Porous Graphitized Carbon

May 19, 2010 - Online Dual Gradient Reversed-Phase/Porous Graphitized Carbon nanoHPLC for Proteomic Applications ... Fax: +4923113924850. E-mail: ... ...
1 downloads 9 Views 2MB Size
Download PDF
Anal. Chem. 2010, 82, 5391–5396

Online Dual Gradient Reversed-Phase/Porous Graphitized Carbon nanoHPLC for Proteomic Applications Urs Lewandrowski* and Albert Sickmann Leibniz Institut fu¨r Analytische Wissenschaften-ISAS e.V., Otto-Hahn-Strasse 6b, 44227 Dortmund, Germany The analysis of proteolytic peptide mixtures is among the dominant tasks within proteomic workflows. In order to limit undersampling effects during mass spectrometric detection, online-coupled liquid chromatography is the method of choice, with reversed-phase chromatography being the most important separation mode. Since hydrophilic compounds such as short peptides and some glycosylated species as well as oligosaccharides from glycoproteomic workflows are commonly not accessible by this analytical setup, we hereby present a dual gradient system combining reversed-phase and porous graphitic carbon retention modes within a single nanoHPLC setup. Samples in the low femtomole range are analyzed consecutively first by reversed phase, and nonretained molecules are directly separated by porous graphitic carbon. Both gradient elution systems allow for online coupled mass spectrometric detection and are demonstrated to enable analysis of protein, peptide, and oligosaccharide mixtures within the same setup. Thereby, the accessible range for proteomic and glycoproteomic applications may be extended far beyond the limits of conventional reversedphase nanoHPLC setups. Current proteomic applications rely fundamentally on separation techniques to reduce the overall complexity of samples prior to fast-scanning mass spectrometric detection for identification and quantification purposes, mostly of tryptic peptides.1,2 Reversedphase (RP) chromatography has been the gold standard for upfront separations, offering both high resolution and buffer systems compatible with bioanalytic mass spectrometry. Commonly, octadecyl derivatized resins are used for peptide separations, while more hydrophobic compounds such as membrane spanning peptides or even intact proteins are easily retained by supports with shorter alkyl chains (C1-C8). However, reversed-phase chromatography also exhibits major drawbacks with respect to very hydrophilic or otherwise modified peptides, e.g., by glycosylations. Hydrophilic peptides that are not retained may elute directly after or within the injection peak or, even worse, may be completely lost due to preconcentration setups commonly em* Corresponding author. Phone: +4923113924142. Fax: +4923113924850. E-mail: [email protected]. (1) Sandra, K.; Moshir, M.; D’Hondt, F.; Verleysen, K.; Kas, K.; Sandra, P. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2008, 866, 48–63. (2) Wolters, D. A.; Washburn, M. P.; Yates, J. R. Anal. Chem. 2001, 73, 5683– 5690. 10.1021/ac100853w  2010 American Chemical Society Published on Web 05/19/2010

ployed for nanoscale reversed-phase applications. Consequently, hydrophilic liquid interaction chromatography can be used to retain hydrophilic analytes such as major glycopeptides or small hydrophilic peptides.3 Nevertheless, separation efficiency may be worse; organic solvent consumption is generally higher, and the application of diluted aqueous samples (e.g., digests or intact proteins) can be problematic. On the contrary, porous graphitic carbon (PGC) stationary phases have been shown to retard hydrophilic substances such as sugars,4,5 hydrophilic peptides,6 or neurotransmitters.7 Therefore, a combined setup for PGC and RP would offer superior separation possibilities for the whole range of peptides to be expected from proteolytic digests. Indeed, a parallel column setup (RP and PGC) has already been introduced by Kim et al8 for analysis of human plasma serum, resulting in detection of peptides and glycans after glycosidic cleavage. Nevertheless, the demonstrated system required prolonged loading times for direct injection and its flow rates (5 µL/min for ∼500 µm ID columns) were still incompatible with common nanospray applications. Furthermore, samples containing peptides and glycans could only be applied to either of the columns, not both. A more sophisticated system consisting of a RP main column followed by a PGC main column was considered on a theoretical basis for analysis of hydrophilic peptides from complex proteomic samples in 2006.9 A potential increase in sequence coverage by retention of small hydrophilic peptides by the additional PGC retention mode was estimated. For a tryptic digest of 50 human proteins, ∼30% of peptides were found to be between 4 and 9 residues in length and, thus, were potentially difficult to cover by RP chromatography alone. However, the design of the intended column system did not allow for convenient nanoscale separations with current preconcentration setups. Herein, we present for the first time a modified dual gradient nanoHPLC setup for successive analysis of proteomic samples by RP and PGC chromatography. The main difference to existing (3) Omaetxebarria, M. J.; Hagglund, P.; Elortza, F.; Hooper, N. M.; Arizmendi, J. M.; Jensen, O. N. Anal. Chem. 2006, 78, 3335–3341. (4) Packer, N. H.; Lawson, M. A.; Jardine, D. R.; Redmond, J. W. Glycoconjugate J. 1998, 15, 737–747. (5) Wilson, N. L.; Schulz, B. L.; Karlsson, N. G.; Packer, N. H. J. Proteome Res. 2002, 1, 521–529. (6) Chin, E. T.; Papac, D. I. Anal. Biochem. 1999, 273, 179–185. (7) Thiebaut, D.; Vial, J.; Michel, M.; Hennion, M. C.; Greibrokk, T. J. Chromatogr., A 2006, 1122, 97–104. (8) Kim, Y. G.; Jang, K. S.; Joo, H. S.; Kim, H. K.; Lee, C. S.; Kim, B. G. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2007, 850, 109–119. (9) Disclosed Anonymously. In Research Disclosure Journal; Kenneth Mason Publications Ltd.: Hants, UK, 2006; RD 510028 20061010.

Analytical Chemistry, Vol. 82, No. 12, June 15, 2010

5391

Figure 1. Dual gradient RP/PGC nanoHPLC system. (A) Fluidic connections for the RP/PGC setup. The depicted status for FLM1 and FLM2 valves is under loading conditions. Upon injection, analytes are first eluted from the RP precolumn over the RP main column to the MS by switching valve FLM1 to the 1_2 position, while the six port valve is set to the 1_2 position. Second, the latter valve is switched to the 1_6 position to allow for separation of the PGC column system over valve FLM2 set to the 1_2 position. (B) Gradient setup for RP (nanopump 1) and PGC gradient (nanopump 2). Tryptically digested bovine serum albumin was consecutively separated by RP/PGC chromatography (black UV trace, 214 nm); a blank injection (blue UV trace, 214 nm) is shown for comparison.

systems is the sequential loading of a single sample volume to (1) a RP precolumn and (2) a PGC precolumn within individual preconcentration setups for each separation mode (Figure 1). Compounds not binding to the C18-RP resin are, therefore, retained by the unique PGC stationary-phase chemistry.10 The setup is shown to enhance detection of hydrophilic peptides, which otherwise may be lost during RP HPLC, as well as for the combined analysis of glycopeptides, glycans, and free oligosaccharides. EXPERIMENTAL SECTION Materials. Bovine serum albumin (tryptic digest) was received from Bruker Daltonics, Bremen, Germany; iodoacetamide, bovine ribonuclease B, sialyllactose and maltoheptaose were obtained from Sigma Aldrich, Steinheim, Germany. Synthetic peptide standards were synthesized by Peptide Specialty Laboratories, Heidelberg, Germany. Sequencing grade trypsin was from Promega (Madison, MA). Glucose homopolymer was bought from ProZyme, Hayward, CA (formerly Glyko). Dithiothreitol and PNGaseF were obtained from Roche. Formic acid and acetonitrile were purchased from Merck; water was drawn from an Elga Labwater system at 18.2 MΩ quality. Sample Preparation. Ribonuclease B was tryptically digested after reduction in 50 mM NH4HCO3 with 5 mM dithiothreitol (DTT) for 15 min at 56 °C and alkylation with 15 mM iodoacetamide for 15 min at room temperature in the dark. Typically, an enzyme to protein ration of 1:20 was used for an overnight incubation at 37 °C. Samples were diluted to appropriate concentrations with 0.1% formic acid in water prior to injection. For deglycosylation, proteins were reduced in 50 mM NH4HCO3 and 5 mM DTT for 15 min at 56 °C and cooled down to room temperature thereafter. Samples were diluted 5-fold with water and 1U PNGaseF (Roche) was added for deglycosylation of 100 µg of protein for 16 h at 37 °C. (10) Knox, J. H.; Kaur, B.; Millward, G. R. J. Chromatogr. 1986, 352, 2–25.

5392

Analytical Chemistry, Vol. 82, No. 12, June 15, 2010

NanoHPLC Separations. Chromatographic separations were performed on a dual gradient U3000 nanoHPLC system (Dionex, Idstein, Germany) consisting of a WPS-3000 autosampler, two DGP-3000 Pump modules (each one loading and one gradient pump), two FLM-3100 column compartments, two SRD-3600 degasser modules, and one VWD-3400 UV-detector. Fluidic connections are shown in Figure 1A; basically, the two custommade precolumns (RP: 100 µm ID × 2 cm length, 5 µm particle size PepMap, Dionex; PGC: 100 µm ID × 2 cm length, 5 µm particle size, Hypercarb, Thermo Fisher Scientific) are connected in-line during the 7 min sample application (flow rate 7 µL/min 0.1% formic acid in water), and samples are successively eluted over the RP main column (75 µm ID × 30 cm length, 3 µm particle size, PepMap, Dionex) and/or the PGC main column (75 µm ID × 30 cm length, 3 µm particle size, Hypercarb, Thermo Fisher Scientific) at flow rates of 220 nL/min (RP) and 260 nL/min (PGC). Temperatures of the column compartments were set to 45 °C (RP) and 60 °C (PGC). The third valve was used to divert the respective flow of the nonseparating main column to waste. The binary gradients consisted of 0.1% formic acid in water for solvent A and 0.1% formic acid, 84% acetonitrile for solvent B, with a slope of 1.43% B/min from 7% B to 50% B for the RP gradient and 1.1% B/min from 7% B to 40% B for the PGC gradient (both 30 min separation time). Both gradients were followed by a short washing step at high content solvent B (90-95%). A common program is shown in Supporting Information (D-1), and a representation of the gradient timing within a typical run is shown in Figure 1B. UV detection was performed at 214 nm by the VWD3400 prior to transfer of samples to the mass spectrometer. Mass Spectrometry. The dual gradient system was connected via the UV outlet to a LTQ XL linear ion trap (Thermo Fisher Scientific) equipped with a nanospray interface using pulled fused silica emitters (FS-360-20-10-D, New Objective, Switzerland), being operated in positive ion mode with a source voltage of 1.9 kV. For detection of analytes, a survey scan (380-1800 m/z) was followed by five tandem MS scans of the five most intensive

precursor ions. Dynamic exclusion was enabled for 20 s with a repeat count of 2. Raw files were converted to Mascot generic format using the extract_msn tool provided by Thermo Fisher Scientific. Database searches against the SwissProt dabase (www. uniprot.org, 26th May 2009, 468 851 sequences) were performed using Mascot 2.2 (MatrixScience, London, UK) with the following settings: Trypsin was chosen as protease with one miscleavage site. Carbamidomethylation of cysteine was set as fixed modification, while oxidation of methionine was allowed as variable modification. Taxonomies were selected according to the respective protein origin (e.g., “other mammalia” for bovine serum albumin). MS and MS/MS tolerance were both set to 0.5 and 1.5 Da, respectively. Only peptides with Mascot scores above 40 (p < 0.05) were considered for manual validation. RESULTS The successive combination of RP and PGC separation modes proved to be of value for several applications such as the separation of peptide mixtures with ranging hydrophobicity as well as separation of peptides/proteins and glycans. A first setup (fluidic connections shown in Supporting Information, Figure 1) consisted of a single nanoHPLC (Famos/Switchos/ Ultimate, Dionex) with two switching valves to achieve a consecutive separation of analyte mixtures by RP/PGC preconcentration setups. Indeed, it was possible to achieve separations on both stationary phases after injection of a single sample. However, the major drawback was the discontinuous flow condition, only one of the main columns was in-line at a time. Although the pressure drop on the second column was partially made up for by the setup (∼50% drop after 1 h), longer equilibration times were necessary to produce stable separations. Therefore, the current enhanced setup was developed as presented (Figure 1A). With the indicated parameters for flow and column hardware, pressure ranged from 160 to 100 bar for the RP and PGC part of the system, respectively. Switching both precolumns in-line with the loading pump generated a total backpressure of 100 bar at 7 µL/min flow rate. Thereby, the system was operating at pressure rates well below 200 bar, rendering it suitable for most HPLC instrumentation. The separation of peptides was first tested with a standard peptide mixture containing seven peptides with a small dynamic range (2-60 fmol/injection with rising concentrations up to 32-960 fmol/injection; for peptide sequences, see Supporting Information Table 1 or Figure 4). Peptides were well retained on the RP column and separated as expected by previous use in one-dimensional RP separations. The sensitivity of the RP system for detection of peptides by mass spectrometry and database searching was determined to be at 2 fmol or below. Furthermore, peptides of the standard mixture were not detectable in the succeeding PGC separation arguing for a reasonable loading capacity of the system. Second, a tryptic digest of carbamidomethylated bovine serum albumin (SwissProt accession P02769) was applied to the dual gradient system (250, 125, 62.5, and 31.25 fmol). Exemplary results for the analysis of the 250 fmol digest are depicted in Figure 2, and

Figure 2. Separation of bovine serum albumin tryptic peptides. (A) Base peak chromatogram of the complete RP and PGC gradients (peptide containing regions are marked gray). (B) Extracted ion chromatogram of the peptide AEFVEVTK, being only partially retained by the RP precolumn; the remaining peptide amount is seen in the PGC separation (marked individually by arrow). (C) Extracted ion chromatogram of the peptide HLVDEPQNLIK being retained completely by the RP resin. (D) Extracted ion chromatogram of the peptide ATEEQLK, which is not retained by the RP resin, but eluted as a defined peak by PGC. Thus, the combined system allows for separation of hydrophilic peptides not retained by RP chromatography.

details of peptide identifications are summarized in Supporting Information Table 1. Furthermore, an UV trace of a 500 fmol bovine serum albumin containing separation is presented in Figure 1B. After database searching, protein sequence coverage was 64% for the RP and 22% for the PGC based separations, respectively, with a combined coverage of 76% for the complete RP/PGC setup. Looking at the number of individual peptides, 37 distinct peptides with a mean length of ∼13 amino acids were found for RP and 16 with a mean length of ∼9 were found for PGC. Using TheorChromo (online version 1.0 beta11) as a prediction tool for peptide retention on RP chromatography systems, the mean retention time of RP retained peptides was calculated to be 24.04 min and, for PGC, it was 15.13 min (see Supporting Information, Table 1) at conditions mirroring the used system (30 min gradient time). Clearly, PGC is able to trap peptides not retained by RP and which are rather small and hydrophilic. Figure 2, therefore, shows the basepeak chromatogram of 250 fmol BSA injected on the columns; the majority of peptides is indeed retained by the first precolumn (RP), and only a minor part is seen retarded by PGC (compare UV/vis spectrum Figure 1B). The extracted ion chromatogram of m/z 461.79 AEFVEVTK (Figure 2B) demonstrates a peptide being only partially retained by RP, while m/z 653.36 HLVDEPQNLIK is retained completely by the RP resin and m/z 409.71 ATEEQLK is not at all retained by RP and will be lost to analysis in a conventional one-dimensional RP setup. Apart from AEFVEVTK, four further peptides show only partial retention on RP and are, thus, found in both runs. Consequently, 11 peptides are found exclusively by the PGC mode. Similar observations were made with tryptically digested and (11) http://theorchromo.ru; Institute of Energy Problems for Chemical Physics: Russia, 2004-2009.

Analytical Chemistry, Vol. 82, No. 12, June 15, 2010

5393

Figure 3. Separation of tryptic bovine ribonuclease B digest. Autoproteolytic trypsin and ribonuclease B peptides are detected between 20 and 45 min during the RP elution. The inset depicts the summed survey spectra from 75 to 80 min during the PGC gradient showing the typical mass distribution of high mannose glycans (GlcNAc2Man5-9) featured by ribonuclease B14 still attached to NLTK, thus demonstrating the separation of small hydrophilic glycopeptides.

carbamidomethylated bovine ribonuclease B, although only a few peptides could be observed at all due to the nature of the glycoprotein. In this context, m/z 535.70 STMSITDCR was exclusively detected in the PGC run, while m/z 1113.02 HIIVACEGNPYVPVHFDASV along with autoproteolytic peptides of trypsin was only observed in the RP fractions (data not shown). Although not fitting for each individual case, predominantly short-length peptides featuring only a few or no strong hydrophobic side chains (P, W, Y) instead of hydrophilic and charged residues, e.g., aspartic or glutamic acid, are weakly or not at all retained by the RP resin. However, the PGC resin still enables the separation of these analytes due to its different retention mode. In turn, peptides being partially trapped by the RP column did contain either at least one bulky hydrophobic residue (P, W, Y) or several smaller hydrophobic side chains (L, I, V) in addition to hydrophilic residues, thereby potentially enabling the residual retention on the RP resin (for plots of hydrophilicity compare Supporting Information Table 1). However, the tryptic digest of bovine ribonuclease B exemplifies another use of the RP/PGC dual gradient system by the retention of small hydrophilic glycopeptides. Figure 3 shows the basepeak chromatogram of ribonuclease B peptides, with the inset depicting the summed MS survey scans within the elution time window from 75 to 80 min. Clearly, five masses (m/z 846.45, m/z 927.43, m/z 1008.42, m/z 1089.43, and m/z 1170.87) are detectable, which correspond to glycan isoforms (GlcNAc2Man5-9) attached to the asparagine of the small peptide moiety NLTK. These isoforms are partially separated by PGC (GlcNAc2Man5 from GlcNAc2Man6-9) but cannot be detected during the RP parts of the separation. Already, efforts have been made regarding glycosylation site analysis on small glycopeptides derived from unspecific proteolytic digests, resulting in short length glycopeptides and single 5394

Analytical Chemistry, Vol. 82, No. 12, June 15, 2010

glycosylated amino acids.12 Those mixtures may be conveniently analyzed by the presented two column system. Although free amino acids are known to bind to the PGC resin,13 they are usually not analyzed along the glycopeptides which feature higher m/z values during mass spectrometric detection. In contrast to free glycans, e.g., those derived from glycosidic cleavages, these small glycopeptides also feature a first possible indication as to the glycosylation site itself and, thus, to the microheterogeneity at individual glycosylation sites. Since PGC has already been extensively used for the separation of oligosaccharides,15,16 the potential use of the RP/ PGC column system for this purpose was explored as well. Starting with model sugars, maltoheptaose, and (2-6)-sialyllactose, no retention was observed for the RP columns and elution by the PGC gradient occurred first for maltoheptaose and late in the gradient for (2-6)-sialyllactose. Within seven consecutive runs, the retention time for MH (85.12 min) varied below 5 s, indicating a good reproducibility of the separation system. The possibility for successive analysis of peptide/oligosaccharide mixtures is further elaborated by results shown in Figure 4. A glucose homopolymer with varying chain lengths was spiked with a seven peptide standard already mentioned. Indeed, the glucose homopolymer was only detectable within the PGC gradient with longer chain length leading to increased retention times. However, homopolymers with chain lengths greater than nine sugar residues were no longer resolved. The sensitivity of the system regarding (2-6)-sialyllactose and maltoheptaose was in the low femtomole region and slightly less sensitive than for peptides (∼100 fmol). Since the mass spectrometric detection of N-glycans often requires the removal of the peptide or protein moiety prior to detailed characterization, the two column system can, furthermore, be used to monitor the progress of deglycosylation reactions and serve as an automated cleanup step. An example is shown within Figure 5 using a PNGaseF digest of intact ribonuclease B. The intact protein without the signal sequence and the cleaved glycans can be observed by the [M + 14H]14+ to [M + 8H]8+ molecular ions within the RP gradient. In contrast, the PGC separation enabled detection of the familiar pattern of ribonuclease B high mannose glycans covering the GlcNAc2Man5-9 isoforms as free glycans in contrast to the already shown glycopeptide form (see inset of Figure 3). DISCUSSION The RP/PGC dual gradient setup was demonstrated to be of value for several applications based on the separation of hydrophilic analytes alongside hydrophobic analytes. Individually, several advantages were gained. For peptide mixtures (12) An, H. J.; Peavy, T. R.; Hedrick, J. L.; Lebrilla, C. B. Anal. Chem. 2003, 75, 5628–5637. (13) Chaimbault, P.; Petritis, K.; Elfakir, C.; Dreux, M. J. Chromatogr., A 2000, 870, 245–254. (14) Mechref, Y.; Novotny, M. V.; Krishnan, C. Anal. Chem. 2003, 75, 4895– 4903. (15) Schulz, B. L.; Packer, N. H.; Karlsson, N. G. Anal. Chem. 2002, 74, 6088– 6097. (16) Thomsson, K. A.; Karlsson, N. G.; Hansson, G. C. J. Chromatogr., A 1999, 854, 131–139.

Figure 4. Separation of peptide standards and glucose homopolymer. Peptides were retained solely by RP mode (retention time (RT) 29.97-40.17 min) while glucose homopolymer was only identified within the PGC separation. The inset shows the summed MS survey scans from 75 to 95 min confirming the presence of Glc4 to Glc9 oligosaccharides. Impurities are marked by rhombi.

Figure 5. Separation of crude ribonuclease B deglycosylation products. After digest by PNGaseF, the processed and deglycosylated ribonuclease B protein backbone (K27 to V150 SwissProt entry P61823) could be monitored during the RP gradient by [M + 14H]14+ to [M + 8H]8+ molecular ions (left inset). The free high mannose glycans passed the RP precolumn and were readily detectable between 80 and 90 min retention time on PGC as singly and doubly charged precursors (right inset, rhombi mark the [M + H]1+ precursors).

derived from proteolytic digests, the system enables higher sequence coverage alongside the exclusive detection of small hydrophilic peptides which otherwise may be lost during separations. This feature is especially interesting with respect to the analysis of, e.g., recombinant pharmaceutical products, where a complete characterization of the protein sequence is required. The parallel analysis of glycopeptides with short peptide moieties, e.g., those derived from unspecific digests using Pronase, may furthermore access glycosylation sites not covered by specific proteases. Furthermore, suppression effects encountered for glycosylated peptides ionized in the presence of unmodified species may be avoided. Lastly, the consecutive analysis of proteins/peptides with glycans enables the glycoproteomic analysis of samples derived from PNGaseF digests, resulting in free N-glycans and peptide moieties; an example, therefore, is given for bovine ribonuclease B.

Besides the already mentioned applicative advantages, the one major asset of the system is its variability. In general, every two stationary-phase systems may be combined, as long as the loading buffer is compatible with both separation conditions. Therefore, other combinations of reversed-phase resins (C1-C18) as well as strong cation exchange chromatography and RP may be applied. Since both main columns are operated independently, buffer systems may be, furthermore, adjusted to distinct analytical problems. While 0.1% formic acid may be the additive of choice for peptides in positive ion mode, 5 mM NH4HCO3 can be more advisible for separation of acidic glycans by PGC in negative ion mode. Apart from changes in stationary phase and buffer systems, further modifications may be performed within the gradient system. Assuming, e.g., a major focus on the glycan profile of a recombinant glycoprotein, the protein/glycan mixture after the PNGaseF Analytical Chemistry, Vol. 82, No. 12, June 15, 2010

5395

digest can be analyzed via a fast RP gradient or step for protein ID, and the PGC gradient can be mutually prolonged to allow for extended glycan analysis. Otherwise, the RP precolumn can be used only for removing the proteins before PGC separation of glycans as a simple cleanup step. Therefore, the PGC column system is guarded against protein contaminations. In this case, the first gradient pump can even be omitted. In conclusion, the RP/PGC dual gradient system has been shown for the first time to provide a versatile and variable analysis tool for the combined analysis of mixtures of proteins, peptides, and oligosaccharides with potential use regarding proteomic and glycoproteomic samples. Its nanoscale design allows for reliable detection of low femtomole analyte

5396

Analytical Chemistry, Vol. 82, No. 12, June 15, 2010

amounts and may be adapted to a broad range of analytical problems. ACKNOWLEDGMENT The financial support by the Ministerium fu¨r Innovation,Wissenschaft, Forschung und Technologie des Landes NordrheinWestfalen and by the Bundesministerium fu¨r Bildung und Forschung is gratefully acknowledged. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review April 1, 2010. Accepted May 8, 2010. AC100853W