Deamidation of -Asn-Gly- Sequences during Sample Preparation for

Anal. Chem. 2006, 78, 6645-6650

Deamidation of -Asn-Gly- Sequences during Sample Preparation for Proteomics: Consequences for MALDI and HPLC-MALDI Analysis Oleg V. Krokhin,*,†,‡ Mihaela Antonovici,† Werner Ens,†,‡ John A. Wilkins,† and Kenneth G. Standing†,‡

Manitoba Centre for Proteomics and Systems Biology, University of Manitoba 799 JBRC, 715 McDermot Avenue, Winnipeg, MB, R3E 3P4, Canada, and Department of Physics and Astronomy, University of Manitoba, Winnipeg, MB, R3T 2N2, Canada

We find that peptides containing -Asn-Gly- sequences typically show ∼70-80% degree of deamidation after standard overnight (∼12 h) tryptic digestion at 37 °C. This emphasizes the need for more detailed information about the deamidation reaction in -Asn-Gly- sequences, in which two deamidated species are produced, one containing an aspartic acid (-Asp-Gly-) residue and the other containing an isoaspartic acid (-βAsp-Gly-) residue. For the peptide SLNGEWR (54-60 β-galactosidase, E. coli), all three components of the reaction mixture were separated by HPLC on C18 300-Å sorbent, with trifluoroacetic acid as an ion-pairing modifier. Their intensity ratios suggested the elution order -βAsp-/-Asn-/-Asp-, which was subsequently confirmed by MALDI MS and MS/MS analysis. The kinetics of the deamidation was studied in detail for the synthetic SLNGEWR parent using RP HPLC with UV detection. The half-life of this peptide was found to be ∼ 8 h under digestion conditions. Analysis of a large pool of peptide retention data shows that the -βAsp-/-Asn-/ -Asp- retention order is normally observed under the above conditions, especially if the original -NG- sequence is surrounded by hydrophobic amino acids. However, changing chromatographic conditions to 100-Å pore size sorbents, or using formic acid as a modifier, increases the retention time of -βAsp- relative to the -Asn-/-Asp- pair, so the order can sometimes be different. Deamidation of Asn and Gln residues is one of the best known and studied posttranslational modifications in proteins.1 Since the degree of deamidation often controls the biological activity and functions of proteins,2 the reaction is particularly important in large-molecule therapeutics. Nonenzymatic deamidation of -Asnresults in gradual formation of isomeric -Asp- and -βAsp- (iso-Asp) residues in a ∼1:3 ratio3 through the intermediate aminosuccin* Corresponding author. Tel: (204) 474 6184. Fax: (204) 474 7622. E-mail: [email protected]. † Manitoba Centre for Proteomics and Systems Biology. ‡ Department of Physics and Astronomy. (1) Aswad, D. W., Ed. Deamidation and isoaspartate formation in peptides and proteins; CRC Press: Boca Raton, FL, 1995. (2) Friedman, A. R.; Ichhpurani, A. K.; Brown, M. D.; Hillman, R. M.; Krabill, L. F.; Martin, R. A.; Zurcher-Neely, H. A.; Guido, D. M. Int. J. Pept. Protein Res. 1991, 37, 14-20. (3) Geiger, T.; Clarke, S. J. Biol. Chem. 1987, 262, 785-794. 10.1021/ac061017o CCC: $33.50 Published on Web 07/21/2006

© 2006 American Chemical Society

imidyl (Asu) derivative (Scheme 1). The process takes place relatively slowly for intact proteins (half-life 1-500 days for Asn, and 100-5000 days for Gln4). However, the rate of deamidation can increase dramatically when the sensitive residues are exposed during proteolytic digestion. The -Asn-Gly- sequence is especially prone to deamidation.3,5,6 In our laboratory overnight tryptic digestion (12 h at 37 °C in ammonium bicarbonate buffer, pH 8.2) is a part of the usual sample preparation prior to MS analysis. This typically results in ∼70-80% conversion of -Asn-Gly- into -Asp-Gly- or -βAsp-Gly. The -Asn-Ser- sequence is also subject to deamidation, but it typically shows a negligible (∼10%) degree of degradation during the ∼12-h digestion, so the problem is acute only for peptides containing the -Asn-Gly- sequence. Such a digestion is part of the majority of MS-based proteomics protocols, so it is important to investigate the impact of the deamidation reaction on MS and HPLC-MS protein analysis. Robinson et al.7 used direct ESI MS to study degree of deamidation in a series of synthetic peptides. As both deamidated components are shifted in mass by +0.984 Da, an error of ∼1 Da is likely to arise in any mass fingerprinting measurement if deamidation is neglected. The m/z spectrum itself may show an isotopic pattern7 that suggests deamidation (see Figure 1a and e), but this is far from definitive. MS/MS measurements are likely to be affected by deamidation as well, since the precursor ion peaks may contain a mixture of the three components. Thus, recently Venable et al.8 applied a cross-correlation algorithm to derive the accurate molecular weight of a peptide directly from corresponding tandem mass spectrum. Using complimentary fragment ions, the authors have significantly improved the mass accuracy for precursor ions in low-resolution ion trap measurements. Application of the algorithm to the yeast whole-cell lysate exhibited a significant group of peptides with +1 Da mass difference, which was properly attributed to deamidation of Asn and Gln residues.8 (4) Robinson, N. E.; Robinson, A. B. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 12409-12413. (5) Bischoff, R.; Kolbe, H. V. J. Chromatogr., B 1994, 662, 261-278. (6) Capasso, S. J. Pept. Res. 2000, 55, 224-229. (7) Robinson, N. E.; Robinson, A. B.; Merrifield, R. B. J. Pept. Res. 2001, 57, 483-493. (8) Venable, J. D.; Xu, T.; Cociorva, D.; Yates, J. R., III. Anal. Chem. 2006, 78, 1921-1929.

Analytical Chemistry, Vol. 78, No. 18, September 15, 2006 6645

Scheme 1

A key element for successful identification and, if needed, relative quantification of all three components of deamidation reaction is complete separation before the MS measurement. Such separation of -Asn- containing proteins or peptides from their -Aspcounterparts can be achieved based on differences in pI9,10 or hydrophobicity. Separation of -Asp- and -βAsp- derivatives on intact ribonuclease A protein by hydrophobic interaction LC has been reported,9 but digestion of a protein coupled with deamidation studies on the resulting peptides offers better detection sensitivity (femtomole to picomole). Decreasing the size of the analyzed species by digestion may also produce better separation between -Asp- and -βAsp- containing species. For example, recently RP HPLC separation of peptides on a C18 300-Å pore size column was used to quantify in vivo deamidation of a monoclonal IgG1 antibody and was found to yield complete separation of all three components in the reaction mixture,11 with ESI-MS/MS detection. In this case, the retention order for the ILPGNGNINYNEK peptide and its degradation products was reported to be -NG-/-βDG-/-DG-, although the authors did not attempt MS/MS differentiation between the two deamidated components and assigned peaks on the basis of their ∼3:1 relative intensity ratio.11 Unmodified peptides can be easily distinguished from deamidated species by mass measurement of the product ions in MS/ MS mode, but distinguishing the -βAsp- and -Asp- deamidated components has been more difficult. However, Carr et al.12 and later Gonzalez et al.13 have reported methods for unambiguous assignment of -βAsp- versus -Asp- containing peptides by high(9) Di Donato, A.; Ciardiello, M. A.; de Nigris, M.; Piccoli, R.; Mazzarella, L.; D′Alessio, G. J. Biol. Chem. 1993, 268, 4745-4751. (10) Capasso, S.; Di Cerbo, P. J. Pept. Res. 2000, 56, 382-387. (11) Huang, L.; Lu, J.; Wroblewski, V. J.; Beals, J. M.; Riggin, R. M. Anal. Chem. 2005, 77, 1432-1439. (12) Carr, S. A.; Hemling, M. E.; Bean, M. F.; Roberts, G. D. Anal. Chem. 1991, 63. 2802-2824.

6646 Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

Figure 1. MALDI MS and HPLC-MALDI MS monitoring of deamidation for the peptides SLNGEWR (54-60 β-galactosidase, E. coli), and VYAENGITR (144-152 glucose-6-phosphate dehydrogenase, yeast). (a, e) Spectra of the relevant parts of the m/z spectra from unseparated digests of β-galactosidase and glucose-6-phosphate dehydrogenase, respectively. Fractions b-d and f-h represent the m/z spectra of the corresponding HPLC fractions, again for SLNGEWR and VYAENGITR, respectively. The data set for peptide retention prediction was initially collected with a Vydac 218 TP C18 column and TFA as the ion-pairing modifier. Calculated m/z values: SLNGEWR 861.422 (862.406 after deamidation); VYAENGITR 1022.527 (1023.511 after deamidation).

energy and low-energy collision-induced dissociation MS/MS, respectively. Gonzalez and co-workers found a particular pair of internal rearrangements only in the -βAsp- containing ions; these yielded either (y′′l-n+1 - 46) ions, or (bn-1 + H2O) ions, depending on the position of the basic residues within the peptide chain.13 Recently we studied in detail another abundant chemical modificationsthe loss of ammonia for N-terminal Gln and Cys alkylated with iodoacetamide.14,15 Deamidation of -Asn-Gly- sequences also falls into the category of reactions that significantly influences the results of MS analysis and are often overlooked by proteomics researchers. In connection with our studies of retention times prediction in HPLC measurements, we have accumulated data on an extensive set of ∼2000 peptides, using an off-line combination of RP HPLC-MALDI MS,16-18 and have (13) Gonzalez, L. J.; Shimizu, T.; Satomi, Y.; Betancourt, L.; Besada, V.; Padron, G.; Orlando, R.; Shirasawa, T.; Shimonishi, Y.; Takao, T. Rapid Commun. Mass Spectrom. 2000, 14, 2092-2102. (14) Geoghegan, K. F.; Hoth, L. R.; Tan, D. H.; Borzilleri, K. A.; Withka, J. M.; Boyd, J. D. J. Proteome Res. 2002, 1, 181-187. (15) Krokhin, O. V.; Ens, W.; Standing, K. G. Rapid Commun. Mass Spectrom. 2003, 17, 2528-2534. (16) Krokhin, O. V.; Ens, W.; Standing K. G. J. Biomol. Tech. 2005, 16, 429440. (17) Krokhin, O. V.; Craig, R.; Spicer, V.; Ens, W.; Standing, K. G.; Beavis, R. C.; Wilkins, J. A. Mol. Cell. Proteomics 2004, 3, 908-919. (18) Krokhin, O. V.; Ying, S.; Craig, R.; Spicer, V.; Ens, W.; Standing, K. G.; Beavis, R. C.; Wilkins, J. A. 52nd ASMS Conference on Mass Spectrometry and Allied Topics, Nashville, TN, 2004; poster.

observed many cases of deamidation. The highest deamidation rate was found where -Asn- is followed by -Gly-, as described above, so we have concentrated our studies on this reaction. This paper summarizes our findings and demonstrates the potential of an off-line combination RP HPLC-MALDI MS approach for separation and identification of the deamidated components. EXPERIMENTAL SECTION Reagents. Deionized water (18 MΩ) and HPLC-grade acetonitrile were used for the preparation of eluents. Trifluoroacetic acid (TFA), formic acid (FA), 2,5-dihydroxybenzoic acid (DHB), dithiothreitol, and iodoacetamide were obtained from SigmaAldrich (St. Louis, MO). Digestion was carried out by sequencinggrade modified trypsin (Promega, Madison, WI). Proteins and Peptides. Commercially available proteins (Sigma-Aldrich) were digested to provide peptide mixtures, which were then separated by HPLC to obtain an extensive data set (∼2000 peptides) for prediction of peptide retention times in RP HPLC/MALDI MS or -(MS/MS) analysis.17,18 Particular peptides examined included the following: SLNGEWR (54-60 β-galactosidase, Escherichia coli), VYAENGITR (144-152 glucose-6-phosphate dehydrogenase, yeast), ELINSWVESQTNGIIR (143-158 ovalbumin, chicken), VGVNGFGR (3-10 glyceraldehyde-3phosphate dehydrogenase, rabbit), DFPIANGER (19-27 carbonic anhydrase II, bovine), and LNGFEVFAR (361-369 integrin alpha V, human). The peptide SLNGEWR, as well as SLDGEWR (containing aspartic acid) and SLβDGEWR (containing isoaspartic acid), were also custom synthesized by BioSynthesis Inc. (Lewisville, TX). Sample Preparation. Protein samples were reduced (10 mM dithiothreitol, 30 min, 57 °C), alkylated (50 mM iodoacetamide, 30 min in the dark at room temperature), dialyzed (100 mM NH4HCO3, 6 h, 7 kDa molecular weight cutoff, Pierce), and digested with trypsin (1/50 enzyme/substrate weight ratio, 12 h, 37 °C). Each unseparated protein digest was mixed 1:1 with DHB matrix solution, deposited on a gold-plated MALDI target, airdried, and subjected to peptide mass fingerprint analysis to confirm protein identity. Chromatography and Fraction Collection. Chromatographic separations were performed with a micro-Agilent 1100 Series system (Agilent Technologies, Wilmington, DE) equipped with a 50-nL flow cell UV-visible detector. Three different sorbents were used: Vydac 218 TP C18, 300-Å pore size, 5 µm (Grace Vydac, Hesperia, CA), Zobrax SB300, 300-Å pore size, 5 µm (Agilent), and C18 PepMap100, 100-Å pore size, 5 µm (Dionex, Sunnyvale, CA). Samples (5 µL, ∼2 pmol of protein digest/injection) were injected directly onto a 150 µm × 150 mm RP column (300 µm × 150 mm for PepMap) and eluted with a linear gradient of 1-80% acetonitrile (0.1% TFA or 0.15% formic acid) in 120 min (0.66% acetonitrile/min) at 4 µL/min flow rate. The column effluent (4 µL/min) was mixed on-line with DHB MALDI matrix solution (0.5 µL/min, 150 mg/mL DHB in water/acetonitrile 1:1), deposited as a set of spots by a computer-controlled robot19 onto a movable gold target at 0.5-min intervals, and then air-dried. Most tryptic peptides eluted within 60 min, so collection was limited to 120 fractions. (19) Krokhin, O.; Qian, Y.; McNabb, J. R.; Spicer, V.; Ens, W.; Standing, K. G. 50th ASMS Conference on Mass Spectrometry and Allied Topics, Orlando, FL, 2002; poster.

TOF Mass Spectrometry. The spots from each individual digest were analyzed by single mass spectrometry (MS) with m/z range 560-5000, and by tandem mass spectrometry (MS/MS) in the Manitoba/Sciex prototype MALDI quadrupole/TOF (QqTOF) mass spectrometer20 with a mass resolving power of ∼10 000 fwhm and 10 ppm mass accuracy. The “m/z” program (Manitoba Centre for Proteomics, www.proteome.ca) was used for peak assignment of MS spectra using a signal-to-noise ratio set to 2.5. RESULTS AND DISCUSSION As noted above, our standard sample preparation includes a ∼12-h tryptic digestion at 37 °C in ammonium bicarbonate buffer (pH 8.2). Typical spectra from samples containing the -NGsequence after this procedure are shown in Figure 1a-d for the peptide SLNGEWR and (e-h) for the peptide VYAENGITR, whose calculated m/z values without deamidation are 861.422 and 1022.527. Panels a and e in Figure 1 show the spectra without chromatographic separation; clearly they contain substantial deamidated components (calculated m/z values 862.406 and 1023.511). The decrease in the unmodified peak intensities, together with the appearance of these new peaks (with calculated mass shifts of +0.984 Da), is likely to cause mistakes in peak assignment during MALDI MS mass fingerprinting experiments. Identification of Degradation Products. Chromatography provides more explicit information, shown in Figure 1b-d and f-h. In each case, the components of interest are present in three fractions. The middle fractions (c and g) contain appreciable amounts of the unmodified peptides, but only deamidated products are found in the other two fractions (43 and 47 for SLNGEWR, 36 and 39 for VYAENGITR). Retention time models17,21 predict shorter retention for Asn-containing peptides than for those containing -Asp-, suggesting that the latest fractions are the ones containing -Asp-, and implying in turn that the earliest fractions contain -βAsp-. Higher peak intensity for deamidated species in Figure 1 (b, f compared to d, h, respectively) also suggests a -βAsp-/-Asn-/-Asp- elution order. This was confirmed directly by measurements on the synthetic peptides SLNGEWR, SLDGEWR, and SLβDGEWR. The peptides and their mixtures (in various proportions) were analyzed by straight MALDI and microflow HPLC, with both MALDI MS (MS/ MS) and UV detection (results not shown). In all these measurements, the deamidated peptide containing -βAsp- was found to elute first, followed by the unmodified peptide (-Asn-), and then by the deamidated peptide containing -Asp-; i.e., the retention order was -βAsp-/-Asn-/-Asp-. MS/MS experiments were also carried out on the two peptides (SLNGEWR) and (VYAENGITR),as well as on four other tryptic peptides that showed similar clear separation into three fractions: (ELINSWVESQTNGIIR), (VGVNGFGR), (DFPIANGER), and (LNGFEVFAR). Low-energy CID MS/MS spectra were acquired for each potential -βAsp-/-Asp- pair. Corresponding spectra for SLβDGEWR/SLDGEWR and VYAEβDGITR/VYAEDGITR are shown in Figure 2. As expected, these spectra were found to be very similar for both peptide pairs. The only distinctive (20) Loboda, A. V.; Krutchinsky, A. N.; Bromirski, M.; Ens, W.; Standing, K. G. Rapid Commun. Mass Spectrom. 2000, 14, 1047-1057. (21) Guo, D.; Mant, C. T.; Taneja, A. K.; Parker, J. M. R.; Hodges, R. S. J. Chromatogr. 1986, 359, 499-517.

Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

6647

Figure 3. Monitoring of deamidation of pure SLNGEWR peptide by MALDI MS and HPLC. An ∼4 pmol/µL solution of SLNGEWR in 100 mM ammonium bicarbonate was incubated at 37 °C and analyzed at different time points by MALDI MS (m/z spectra shown in the insets) and micro-HPLC with UV detection at 214 nm. Figure 2. Distinguishing between -Asp- and -βAsp- deamidation products by low-energy collision-induced dissociation MS/MS. MS/ MS spectra: (a) from the SLβDGEWR precursor ion (shown in Figure 1b); (b) from the SLDGEWR precursor ion (shown in Figure 1d); (c) from the VYAEβDGITR precursor ion (shown in Figure 1f); (d) from the VYAEDGITR precursor ion (shown in Figure 1 h). The corresponding y′′l-n+1 - 46 fragments are shown in the insets.

feature is the presence of a small y′′l-n+1 - 46 peak in all the earlyeluting peptides; they therefore contain -βAsp-, according to the results of Gonzalez et al.13 (for peptides with a basic residue (Lys, Arg) at the C-terminal). For example, the VYAEβDGITR peptide consists of nine amino acids with -βAsp- in the fifth position. The corresponding y5 - 46 ion (9 - 5 + 1 ) 5) appears in this spectrum at m/z ) 515.297 (Figure 2c). MS/MS for all pairs were collected using the same collision energy of ∼50 V per 1000 Da. Although the relative abundances of the major product ions seem identical, we found that for all -βAsp-/-Asp- pairs the relative height of the precursor ion compared to the product ions was always smaller for -βAsppeptides (Figure 2). This may serve as one more feature to distinguish the two deamidation products. Therefore, the three independent measurements, relative intensities of deamidated species, analysis of synthetic peptides (SLNGEWR, SLDGEWR, SLβDGEWR) by RP HPLC with UV detection, and low-energy CID, all confirm the -βAsp-/-Asn-/-Aspretention order for peptides discussed above. Study of Deamidation rate at 37 °C, pH 8.2. Since deamidation produces such a large effect on the MS profile for peptides containing the NG sequence, it is worthwhile to study the kinetics of this reaction. A solution of pure SLNGEWR peptide was prepared in 100 mM ammonium bicarbonate and stored at 37 °C for 16 h. Portions of the solution were extracted at different times (0, 1, 2, 4, 6, 8, 10, 12, 14, and 16 h) and analyzed by both MALDI MS and microflow HPLC with UV detection (Figure 3). Corresponding MALDI MS spectra for five time points are shown in the insets in Figure 3. The gradual transformation of m/z 861.422 into m/z 862.406 is consistent with the results described 6648 Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

Figure 4. Monitoring of the deamidation reaction for the SLNGEWR peptide micro-HPLC with UV detection. Conditions as in Figure 3.

above for the corresponding peptide in the real β-galactosidase digest (compare the 12-h points in Figure 3 and Figure 1a). Chromatographic UV profiles are more appropriate for the quantitative study of the deamidation process than mass spectra, because deamidation is likely to decrease the MALDI response for a particular peptide, while UV absorbance likely remains the same. In addition, RP HPLC can often separate all three components of the reaction mixture. The five chromatograms in Figure 3 indicate gradual transformation of the intact peptide (-NG-) into two deamidation products containing (-βDG- and -DG-). The ratios of individual peak areas to the total peak area of the three components were calculated for all time points (Figure 4). Thus, a half-life of ∼8 h was found for the SLNGEWR peptide in 100 mM NH4HCO3 solution at 37 °C, a value in good agreement with the data of the corresponding peptide in the real β-galactosidase digest, which shows 70-80% degradation in ∼12 h. Chromatographic Retention of -βAsp-/-Asn-/-Asp- Triads in RP HPLC (the Bigger Picture). The results just presented suggest that the retention order for C18 300-Å pore size sorbents with TFA as modifier will always be -βAsp-/-Asn-/-Asp-. However,

Figure 5. Separation selectivity of the SLNGEWR peptide and its deamidation products on three different C18 sorbents with trifluoroacetic acid and formic acid as ion-pairing modifiers. (a, d) Vydac 218 TP C18, 300-Å pore size; (b, e) Zobrax SB300, 300-Å pore size; (c, f) PepMap100, 100-Å pore size. (a-c) 0.1% trifluoroacetic acid; (d-f) 0.15% formic acid. The SLNGEWR peptide was incubated for 12 h, as in Figure 3. Chromatographic conditions are described in Experimental Section.

the results apply only to those clear-cut cases selected above for complete separation of the three components. Actually, our set of ∼2000 peptides included some cases of coelution of -βAsp- and -Asn- or even reverse elution of this pair. A recent paper by Huang et al.11 also indicated an -Asn-/-βAsp-/-Asp- elution order for the ILPGNGNINYNEK peptide on C18 300-Å pore size sorbents with formic acid as modifier. This stimulated us to summarize the retention data for our larger set of -NG- containing peptides and to try different chromatographic conditions. Our ∼2000-peptide data set was found to contain 48 -NGcontaining peptides. We found that all peptides carrying -NGsequences undergo greater than 50% deamidation under the conditions described above. Since a small change in amino acid composition is unlikely to affect the retention times of large peptides, 10 species with mass higher than 3000 Da were excluded from further consideration. Out of the remaining 38, the -βAsp-/ -Asn-/-Asp- order was established for 22 cases, whereas the remaining 16 showed mostly coelution of the -βAsp-/-Asn- pair. Changing chromatographic conditions may alter the retention characteristics of peptides as well. A sample of the SLNGEWR peptide after 12-h digestion was separated on different columns and with formic acid as an ion-pairing modifier. Figure 5 shows separations on three different columns Vydac TP218 (a, d), Zorbax SB300 (b, e), and PepMap100 (c, f) with two different modifiers 0.1% TFA (a-c) and 0.15% FA (d-f). Separation selectivity on two 300-Å pore size C18 sorbents with TFA in the eluent was found to be virtually identical (a, b). Application of a 100-Å pore size column under the same conditions increased retention of the -βAsp- derivative compared to -Asn- but left the retention order unchanged (Figure 5c). However, application of FA dramatically increased the relative retention of the SLβDGEWR peptide for all columns. For example, the -βAsp-/-Asn- pair coelutes at these

conditions on a 300-Å pore size column (d, e), and reversed retention order -Asn-/-βAsp- was found for PepMap100 (Figure 5f). Impact of Deamidation on Proteomics Analysis. Nonenzymatic deamidation of -NG- sequences during proteolytic digestion affects results of both MS and MS/MS analyses, as m/z values for the products of this reaction are shifted by +0.984 Da compared to the original peptide. Thus, a significant degree of deamidation may cause problems with correct mass assignments and consequent misidentification of the observed peptides. Moreover, quantitative studies require the degree of nonenzymatic deamidation during sample preparation to be closely controlled. For example, the starting point for in vivo studies of monoclonal IgG1 antibody exhibited ∼25% deamidation,11 produced, to a large extent, during ∼3-h digestion at 37 °C. Quantitative assessment of the degree of deamidation requires complete separation of the original peptide and its products. As one possibility of alleviating these problems, the digestion time could be shortened to 3-4 h. This will still produce a mixture of intact peptide and deamidation products, but a lower degree of transformation (∼25%), which should make peak assignments more reliable. However, shorter digestion times may require an application of bigger enzyme/substrate ratios to reduce the number of missed cleavage sites. Moreover, highly deamidated proteins are sometimes found in the sample before the digestion, depending on the degree of in vivo deamidation and the sample history. In that situation, it is probably advantageous to allow explicitly for the presence of deamidation, i.e., to perform a long digestion as usual, but to include deamidation of -NG- sequences in the list of complete chemical modifications that are considered in the following analysis. Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

6649

Analysis of deamidation products is an example where chromatographic methods may provide better results than a pure MS approach. If the three components can be separated by HPLC, they can be assigned first by mass, and second by ratio of peak intensities,11 or else by MS/MS,13 to distinguish between -Aspand -βAsp- derivatives. Our results suggest that using TFA as modifier on 300-Å pore size C18 columns will likely provide -βAsp-/-Asn-/-Asp- retention, but analysis of peptide sequences with this retention behavior showed that most of them feature -NG- sequences surrounded by hydrophobic amino acid residues. Hydrophilic neighbors, such as proline residues, provide equal or higher retention of -βAspcompared to -Asn-. Such discrepancies point out once again that retention of peptides in RP HPLC cannot be predicted accurately based on additive retention models; sequence-specific algorithms17,18 seem to be more promising. Switching to formic acid and 100-Å pore size increases the retention of -βAsp- compared to the -Asn-/-Asp- pair. Knowing such rules helps to optimize chromatographic conditions for each particular case of deamidation being studied. CONCLUSIONS Deamidation of -Asn-Gly- sequences during sample preparation significantly affects peptide and protein identification during

6650

Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

standard proteomic protocols. The rate of decomposition at 37 °C, pH 8.2 was studied for a synthetic peptide (SLNGEWR); 50% decomposition was found for 8-h incubation, and a typical overnight (12 h) digestion resulted in ∼80% decay. This result is typical for a number of real digests observed in our studies. All three components of the reaction mixture were separated by ionpair RP HPLC on 300-Å pore size C18 sorbent and unequivocally assigned by MS and low-energy CID measurements. The retention order -βAsp-/-Asn-/-Asp- was demonstrated for this type of sorbent and trifluoroacetic acid as modifier. However, the separation selectivity could be changed by varying the pore size of the separation media and the ion-pairing modifier. ACKNOWLEDGMENT This work was supported by grants from the Canadian Institutes of Health Research (J.A.W.), from the Natural Sciences and Engineering Research Council of Canada (K.G.S., W.E.), and from the U.S. National Institutes of Health (GM 59240, K.G.S.).

Received for review June 2, 2006. Accepted June 20, 2006. AC061017O