Comparison of Vacuum Matrix-Assisted Laser Desorption/Ionization (MALDI) and Atmospheric Pressure MALDI (AP-MALDI) Tandem Mass Spectrometry of 2-Dimensional Separated and Trypsin-Digested Glomerular Proteins for Database Search Derived Identification Corina Mayrhofer,†,‡ Sigurd Krieger,‡ Emmanuel Raptakis,§ and Gu1 nter Allmaier*,† Institute of Chemical Technologies and Analytics, Vienna University of Technology, A-1060 Vienna, Austria, Clinical Institute of Pathology, Medical University of Vienna, A-1090 Vienna, Austria, and Shimadzu Biotech Kratos Analytical, Manchester, M17 1 GP, United Kingdom Received April 12, 2006
Mass spectrometric based sequencing of enzymatic generated peptides is widely used to obtain specific sequence tags allowing the unambiguous identification of proteins. In the present study, two types of desorption/ionization techniques combined with different modes of ion dissociation, namely vacuum matrix-assisted laser desorption/ionization (vMALDI) high energy collision induced dissociation (CID) and post-source decay (PSD) as well as atmospheric pressure (AP)-MALDI low energy CID, were applied for the fragmentation of singly protonated peptide ions, which were derived from two-dimensional separated, silver-stained and trypsin-digested hydrophilic as well as hydrophobic glomerular proteins. Thereby, defined properties of the individual fragmentation pattern generated by the specified modes could be observed. Furthermore, the compatibility of the varying PSD and CID (MS/MS) data with database search derived identification using two public accessible search algorithms has been evaluated. The peptide sequence tag information obtained by PSD and high energy CID enabled in the majority of cases an unambiguous identification. In contrast, part of the data obtained by low energy CID were not assignable using similar search parameters and therefore no clear results were obtainable. The knowledge of the properties of available MALDI-based fragmentation techniques presents an important factor for data interpretation using public accessible search algorithms and moreover for the identification of two-dimensional gel separated proteins. Keywords: kidney proteomics • two-dimensional gel electrophoresis • vacuum matrix-assisted laser desorption/ ionization • atmospheric pressure matrix-assisted laser desorption/ionization • post-source decay • low and high energy collision-induced dissociation
Introduction The analysis of tissues, body fluids, cells, or cell organelles on the proteome level enables a comprehensive understanding of the expression, modifications, interactions, and regulation of proteins and thereby provides a deeper insight into biological processes. Such proteomic based approaches include typically the following steps, protein extraction, protein fractionation and separation, and finally, protein identification. Twodimensional gel electrophoresis is a widely used technique for separation of proteins and in the process of protein identification mass spectrometry is an integral part of the whole strategy. Since its introduction in the late 1980s vacuum matrix-assisted * To whom correspondence should be addressed. Institute of Chemical Technologies and Analytics, Vienna University of Technology, Getreidemarkt 9/164-AC, A-1060 Vienna, Austria. Tel: +43-1-58801-15160. Fax: +43-158801-15199. E-mail:
[email protected]. † Vienna University of Technology. ‡ Medical University of Vienna. § Shimadzu Biotech Kratos Analytical. 10.1021/pr060165s CCC: $33.50
2006 American Chemical Society
laser desorption/ionization (vMALDI)1,2 has become a powerful desorption/ionization technique. vMALDI has especially been used in conjunction with time-of-flight (TOF) mass analyzers and further developments of the technique, including time delayed extraction,3-7 the use of a reflectron8 as TOF analyzer (RTOF), as well as the coupling to tandem mass spectrometry9-12 has established vMALDI-TOF mass spectrometry as an indispensable tool for protein and peptide analysis. Other extensions of the technique include the coupling of vMALDI ion sources to other mass analyzers, such as Fourier transform ion cyclotronresonance instruments,13 sector-field instruments14 and ion traps.15-18 A relatively new development of the MALDI technology is the generation of ions at atmospheric pressure (AP) using a time-of-flight mass analyzer19 and the coupling of the APMALDI ion source with an ion trap.20 The general process for the mass spectrometric based identification of the proteins of interest, independent of the applied technique, comprises two steps, namely data acquisition and data interpretation. The Journal of Proteome Research 2006, 5, 1967-1978
1967
Published on Web 07/20/2006
research articles generation of a peptide mass fingerprint (PMF) of either enzymatically digested or chemically cleaved proteins is a widely used method for protein identification.21-25 These data can be completed and/or verified by additional sequence information on selected peptides. Due the uniqueness of peptide sequences, sequence tags allow in many cases the unambiguous identification of the protein. Using MALDI-RTOF instruments sequence information can be obtained by postsource decay (PSD).26,27 Furthermore, ion activation employed in tandem mass spectrometers can be performed by collisions of the selected precursor ion with inert gas in a collision cell.28-30 The collision induced dissociation (CID) processes occurring routinely can be divided by the translational energy of the precursor ion. Low-energy collisions, common in quadrupole and ion trap instruments, occur in the 1-100 eV range and high-energy collisions, common in TOF/TOF and foursector instruments, are in the kiloelectronvolt range. The obtained data from the PSD and MS/MS experiments can then be analyzed by various database search engines available on the Internet. The characteristics of fragmentation processes based on vMALDI ionization have been the objective of a number of studies.31-34 The recent availability of AP-MALDI in combination with analyzers allowing low energy CID (e.g., ion trap) extends further the scope of possible fragmentation techniques.35 In contrast to vMALDI mass spectrometry, up to now only few applications of AP-MALDI for identification of proteins separated by gel-based methods are described.36-39 In this study, the impact of different modes of ion dissociation combined with vMALDI and AP-MALDI as precursor ion generating techniques on the database search derived identification of proteins separated by two-dimensional gel electrophoresis was evaluated. The comparative study included the application of high energy CID as well as PSD, both performed on a vMALDI TOF/curved field reflectron (CFR) mass spectrometer and low energy CID, performed on a 3D ion trap instrument equipped with an AP-MALDI source to gain peptide specific sequence information. Silver stained cytosolic as well as membrane glomerular proteins were enzymatically in-gel digested, followed by peptide mass fingerprinting (vMALDI and AP-MALDI) and the selected singly charged tryptic peptides (precursor ions: protonated molecular ions) were fragmented using the three fragmentation techniques as specified above. For data interpretation, two commonly used public available search algorithms were chosen. The evaluation of the impact of different fragmentation techniques in combination with vMALDI and AP-MALDI (for singly charged precursor ion generation) on the unambiguous identification process will be useful for further proteomic approaches.
Experimental Section Sample Preparation. Glomeruli were obtained by graded sieving of minced rat kidney cortex suspensions on ice as described previously.40 Glomerular proteins were extracted and fractionated by means of the Triton X-114 phase separation kit (2-D Sample Prep for Membrane Proteins, Pierce, Rockford, IL). Two-Dimensional Gel Electrophoresis. Extracted glomerular hydrophilic as well as hydrophobic proteins were solubilized in rehydration buffer containing 7 M urea, 2 M thiourea, and 4% CHAPS. In particular cases, 2% octyl-β-D-glucopyranoside was added. Finally, 15 mM dithiothreitol (DTT) and 0.5% IPGbuffer were added. In the case of hydrophilic proteins, commercial immobilized pH gradient gels (Immobiline Dry Strips, 1968
Journal of Proteome Research • Vol. 5, No. 8, 2006
Mayrhofer et al.
pH 3-10 NL, GE Healthcare, Uppsala, Sweden) were rehydrated with the sample solutions for 12 h at 20 °C. Hydrophobic proteins were applied to the immobilized pH gradient gels (IPG, Immobiline Dry Strips, pH 3-10 NL, GE Healthcare) by in-gel rehydration for 6 h, followed by an active rehydration step for 6 h at 20 V. Isoelectric focusing was performed using an IPGphor system (GE Healthcare) according to manufacturer’s instructions. For subsequent sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) IPG-strips containing hydrophilic proteins were equilibrated with 1% DTT in equilibration buffer containing 7 M urea, 2 M thiourea, 30% (87% v/v) glycerol, 0.05 M Tris-HCl, pH 8.8 and 2% SDS for 10 min followed by 4% iodoacetamide (IA) in equilibration buffer for additional 10 min. IPG-strips containing hydrophobic proteins were equilibrated with 1% DTT in equilibration buffer containing 7 M urea, 2 M thiourea, 30% (87% v/v) glycerol, 0.05 M Tris-HCl, pH 8.8 and 4% SDS for 10 min followed by 4% IA in equilibration buffer for additional 10 min. SDS-PAGE was performed using vertical self-made 3.6-15% polyacrylamide gels (PROTEAN II xi Cell, Mini-PROTEAN Cell, Biorad, Hercules, CA). Visualization of Proteins. Gels were silver stained according to Blum.41 In-Gel Tryptic Digestion of Proteins. The protein spots were manually excised and the silver stained gel pieces were destained with 20 µL of aqueous solution containing 15 mM potassium ferricyanide and 50 mM sodium thiosulfate42 and washed twice for 5 min with distilled water. The gel pieces were then equilibrated for 10 min with 25 mM ammonium bicarbonate, washed twice for 10 min with 50% acetonitrile (ACN) in 25 mM ammonium bicarbonate and subsequently dried in a vacuum centrifuge. The dried gel pieces were rehydrated in 5 µL trypsin working solution containing 12.5 ng/µL sequencing grade trypsin (Roche, Mannheim, Germany) in 25 mM ammonium bicarbonate for 35 min at 4 °C. Excess trypsin working solution was removed and the gel pieces were washed once with 25 mM ammonium bicarbonate and finally 15 µL 25 mM ammonium bicarbonate was added to each sample. The samples were incubated either for 18 h at 37 °C or for 10 min using a microwave oven at 210 W. Peptides were extracted once with 20 µL H2O/0.1% trifluoroacetic acid (TFA) for 20 min and twice with 20 µL 50% ACN/5% TFA acid for 20 min with gentle shaking at room temperature. Subsequently, the combined peptide extracts were concentrated in a vacuum centrifuge, redissolved in 10 µL 0.1%TFA and purified by ZipTip RP-18 technology (Millipore, Bedford, MA). Elution of the peptide solution was performed with 5 µL of 60% ACN/40% 0.1% TFA. MALDI-MS Sample Preparation. As matrix 0.45 µL of 10 mg/mL of R-cyano-4-hydroxy-cinnamic acid (type C-2020 Sigma-Aldrich) in acetone was applied to either a stainless steel target in the case of vMALDI or to a gold coated sample plate (Mass Tech, Columbia, MD) in the case of AP-MALDI using the thin-layer technique.43 One microliter of the purified peptide mixture was deposited on the microcrystalline matrix layer and dried at room temperature in a gentle stream of air. Vacuum MALDI-PSD (RTOF) and High Energy CID (TOF/ CFR, MS/MS). vMALDI-MS, PSD, and high energy CID (MS/ MS) experiments were performed on a Shimadzu Biotech Kratos Analytical TOF2 vMALDI-TOF/curved field reflectron (CFR) mass spectrometer (Manchester, UK) equipped with a high energy collision cell and a nitrogen laser operating at a wavelength of 337 nm (10 Hz). All mass spectra were acquired in positive reflectron ion mode using delayed extraction. The
research articles
V-MALDI and AP-MALDI-MS/MS of Glomerular Proteins
accelerating voltage was set to 20 kV in all cases and no deceleration or reacceleration of the ion beam was necessary, allowing 20 keV laboratory frame collisional energy for high energy CID experiments. For high energy CID experiments helium was used as collision gas and the gas pressure in the collision cell was set to 5 × 10-6 mbar by means of differential pumping of the cell, whereas in the MS mode and in the PSD mode the gas pressure was typically around 6.5 × 10-8 mbar in the complete analyzer system (the collision cell was free of any collision gas). For high energy CID experiments, the laser power was set slightly above the threshold of precursor ion formation. For PSD experiments the laser power was increased until product ions with sufficient signal/noise (S/N) ratio were observed. The precursor ions were selected with an isolation width of (3 to 5 Th. For spectra acquisition between 3000 and 6000 unselected laser pulses were summed (for the correct comparison with AP-MALDI IT experiments). For the comparative studies equal laser shots were acquired for the CID and the PSD spectra of each selected precursor ion. External mass calibration was performed using the following standard peptides, namely Bradykinin (Sigma B-4181) and ACTH, fragment 18-39 (Sigma A-8346). The given vMALDI PSD and high energy CID mass spectra were smoothed using the company supplied averaging algorithm and the baseline was subtracted. Atmospheric Pressure MALDI Low Energy CID (IT, MS/MS). Positive-ion AP-MALDI mass spectra and low energy CID spectra (MS/MS, MS2) were obtained on a Bruker Daltonics HCTplus (Bremen, Germany) 3D-ion trap mass spectrometer equipped with a second generation AP-MALDI-pulsed dynamic focusing (PDF)-source (Agilent Technologies, Palo Alto, CA) with a nitrogen laser (λ ) 337 nm) operating at 10 Hz repetition rate. Full scan mass spectra as well as CID-spectra were acquired under the following experimental parameters. The dry gas temperature was set to 350 °C. The accumulation time was set to 100 ms and the acquisition times of the final spectra ranged between five and 10 minutes, corresponding to 30006000 laser pulses. For CID-spectra, precursor ions were selected with an isolation width and activation width of (2 Th. The low mass-cutoff (q value) was set in all cases to 27%. Both, collisional cooling and low energy CID were performed utilizing helium gas. External mass calibration was performed using a mixture containing the following peptides, namely Bradykinin ([M + H]+ 757.3997), Angiotensin II ([M + H]+ 1046.54) Somatostatin ([M + H]+ 1637.72) and ACTH fragment 18-39 ([M + H]+ 2465.1989). The shown AP-MALDI low energy CID spectra are without any smoothing. Database Searches. The obtained data were analyzed applying two different public available search algorithms, namely MASCOT (http://matrixscience.com) and ProteinProspector MS-Tag (http://prospector.ucsf.edu). Searches were carried out using either the Swiss-Prot or the NCBI-non redundant protein sequence database using the parameters described in Table 2.
Results and Discussion PMF by vMALDI and AP-MALDI for Identification of Glomerular Proteins. The nature of proteins being relevant in biological processes and therefore of interest are manifold. One group of proteins attracting great attention, due their fundamental cellular functions, such as cell signaling and ion transport, are membrane proteins. Membrane proteins are generally under-represented compared to soluble proteins and for the purpose of analyzing both, cytosolic and membrane
Table 1. Number of Fragment Ions of Selected Tryptic Peptides, Derived from Hydrophilic (cytosolic) as Well as Hydrophobic (membrane) Proteins Extracted from Rat Glomeruli, Generated by the Different Applied Mass Spectrometric Techniques and Further Used for Database Searches no. of fragment ions nature of protein
precursor ion (selected) [MH]+a
vMALDI high energy CID
vMALDI PSD
AP-MALDI low energy CID
cytosolic cytosolic cytosolic found in both fractions found in both fractions membrane membrane membrane membrane membrane found in both fractions found in both fractions
1206.7 1529.9 1556.7 1566.9
23 26 30 30
15 13 24 23
16 15 9 49
1588.9
30
10
28
1396.9 1650.9 1919.1 1606.9 2929.6 1954.1
24 29 40 15 28 30
17 16 31 11 22 21
24 50 29 13 3 25
1790.8
35
20
20
a
m/z derived from vMALDI PMF analysis.
proteins, glomerular proteins were fractionated by the means of TritonX-114 phase separation. Phase separation resulted in a detergent phase enriched in membrane proteins and an aqueous phase,44 in which the majority of cytosolic proteins were recovered. Both fractions were then individually separated by two-dimensional gel electrophoresis. A number of silver stained either cytosolic or membrane proteins were selected for the comparative mass spectrometric investigation. To evaluate the usefulness of the fragmentation techniques and the quality of the resulting data based on vMALDI and APMALDI as desorption/ionization techniques, proteins of varying abundance and of varying molecular weight were chosen. Figure 1 gives an overview of the applied strategy. Examples of a selected low abundant cytosolic protein and an abundant membrane protein are marked in Figure 2A and Figure 2B, respectively. The first step was the acquisition of a PMF using a vMALDI-TOF/CFR mass spectrometer and an AP-MALDI ion trap instrument. Figure 2C shows the PMF of the marked ingel digested low abundant cytosolic protein acquired by vMALDI tandem instrument. The full scan spectrum of the same tryptic peptide mixture obtained on an AP-MALDI high capacity 3D ion trap mass spectrometer is given in Figure 2E. Figure 2D,F shows the tryptic peptide molecular ions derived from an in-gel digested abundant membrane protein (as marked by an arrow and rectangular area cut out in Figure 2B), generated by vMALDI (Figure 2D) and AP-MALDI (Figure 2F). It is obvious from the shown mass spectra that the PMFs of the proteins which differ in their abundances and molecular weights result in different numbers of peptide molecular ions, and furthermore the number is depending on the chosen desorption/ionization method. Mostly, PMFs consisting of only few peptides do not enable to get a reliable identification of the proteins. Anyway, in such cases, the proteins of interest have to be unambiguously identified by peptide specific sequence tags, which can also be used to verify the PMF data afterward. To obtain such peptide specific sequence tags PSD and tandem mass spectra (MS/MS) of selected peptides were Journal of Proteome Research • Vol. 5, No. 8, 2006 1969
research articles
Mayrhofer et al.
Table 2. Parameters Used for Database Searches of Fragment Ion Data vMALDI high energy CID parameters
database enzyme taxonomy instrument type name max. no. of missed cleavages cysteine modification variable modificationa max. % of unmatched ionsb precursor ion tolerance fragment ion tolerance peptide chargea a
MASCOT
AP-MALDI low energy CID
vMALDI PSD
MS-Tag
MASCOT
MS-Tag
MASCOT
MS-Tag
Swiss-Prot/NCBInr trypsin Rattus norvegicus MALDI-TOF-TOF
Swiss-Prot/NCBInr trypsin Rattus norvegicus MALDI-TOF-PSD MALDI-PSD
Swiss-Prot/NCBInr trypsin Rattus norvegicus ESI-TRAP ESI-ION-TRAP
2
2
2
carbamidomethylation
carbamidomethylation
carbamidomethylation
oxidation of methionine
oxidation of methionine
oxidation of methionine
10
10
10/50
(0.2 Da
( 0.2 Da
(0.2 Da
max ( 1.2 Da 1+
( 0.2 Da
(0.2 Da
( 0.2 Da
max ( 1.2 Da
max ( 1.2 Da
1+
1+
b
Parameter selectable using the MASCOT search engine. Parameter selectable using the MS-Tag search algorithm.
Figure 1. Schematic representation of the mass spectrometricbased analytical strategy. After fractionation of the glomerular proteom into a hydrophilic (cytosolic proteins) and hydrophobic (membrane proteins) subproteom and separation by twodimensional gel electrophoresis, selected proteins were in-gel tryptic-digested and the resulting purified peptide mixtures were used for mass spectrometric analysis using vMALDI and APMALDI as precursor ion generating techniques. After generation of a peptide mass fingerprint, peptide sequencing was performed by means of vMALDI high energy CID as well as PSD and APMALDI low energy CID and the obtained data were used for database search derived identification.
acquired. Thereby only peptide ions generated by both desorption/ionization techniques were selected. Comparison of the Fragmentation Patterns Obtained from PSD and Tandem Mass Spectrometry Techniques. Twelve 1970
Journal of Proteome Research • Vol. 5, No. 8, 2006
singly protonated peptide ions derived from tryptic digestion of hydrophilic (cytosolic) and hydrophobic (membrane) proteins in the m/z range from 1206.7 to 2929.6 were selected for the comparative analysis. For this purpose, ion activation was performed using three different fragmentation methods, including high energy CID as well as PSD, both obtained on a vMALDI-TOF/CFR mass spectrometer and low energy CID, performed on a 3D ion trap instrument equipped with an APMALDI source. After data acquisition, the selected fragment ions useable for database searches were evaluated. The spectra and resulting properties of the different ion activation methods are illustrated and exemplified by the means of two different precursor ions, which derived from tryptic digestion of a cytosolic as well as a membrane protein. Figure 3 shows the PSD and two MS/MS spectra of a tryptic peptide produced from the PMF analysis of the low abundant protein, as shown in Figure 2C,D and marked with an asterisk. The fragment ion spectra of the precursor ion with m/z 1919.1, which derived from tryptic digestion of an abundant membrane protein (marked in Figure 2B), are given in Figure 4. Both figures show (A) the PSD, (B) the high energy CID, and (C) the low energy CID spectrum in the same sequence. The PSD spectra and the high energy CID spectra were acquired under identical conditions, except that helium was allowed into the collision cell at a pressure of 5 × 10-6 mbar and the laser power was decreased in the latter case in order to generate similar S/N ratios. As is evident from Figures 3 and 4 the high energy CID spectra differed clearly in the appearances of fragment ions in the low mass range (e m/z 200). Most of the peaks appearing in the higher mass range of the PSD and high energy CID spectra were identical, but also here some additional ions could be exclusively found under high energy CID conditions. The absolute fragment ion intensities were higher using high energy CID, whereas the relative intensities within the fragment ions (one fragment ion is defined as base peak) were comparable using the two methods. The achieved mass resolution (Rfwhm) for the fragment ions in the high energy CID as well as PSD spectra ranged from 300 to 600 (after smoothing was applied) up to a mass range of m/z 3000 (without smoothing and at a lower number of laser pulses the achievable resolution lies between
V-MALDI and AP-MALDI-MS/MS of Glomerular Proteins
research articles
Figure 2. Application of vMALDI and AP-MALDI for identification of cytosolic as well as membrane proteins. (A) Detail of the silver stained two-dimensional pattern of cytosolic proteins. The arrow marks a low abundant cytosolic protein (black ellipsoid was cut out), with approximate molecular mass 25 kDa, subjected to in-gel tryptic digestion. The tryptic digestion was performed overnight at 37 °C. (B) Section of a representative silver stained two-dimensional pattern of membrane proteins enriched by TritonX-114 phase separation. The arrow marks an abundant protein (white rectangle was cut out), with approximate molecular mass 55 kDa, subjected to mass spectrometric analysis. Tryptic digestion was performed for 10 minutes using a microwave oven. (C,D) Peptide mass fingerprints of the marked cytosolic protein (C) as well as the marked membrane protein (D) acquired on a vMALDI TOF/CFR instrument. 100% intensity corresponds to (C) 660 mV and (D) 57 mV. (E, F) Full scan mass spectra of the identical tryptic peptide mixtures (aliquot) acquired on an AP-MALDI ion trap instrument. The peaks denoted with asterisks were selected for further PSD and MS/MS experiments. Autolysis products of trypsin are marked with T. The S/N ratios for the peak with m/z 1206.7 are (C) 26.5 and (E) 3.5. The S/N ratios for the peak with m/z 1919.1 are (D) 56.2 and (F) 35.7.
800 and 1800). The corresponding low-energy CID fragmentation patterns (starting from a singly charged precursor ion, which is usually generated by vMALDI and AP-MALDI) differed in several points. First, as a consequence of the technical characteristics of a 3D ion trap, fragment ions in the m/z range below approximately 30% of the m/z of the precursor ion were absent in the spectra. Second, in contrast to the spectra acquired on the vMALDI-TOF/CFR instrument, more ions in the upper m/z range could be obtained by low energy CID (due the q-factor which is optimized for higher m/z value ions). Furthermore, the peak intensities were generally lower compared to the vMALDI fragmentation techniques. An advantage of the ion trap instrument on the other hand was the higher mass resolution (2500-3500) up to the m/z 2000 range of the fragment ions compared to the TOF/RTOF instrument and a narrower precursor ion selection window. These observed properties and resulting spectrum qualities of each fragmenta-
tion method were independent of the selected precursor ions. But as it is evident from the examples in Figures 3 and 4, fragmentation of a peptide derived from tryptic digestion of an abundant protein generally generated a higher number of fragment ions. Table 1 summarizes the number of fragment ions obtained by the different instrumental techniques and which were further used for database searches of each selected precursor ion. In nine cases, high energy CID yielded, independent of the nature of the protein, the highest number of generated fragment ions. Using the PSD mode, the number of gained fragment ions turned out to be as expected lower. The number of fragment ions obtained by low energy CID was quite varying. In six cases, the number of fragment ions generated with the AP-MALDI instrument was comparable to the number of fragment ions obtained by PSD. Low-energy CID fragmentation of two further precursor ions, 1566.9 and 1650.9, derived from tryptic digestion of a protein recovered in both fractions Journal of Proteome Research • Vol. 5, No. 8, 2006 1971
research articles
Mayrhofer et al.
Figure 3. Comparison of the (A) vMALDI PSD spectrum, the (B) vMALDI high energy CID spectrum and the (C) AP-MALDI low energy CID spectrum of the singly protonated precursor ion at m/z 1206.7 produced from the PMF analysis of the protein spot marked in Figure 2A. For the PSD as well as the high energy CID spectrum 4000 unselected laser shots were summed. 100% intensity corresponds to (A) 7.8 and (B) 8.1 mV. The accumulation time for the low energy CID spectra was 7.7 min corresponding to 4620 laser pulses. The S/N ratios for the peak with m/z 428 are (A) 7.0, (B) 9.5, and (C) 5.7.
and a protein recovered in the detergent (hydrophobic) phase, respectively, generated clearly the highest number of fragment ions which were subsequently used for further database 1972
Journal of Proteome Research • Vol. 5, No. 8, 2006
searches. An exceptional case was the TOF/RTOF mass spectrometric analysis of a peptide precursor ion with m/z 2929.6. Whereas PSD and high energy CID gained sufficient fragment
V-MALDI and AP-MALDI-MS/MS of Glomerular Proteins
research articles
Figure 4. Fragmentation spectra of the tryptic peptide [M + H]+ at m/z 1919.1, produced from in-gel digestion of the protein marked in Figure 2B. The spectra have been acquired using (A) vMALDI PSD, (B) vMALDI high energy CID and (C) AP-MALDI low energy CID conditions. For the PSD as well as the high energy CID spectra, 3000 unselected laser shots were summed. 100% intensity corresponds to (A) 2.6 mV and (B) 4.1 mV. The accumulation time for the low energy CID spectra was 7.2 min corresponding to 4320 laser pulses. The S/N ratios for the peak with m/z 925 are (A) 15.5, (B) 17.1 and (C) 5.2.
ions, low energy CID resulted in hardly any fragmentation. This points out the limit of ion trap instruments for analyzing ions with higher m/z values as well as singly charged precursor ions
and may be a drawback for analyzing membrane proteins, because tryptic digestion of proteins containing hydrophobic domains often result in large peptides with higher m/z values. Journal of Proteome Research • Vol. 5, No. 8, 2006 1973
research articles
Mayrhofer et al.
Table 3. Comparison of Database Search Results for High Energy CID, PSD and Low Energy CID Spectra Using Two Public Available Search Engines vMALDI high energy CID precursor ion (observed) [MH]+a
1206.7 1529.9 1556.7 1566.9 1588.9 1396.9 1650.9 1919.1 1606.9 2929.6 1954.1 1790.8
peptide sequenceb
(K) HLSVNDLPVGR (S) (R) FLQEYFDGNLKR (Y) (K) TEGVYRVELDTK (S) (R) ITPSYVAFTPEGR (L) (K) KSDIDEIVLVGGSTR (I) (R) ILFRPVASQLPR (I) (R) LVLEVAQHLGESTVR (T) (K) VLDSGAPIKIPVGPETLGR (I) (R) KLEAAEDIAYQLSR (S) (K) VSDAISTQYPVVDHEFDAVVVGAGGAGLR (A) (R) VAPEEHPVLLTEAPLNPK (A) (K) SYELPDGQVITIGNER (F)
no. of ions/ no. of unmatched ionsc
scored
vMALDI PSD no. of ions/ no. of unmatched ionsc
scored
AP-MALDI low energy CID no. of ions/ no. of unmatched ionsc
scored
23/0 rank 1/1
49
15/0 rank1/1
55
16/6 rank 1/2
18e
26/1 rank 1/1
37
13/0 rank 1/1
32
15/4 rank 1/4
14e
30/3 rank 1/1
29
24/2 rank 1/1
33
9/3 rank 2/10
no ID
30/3 rank 1/1
33
23/2 rank 1/1
32
49/23 rank 1/1
32
30/0 rank 1/1
55
10/0 rank 1/1
53
28/2 rank 1/1
95
24/2 rank 1/1
13e
17/1 rank 1/1
12e
24/12 rank 1/1
8e
29/1 rank 1/1
59
16/1 rank 1/1
56
50/20 rank 1/1
55
40/2 rank 1/1
56
31/2 rank 1/1
83
29/7 rank 1/1
64
15/0 rank 1/1
21e
11/0 rank 1/1
60
13/5 rank 1/1
15e
28/2 rank 1/1
90
22/2 rank 1/1
83
3
30/0 rank 1/1
75
21/2 rank 1/1
64
25/4 rank 1/1
58
35/2 rank 1/1
48
20/2 rank 1/1
69
20/4 rank 1/1
44
no ID
a m/z derived from vMALDI PMF analysis. b Peptide sequence identified by search engines. c Data retrieved by the MS-Tag search algorithm. d Probabilitybased Mowse score reported by the MASCOT search algorithm. e No significant Mowse score.
The applicability of the selected fragmentation methods in terms of a real high throughput approach is limited at the moment, because the acquisition of CID spectra is rather timeconsuming with the used low repetition rate laser systems. Database Search Derived Identification of Proteins. To evaluate the influence of different fragmentation patterns on database search derived identification of proteins, the obtained data were searched against the Swiss-Prot or NCBInr database using the MASCOT as well as the MS-Tag search algorithm. The applied search parameters are described in detail in Table 2 and the generated results in Table 3. The determined peptide sequences and the probability-based Mowse scores using the MASCOT search algorithm for each precursor ion have been included in the table. Also included were the number of input fragment ions and the number of ions that were not matched to the peptide sequence using the MS-Tag search algorithm. In summary, the database searches with the PSD and the corresponding high energy CID spectra using the MS-Tag search algorithm led to the identification of the same peptide sequence as the unique hit for each analyzed precursor ion. Using similar parameters for the low energy CID spectra, the data could not be matched to any peptide sequence, with the exception of one case. Only after increasing the number of 1974
Journal of Proteome Research • Vol. 5, No. 8, 2006
possible unmatched ions to 50%, the corresponding low energy CID spectra enabled in 10 cases the identification of the same peptide sequence ranked as the first hit as well. In one case the peptide sequence (K)TEGVYRVELDTK(S) was ranked only as the second hit and in one case the low energy CID spectrum was not useable due the lack of fragment ions. Using the MASCOT search program, 10 high energy CID spectra allowed the identification of the peptide sequences with significant probability-based Mowse scores. The high energy CID data of two precursor ions (m/z 1396.9 and m/z 1606.9), both derived from tryptic digestion of a membrane protein led unfortunately to scores below the 95% significance level given by the MASCOT program. With one exception, the PSD spectra led to the unambiguous identification of the peptide sequences with significant MASCOT scores. The interpretation of six low energy CID spectra yielded either no results, due the lack of fragment ions or no significant MASCOT scores as examplified in Figure 5C. The assigned fragment ions, as indicated by the MS-Tag and the MASCOT search algorithm of (A) the PSD, (B) the high energy CID, and (C) the low energy CID spectrum of the precursor ion with m/z 1206.7 (the peptide precursor ion was derived from tryptic digestion of a cytosolic protein (marked in Figure 2A)), are shown. The probability-based
V-MALDI and AP-MALDI-MS/MS of Glomerular Proteins
research articles
Figure 5. Comparison of fragment ions generated by (A) PSD, (B) high energy CID and (C) low energy CID conditions of the tryptic peptide ([M + H]+ at m/z 1206.7). Fragment ions are labeled according to the MS-Tag search results. The peptide sequence (K)HLSVNDLPVGR(S) was retrieved as the top hit. The corresponding MASCOT scores are shown for each spectrum.
Mowse scores were 49 and 55 for the high energy CID and the PSD spectrum, respectively. These scores were high enough for database search derived identification of the low abundant silver stained protein. However, the corresponding low energy CID (MS/MS) experiment performed on an ion trap instrument led to a MASCOT score (18) below the 95% significance region. Obviously, the total number of fragment ions used for the database search seems to have no significant influence on the search results, since the number of fragment ions obtained by low energy CID was comparable to that number of fragment ions obtained by PSD. Interestingly, 6 of the 16 fragment ions were missed in the search performed with the low energy CID spectrum, which might be the reason for the rather low score. This consideration is supported by two observations. First, also
in other low energy CID experiments low MASCOT scores were linked to a relatively high number of unassigned fragment ions, as indicated in Table 3. Second, in the case of the low energy CID spectrum of the precursor ion m/z 1588.9 nearly all fragment ions were assigned and thereby the highest MASCOT score was achieved compared to the other fragmentation techniques. To exclude the possibility that the number of unassigned fragment ions were linked to the selected instrument type in the program, which determines already the considered ion types, the searches with the low energy CID spectra were also performed selecting “TOF/TOF” or alternatively “TOF-PSD” as instrument type. But this change of the parameter setting did not influence the outcome of the searches regarding the Mowse score. Database searches using the MSJournal of Proteome Research • Vol. 5, No. 8, 2006 1975
research articles
Mayrhofer et al.
Table 4. Types of Fragment Ions Generated by the Three Different Applied Mass Spectrometric Techniques and Assigned by Public Available Search Engines vMALDI high energy CID
vMALDI PSD
AP-MALDI low energy CID
precursor ion (observed) [MH]+a
main types of fragment ions
main types of fragment ions
main types of fragment ions
1206.7
immonium ions, internal fragment ions, a, y, b ions, side chain cleavage ions immonium ions, internal fragment ions, a, b, y ions, side chain cleavage ions immonium ions, internal fragment ions, b, y, ions, side chain cleavage ions immonium ions, internal fragment ions, b, y ions, side chain cleavage ions immonium ions, internal fragment ions, a, b, y ions, side chain cleavage ions immonium ions, a, b, y ions side chain cleavage ions immonium ions, internal fragment ions, b, y ions, side chain cleavage ions immonium ions, internal fragment ions, b, y ions, side chain cleavage ions immonium ions, internal fragment ions, b, y ions immonium ions, internal fragment ions, b, y ions immonium ions, internal fragment ions, a, b, y ions immonium ions, internal ions, a, b, y, ions, side chain cleavage ions
a, b, y ions neutral loss ions
a, b, y ions neutral loss ions
b, y ions
b, y ions neutral loss ions
internal fragment ions a, b, c, y ions
b, y ions neutral loss ions
internal fragment ions y ions
b, y ions neutral loss ions
a, b, y ions
b, y ions neutral loss ions
internal fragment ion, b, y ions b, y ions
b, y ions neutral loss ions b, y ions neutral loss ions
internal fragment ions, b, y ions, neutral loss ions
b, y ions neutral loss ions
b, y ions
b, y ions neutral loss ions
1529.9
1556.7
1566.9
1588.9
1396.9 1650.9
1919.1
1606.9 2929.6 1954.1 1790.8
a
internal fragment ions y ions internal fragment ions, b, y ions internal fragment ions, b, y ions
b, y ions neutral loss ions b, y ions neutral loss ions
m/z derived from vMALDI PMF analysis.
Tag algorithm revealed that a few additional internal fragment ions were assigned to the low energy CID data in some particular cases (data not shown). Therefore, one explanation for this observation might be that part of these unassigned fragment ions result from multiple losses of water, CO or ammonium which are not considered in the search algorithms. Accordingly, the main types of fragment ions assigned by the MS-Tag search algorithm have been evaluated. Table 4 gives a summary of the main fragment ion types generated by the different applied ion dissociation methods of each precursor ion. Figure 6 exemplifies the assigned fragment ions of (A) the PSD, (B) the high energy CID, and (C) the low energy CID spectrum of a precursor ion with m/z 1919.1 (which was generated from tryptic digestion of an abundant membrane protein (marked in PMF with an asterisk in Figure 2D,F)). The PSD spectrum contained predominantly b- and y-type ions, including neutral losses and internal fragment ions. High energy CID of this precursor ion with an inert gas generated preferentially b- and y-type ions and further low molecular weight ions, namely immonium ions, which indicate the presence of specific amino acids in the peptide sequence. Additionally, the vMALDI high energy CID spectrum contained internal fragment ions and side chain cleavage fragment ions, such as w- and v-type ions. The CID spectra acquired on the AP-MALDI IT instrument contained preferentially b- and y-type ions including a high number of ions showing neutral losses. 1976
Journal of Proteome Research • Vol. 5, No. 8, 2006
In this case, all three spectra allowed the unambiguous identification of the peptide sequence. Interestingly, despite high energy CID produced a more complex spectrum which provided more sequence specific information the PSD spectrum clearly yielded the highest MASCOT score. The described observation can be attributed to the applied commercial search algorithms, which are not designed for high energy CID spectra. This effect could also be observed by fragmentation of two other peptide ions (m/z 1606.9 and 1790.8). Similar observations have been described for the laser-induced dissociation (LID) and CID spectra of model peptides with higher m/z by Macht et al.32 It has been suggested that this effect might be a result of many additional signals gained by CID which cannot be ascribed properly by the applied search algorithm. As it is shown in Figure 6B and Figure 5B, a number of fragment ions obtained by high energy CID could not be unambiguous assigned by the search algorithm and accordingly the peaks were labeled with multiple structural possibilities. But also some fragment ions in the PSD spectra (Figure 5A and Figure 6A) could not be unambiguous assigned by the MS-Tag search algorithm. Nevertheless, a higher MASCOT score was yielded. According to this, another explanation might be that although the selected MALDI-TOF/TOF specific parameter setting considers more ion types, the assigned immonium ions, which indicate the presence of specific amino acids in the peptide sequence, and side chain cleavage ions have less impact on
research articles
V-MALDI and AP-MALDI-MS/MS of Glomerular Proteins
Figure 6. Fragment ions obtained by (A) vMALDI PSD, (B) vMALDI high energy CID and (C) AP-MALDI low energy CID fragmentation of the tryptic peptide ([M + H]+ at m/z 1919.1). Fragment ions are labeled according to the MS-Tag search results. The peptide sequence (K)VLDSGAPIKIPVGPETLGR(I) was retrieved as the only significant hit. The corresponding MASCOT scores are shown for each spectrum.
the search result than the b- or y- type ions, which are in both spectra the main fragment ion types. The immonium ions generated by vMALDI high energy CID might be more helpful for de novo sequencing than for database search identification. AP-MALDI low energy CID fragmentation generated preferentially b- and y-type ions, but as it has become evident from this study a major limitation in some cases regarding data interpretation is that a number of generated fragment ions, which might result from multiple neutral losses, are not considered using the selected public accessible search algorithms. Using the MASCOT search algorithm, this fact (multiple loss of neutrals) has a higher influence on the score, whereas
in the case of MS-Tag it has to be incorporated into the setting of the parameters. Therefore, the AP-MALDI data enabled in 50% of the analyzed peptides either no or ambiguously identification due to the lack of significant scores using the MASCOT search algorithm.
Conclusions Proteomic approaches are now widely established, enabling a profound knowledge of biological processes. Thereby, the unambiguous identification of the proteins, which provides the basis for further biological or medical research, is an essential step and thereby MALDI-MS in its various forms has become Journal of Proteome Research • Vol. 5, No. 8, 2006 1977
research articles an indispensable tool. Here, we have presented the application of different laser-based desorption/ionization techniques combined with ion activation methods for the database search derived identification of hydrophilic as well as hydrophobic two-dimensional gel electrophoretic separated and trypsin digested glomerular proteins. Defined differences in the individual fragmentation patterns, which could be observed, include the number of generated fragment ions, the main types of fragment ions and the resolution of the fragment ions. The observed properties of each fragmentation method were independent of the origin of the selected precursor ions. Using two public accessible search algorithms and databases, minor differences in the search results using PSD and high energy CID data could be observed. These differences are mainly expressed in slightly varying scores for individual peptides. Remarkably, although high energy CID has provided more additional sequence specific information, the information obtained by PSD seemed to be sufficient and in some cases even more successful for database search derived identification due to the fact that available search algorithms were until now not optimized for high energy CID data. Nevertheless, both represent appropriate tools for the generation of sequence tags from singly protonated ions even of high mass peptides. The performed AP-MALDI low energy CID experiments of tryptic peptides present definitively an alternative to vMALDI-MS/ MS experiments, particularly considering that AP-MALDI sources can easily be retrofitted to existing ESI-IT instruments. Furthermore, this technique is also applicable to the analysis of two-dimensional separated silver stained proteins. But, obviously, a major limitation using AP-MALDI low energy CID for database search derived identification is that a number of fragment ion types derived from monoprotonated precursor ions are not considered in the public available search algorithms and thereby allow no straightforward automatic identification of the protein due peptide sequence tags. This feature has to be kept in mind using CID combined with AP-MALDI as precursor ion generating technique and highlights further the relevance of the selected search parameters and of necessary software improvements. Because the definition of the types of fragment ions is of great importance in matching the data to a sequence, the characteristic fragment ion types generated by different methods represent an important factor for database search derived identification. In addition, the interpretation of the individual fragmentation pattern is qualified by the properties of the chosen search algorithm. Knowledge of the idiosyncrasies of each of the three applied techniques in combination with publicly available database search algorithms will be useful for data interpretation and for further proteomic approaches.
Acknowledgment. This work was supported by Grant No. P15008 (to G. A.) from the Austrian Science Foundation and the EC Project QLG1-CT2002-01215. References (1) Tanaka, K. I.; Y.; Akita, S.; Yoshida, Y.; Yoshida, T. In Proceedings of the 2nd Japan-China Joint Symposium on Mass Spectrometry, Osaka, Japan, 1987; 185-187. (2) Karas, M.; Hillenkamp, F. Anal. Chem. 1988, 60 (20), 2299-2301. (3) Brown, R. S.; Lennon, J. J. Anal. Chem. 1995, 67 (13), 1998-2003. (4) Colby, S. M. K. T. B.; Reilly, J. P. Rapid Commun. Mass Spectrom. 1994, 8 865-868. (5) Vestal, M. J.; P.; Martin, S. A. Rapid Commun. Mass Spectrom. 1995, 9 1044-1050.
1978
Journal of Proteome Research • Vol. 5, No. 8, 2006
Mayrhofer et al. (6) Whittal, R. M.; Li, L. Anal. Chem. 1995, 67 (13), 1950-1954. (7) Jensen, O. N.; Podtelejnikov, A.; Mann, M. Rapid Commun. Mass Spectrom. 1996, 10 (11), 1371-1378. (8) Mamyrin, B. A.; Karatajev, V. J.; Shmikk, D. V.; Zagulin, V. A. Sov. Phys. JETP 1973, 37, 45-48. (9) Medzihradszky, K. F.; Campbell, J. M.; Baldwin, M. A.; Falick, A. M.; Juhasz, P.; Vestal, M. L.; Burlingame, A. L. Anal. Chem. 2000, 72 (3), 552-558. (10) Doroshenko, V. M.; Cotter, R. J. J. Mass Spectrom. 1998, 33 (4), 305-318. (11) Shevchenko, A.; Loboda, A.; Shevchenko, A.; Ens, W.; Standing, K. G. Anal. Chem. 2000, 72 (9), 2132-2141. (12) Lee, H.; Lubman, D. M. Anal. Chem. 1995, 67 (8), 1400-1408. (13) Castoro, J. A.; Koster, C.; Wilkins, C. L. Anal. Chem. 1993, 65 (6), 784-788. (14) Annan, R. S.; Kochling, H. J.; Hill, J. A.; Biemann, K. Rapid Commun. Mass Spectrom. 1992, 6 (4), 298-302. (15) Doroshenko, V. M.; Cotter, R. J. Anal. Chem. 1995, 67 (13), 21802187. (16) Qin, J.; Chait, B. T. Anal. Chem. 1996, 68 (13), 2108-2112. (17) Qin, J.; Chait, B. T. Anal. Chem. 1996, 68 (13), 2102-2107. (18) Qin, J.; Ruud, J.; Chait, B. T. Anal. Chem. 1996, 68 (10), 17841791. (19) Laiko, V. V.; Baldwin, M. A.; Burlingame, A. L. Anal. Chem. 2000, 72, (4), 652-657. (20) Laiko, V. V.; Moyer, S. C.; Cotter, R. J. Anal. Chem. 2000, 72 (21), 5239-5243. (21) Henzel, W. J.; Billeci, T. M.; Stults, J. T.; Wong, S. C.; Grimley, C.; Watanabe, C. Proc. Nat’l. Acad. Sci., U.S.A. 1993, 90 (11), 50115015. (22) James, P.; Quadroni, M.; Carafoli, E.; Gonnet, G. Biochem. Biophys. Res. Commun. 1993, 195 (1), 58-64. (23) Mann, M.; Hojrup, P.; Roepstorff, P. Biol. Mass Spectrom. 1993, 22 (6), 338-345. (24) Pappin, D. J.; Hojrup, P.; Bleasby, A. J. Curr. Biol. 1993, 3 (6), 327-332. (25) Yates, J. R., 3rd; Speicher, S.; Griffin, P. R.; Hunkapiller, T. Anal. Biochem. 1993, 214 (2), 397-408. (26) Spengler, B.; Kirsch, D.; Kaufmann, R.; Jaeger, E. Rapid Commun. Mass Spectrom. 1992, 6 (2), 105-108. (27) Kaufmann, R.; Chaurand, P.; Kirsch, D.; Spengler, B. Rapid Commun. Mass Spectrom. 1996, 10 (10), 1199-1208. (28) Johnson, R. M.; S. A.; Biemann, K.; Stults, J. T.; Watson, J. T. Anal. Chem. 1987, 59 (21), 2621-2625. (29) Johnson, R. M.; S. A.; Biemann, K. Int. J. Mass Spectrom. Ion Processes 1988, 86, 137-154. (30) Cramer, R.; Corless, S. Rapid Commun. Mass Spectrom. 2001, 15 (22), 2058-2066. (31) Kaufmann, R.; Spengler, B.; Lutzenkirchen, F. Rapid Commun. Mass Spectrom. 1993, 7 (10), 902-910. (32) Macht, M.; Asperger, A.; Deininger, S. O. Rapid Commun. Mass Spectrom. 2004, 18 (18), 2093-2105. (33) Rouse, J. Y.; W.; Martins, S. A. J. Am. Soc. Mass Spectrom. 1995, 6 (9), 822-835. (34) Kaufmann, R.; Kirsh, D.; Spengler, B. Int. J. Mass Spectrom. Ion Proc. 1994, 131, 355. (35) Doroshenko, V. M.; Laiko, V. V.; Taranenko, N. I.; Berkout V. D.; Lee H. S. Int. J. Mass Spectrom. 2002, 221, 39-58. (36) Levander, F.; James, P. J. Proteome Res. 2005, 4 (1), 71-74. (37) Mehl, J. T.; Cummings, J. J.; Rohde, E.; Yates, N. N. Rapid Commun. Mass Spectrom. 2003, 17 (14), 1600-1610. (38) Wang-Su, S. T.; McCormack, A. L.; Yang, S.; Hosler, M. R.; Mixon, A.; Riviere, M. A.; Wilmarth, P. A.; Andley, U. P.; Garland, D.; Li, H.; David, L. L.; Wagner, B. J. Invest. Ophthalmol. Vis. Sci. 2003, 44 (11), 4829-4836. (39) Keough, T.; Lacey, M. P.; Strife, R. J. Rapid Commun. Mass Spectrom. 2001, 15 (23), 2227-2239. (40) Breiteneder-Geleff, S.; Matsui, K.; Soleiman, A.; Meraner, P.; Poczewski, H.; Kalt, R.; Schaffner, G.; Kerjaschki, D. Am. J. Pathol. 1997, 151 (4), 1141-1152. (41) Blum, H.; Beier, H.; Gross, H. J. Electrophoresis 1987, 8, 93-99. (42) Gharahdaghi, F.; Weinberg, C. R.; Meagher, D. A.; Imai, B. S.; Mische, S. M. Electrophoresis 1999, 20 (3), 601-605. (43) Vorm, O.; Roepstorff, P.; Mann, M. Anal. Chem. 1994, 66, 32813287. (44) Mayrhofer, C.; Krieger, S.; Allmaier, G.; Kerjaschki, D. Proteomics 2006, 6 (2), 579-585.
PR060165S
