was applied to accomplish ‘ion parking′20,38 to concentrate the ions of higher charge states into that of a lower charge state of interest. Subsequent precursor ion isolation and CID was performed as described with the peptides and the product ions were then subject to ion-ion reactions to reduce the multiply charged products to predominantly their singly charged forms for simplification of spectral interpreta48
Journal of Proteome Research • Vol. 3, No. 1, 2004
Pitteri et al.
tion. The post ion/ion MS/MS spectra shown for protein ions are typically the average of 250 scans.
Results and Discussion Charge State and Chemical Modification Dependent Fragmentation of Model Peptide Ions. The fragmentation behavior of the singly, doubly, and, in some cases, triply charged ions of a suite of model peptides of the general form XGAILYGAILR (where X and Y are variable amino acid residues) were examined to establish the roles of charge state and chemical modification on the dissociation of relatively simple peptide systems. The model peptide GAILDGAILR illustrates the dramatic role that parent ion charge state can play in peptide dissociation. The MS/MS spectrum of singly protonated GAILDGAILR (Figure 1A) shows almost exclusive fragmentation to yield the y5+ fragment, which arises from cleavage C-terminal to the aspartic acid residue. The next most prominent loss is that of neutral water and/or ammonia. Within the context of the mobile proton model25, this behavior is readily rationalized on the basis of charge sequestration by the highly basic C-terminal arginine residue, causing the ionizing proton to be “nonmobile”. Thus, the energy required to mobilize the proton to initiate nonspecific fragmentations along the polypeptide chain is greater than that required for the C-terminal aspartic acid cleavage. Hence, the aspartic acid cleavage tends to dominate.39,40 The MS/MS spectrum of the doubly protonated peptide (Figure 1B) shows dramatically different fragmentation behavior. Here, every amide bond cleavage is observed, in addition to the neutral water loss. This behavior is also consistent with the mobile proton model25, where one proton is expected to be strongly associated with the arginine residue, whereas the other is considered to be “mobile”. Therefore, the mobile proton is able to catalyze nonspecific peptide backbone cleavages during ion activation. The y5+ ion, corresponding to cleavage C-terminal to the aspartic acid residue is still present, as with the singly charged MS/MS spectrum. However, it is no longer the most prominent cleavage. The b5+ ion complement of the y5+ ion is also observed in the doubly charged MS/MS spectrum because there are now two charges to support the observation of both products. The addition of a lysine residue at the N-terminus of GAILDGAILR to give KGAILDGAILR provides another moderately basic site to the peptide to affect dissociation behavior. Similar to that of singly protonated GAILDGAILR, the most abundant product ion in the dissociation of singly protonated KGAILDGAILR (Figure 2A) is the C-terminal aspartic acid cleavage. This is consistent with the proton being once again sequestered on the arginine residue. The loss of a small neutral molecule from the parent is again observed, which is also consistent with the MS/MS data of singly charged GAILDGAILR. The y10+ ion corresponds to cleavage occurring C-terminal to the lysine residue, which is another fragmentation behavior that has previously been observed with low charge states of protein ions41 and simple lysine containing peptides.42 The MS/MS spectrum of doubly protonated KGAILDGAILR (Figure 2B) shows somewhat similar fragmentation behavior to that of the singly protonated ion. The most prominent cleavage in this case is still the C-terminal aspartic acid cleavage. In this case the b6+ complement to the y5+ ion is also observed because an excess proton is available to support the observation of both products. The abundances of the Cterminal aspartic acid cleavage product ions also suggest that the second proton is less mobile than is the case with doubly
Guanidinated Peptide/Protein Dissociation
research articles
Figure 1. Ion trap CID MS/MS product ion spectra of (a) singly protonated GAILDGAILR and (b) doubly protonated GAILDGAILR.
Figure 2. Ion trap CID MS/MS product ion spectra of (a) singly protonated KGAILDGAILR and (b) doubly protonated KGAILDGAILR.
protonated GAILDGAILR. Although the C-terminal lysine specific y10+ ion, and the neutral loss product is also observed in this dissociation, the increased abundance of nonspecific cleavages relative to that observed in the singly protonated KGAILDGAILR ion suggests that addition of the second proton affords an enhanced degree of proton mobility for nonspecific
dissociation, but such dissociation is still not highly competitive with aspartic acid cleavage. The fragmentation behavior of KGAILDGAILR can be altered by chemical modification of the peptide. Guanidination affects the ability of lysine residues to sequester protons by increasing the basicity of the side chain (the proton affinity of lysine is Journal of Proteome Research • Vol. 3, No. 1, 2004 49
research articles
Figure 3. Ion trap CID MS/MS product ion spectra of (a) singly protonated guanidinated KGAILDGAILR, (b) doubly protonated guanidinated KGAILDGAILR and (c) triply protonated guanidinated KGAILDGAILR.
238.0 kcal/mol whereas that of arginine is 251.2 kcal/mol43). The number of basic residues (1 Arg and 1 HomoArg) in singly protonated guanidinated KGAILDGAILR exceeds the number of protons, making it a peptide analogue to a protein ion with net charge less than the number of arginine residues.23 Here, the most abundant product ion results from the neutral loss of a water molecule (Figure 3A). Although there is also some evidence of C-terminal aspartic acid cleavage in the spectrum (y5+/b6+), it clearly does not compete favorably with the neutral loss. This dissociation behavior differs dramatically from that observed in the unmodified singly protonated KGAILDGAILR, where the most facile cleavage was the C-terminal aspartic acid cleavage (Figure 2A). It has been shown previously that solvation of a proton by several basic residues can facilitate the loss of small neutral molecules,26 which is consistent with the behavior noted here. There is also the possibility for interaction of a neutral basic site with the acidic side chain of the aspartic acid residue that could serve to inhibit the mechanism for C-terminal aspartic acid cleavage.44 The doubly protonated guanidinated KGAILDGAILR CID MS/MS product ion spectrum (Figure 3B) shows a strong preference for C-terminal aspartic acid cleavage. In this case, there are two protons and two strongly basic residues so that the excess protons are presumably associated with the homoarginine and arginine residues. This dissociation behavior is consistent with the fragmentation of low protein ion charge states and can be rationalized by low proton mobility arising from the close association of the charges with the strongly basic side chains of the arginine and homoarginine residues. Furthermore, as each basic site is expected to be occupied by a proton, the potential for intramolecular solvation of an individual charge site by a second strongly basic residue is inhibited. Also, any potential interaction between neutral basic and acidic sites are precluded by occupying the basic sites with 50
Journal of Proteome Research • Vol. 3, No. 1, 2004
Pitteri et al.
protons. It is interesting to compare the MS/MS spectra of doubly protonated KGAILDGAILR (Figure 2B) with that of the doubly protonated guanidinated KGAILDGAILR. By guanidinating the peptide, the doubly protonated ion has essentially been converted from a “low” charge state to a “very low” charge state, as exemplified by the fragmentation behavior. The triply protonated guanidinated KGAILDGAILR ion was observed to fragment nonspecifically (Figure 3C), which is dramatically different from that of either the singly or doubly protonated ions. In this case, there is no apparent preference for C-terminal aspartic acid cleavage and fragmentation is observed at every amide bond. For the triply protonated ion, two of the protons are expected to be accommodated by the homoarginine and arginine residues, respectively, whereas the third proton does not have a strongly preferred site for association, and is therefore presumably relatively free to catalyze nonspecific cleavages along the peptide backbone upon activation. The three charge states of guanidinated KGAILDGAILR exhibit most of the generic fragmentation behavior thus far observed for whole protein ions. For example, the loss of small neutral molecules is maximized when there is a high likelihood for proton solvation by multiple strong basic sites and/or interactions between strong neutral acidic and basic sites. This is the case for the singly protonated ion of guanidinated KGAILDGAILR. When charge state is sufficiently high to minimize intramolecular charge solvation, but too low to allow for a high degree of proton mobility, preferred cleavage at aspartic acid residues is noted. This is the case for doubly protonated guanidinated KGAILDGAILR. When the total number of protons exceeds the number of strongly basic sites, nonspecific cleavages are maximized. This condition appears to be satisfied with the triply protonated ion of guanidinated KGAILDGAILR. At higher charge states for protein ions, nonspecific cleavages of protein ions tend to give way to specific cleavages N-terminal to proline residues, if present. At even higher charge states, preferred cleavage at proline residues can give way to a limited set of prominent cleavages that, thus far, cannot be predicted or readily rationalized.41,45 However, in the case of guanidinated KGAILDGAILR, the possibility for the latter fragmentation behaviors is absent. Charge State and Chemical Modification Dependent Fragmentation of Ubiquitin Ions. The data for the model peptides clearly indicate that guanidination of lysine residues can significantly affect the fragmentation behavior of a given peptide ion charge state. The results suggest that these effects arise from the effect that conversion of a lysine residue to a homoarginine residue has on proton mobility. Given the similar fragmentation behavior of the various charge states of guanidinated KGAILDGAILR ions with the charge state dependent fragmentation behavior noted thus far for whole protein ions, it was of interest to examine the effect that protein guanidination may have on whole protein ion dissociation. Therefore, we have examined the effect of guanidination on bovine ubiquitin, a protein for which CID data have been collected over a wide range of charge states.21,23 Of particular interest for protein identification/characterization purposes is the preferred cleavage C-terminal to aspartic acid residues. In analogy with the model peptide ion fragmentation behavior outlined above, whole protein guanidination might be expected to direct fragmentation more strongly to aspartic acid residues by minimizing proton mobility and thereby diminishing contributions from nonspecific cleavages. In such a scenario, the
Guanidinated Peptide/Protein Dissociation
research articles
Figure 4. Mass spectra of (a) bovine ubiquitin and (b) guanidinated bovine ubiquitin with 200 ms ionization time.
Figure 5. Post ion/ion reaction CID MS/MS product ion spectra of (a) the [M + 10H]10+ ion of ubiquitin and (b) the [M + 10H]10+ ion of guanidinated ubiquitin.
analogy between gas phase cleavage and specific enzymatic condensed phase cleavage is most strongly demonstrated. Such
a situation would enable protein search strategies based exclusively on cleavages occurring at aspartic acid residues. Journal of Proteome Research • Vol. 3, No. 1, 2004 51
research articles
Pitteri et al.
Figure 6. Post ion/ion reaction CID MS/MS product ion spectra of (a) the [M + 8H]8+ ion of ubiquitin and (b) the [M + 8H]8+ ion of guanidinated ubiquitin.
Although the effect of whole protein guanidination on protein ion dissociation behavior is of primary interest in this study, it is also of relevance to note the effects of guanidination on signal levels and charge state distributions for the system studied here. The main rationale used to date for guanidination of tryptic peptides has been to enhance signal response in matrix-assisted laser desorption ionization by converting peptides with no arginine residues to peptides with at least one homoarginine residue. In the case of most proteins, including ubiquitin, arginine residues are present in the unmodified system. Furthermore, it has been shown that the electrospray responses for positively charged proteins are similar on a charge normalized basis when the proteins are expected to carry a net positive charge in solution.46 Therefore, the effect of protein guanidination is expected to affect signal response only insofar as it affects the average charge of the proteins. Figure 4 shows the electrospray mass spectra of bovine ubiquitin and guanidinated bovine ubiquitin acquired under nominally the same electrospray conditions. Guanidination gives rise to a small but measurable increase in average abundance weighted charge. The average charge state for ubiquitin is approximately 10.4, whereas the average charge state for guanidinated ubiquitin is approximately 11.2. This increase would suggest that guanidination would give rise to only a small increase in electrospray response. Although careful comparisons of protein response versus concentration have not been made for both modified and unmodified ubiquitin, where approximate comparisons could be made, only small increases in electrospray response at most were noted for the guanidinated ubiquitin ions. 52
Journal of Proteome Research • Vol. 3, No. 1, 2004
In this work, CID was performed on three charge states of guanidinated ubiquitin and the products were subjected to ion/ ion reactions to reduce their charge states to predominantly +1. The CID products of guanidinated ubiquitin ions were compared to the CID products of unmodified ubiquitin ions. The three charge states were selected to represent the behavior of high charge state ions (i.e., 10+-12+ for ubiquitin), intermediate charge state ions (5+-9+), and low charge state ions (i.e.,
