Simple and Robust Two-Layer Matrix/Sample Preparation Method for MALDI MS/MS Analysis of Peptides Jing Zheng,† Nan Li,† Marc Ridyard,‡ Hui Dai,† Stephen M. Robbins,‡ and Liang Li*,† Department of Chemistry, University of Alberta, Edmonton, Alberta, Canada T6G 2G2, and Departments of Oncology and Biochemistry & Molecular Biology, University of Calgary, Calgary, Alberta, Canada T2N 4N1 Received May 29, 2005
Recent advances in MALDI MS/MS instrumentation allow a high degree of automation in the efficient detection of peptide fragment ions that can be used for protein identification. However, the performance of the technique is dependent on the MALDI sample preparation. We present a simple and robust two-layer sample preparation method tailored for sensitive and reproducible generation of MALDI MS/ MS data. This method produces a strong and uniform crystal layer which allows acquisition of high quality MS/MS spectra over the entire sample surface area. Furthermore, due to its crystal strength, the matrix/sample layer can be washed extensively on target, enabling direct analysis of samples containing impurities, such as salts and surfactants. This method is demonstrated to be very useful in routine analysis of in-gel tryptic digests of silver-stained protein gel spots, without the need of desalting steps or hunting for “hot” spots. As an example, seven threonine-phosphorylated proteins involved in signal transduction in response to growth factor stimulation within the lipid raft fractions of the IMR5 neuroblastoma cells have been identified using differential gel display, in-gel digestion and MALDI MS/MS with the new two-layer sample preparation method. Some of these proteins have the functions of maintaining raft structure or cell signaling. Keywords: MALDI MS/MS • matrix/sample preparation • two-layer • dried-droplet • lipid raft proteome
Introduction With the availability of commercial instrumentation for matrix-assisted laser desorption/ionization (MALDI) tandem mass spectrometry (MS/MS), the use of MALDI MS/MS for peptide sequencing has become a valuable tool for proteome analysis.1-13 Having different ionization and fragmentation mechanisms, MALDI can provide complementary information and sometimes unique advantages over electrospray ionization (ESI) MS/MS.14-17 However, unlike ESI, where precursor ion signals are usually stable during fragment ion spectrum acquisition, a large variation in ion signals can occur during the MALDI ionization step. This is due to shot-to-shot signal variations caused mainly by inhomogenity of the analyte distribution in matrix/sample crystals prepared on the MALDI target.18 Development of a robust sample preparation method to overcome this problem would therefore facilitate MALDI MS/MS spectra acquisition. There are many different methods reported for MALDI sample preparation, mainly developed for MALDI MS experiments.19-36 For MALDI MS/MS, dried-droplet methods and, to a lesser extent, a two-layer method, have been used.37,38 In the dried-droplet methods, either 2,5-dihydroxybenzoic acid (DHB) or R-cyano-4-hydroxycinnamic acid (CHCA) is mixed * To whom correspondence should be addressed. E-mail: Liang.Li@ ualberta.ca. † University of Alberta. ‡ University of Calgary. 10.1021/pr050157w CCC: $30.25
2005 American Chemical Society
with a sample at an optimal ratio and then directly spotted onto the MALDI target plate. The major drawback of this method is that it produces uneven sample/matrix crystals and thus requires the search for “hot” spots to generate good quality spectra. In the two-layer method, a thin CHCA layer is first spotted onto the MALDI target. Then matrix/sample mixture is deposited on top of this thin layer to form the second layer. The resultant sample spots are smooth and analyte molecules are evenly distributed on the entire sample surfaces, leading to a more reproducible MS detection. This two-layer method was originally designed for MALDI MS analysis of peptides and proteins.18 However, in MALDI MS/MS experiments, where a much stronger laser power is used in order to generate a sufficient number of precursor ions for fragmentation, the matrix/sample spots formed through this original two-layer method are easily consumed. This results in a reduction of the overall detection sensitivity and therefore the sequencing efficiency for peptides. In this work, the original two-layer method has been modified with a new second-layer solution formulation tailored to MALDI MS/MS sequencing experiments. Instead of using methanol as the organic solvent in the second layer solution used for MALDI MS, acetonitrile (ACN) is now employed. The effects of solvent conditions on the performance of the twolayer method for MALDI MS/MS are illustrated. Comparisons of this new method with the commonly used dried-droplet method and the original two-layer method are presented. This new method is demonstrated to offer unique advantages in Journal of Proteome Research 2005, 4, 1709-1716
1709
Published on Web 08/25/2005
research articles analyzing “dirty” samples without any further purification steps. We also foresee wide applications of this protocol in automated MALDI MS/MS.
Experimental Section Materials. Matrices used in MALDI MS and MS/MS analysis were 2,5-dihydroxybenzoic acid (DHB) and R-cyano-4-hydroxycinnamic acid (CHCA). DHB was obtained from Aldrich (Milwaukee, WI) with 99% purity. CHCA, NH4HCO3, bovine trypsin and bovine serum albumin (BSA) were obtained from Sigma (St. Louis, MO). CHCA was purified by ethanol recrystallization. Trifluoroacetic acid (TFA) and CaCl2 were purchased from Fisher (Fair Lawn, NJ). Sodium dodecyl sulfate (SDS) and all reagents for gel electrophoresis were from Bio-Rad Laboratories (Hercules, CA). All organic solvents were HPLC grade from Fisher. The water used for all the experiments was obtained from a MilliQ UV plus water purification system from Millipore (Bedford, MA). Standard peptide, Glu-Fibrinopeptide B (Glu-Fib, EGVNDNEEGFFSAR), was from MDS Sciex (Toronto, ON, Canada). Isolation of Detergent-Insoluble Low-Density Fractions. IMR5 neuroblastoma cells were grown to confluence on 150 mm diameter dishes. Cells were rinsed in ice cold PBS and lysed in 1% Triton buffer (1% Triton X-100, 25 mM MES, 150 mM NaCl, pH 6.5). The cells were fractionated on sucrose gradients and low-density, detergent-insoluble fractions were isolated as described in detail previously.39 Two-Dimensional Gel Electrophoresis. Lipid raft fractions isolated from the sucrose gradient experiment were solubilized in thiourea lysis buffer (7 M urea, 2 M thiourea, 4% CHAPS, 1% DTT, 2% ampholytes). After brief vortex, samples were centrifuged at 100 000 × g for 15 min and the supernatants were run on 3.4 mm × 100 mm isoelectric focusing tube gels using the BioRad Model 175 Tube Cell system. Tube gels were made with 5% acrylamide, 2% NP-40, 2% ampholytes and 9 M urea. 0.01 M H3PO4 was used as anolyte and 0.1 M NaOH was used as catholyte. Samples were loaded at the gel cathode and focused for 12 000 Vhr. After focusing, tube gels were equilibrated with SDS buffer (0.05 M Tris-HCl, 6 M urea, 30% glycerol, 2% SDS, 65 mM DTT) for 45 min followed by the second dimension on 10% SDS-PAGE gels. Gels were Western blotted using 1:1000 dilution of anti-phosphothreonine-proline antibody (New England Biolabs, Pickering, ON, Canada) or silver stained. In-Gel Digestion. Protein gel bands/spots were excised and placed into the wells of a Millipore multiscreen filter plate (Millipore, Bedford, MA). The gel pieces were washed by double distilled (dd) H2O, shrunk by ACN and dried in a SpeedVac. The gel pieces were then covered with 9 mM DTT in 100 mM NH4HCO3 and reduced at 50 °C for 30 min followed by carbamidomethylation with 20 mM iodoacetamide in 100 mM NH4HCO3 for 30 min in dark. Shrinking and drying steps were then repeated. The dried gel pieces were finally covered with 2 mM CaCl2 and 10 ng/µL trypsin in 100 mM NH4HCO3. The proteins were digested overnight at 30 °C. Peptides in the gel were extracted and eluted into a v-bottom multititer plate (Corning Incorporated, Corning, NY) by 50% ACN in 0.15% TFA three times with 20 min shaking. The pooled extract was dried in a SpeedVac. Matrix/Sample Preparation. Both two-layer and drieddroplet sample/matrix preparations were employed in this study. In the dried-droplet method, samples were mixed with CHCA or DHB matrix solutions first and 1 µL sample/matrix 1710
Journal of Proteome Research • Vol. 4, No. 5, 2005
Zheng et al.
mixture was deposited onto the sample plate and let dry. The matrix composition in the final sample/matrix mixture was 10 mg/mL CHCA in 50% ACN/water (v/v) or 40 mg/mL DHB in 25% ACN/water (v/v). In the two-layer preparations, only CHCA was used. The twolayer method is not applicable to DHB because DHB does not dissolve in an appropriate organic solvent, such as acetone, to form a thin layer by fast evaporation. This method involves the use of two matrix solutions. The first-layer matrix solution in all two-layer preparations was composed of 20 mg/mL CHCA in 90% acetone/methanol (v/v). The second layer matrix solution of the conventional two-layer method used for MALDI MS consisted of saturated CHCA in 30-40% methanol/water (v/v) in the final sample/matrix mixture. To search for an optimal two-layer method tailored for MALDI MS/MS work, solvent type and solvent composition of the second layer solution were varied (see Results and Discussion). In all cases, 0.5 µL of the first layer matrix solution was first applied to a MALDI plate to form a thin matrix layer. Then 1 µL of the second layer matrix solution mixed with the sample was deposited onto the first layer and allowed to dry. All spots prepared from the two-layer method, except the one for the comparison of washing effect, were washed on target at least three times with water. It should be noted that the sample-tomatrix ratio can be altered as long as the final composition is kept as described above or in Results and Discussion. In the analysis of peptides extracted from the gels, peptide extracts placed in the multititer plate were first SpeedVaced to dryness. Then 10-30 µL of 50% ACN saturated by CHCA was added into plate wells to reconstitute the peptides by 20 min agitation. After adding 2-6 µL of H2O to the vial and spinning down the matrix crystal, the supernatant was spotted onto the first-layer and allowed to dry, followed by the on-target wash steps. MALDI MS and MS/MS. MALDI MS experiments were carried out on a Bruker REFLEX III time-of-flight (TOF) mass spectrometer (Bremen/Leipzig, Germany). MALDI MS/MS analyses were performed on a MDS Sciex API-QSTAR pulsar QqTOF mass spectrometer with a MALDI source (MDS Sciex, Toronto, ON, Canada). Microscope Imaging. Microscope imaging was carried out on a QX3 computer microscope with 60× magnification (Mattel Inc., China).
Results and Discussion Matrices and Sample Preparation Methods. Over the years numerous matrix candidates have been evaluated for MALDI applications. However, DHB and CHCA have become the most commonly used matrices for the analysis of peptides.40,41 In this work, we mainly focus on the comparison of these two matrices for MALDI MS/MS. For matrix/sample preparation, because of the ease of solution preparation and sample deposition, the dried-droplet method is widely used for MALDI MS/MS. On the other hand, the application of the two-layer sample preparation method in MALDI MS/MS is limited due to the low matrix crystal strength, despite its advantages such as providing good signal reproducibility. Other methods have been developed to alleviate the shot-to-shot reproducibility problem associated with the dried-droplet method.19-36 In the present work, the conventional two-layer method designed for MALDI MS is further developed to meet the unique requirements of MALDI MS/MS of peptides. This method is compared to the widely used dried-droplet method. No attempt was made
Two-Layer Matrix/Sample Preparation Method
research articles
Figure 1. MALDI MS/MS spectra of 5 fmol Glu-Fib obtained by using (A) dried-droplet sample preparation with DHB as matrix and (B) two-layer sample preparation with CHCA as matrix. In the dried-droplet method, the final matrix composition was 40 mg/mL DHB in 25% ACN/H2O (v/v). In the two-layer method, the second-layer composition was saturated CHCA in 30% methanol. The fragment ion peaks which are labeled in the spectra can be distinguished from the noise peaks by examining their peak shapes; noise peaks is random lines and a fragment ion peak is a cluster of inter-connected lines centered at the m/z value of the peak.
to compare the two-layer method with other reported methods, such as the matrix-mixture preparation method,36 which uses a mixture of DHB and CHCA and was found to provide better performance for MALDI MS analysis of peptides and proteins than using either matrix alone. However, the matrix-mixture method was reported to require twice the laser power to generate ions from the rim of the sample spot than that used for generating ions from the center.36 The requirement of different laser powers to generate ions from the same sample spot would present a major challenge in automating MALDI MS/MS spectral acquisition. In addition, since DHB is used in the matrix mixture, on-probe washing, which is critical for reducing interference when analyzing real world sample including samples containing surfactants, cannot be used in the matrix-mixture preparation method. It is well-known that different matrices and different preparation methods can result in very different crystal morphology.18,36,42 For the DHB dried-droplet prepared sample, microscopic examination revealed large nonuniformities of crystal morphology and localization. In most cases, the long, sharp crystals existed mostly at the outside edge of the sample spot, whereas the spot interior was composed of fewer crystals. In general, smaller crystals and a more uniform matrix/sample layer were seen in the two-layer prepared samples. The
morphology of sample spots prepared from the CHCA drieddroplet method falls in between. It is worth noting that, although the morphology differences affect the laser power required for MS analysis of peptides and govern whether searching for the “hot” spots is required to obtain high-quality MS spectra, we observe similar MALDI MS detection sensitivity for peptides in these three preparations. In terms of mass resolution and accuracy, in the Qq-TOF instrument with an orthogonal MALDI source that can decouple the ionization event from the ion extraction and detection processes,2-4 they remain the same for all three preparations. However, in MALDI MS analysis using a TOF mass spectrometer, the resolution and mass measurement accuracy are generally inferior for the drieddroplet prepared samples, compared to the two-layer prepared samples.18 MALDI MS vs MALDI MS/MS. In contrast to the MALDI MS analysis of peptides, the performances of these matrix/sample preparation methods in MALDI MS/MS sequencing experiments are very different. This can be shown in the analysis of 5 fmol Glu-Fib spotted on a MALDI plate. In this experiment, an individual spot was totally consumed to determine how many good quality MS/MS spectra could be obtained from a spot. Panels A and B of Figure 1 show the representative MS/ MS spectra obtained by the DHB dried-droplet method and Journal of Proteome Research • Vol. 4, No. 5, 2005 1711
research articles
Zheng et al.
the original two-layer method, respectively, which are considered to be of good quality in terms of their fragment ion patterns and signal-to-noise ratios from the MS/MS database search point of view. It should be noted that much better quality spectra than those shown in Figure 1 could be readily obtained by adding more laser shots to each spectrum. However, increasing the number of the laser shots used to generate one spectrum would reduce the total number of spectra of similar quality that could be collected from one sample spot. Thus, in any proteome analysis work where sample amounts are limited, a minimum number of laser shots are used to generate a MALDI MS/MS spectrum from a peptide ion that is deemed to be of sufficient quality for MS/MS database searching. As soon as a database searchable MS/MS spectrum is generated, a new peptide ion is selected from the same spot for MS/MS. This process continues until the sample is consumed. In this study, all spectra collected from the same sample spot should have the same quality as those shown in Figure 1. From replicate experiments of 10 spots, with each spot containing 5 fmol Glu-Fib, 15 to 20 MS/MS spectra could be generated from one 5 fmol Glu-Fib spot prepared by the DHB dried-droplet method, whereas 5 to 8 MS/MS spectra could be obtained in the CHCA dried-droplet method, and only 2 to 6 spectra could be generated in the case of the CHCA twolayer method. This striking performance difference can be explained by the very different operating mechanisms in MALDI MS/MS compared to a MALDI MS experiment. In MALDI MS with a TOF mass spectrometer, by keeping the laser power low, usually just above the desorption/ionization threshold, and the laser beam focused to a small spot, optimal resolution can be achieved. In MALDI MS/MS, a much stronger laser power is used, and the laser beam size is much larger, at least in the Qq-TOF instrument. These measures are necessary to ensure that a sufficient number of peptide ions are being generated to pass onto a collision cell for dissociation. If only a small number of peptide ions are generated by the laser, then ion loss during transition and inefficiency of singly charged peptide ion fragmentation will result in a very small number of fragment ions that may not be registered as an analyte signal in fragment ion detection. Since higher laser power and larger laser beam size are used in MALDI MS/MS, the sensitivity of the MS/MS experiment is not only dependent on the success of cocrystallization of the sample and matrix, but also heavily dependent on the strength of the crystals. The original two-layer method with CHCA performs poorly because the sample/matrix spots generated from it are easily penetrated and consumed by the strong laser power. On the other hand, the DHB dried-droplet sample spots can withstand the high laser power, resulting in improved sensitivity in MALDI MS/MS. However, there are two major drawbacks with the DHB dried-droplet method for MALDI MS/MS. First, DHB is watersoluble and thus sample spots prepared with DHB cannot be washed on-target for salt removal. Therefore, reducing the salt content in the sample, such as by the use of a Ziptip, prior to MS detection is critical to acquire mass spectra successfully using DHB as the matrix. Second, it is often necessary to search for a “hot” spot that gives good analyte signals. As a result, it is crucial to develop a matrix/sample preparation method tailored to MALDI MS/MS sequencing experiments that provides high sensitivity and signal reproducibility. Our approach to achieve this goal was to modify the original CHCA two-layer method. 1712
Journal of Proteome Research • Vol. 4, No. 5, 2005
Figure 2. Effects of the second-layer solvent composition on MALDI sample preparation. The second-layer compositions were (A) saturated CHCA in 30% methanol, (B) saturated CHCA in 15% methanol and 15% ACN, (C) saturated CHCA in 30% ACN, and (D) saturated CHCA in 1: 2: 3 formic acid: 2-propanol: water. (E) and (F) show the images of the sample spots after acquiring 4 MS/MS spectra. The second-layer compositions are (E) saturated CHCA in 30% methanol and (F) saturated CHCA in 30% ACN.
New Two-Layer Method. In the original two-layer method designed for MALDI MS, the first layer is prepared by dissolving the CHCA matrix in 90% acetone/methanol (v/v) to a concentration of 20 mg/mL. This first layer is employed merely to establish a thin, but densely packed microcrystal layer on the MALDI plate, which can act as crystalline seeds for the growth of matrix/sample crystals after the addition of the second layer matrix/analyte solution. Casting a densely packed first layer can be readily achieved using the above matrix solution, so long as the surface of the MALDI plate is clean. Thus, our efforts in developing a two-layer method tailored to MALDI MS/MS were devoted to the modification of the second layer solution composition. We have examined a number of different solution compositions and some of the results are highlighted here. Figure 2A shows a sample spot prepared by the two-layer method where the second layer solution did not contain ACN, i.e., the original two-layer method. When ACN was added into the methanolbased second layer, the crystals of the sample spot were found to become more densely packed (see Figure 2B). When only ACN was employed as the organic solvent for the second layer preparation, the crystal layer was even denser and thicker (see Figure 2C). This ACN-based second layer was also found to be better than other second layer preparations, such as the “FPW” recipe (formic acid, 2-propanol, water)32,33 (see Figure 2D), in providing a densely packed and uniform second layer. The ability of these crystals to withstand high power laser irradiation, referred to as “crystal strength” in this discussion, was tested in MALDI MS/MS. Figures 2E and 2F show the sample spots after four MS/MS sequencing experiments. Each
Two-Layer Matrix/Sample Preparation Method
research articles
Table 1. MS/MS Performance Comparison with Different ACN Percentages in the Second-layer Solution ACN (%)
no. of MS/MS spectra from one 5 fmol Glu-Fib spot
45
6-8, spots are easily penetrated 10-14 15-20
