ionization tandem mass spectrometry for high-throughput proteomic analysis

Journal of Chromatography A, 1139 (2007) 191–198

Multidimensional capillary array liquid chromatography and matrix-assisted laser desorption/ionization tandem mass spectrometry for high-throughput proteomic analysis Chunli Liu a,b , Xiangmin Zhang b,∗ a

Department of Biochemistry and Molecular Biology, Second Military Medical University, Shanghai 200433, China b Department of Chemistry and Research Center for Proteome, Fudan University, Shanghai 200433, China Received 26 April 2006; received in revised form 2 November 2006; accepted 3 November 2006 Available online 28 November 2006

Abstract A two-dimensional capillary array liquid chromatography system coupled with matrix-assisted laser desorption/ionization (MALDI) mass spectrometry (MS) was developed for high-throughput comprehensive proteomic analysis, in which one strong cation-exchange (SCX) capillary chromatographic column was used as the first separation dimension and 10 parallel reversed-phase liquid chromatographic (RPLC) capillary columns were used as the second separation dimension. A novel multi-channel interface was designed and fabricated for on-line coupling of the SCX to RPLC column array system. Besides the high resolution based on the combination of SCX and RPLC separation, the developed new system provided the most rapid two-dimensional liquid chromatography (2D-LC) separation. Ten three-way micro-splitter valves used as stop-and-flow switches in transferring SCX fractions onto RPLC columns. In addition, the three-way valves also acted as mixing chambers of RPLC effluent with matrix. The system enables on-line mixing of the LC array effluents with matrix solution during the elution and directly depositing the analyte/matrix mixtures on MALDI plates from the tenplexed channels in parallel through an array of capillary tips. With the novel system, thousands of peptides were well separated and deposited on MALDI plates only in 150 min for a complex proteome sample. Compared with common 2D-LC system, the parallel 2D-LC system showed about 10-times faster analytical procedure. In combination with a high throughput tandem time of flight mass spectrometry, the system was proven to be very effective for proteome analysis by analyzing a complicated sample, soluble proteins extracted from a liver cancer tissue, in which over 1202 proteins were identified. © 2006 Elsevier B.V. All rights reserved. Keywords: Multidimensional liquid chromatography; Array; Proteomic analysis

1. Introduction Proteomics, the systematic study of all proteins expressed in cells, tissues or organisms, has placed tremendous demand for highly efficient analytical platforms for protein profiling of complex biological samples. Current proteomics is mostly based on two-dimensional gel electrophoresis (2D-GE) separations of proteins followed by mass spectrometric analysis [1–4]. 2D-GE is a powerful separation technique, which fulfils the two-dimensional orthogonal separation of proteins by isoelectric point and molecular weight, and allows the separation of

∗

Corresponding author. Tel.: +86 21 25070306 E-mail addresses: [email protected] (C. Liu), [email protected] (X. Zhang). 0021-9673/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2006.11.019

thousands of proteins in a single analysis. However, the technical limitations associated with 2D-GE, such as labor-intensive, relatively low throughput and difficulties in identifying lowabundance, more basic and large proteins, impeded thorough proteome characterization [5]. Chromatography is a highly efficient separation technique and can resolve most problems met in gel-based proteomic analyses. In recent years, many multidimensional analysis platforms based on liquid chromatography (LC) or capillary electrophoresis (CE) were investigated, including LC–LC [6–12], LC–CE/CE–LC [13–16], and CE–CE [17–19]. Mass spectrometry (MS) is playing a prominent role for protein identification, as a result of the development of two new ionization methods, electrospray ionization (ESI) and matrixassisted laser desorption/ionization (MALDI). Based on direct LC separation of proteolytic digests of proteins and tandem

192

C. Liu, X. Zhang / J. Chromatogr. A 1139 (2007) 191–198

mass spectrometry (MS/MS) analysis of peptides, shotgun proteomics approach has been proved to be a powerful tool to proteome profiling. Using 2D-LC coupled to ESI MS/MS, Yates’ group pioneered an automated technique for shotgun proteomics, termed multidimensional protein identification technology (MUDPIT) [20]. In MUDPIT, a biphasic column packed with two different types of packings (SCX and RPLC particles) was used for 2D-LC separation of protein digests, and ESI MS/MS was used for the proteins identification. By 15 SCX and RPLC elution cycles for each run of each fraction, in 83 h for three fractions, thousands of proteins were identified in the S. cerevisiae proteome [21]. Because ESI-MS can be directly coupled to LC, ESI-MS in conjunction with 2D-LC is increasingly applied for proteomics research [12,16,20–23]. For the relatively simple proteome, such as microbial and lower eukaryotic proteomes, the protein identification capacity of 2D-LC-ESI/MS is sufficient for the direct achievement of the comprehensive proteins characterization from the proteins mixture. However, for the extreme complex proteome, such as mammalian proteome, considerable co-elution of peptides in 2D-LC, ion suppression and limited spectral recording duty cycle of MS may yield an incomplete analysis. The approach to repetitive 2D-LC/ESI-MS analyses for one sample was taken to increase the coverage of proteome and achieve more comprehensive analysis [24], which led to increased sample consumption, and the analysis throughput further reduced. MALDI [25,26] has shown to be complementary to ESI in producing biomolecular ion for MS analysis [27]. In addition, MALDI–MS offers higher tolerance toward sample contaminants such as buffers, salts, and surfactants. In recent years, many coupling approaches for LC and MALDI have been reported [28–32]. By deposition of LC effluents as discrete spots or track on a plate, LC separation can be easily combined with MALDI–MS. For the offline coupling of LC separation to MALDI–MS, LC-MALDI-based shortgun approach enables more MS/MS analyses at a given point in a separation because the extended period can be devoted to the analysis, which facilitates improved proteome coverage. However, the highefficiency separations of complex sample prior to MALDI analysis achieved by the current 2D-LC are usually performed first by SCX fractionation followed by sequential RPLC elutions in the second dimension. The approach is time-consuming, and its throughput is significantly limited by running LC one by one [12,20–23]. The advantage of the multiple-column LC system recently became increasingly attractive in single-dimension separation [33–38]. To increase the throughput for multidimensional LC analysis, an online 2D array LC system using 10 capillary columns in parallel as the second separation dimension was first developed in this work. The 2D-LC array system allows the concurrent gradient elution of 10 fractions displaced in the first separation dimension, and therefore, the total chromatographic separation time for a proteomic sample was reduced by ca. 10-fold. In addition, the system enables on-line mixing of the effluents of tenplexed LC with matrix and depositing of the analyte/matrix mixtures from 10 channels in parallel by one spotting

system. With MALDI-time-of-flight (TOF)-TOF-MS, the performance of the system for high-throughput and high-resolution proteome analysis was demonstrated using the soluble fraction of D20 liver cancer tissue extracts. 2. Experimental 2.1. Materials and reagents Fused-silica capillaries (50 ␮m I.D., 375 ␮m O.D.; 75 ␮m I.D., 375 ␮m O.D.; 250 ␮m I.D., 375 ␮m O.D.; and 320 ␮m I.D., 450 ␮m O.D.) were from Yongnian Optical Fiber Factory (Yongnian, Hebei, China). Packing materials, 5 ␮m spherical silica gel Zorbax BP-Sil and 5 ␮m Zorbax 300 SB-C8 were obtained from DuPont (Wilmington, DE, USA), and Agilent Technologies (Palo Alto, CA, USA), respectively. Poros SCX packing was kindly supplied by PerSeptive Biosystems (Framingham, MA, USA). HPLC-grade acetonitrile (ACN) and trifluoroacetic acid (TFA) were from Merck (Darmstadt, Germany). HPLC grade ammonium acetate (CH3 CO2 NH4 ) was from Tedia (USA). MALDI matrix a-cyano-4-hydroxycinnamic acid (CHCA) was from Aldrich (Milwaukee, WI, USA). All protein standards and sequencing grade trypsin were obtained from Sigma. 2.2. Preparation of samples Cytochrome c was dissolved in 100 mmol/L NH4 HCO3 buffer at a concentration of 3 mg/mL and denatured by boiling for 15 min. The protein solution was digested overnight at 37 ◦ C with sequence grade trypsin at a ratio of 25:1 (w/w). The liver cancer tissue of D20, a human hepatocellular carcinoma model in nude mice with high metastatic potential [39,40], was obtained from Liver Cancer Institute in Zhongshan Hospital, Fudan University. D20 liver cancer tissue was cleaned with Milli-Q water to remove contaminants, cut into small pieces, and homogenized in 1 mmol/L phenylmethanesulfonyl fluoride (PMSF) aqueous solution using glass homogenization vessel in ice bath. The resulting homogenate was swirled for 20 min and centrifuged for 10 min at 12,000 × g. The supernatant was collected, fractionated in aliquots and stored at −20 ◦ C till further analysis. The protein concentration was measured by a Bio-Rad assay using bovine serum albumin (BSA) as standard. The solution containing 1 mg of proteins was reduced and denatured by boiling at 100 ◦ C for 15 min, and then was digested overnight at 37 ◦ C with sequence grade trypsin at a ratio of 50:1 (w/w). The digests were acidified by adding TFA before analyses. 2.3. Two-dimensional capillary array liquid chromatography system A diagram of the automated two-dimensional capillary array liquid chromatography system is shown in Fig. 1. The 2D-LC system consists of two Agilent 1100 series capillary pumping systems (Agilent Technologies), a Famos micro-autosampler (LC Packings, Dionex), an SCX column, a 10 capillary RP columns, a multi-channel interface, 10 thermo expansion pumps [41] (designed and constructed in our laboratory), and a Probot

C. Liu, X. Zhang / J. Chromatogr. A 1139 (2007) 191–198

193

Fig. 1. (a) Schematic diagram of column array-based 2D-LC system, (b) post-column micro valve was open, (c) post-column micro valve were close.

microfraction collector (LC Packings, Dionex). One pumping system was used for delivery of ion-exchange chromatography (IEC) loading buffer (5% ACN/0.1% TFA) to the autosampler and 10-step salt gradient of NH4 AC (0, 25, 50, 75, 100, 150, 200, 250, 500, and 1000 mM NH4 AC in loading buffer, respectively) was used for SCX fractionation. The other pumping system was the reverse phase pump used for gradient elution of the array. Mobile phase A consisted of 5% ACN/0.1% TFA and mobile phase B consisted of 80% ACN/0.1% TFA. The solvent gradient used was 0–50% B in 80 min, 50%–100% B from 80–80.1 min, 100% B from 80.1–85 min, and 100%-0 B from 85–105 min. The multi-channel interface was used to couple IEC and RPLC separation systems, and the micro-autosampler was used for sample injection. On-column UV detection was carried out using a Waters 484 tunable absorbance detector (Waters, Milford, MA, USA) with modification. Multiple-channel interface was fabricated with polyether ether ketone (PEEK) tube and fused-silica capillaries. On the one side of PEEK tube (8 cm × 500 ␮m I.D.), 10 holes with diameter of 380 ␮m were pierced in an interval of 0.5 mm in the middle of the PEEK tube. 10 fused-silica capillaries (10 cm × 50 ␮m I.D. × 375 ␮m O.D.) were inserted into the holes, and the epoxy was applied around the outside of the PEEK/fused-silica capillary boundaries. The epoxy was allowed to dry for about 24 h. The interface was tested with the pressure up to 200 bar and employed as interface for the 2D-LC array system. One end of the PEEK tube of the multiple-channel interface was connected to an SCX column (7 cm × 320 ␮m I.D. 20 ␮m polymeric beads, Applied Biosystem Inc.) and a loading pump, and the other was connected to the gradient pump. The outlets of 10 fused-silica capillaries are connected to 10 corresponding capillary RPLC columns (25 cm × 250 ␮m I.D., 5 ␮m C8 particles). Three-way micro-splitter valve (Upchurch, Sulpecol) was placed behind each capillary RPLC column and acted as on-off of the column effluent and the mixer of effluents with the matrix. The upper and the lower ways of the micro-splitter valves were connected to the capillary RPLC columns and spotting capil-

laries (10 cm × 75 ␮m I.D. × 375 ␮m O.D.), respectively. The side ways of the valve were individually connected to thermo expansion pumps with 30 cm × 100 ␮m I.D. × 375 ␮m O.D. fused-silica capillary. The lower ways of the spotting capillaries were fixed vertically by a clamp at the same interval of 4 mm, and their ends were located 10 cm above the robotic moving flat of the Probot microfraction collector. With the standard ABI MALDI plate, a 10 × 20 spotting pattern was used. 2.4. Mass spectrometry AB 4700 Proteomics Analyzer (Applied Biosystems, USA) was used to acquire both MALDI/MS and MALDI/MS/MS spectra. The mass spectrometer was set to acquire positive ion MS survey scans with a laser (337 nm) as desorption ionization source. MS acquisition was performed with 1500 laser shots/spot (10 sub-spectra accumulated from 150 laser shots each), and MS scanning range was 700–3500 Da. Data dependent acquisitions were performed with up to eight precursors selected for interrogation from each MS survey scan. Precursor selection was based upon ion intensity and MS/MS acquisition was performed with 2500 laser shots/spot (10 sub-spectra accumulated from 250 laser shots each). The mass calibrations were done externally on all the targets with myglobin digest peptides. 2.5. Database searching The MS/MS data obtained from the TOF/TOF were searched against NCBI nonredundant protein sequence database with Mascot (version 1.6, Matrix Sciences, London, UK) as the search engine. GPS Explorer (version 1.0, Applied Biosystems, USA) was used for submitting data acquired from the TOF/TOF for database searching. The Homo sapiens [human] sub-database was used and one missing cleavage of trypsin was allowed for the database searching. The mass tolerance of MS and MS/MS were set to 0.1 and 0.2 Da, respectively, and the S/N of MS/MS was set to 10. At least a peptide match of ion score more than

194

C. Liu, X. Zhang / J. Chromatogr. A 1139 (2007) 191–198

24 was arbitrarily set as the threshold for acceptance. In cases where the protein was identified by a single peptide match of ion score lower than 24, the original MS/MS spectra were manually inspected for notable sequence tag. If the peptide match was based on a sequence tag of at least three or more consecutive amino acids, then the peptide candidate was selected as identified. 3. Results and discussion 3.1. Systematic considerations of the array 2D-LC system In the conventional LC–LC systems, one RPLC column is generally implemented for the further separation of fractions eluted from the primary dimension. For global proteome profiling, the chromatographic separation throughput was greatly limited by multiple LC runs that must be carried on one by one and column regeneration between two runs. As shown in Fig. 1, a capillary array LC–LC system was constructed, in which 10 capillary RP columns were used as the second separation dimension, to enable the simultaneous elutions of 10 fractions eluted from the first dimension. The whole 2D-LC system has to be well matched in parameters of sample size, column sizes, flow rates, spot sizes deposited on MALDI plate, and so on. MALDI spot size was depended on spotting frequency, the LC and matrix flow rates, while the spotting frequency was further depended on the RPLC eluting peak width and solvent flow rate. Therefore, RPLC column size is one of the key parameters to be selected for system optimization. Meanwhile, RPLC column size is also related to total analysis time of 2D-LC in transferring sample fractions from SCX column, the first dimension, to the RPLC columns. In coupling of LC to ESI-MS, the smaller inner diameter column can ensure a less sample dilution and greatly benefit the sensitivity of LC/MS, and nanoscale capillary column with 75 ␮m I.D. is frequently used for achieving high sensitivity analysis. Its typical flow rate is lower than 200 nL/min. It is unnecessary to be used in MALDI–MS since the solvent of LC effluents is completely dried after mixed with matrix. Sample concentration of LC effluents was not important for MALDI–MS analysis any more in this case, and wider bore column could be employed in LC–MALDI–MS system. In as-constructed 2D-LC system, the column array was consisted of 10 capillary analytical columns (250 ␮m × 25 cm) packed with 5 ␮m particles. The larger inner diameter of analytical columns could allow larger sample loading which is favorable for the detection of low-abundance proteins. Furthermore, the larger inner diameter of columns could allow the higher flow rate (5 ␮L/min) for SCX fracton transfer from SCX column to the RPLC column at a moderate back pressure, which could be fairly compatible to 0.2–2 ␮L dead volumes of the multi-channel interface. In addition, analytical column length of 25 cm produced a satisfactory separation resolution. Capillary SCX column of 320 ␮m × 7 cm packed with 20 ␮m Poros SCX particles was employed as the first separation dimension. The unique perfusional pore structure of poros particles substantially reduced the mass transfer resistance in particles,

which not only help to speed up separation on SCX column, but also provide a dramatically low pressure drop in transferring sample fractions from first dimension to the second dimension. 3.2. Parallel performance of the array columns Parallel performance of the array columns is very important for the array 2D-LC system. The 10 RPLC columns should be identical in pressure drop/permeability and resolution since a single pump was employed to deliver gradient solvents in simultaneous elutions. The unparallel in column performance could result in diversity of flow rate, resolution differences, and reproducible problems. In this work, a batch of 40 capillary RPLC columns was carefully prepared according to the procedure described previously [42]. Besides identical packing material and capillary length, the same packing pressure (400 bar), sonicating time, and the same solvent conditioning were applied for the preparation of RPLC columns. Ten columns with theoretical plate number of 5.8–6.5 × 104 plates/m and backpressure 35 ± 3 bar at 2 ␮L/min flow rate of 70% ACN were selected to construct array RPLC system. The performance of each column was evaluated by separating tryptic digest peptides of pure protein, cytochrome c. A comparison of 10 reversed-phase capillary columns used in the array is shown in Fig. 2. The standard errors for elution time were calculated as 1.37% for 9 peaks eluting between 20 and 50 min. The results indicated that even flow rates and similar separations across the channels could be obtained by the tenplexed channel array system integrated by these columns. The evaluation for the columns was repeated periodically to monitor their performance during uses. 3.3. The multiple channel union and the operation of the 2D-LC system It is critical to minimize the dead volume of the interface and connection for fractions transfer with high efficiency from the first dimension to the next dimensions. For a single RPLC column 2D-LC system [8–10], valves switching is the best choice for sample transfers. However, for column array-based 2D-LC, a simple multi-channel interface was designed as described in Section 2. In order to minimize the dead volume between SCX and each RPLC column, 0.5 mm I.D. PEEK tube and an interval of 0.5 mm for each channel were chosen for accessible conventional fabrication. To minimize the connection volumes, SCX column was inserted into one end of the PEEK tube of the interface and fixed outlet of the column at first channel position so that SCX fraction could immediately go into the first channel of array columns. The dead volumes between SCX and RPLC columns were about 0.2–2.0 ␮L. The maximum delay time was less than 0.4 min at flow rate of 5 ␮L/min. In this case, an SCX fraction transfer could be completed in less than 4.5 min. RPLC gradient pump was connected at the other end of the PEEK tube. The solvent gradient delay was not significant because 10 RPLC columns needed a total 18 ␮L/min flow rate. Thus, the solvent

C. Liu, X. Zhang / J. Chromatogr. A 1139 (2007) 191–198

195

Fig. 2. The separations of the tryptic digest of cytochrome c onto 10 different C8 columns in single-dimensional singe-column runs. Conditions: 25 cm × 250 ␮m I.D. capillary columns packed with 5 ␮m C8-bonded particles at a flow rate of 1.8 ␮L/min. UV detection at 215 nm. Mobile phase A consisted of 5% ACN/0.1% TFA, and mobile phase B consisted of 80% ACN/0.1% TFA. The solvent gradient used was 0–50% B within first 40 min, then 50% B during the period of 40.1–45 min, and 50%-0 B for the period of 45.1–70 min.

196

C. Liu, X. Zhang / J. Chromatogr. A 1139 (2007) 191–198

gradient delay was very small at a relative high velocity of flow in the connecting tube. Before analysis, the 2D-LC array system was equilibrated with loading buffer (5% ACN and 0.1% TFA (v/v)). Sample was loaded onto SCX column by the autosampler at a flow rate of 5 ␮L/min. SCX fractions eluted by discrete salt solutions transferring into each column of the RPLC array were controlled by on-off switching of the post-column 3-way micro valves. During the sample loading, the post-column micro valve behind the RP column 1 was open (Fig. 1b), while others were closed (Fig. 1c). For the case, the SCX column was online with the RP column 1, and the flowthrough peptides was transferred into the RP column 1. After fractionating, post-column micro valve 1 was closed and micro valve 2 was opened. The salt solution of 25 mM NH4 AC was used for fractionation, and the fraction eluted was transferred into the RP column 2. By sequentially switching the valves the SCX column was individually online with each RPLC column, and the sample fractions eluted by salt solutions were transferred into corresponding RPLC column. Peptides in the fractions were trapped at corresponding head of columns while buffer salts in the faction were washed away as unretained elution peak in RPLC separations. In the work, 10-step salt gradient of NH4 AC (0, 25, 50, 75, 100, 150, 200, 250, 500, and 1000 in loading buffer, respectively) was used for SCX fractionation. After fractionation of the first dimension, all post-column valves were opened and the gradient elutions of the 10 channels were concurrently performed by reverse phase pump. 3.4. The interface for LC array effluents mixing with matrix and depositing on MALDI plates The 3-way micro-splitter valves were acted as mixing chambers of LC effluent with matrix solution. To minimize the resolution loss of LC effluents and achieve a fairly mixing efficiency of matrix with LC efflent, 10 cm × 75 ␮m I.D. fusedsilica capillary (<0.45 ␮L dead volumes) was used as depositing tip. The lower part of the spotting capillaries was fixed with a clamp so that the mixtures of the 10 channels could be deposited uniformly in parallel on the same sample plate. The same length of spotting capillaries and interval between spotting capillaries are the keys to achieve uniform sample deposition spot which could allow the high-throughput automated MALDI–MS analysis. The length as well as the interval was finely adjusted. The interval was dependent on spot size, and 4 mm interval was used in the work. Matrix solution was delivered by 10 thermo expansion pumps in parallel. All flow rates of the pumps were set at 1.8 ␮L/min. Matrix, CHCA, and effluents from each RP capillary column were mixed with the optimized ratio of 1:1 for MALDI–MS analysis. A spotting frequency of 15 s was finally employed. This frequency offered advantages for concentrating the sample and decreasing the total number of MS spectra to be scanned while the chromatographic performance was maintained. During the deposition of the sample, CHCA would be crystallized and eventually clogged in the tips of the capillary due to solvent evaporation. To prevent the solvent evaporation on

the tips, all the capillary tips were coated with a thin layer of hydrophobic material, e.g. Teflon, wax or Vaseline. The coating also precisely facilitated the positioning of sample spots. No interference signal and influence of ionization of peptides was confirmed. 3.5. Performance validation of the array 2D-LC coupling to MALDI–MS-MS system To verify the effectiveness of the 2D-LC array system for fast proteomic separation, MALDI/TOF/TOF MS was applied to profile soluble proteins in a liver cancer tissue of HCC D20. Extracts of soluble proteins in liver tissue were digested with trypsin. About 300 ␮g of the digests were loaded onto the SCX column. With a loading pump to deliver IEC loading buffer at a flow rate of 5 ␮L/min, discrete 10-step salt gradient of NH4 AC from 0 to 1000 mM (0, 25, 50, 75, 100, 150, 200, 250, 500, and 1000 mM NH4 AC in loading buffer, respectively) was used for SCX fractionation. Peptides eluted in each salt-step were trapped on the corresponding RPLC column head. After fractionations of the first dimension, the simultaneous RPLC gradient elutions of the 10 channels were carried out. Meanwhile, effluents of the second dimension mixed on-line with matrix were deposited onto target plates in series. With the spotting pattern, 200 spots (20 spots/column) were deposited on each plate. After half an hour drying, all plates were loaded on mass spectrometer to carry out MS and MS/MS analysis. One SCX fractionation was for about 5 min at a flow rate of 5 ␮L/min, while 10 fractionations were for around 50 min. RPLC gradient elution time was about 100 min. Thus, the separation time for the array 2D-LC was totally ca. 150 min. The analysis was almost 10-times faster than those of sequential 2DLC systems. During a separation with solvent gradient, 300 spots per column were deposited at target plates. MALDI TOF/TOF mass spectrometer can offer automated high-throughput MS analysis. In this work, a high-throughput MS, 4700 Proteomics Analyzer with 200 laser shuts/s, was used for peptides sequencing. To increase the analysis throughput of MS, all sample plates were automatically loaded by an autoloader which made plates rapidly transfer under vacuum (plate to plate, <5 s). The sample spots were at first scanned in the MS mode to quickly produce an overview of the whole sample. Based on the overview, the ions of peptides were then selected for interrogation by MS/MS, while ions from matrix were excluded by defining the peak monitor window. The mass accuracy in the range of 100 ppm was observed for peptide assignment. Combining all 10 fractions, 2138 different peptides were assigned, and a total of 1202 proteins were identified from soluble extract of the liver tissue, with an average of 1.78 peptides per protein (see the Supporting Information). Of the identification, 43% were based on two or more unique peptides per protein and 57% of the proteins were assigned by single peptide hits. The acid protein with pI < 4.3 and the basic protein with pI > 11 were identified. The most acidic protein identification was lecuine-rich acidic protein with a pI of 3.78, and the most basic one was similar to probable mucin DKFZp434C196.1 with a pI of 13.55. In the study, the pro-

C. Liu, X. Zhang / J. Chromatogr. A 1139 (2007) 191–198

197

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.chroma.2006.11.019. References

Fig. 3. 3D plot of the resulting mass spectra for the 25 mM fraction obtained by the 2D-LC/MALDI–MS analysis of a tryptic digest of the soluble proteins in D20.

tein with MWs <10,000 and >200,000 were all represented; however, the protein with MWs between 10,000 and 200,000 accounted for the largest percentage (85%). For example, 169 protein with a MW in excess of 200 kDa were identified, the largest being Titin with a MW of 3813 kDa and the smallest one was pyruvate dehydrogenase E1-alpha subunit with a MW of 2029 Da. A three-dimensional (3D) plot obtained from the fraction of 25 mM salt-step was shown in Fig. 3. Actually, the 2D-LC system is not limited to 10 columns in the second dimension. Twenty or even more RPLC column could be applied. However, in this paper, the number of columns was limited by the gradient pump for delivering solvents. 4. Conclusions A novel 2D-LC system based on 10 parallel LC columns and its application for high-throughput 2D-LC/MALDI analysis were demonstrated. The most significant advantage of the 2D-LC system is its high-throughput for complex proteome separation. In fact, the 10 parallel column array is easily extended to 20 or even more columns to perform separations of more SCX fractions of complex proteomic samples simultaneously. Also, based on such a fast separation system, a setup of a threedimensional system using size exclusion chromatography as the first dimension is under investigation. It is possible to separate a complex sample from the investigated 3D chromatography within 24 h, instead of weeks. Thus, the complexity of a sample is highly reduced and the simplified fractions can be fairly separated and identified. It is very important for large-scale proteomic analysis. Acknowledgment This research work was supported by the National Natural Science Foundation of China (Project 20305005), the National 863-project 2002AA2Z2042.

[1] L.J. Thompson, F. Wang, A.D. Proia, K.G. Peters, B. Jarrold, K.D. Greis, Proteomics 3 (2003) 2258. [2] G. Mamone, S. Caira, G. Garro, Electrophoresis 24 (2003) 2824. [3] J. Kim, S.H. Kim, S.U. Lee, Electrophoresis 23 (2002) 4142. [4] C.S. Giometti, T. Khare, S.L. Tollaksen, Proteomics 3 (2003) 777. [5] S. Beranova-Giorgianni, Trends. Anal. Chem. 22 (2003) 273. [6] M.M. Bushey, J.W. Jorgenson, Anal. Chem. 62 (1990) 161. [7] H.A. Holland, J.W. Jorgenson, Anal. Chem. 67 (1995) 3275. [8] G.J. Opiteck, K.C. Lewis, J.W. Jorgenson, R.J. Anderegg, Anal. Chem. 69 (1997) 1518. [9] G.J. Opiteck, J.W. Jorgenson, R.J. Anderegg, Anal. Chem. 69 (1997) 2283. [10] M.T. Davis, J. Beierle, E.T. Bures, T.D. Lee, J. Chromatogr. B 752 (2001) 281. [11] K. Wagner, T. Miliotis, G. Marko-Varga, R. Bischoff, K.K. Unger, Anal. Chem. 74 (2002) 809. [12] A.J. Link, J. Eng, D.M. Schieltz, E. Carmack, G.J. Mize, D.R. Morris, B.M. Garvik, J.R. Yates III, Nat. Biotechnol. 17 (1999) 676. [13] T.F. Hooker, J.W. Jorgenson, Anal. Chem. 69 (1997) 4134. [14] A.W. Moore, J.W. Jorgenson, Anal. Chem. 67 (1995) 3448. [15] A.W. Moore, J.W. Jorgenson, Anal. Chem. 67 (1995) 3456. [16] M.T. Kachman, H. Wang, D.R. Schwartz, K.R. Cho, D.M. Lubman, Anal. Chem. 74 (2002) 1779. [17] R.D. Rocklin, R.S. Ramsey, J.M. Ramsey, Anal. Chem. 72 (2000) 5244. [18] D. Mohan, L. Pasa-Tolic, C.D. Masselon, N. Tolic, B. Bogdanov, K.K. Hixson, R.D. Smith, C.S. Lee, Anal. Chem. 75 (2003) 4432. [19] D. Mohan, C.S. Lee, Electrophoresis 23 (2002) 3160. [20] D.A. Wolters, M.P. Washburn, J.R. Yates III, Anal. Chem. 73 (2001) 5683. [21] M.P. Washburn, D. Wolters, J.R. Yates III, Nat. Biotechnol. 19 (2001) 242. [22] K.G. Mawuenyega, H. Kaji, Y. Yamauchi, T. Shinkawa, H. Saito, M. Taoka, N. Takahashi, T. Isobe, J. Proteome Res. 2 (2003) 23. [23] J.M. Peng, J.E. Elias, C.C. Thoreen, et al., J. Proteome Res. 2 (2003) 43. [24] A. Premstaller, P.J. Oefner, H. Oberacher, C.G. Huber, Anal. Chem. 74 (2002) 4688. [25] A. Premstaller, H. Oberacher, A. Rickert, C.G. Huber, P.J. Oefner, Genomics 79 (2002) 793. [26] V. De Biasi, N. Haskins, A. Organ, R. Bateman, K. Giles, S. Jarvis, Rapid Commun. Mass Spectrom. 13 (1999) 1165. [27] W.M. Bodnar, R.K. Blackburn, J.M. Krise, M.A. Moseley, J. Am. Soc. Mass Spectrom. 14 (2003) 971–979. [28] R.M. Whittal, L.M. Russon, L. Li, J.Chromatogr. A 794 (1998) 367– 375. [29] X. Zhang, D.A. Narcisse, K.K. Murray, J. Am. Soc. Mass Spectrom. 15 (2004) 1471–1477. [30] Q. Zhan, A. Gusev, D.M. Hercules, Rapid Commun. Mass Spectrom. 13 (1999) 2278–2283. [31] R. Mukhopadhyay, Anal. Chem. 77 (2004) 150A–152A. [32] J.B. Young, L. Li, J. Am. Soc. Mass Spectrom. 17 (2006) 325–334. [33] C.K. Van Pelt, T.N. Corso, G.A. Schultz, S. Lowes, J. Henion, Anal. Chem. 73 (2001) 582. [34] B.B. Feng, M.S. McQueney, T.M. Mezzasaima, J.R. Slemmon, Anal. Chem. 73 (2001) 5691. [35] K.F. Medzihradszky, J.M. Campbell, M.A. Baldwin, A.M. Falick, P. Juhasz, M.L. Vestal, A.L. Burlingame, Anal. Chem. 72 (2000) 552. [36] A. Shevchenko, A. Loboda, W. Ens, K.G. Standing, Anal. Chem. 72 (2000) 2132.

198

C. Liu, X. Zhang / J. Chromatogr. A 1139 (2007) 191–198

[37] H. Lee, T.J. Griffin, S.P. Gygi, B. Rist, R. Aebersold, Anal. Chem. 74 (2002) 4353. [38] C. Ericson, Q.T. Phung, D.M. Horn, E.C. Peters, L.M. Brill, A. Brock, Anal. Chem. 75 (2003) 2309. [39] F.X. Sun, Z.Y. Tang, K.D. Liu, et al., J. Cancer Res. Clin. Oncol. 122 (1996) 397–402.

[40] F.X. Sun, Z.Y. Tang, K.D. Liu, et al., Int. J. Cancer 66 (1996) 239– 243. [41] X.M. Zhang, Q. Wu, Chin. Pat. ZL02217871.6 (26 March 2003). [42] X.M. Zhang, S. Huang, J. Chromatogr. A 910 (2001) 13.