Created: 1st March 2000, last updated: 30th May 2000, © 2000 ABRF
Felicia Rusnak, Jie Zhou, and Gary M. Hathaway
California Institute of Technology, Beckman Institute, Pasadena, CA 91125
A method is described for constructing spin columns for reverse-phase centrifugal desalting of proteolytic digests. The technique employs small, self-packed, reusable cartridges and required less than 30 minutes to process six samples, making the procedure useful as a parallel technique. Up to 15 µL of sample could be loaded and eluted with 2 to 7 µL of a solvent compatible with electrospray ionization. The method was not limited to large-diameter resins or short column heights; a relative centrifugal driving force as low as 30 (500 rpm) applied for 1 to 3 minutes usually was sufficient for sample loading. Subsequently, data were obtained with 1 X 5-mm columns of 3- 5-, and 10-µm silica resins. The efficiently of recovery in the range of 0.5 to 250 pmol of peptides was measured as 60% to 90%, depending on resin type and sample load. Successful nanospray data were obtained with peptides that had been adulterated with 2 M urea and processed with a spin column. Matrix-assisted laser desorption and ionization/ time-of-flight mass spectrometry data greatly improved after desalting of an in-gel digest of a 280-kd protein. Data are presented on the preparation of columns, optimization of procedures, the use of various types of C18 resins, and the efficiency of peptide recovery. The effect of rotor speed and the rate of sample processing are discussed. (J Biomol Tech 2000;11:12-19)
Key Words: digest, desalting, mass spectrometry.
Address correspondence and reprint requests to: Gary M. Hathaway, California Institute of Technology, The Beckman Institute, Mail Code 139-74, 400 S. Wilson Avenue, Pasadena, CA 91125.
The presence of nonvolatile solutes such as chaotropes, buffer salts, reductants, and alkylating agents used in proteolytic digestions usually leads to poor or no results in postdigest peptide analysis by mass spectrometry.1-4 Consequently, various procedures based on reverse-phase,5-7 ion exchange,8 and hydrophilic interaction9 chromatography have been investigated to remove these substances while minimizing sample loss. For experiments that involve digestion of proteins in denaturing polyacrylamide gels, reverse-phase cleanup has become the method of choice because most of the contaminants encountered in experiments involving in-gel digestion are readily removed, and samples are recovered in small volumes of solvents compatible with mass spectrometry. This maximizes signals in the mass spectrometer and minimizes sample loss. Resin-filled Geloader pipette tips5 are particularly attractive for this procedure, because they are relatively inexpensive to produce and, being self-fabricated, are readily customized particularly for high-sensitivity applications such as nanospray mass spectrometry.5,6 However, we encountered certain difficulties in their use. Sample application and elution were difficult to control, and the choice of reverse-phase resin was severely limited because of the method's intolerance of high backpressure. These constraints, together with the fact that it is a relatively slow, labor-intensive technique, prompted us to develop a parallel process, centrifugal method.
The technique uses empty miniature guard cartridge inserts that, when filled with various reverse-phase resins, may be used to desalt or concentrate peptide digests. The procedure avoids the problem of backpressure by using radially generated centrifugal pressure as the driving force for sample loading and elution. Because the method is relatively insensitive to backpressure, it allowed a wide choice of column resins and the parallel processing of samples.
Reverse-phase spin columns were constructed in empty Optiguard inserts that contained only a lower frit. A slurry (1:5 settled volume in 100% methanol) of 3-µm Vydac, 5-µm Stablebond (Hewlett-Packard, USA), or 10-µm R2-10 Poros (Perseptive Biosystems, Foster City, CA) C18 resins was centrifuged at low speed. Final resin column height was based on calculations using the physical dimensions listed in the Materials section. All operations were carried out in a swinging bucket rotor (Brinkmann, Westbury, NY; Eppendorf model 5417 tabletop centrifuge) at room temperature with the spin cartridge placed inside a 0.650-mL or 1.5-mL microcentrifuge tube. Column construction, sample loading, and elution were performed at low centrifugal force (30 to 110 rcf; 500 to 1000 rpm). One of two techniques was employed, depending on whether sample concentrating or desalting was the objective. Cartridges were regenerated by attaching a short piece of tubing to the column bottom, filling the tubing with solvent, and centrifuging the inverted column.
In method I for concentrating samples, the resin slurry was loaded to give a column of the desired height. Then 10 µL of 2% acetonitrile (ACN) with 0.05% trifluoroacetic acid (TFA) were centrifuged through the column, followed by the sample in the low organic buffer. The eluate was removed. The sample was then eluted with two 5- to 7-µL aliquots of 75% ACN with 5% formic acid or 50% MeOH with 10% n-propanol.
Method II was sample desalting. Because of the hydrophobic nature of reverse-phase resin, it was critical that the column remain wetted during this procedure. To accomplish this, liquid is added external to the column to prevent the sample solution meniscus from falling below the top of the inner resin column. Because the outer solvent meniscus rose only 4% relative to the inside height above the column top, the outside level did not need to be adjusted after each wash cycle. After sample washing, the outer liquid was removed, and the sample was eluted with two 2- to 7-µL aliquots of high organic buffer. The method is illustrated in the drawing of Figure 1.
For proteolytic digestions, horse apomyoglobin peptides were prepared by denaturing 5 nmol of the protein in freshly deionized 8 M urea. The solution was diluted to a final concentration of 2 M urea in 0.1 M ammonium bicarbonate buffer at pH 8.5 (ie, buffer 1). Trypsin (1%) was added and incubated overnight at 30°C. Peptides were desalted using a Brownlee C18 guard cartridge, aliquoted, lyophilized, and stored dry at -15°C. Total peptides were quantitated from reconstituted aliquots by amino acid analysis in triplicate.
FIGURE 1. The spin column was filled with C18 resin to a height of 1 to 5 mm. For method I, samples were loaded in low organic buffer and centrifuged at the lowest setting of the rotor (500 rpm, 30 rcf). Usually, a period of 60 seconds was sufficient. For method II desalting, before loading the sample, the outer container was filled with approximately 250 µL of wash buffer(2% ACN/0.05% TFA) to a level just above the inner column bed. The sample was then loaded and washed two times with 10 to 15 µL of wash buffer. The outer liquid was removed, and the sample eluted twice with 2 to 7 µL of 75% methanol/1% formic acid for electrospray or 50% acetonitrile/10% propanol/0.05% TFA for MALDI-TOF.
In-gel digestion was performed with lysyl endopeptidase in 25 mM Tris HCl buffer at pH 9.2 (ie, buffer 2). The procedure of Hellman et al.10 was modified as follows. Neutralized tris(2-carboxyethyl)phosphine (2 mM in 50% ACN/buffer 2) was substituted for dithiothreitol and incorporated into the first dye extraction step to minimize extractive losses. The gel piece was rinsed once with buffer 2, and alkylation with iodoacetamide was carried out incorporating the reagent (25 mM in 50% ACN/buffer 2) into a second extraction step (room temperature and in the dark). After the gel-drying step, digestion was initiated by adding 0.25 µg of protease to the dried gel piece in 5 µL of buffer 2. This was repeated once. After overnight incubation at 30°C, peptides were extracted as outlined in Hellman et al.,10 dried under vacuum, and reconstituted to a final volume of 25 µL with 2% ACN and 0.05% TFA. A detailed description of the digest procedure may be found at our website (http:// www.its.caltech.edu/~Ppmal).
Electrospray mass spectrometric analysis was performed with a model API365 triple quadrupole (Perkin Elmer/Sciex, Ontario, CA). Matrix-assisted laser desorption and ionization/time-of-flight (MALDI-TOF) mass spectrometry was carried out with a Perseptive Biosystems' Voyager Elite equipped with delayed extraction optics (Foster City, CA). For MALDI-TOF mass spectrometry, the digest extract was mixed directly with an equal volume of alphacyano-4-hydroxycinnamic acid matrix (alphacyano) before and after desalting by method II.
Peptides recovered from spin column eluates were quantitated by integration of UV absorbance recordings from microcapillary HPLC. Samples from methods I and II along with controls were obtained in triplicate, dried in a Speedvac Concentrator, reconstituted with 25 µl of buffer A (2% ACN/0.05% TFA), and 20 µl was injected onto a 0.5 X 150-mm Reliasil C18 column (Column Engineering, Ontario, CA). Three to seven, well-separated peptides were chosen for integration, and total peak heights were averaged and compared with the control. Triplicate measurements gave peak heights reproducible to ±5%. For experiments carried out in 2 M urea, yields were estimated by electrospray mass spectrometry from a standard curve of apomyoglobin peptide concentration using peak height ratios with bradykinin as internal standard.
The 12-place swinging bucket rotor and Eppendorf centrifuge (model 5417) were purchased from Brinkmann (Westbury, NY). The model 172A HPLC obtained from Applied Biosystems (Foster City, CA) was modified with an LC Packings precolumn splitter and 35-nL flow cell (San Francisco, CA). EZchrom integrating software (Scientific Software, San Ramon, CA) was used to quantitate yields from the spin column experiments. Reliasil 0.5 X 150 mm, high-performance liquid chromatography (HPLC) columns were obtained from Column Engineering (Ontario, CA). Single-frit, empty Optiguard columns were obtained from Western Analytical (cat. no. 10-02-02179, Temecula, CA). Optiguard column specifications were measured as follows: upper chamber of 3-mm OD X 4.5-mm high; overall column height of 21 mm; stem height of 16 mm, frit height of 1.1 mm; stem volume of 13 µL; 1.00-µL column height of 1.2 mm. Modified trypsin was obtained from Promega Corporation (Madison, WI) and lysyl endopeptidase from Achromobacter lyticus was purchased from WAKO Fine Chemicals (Richmond, CA). Geloader 1- to 10-µL pipette tips (cat. no. 22 35 165-6) were obtained from Eppendorf (Madison, WI). Poros R2-10 C18 resin (cat. no. 820966-322) and TFA were obtained from Perseptive Biosystems (Foster City, CA). Tris(2-carboxyethyl)phosphine hydrochloride was purchased from Molecular Probes, Inc. (Eugene, OR). Zorbax Stablebond C18 was a kind gift from Hewlett-Packard, and Vydac 3-µm resin was obtained from Vydac (V3BZ-C18, Hesperia, CA). Microcentrifuge tubes, 0.65 mL and 1.5 mL, and ACN were purchased from VWR Scientific Products (San Dimas, CA).
To estimate the time required for sample loading and elution, a 5-mm column of Zorbax Stablebond C18 was constructed and centrifuged according to the procedure of method I. Solvent flow rates were measured gravimetrically after centrifuging for 2 to 3 minutes. As expected, flow rates were a linear function of relative centrifugal force (Fig. 2). The standard curve could then be used to predict the time needed to apply a given volume to the column.
FIGURE 2. Dependence of flow rate on relative centrifugal force (rcf). Flow was measured gravimetrically at intervals of 2 to 3 minutes. Packing was Zorbax Stablebond C18.
Initially, the amount of eluted peptide and the amount in the flow-through during sample application were measured, and the sum of the two recoveries was compared with the amount of applied sample. The results suggested that binding of the sample to the resin during sample application was critical (data not shown). To measure the effect of column height and resin type on sample recovery, the height and type of the resin bed were varied, and the amount of peptide recovered was measured after applying the sample at the lowest possible rotor speed. For these experiments, a relatively high load of 13 pmol of peptides was used to minimize error in measuring sample recovery by peak integration. Results were always compared with a control, which was run in parallel to samples applied to the column to eliminate the effect of sample loss from binding to the plastic sample containers and pipette tips used in the experiment.
Figure 3 shows the results of measuring sample recovery for three types of reverse-phase resins at various column heights. Quantitative data were obtained in triplicate from the integrated UV absorbance traces of five peptides in a postrun HPLC experiment, as outlined in the methods section. Recoveries were seen to decline for the shorter columns, and sample to sample variability increased. Peptide recovery for the three different resin types is also shown in Figure 3, with the 3- and 5-µm resins being slightly more efficient than the 10-µm resin. Based on these results, a column height of 5 mm was chosen as the best compromise.
FIGURE 3. Peptide recovery is compared with spin column bed volume and resin type. Recovery was measured for 13 pmol of apomyoglobin tryptic peptides using R2-10 Poros C18 resin (circles), Zorbax 5-µm C18 (triangles), and Vydac 3-µm C18 (squares) at various column heights. Data were determined by integrating the trace obtained by postrun chromatography of the eluted sample as outlined in the methods section. Error bars indicate the standard deviation obtained for three to six independent measurements.
To examine whether some peptides were being lost preferentially, HPLC tracings at a sample load of 1 pmol were compared with those of the control. Results (Fig. 4) indicated that some peptides were being lost preferentially, and as might be expected, the peptides lost to the greatest degree were poor binders as evidenced by their early elution in conventional reverse-phase chromatography (Fig. 4, arrows). Channeling was excluded as the major reason for reduced sample binding.
FIGURE 4. Absorbance tracings at 200 nm from the HPLC chromatography of a 1-pmol sample of apomyoglobin tryptic peptides loaded to a 5-mm column of Zorbax C18. The control sample was a tryptic digest of apomyoglobin. The 1-pmol control sample was dried in a 0.65-mL microcentrifuge tube, reconstituted in 2% ACN/0.05% TFA, and injected (upper trace). A 1-pmol sample was loaded to the spin column, eluted with 2 X 7 µL of 50% ACN/10% propanol/0.05% TFA, dried, and reconstituted as in the control and injected (lower trace). Notice the reduced yields of some peptides (arrows).
In an attempt to measure irreversible sample loss, peptide recovery in the flow-through and eluate were measured, and their combined value was plotted as a function of applied sample load. If no irreversible binding were occurring, the plot of the data from such an experiment would be expected to extrapolate to zero recovery at zero sample applied. On the other hand, irreversible binding should appear as the intercept at some positive value on the abscissa. This value estimates the minimum amount of sample required to overcome such losses and appear as recovered sample. The results shown in Figure 5 gave an abscissa intercept at an applied sample load equivalent to 185 ± 120 fmol.
FIGURE 5. Nonspecific losses from spin column concentrating. Various amounts of a tryptic digest of apomyoglobin were loaded to a 5-mm spin column as described in Figure 4. Experiments were performed in triplicate (error bars shown). A least squares fit was used to estimate the nonspecific sample loss from the extrapolated intercept on the abscissa for the value y = 0. A value of 185 ± 120 fmol was estimated.
To demonstrate the efficiency of the method's parallel processing, we timed ourselves with the Geloader pipette tip method and with the spin column. Three samples were processed in tandem with Geloader tips, including column construction, sample application, washing, and elution. The procedure took 90 minutes for the three Geloader columns. Thirty minutes were required to assemble and process six samples with spin columns. Reproducibility and recoveries for the spin column experiment are shown in the data of Figure 6.
FIGURE 6. Apomyoglobin tryptic peptides (13 pmol) were loaded to columns as described in Figure 4. Samples were processed in parallel, and 80% of the recovered sample was analyzed by HPLC chromatography. The control was dried and reconstituted but not applied to the spin column (C trace). Five replicate samples are shown (traces 1 through 5).
Next, experiments were carried out applying method II for sample desalting and mass spectrometry. Method II is illustrated in Figure 1, and 5-mm spin columns were used with Zorbax C18 resin. Because of the hydrophobic nature of reverse-phase resins, method II was devised to ensure that the resin would remain solvated during sample application and washing. The centrifugal method suggested that a simple way to achieve this was by matching the centrifugal force on two liquid columns. One column represents the liquid above the inner resin top, and the second column is a liquid placed external to the column device. The internal and external menisci reach equivalent levels at centrifugal equilibrium, preventing the inner liquid from falling below the resin. Because of the difference in diameter of the inner and outer liquid columns, the linear rise of the outer column was only 4% that of the inner column. Multiple washes could be carried out without removing any liquid from the outer solution. When washing was complete, the outer liquid was removed before the final elution step.
Figure 7 shows results comparing an untreated apomyoglobin tryptic digest infused at a concentration of 500 fmol/µL with one deliberately adulterated with 2 M urea and then desalted by method II. Bradykinin was included as an internal standard. The number of peptides recovered for the treated sample (Fig. 7, lower panel) compared well with the untreated peptides (Fig. 7, upper panel). Recovery was 25% for the urea-treated sample using measured peak heights/bradykinin ratios compared with a standard curve.
FIGURE 7. Electrospray mass spectrometry of apomyoglobin tryptic peptides recovered from 2 M urea using a 5-mm spin column of Zorbax C18. The untreated control was electrosprayed from 70% methanol/5% formic acid, infusing at 100 nL/minute, and at a peptide concentration of 500 fmol/1 µL, with bradykinin used as an internal standard (upper trace). A 5-pmol sample of peptides was applied in 2 M urea to the spin column (5-mm Zorbax), washed twice, and then eluted with 70% methanol/5% formic acid. The desalted sample was reconstituted to a volume of 10 µL and infused at 100 nL/min, with bradykinin used as an internal standard (lower trace). Notice that the peak at m/z 564 was not a peptide. Recovery was measured to be 25% from the peak ratios to the internal standard. No peaks were obtained for the sample in the presence of 2 M urea (data not shown).
To demonstrate the usefulness of the method for desalting in-gel digests for MALDI-TOF, an experiment was carried out digesting a 280-kd unknown protein with Achromobacter endolysl peptidase. The digested sample, contained in an SDS polyacrylamide gel slice was extracted, and 0.5 µL (2% of the total digest) was mixed with an equal volume of alphacyano matrix and applied to the target (Fig. 8, upper panel). An aliquot (30%) of the digest was desalted, the sample washed, and 0.5 µL of the eluate (1% of the total digest) was mixed with matrix and spotted to the target. As shown by the data (Fig. 8, lower panel), substantially more peaks were seen after the desalting procedure, particularly at higher m/z values, even though one half of the amount of peptide was spotted compared with the untreated sample. Both analyses gave a greater total number of peptide masses than would otherwise have been measured with either analysis alone.
FIGURE 8. MALDI-TOF of an endolysyl-C in-gel digest of a 280-kd unknown protein. The upper trace shows 2% of the extracted peptides, and the lower trace shows 1% of the extracted peptides after desalting by method II.
The method described here is a robust technique for rapidly desalting samples for electrospray and MALDI-TOF mass spectrometry. Yields exceeding 90% for tryptic peptides could be obtained provided 5-mm columns were used. Yields fell to as low as 60% at loads below 1 pmol. Some sample loss appeared to result from nonspecific binding, and an amount equal to 0.19 ± 0.12 pmol was experimentally measured. This result indicated that some sample loss is unavoidable in the operation, and although it may be as low as 70 fmol, the data caution against the use of the method for sample amounts of less than 0.5 pmol. Results indicated that the minimum amount of sample probably also depends on the chemical nature of the peptide.
Although we used new resin in this study, the method has the advantage of being able to use resins scavenged from HPLC columns that are no longer useful for high-resolution chromatography. This further reduces costs associated with processing multiple samples. Six columns were prepared and the sample loaded, washed, and eluted in the same time required to process a single Geloader pipette tip. Up to 12 samples could be processed in parallel with the centrifugal technique in the swinging bucket rotor. Spin columns could be constructed in advance and left overnight without degradation, provided the columns were "rewet" with solvent before use (data not shown). Although the 3-µm resin functioned no better than the 5-µm resins, it illustrated the method's relative insensitivity to backpressure. Investigation of the method's applicability to a wide range of resin types including ion-exchange and even hydroxyapatite is warranted.
In conclusion, centrifugal desalting has been demonstrated to be a simple, robust, parallel technique, useful for applying self-constructed desalting columns to proteolytic digests before electrospray- or matrix-assisted laser desorption and ionization mass spectrometry.
Support for this research was provided by The Caltech Beckman Institute.
1. Lee TD, Shively JE. In McCloskey JA (ed): Methods in Enzymology, vol 193: Mass Spectrometry. San Diego: Academic Press, 1990:361-374.
2. Jeannot MA, Zheng J, Li L. Observation of sodium gel-induced protein modifications in dodecylsulfate polyacrylamide gel electrophoresis and its implications for accurate molecular weight determination of gel-separated proteins by matrix-assisted laser desorption ionization time-of-flight mass spectrometry. J Am Soc Mass Spectrom 1999;10:512-520.
3. Xiang F, Beavis RC. A method to increase contaminant tolerance in protein matrix-assisted laser desorption/ ionization by the fabrication of thin protein-doped polycrystalline films. Rapid Commun Mass Spectrom 1994; 8:199-204.
4. Johnson KL, Veenstra TD, Londowski JM, Tomlinson AJ, Kumar R, Naylor S. On-line sample clean-up and chromatography coupled with electrospray ionization mass spectrometry to characterize the primary sequence and disulfide bond content of recombinant calcium binding proteins. Biomed Chromatogr 1999;13: 37-45.
5. Wilm M, Shevchenko A, Houthaeve T, Breit S, Schweigerer L, Fotsis T, Mann M. Femtomole sequencing of proteins from polyacrylamide gels by nanoelectrospray mass spectrometry. Nature 1996;379:466-469.
6. Erdjument-Bromage H, Lui M, Lacomis L, Grewal A, Annan RS, McNulty DE, Carr SA, Tempst P. Examination of micro-tip reversed-phase liquid chromatographic extraction of peptide pools for mass spectrometric analysis. J Chrom A 1998;826:167-181.
7. Chin ET, Papac DI. The use of a porous graphitic carbon column for desalting hydrophilic peptides prior to matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. Anal Biochem 1999;273:179-185.
8. Sheer DG, Kellard E, Kopaciewicz W. A new centrifugal device used in sample cleanup and concentration of peptides. Tech Protein Chem 1997;8:133-142.
9. Swiderek KM, Klein M, Hefta SA, Shively J. Strategies for the removal of ionic and nonionic detergents from protein and peptide mixtures for on- and off-line liquid chromatography mass spectrometry (LCMS). Tech Protein Chem 1995;6:267-275.
10. Hellman U, Wernstedt C, Gonez J, Heldin CH. Improvement of an "in-gel" digestion procedure for the micropreparation of internal protein fragments for amino acid sequencing. Anal Biochem 1995;224:451-455.