Matrix Suppression as a Guideline for Reliable Quantification of

Aug 23, 2013 - Medical Proteomics Research Center, KRIBB, Daejeon 305-806, Korea ... suppression becomes a guideline for reliable quantification, rath...
0 downloads 0 Views 697KB Size
Download PDF
Article pubs.acs.org/ac

Matrix Suppression as a Guideline for Reliable Quantification of Peptides by Matrix-Assisted Laser Desorption Ionization Sung Hee Ahn,† Yong Jin Bae,† Jeong Hee Moon,‡ and Myung Soo Kim*,† †

Department of Chemistry, Seoul National University, Seoul 151-742, Korea Medical Proteomics Research Center, KRIBB, Daejeon 305-806, Korea

‡

ABSTRACT: We propose to divide matrix suppression in matrixassisted laser desorption ionization into two parts, normal and anomalous. In quantification of peptides, the normal effect can be accounted for by constructing the calibration curve in the form of peptide-to-matrix ion abundance ratio versus concentration. The anomalous effect forbids reliable quantification and is noticeable when matrix suppression is larger than 70%. With this 70% rule, matrix suppression becomes a guideline for reliable quantification, rather than a nuisance. A peptide in a complex mixture can be quantified even in the presence of large amounts of contaminants, as long as matrix suppression is below 70%. The theoretical basis for the quantification method using a peptide as an internal standard is presented together with its weaknesses. A systematic method to improve quantification of high concentration analytes has also been developed.

S

irreproducibility remains.20 Recently, we observed that irreproducibility could be largely eliminated by selecting spectra associated with the same effective temperature in the early plume (Tearly) where in-source decay (ISD)21 occurred.20,22 Then, the peptide concentration in a solid sample became the only factor that determined ion abundances in the MALDI spectrum. Our interpretation of the above correlation is that the temperature of the irradiated surface determines both Tearly and the ion abundances in a spectrum, regardless of the experimental condition. On the basis of this finding, we devised a method named “total ion count (TIC) control”23 to generate quantitatively reproducible spectra. It has been known that the presence of an analyte at a sufficiently high concentration in a sample reduces the matrix ion signal (matrix suppression)24,25 and that of another analyte(s) in the sample (analyte suppression).25,26 In our study of the thermal determination of MALDI spectra, we found that the increase in the total abundance of analytederived ions was matched by the corresponding decrease in that of the matrix-derived ions, a quantitative example of matrix suppression.22,27 From the spectra acquired at the same Tearly, we also measured the reaction quotient (Q) for the matrix ([M + H]+)-to-peptide (P) proton transfer.20

ince most biological processes are controlled by proteins, the identification and quantification of most of the proteins in a proteome are important in systems biology.1−4 Even though mass spectrometry is a powerful technique for identification and de novo sequencing of proteins,5 it is not a routine quantification tool for proteins.6−8 For quantification of a protein, it is cleaved to peptides by a proteolytic reagent such as trypsin.9,10 Then, the resulting peptides are quantified by methods that use internal standards prepared in various ways. In the simplest and most powerful approach called “absolute quantification” (AQUA),11 an analogue of the peptide to be analyzed that is multiply labeled with stable isotopes such as 13C, 15N, and 18O is used as the internal standard. AQUA can be costly in both time and money. Chemical tagging1,2 and metabolic incorporation2,6,12 of isotopes are the attempts to devise strategies that are less costly. The least costly strategy is to use a peptide with properties similar to those to be analyzed as the internal standard.13 Liquid chromatography (LC) combined with electrospray ionization mass spectrometry (ESI-MS) and matrix-assisted laser desorption ionization (MALDI) of samples separated by a two-dimensional gel are widely used for protein quantification.2,14,15 In both cases, suppression of analyte ion signals by contaminants is one of the difficulties. Compared to ESI, MALDI is more tolerant of contaminants in a sample16,17 and hence may be useful to quantify an analyte in a complex mixture. With regard to the quantification of a peptide, the main problem for MALDI is its poor reproducibility.18,19 Even for homogeneous samples prepared by vacuum drying of a solution containing α-cyano-4-hydroxycinnamic acid (CHCA) as matrix, © 2013 American Chemical Society

[M + H]+ + P → M + [P + H]+

(1)

Q = {I([P + H]+ )/I([M + H]+ )} × {I (M)/I (P)}

(2)

Received: June 30, 2013 Accepted: August 23, 2013 Published: August 23, 2013 8796

dx.doi.org/10.1021/ac401967n | Anal. Chem. 2013, 85, 8796−8801

Analytical Chemistry

Article

Figure 1. Calibration curves (a solid line with filled circles) for Y5K plotted as (a) I([P + H]+)/I([M + H]+) vs c(P) and (b) I([P + H]+) vs c(P) and for YLYEIAR plotted as (c) I([P + H]+)/I([M + H]+) vs c(P) and (d) I([P + H]+) vs c(P). Percentages of matrix suppression are shown as open circles. TIC of 3000 particles per shot was used for temperature control.

H]+) versus c(P) curve deviated from linearity. We will call this anomalous suppression. It is obvious that the quantification results for the above method will become erroneous when the anomalous matrix suppression sets in. In this work, we attempted to find an empirical rule to judge the reliability of a quantification result. The extent of matrix suppression turned out to be an excellent guideline that we sought. Its validity will be demonstrated by presenting the quantification results for various samples. When the extent of suppression is very high and suggests erroneous quantification, accuracy can be improved by dilution. A systematic method to determine the dilution factor and hence to improve the quantification accuracy will be given.

I denotes the abundance of an ion or neutral in the plume. The peptide-to-matrix neutral abundance ratio in the plume, I(P)/ I(M), was approximated by the same ratio in the solid sample, that is, the peptide concentration c(P). We found that Q values thus estimated were nearly the same regardless of the peptide concentration, indicating that the matrix-to-peptide proton transfer was in quasi-equilibrium. The expression for Q can be converted to the following form. I([P + H]+ )/I([M + H]+ ) = Q c(P)

(3)

That is, when Q is constant, the peptide-to-matrix ion abundance ratio is proportional to the peptide concentration in the solid sample. It is a linear calibration curve that can be used for peptide quantification.28 The method is also an inexpensive method of peptide quantification that does not require any isotopically substituted internal standard. In principle, all the peptide- and matrix-derived ions must be included in the estimation of I([P + H]+) and I([M + H]+), respectively. This is unnecessary in actual quantification because the relative abundances of all the ions are fixed at a fixed Tearly.22,28 For all the peptide samples that we studied,28 good direct proportionality between I([P + H]+)/I([M + H]+) and c(P) was observed over wide dynamic ranges when the temperature selection or control was made.23,28 Here, the abundance of the matrix ion decreases as that of the analyte ion increases according to the expression for Q. We will call this normal suppression. At high peptide concentration and hence at high matrix suppression, we observed that the I([P + H]+)/I([M +

■

EXPERIMENTAL SECTION The home-built MALDI-TOF instrument and its operation were reported previously.20,27 A 337 nm output from a nitrogen laser (MNL100, Lasertechnik Berlin, Germany) was used as the light source. A technique called TIC control23 was used to acquire spectra associated with a particular effective temperature in the early plume (Tearly). Here, the total number of particles appearing in the MALDI spectrum generated by a single laser shot was counted. When this was smaller (or larger) than the preset value, we raised (or lowered) the laser pulse energy by properly rotating a neutral density filter and resumed spectral acquisition. The method to estimate Tearly from a spectral pattern was reported previously.28,29 Sample Preparation. CHCA, sucrose, adenosine 5′triphosphate disodium salt (ATP), and insulin were purchased 8797

dx.doi.org/10.1021/ac401967n | Anal. Chem. 2013, 85, 8796−8801

Analytical Chemistry

Article

matrix-to-peptide proton transfer reactions, [M + H]+ + P1 → M + [P1 + H]+ and [M + H]+ + P2 → M + [P2 + H]+, both of which are in quasi-equilibrium. Simultaneous equilibria for the two reactions dictate that the abundance of [P1 + H]+ will get smaller in the presence of P2 compared to the case in which P1 is the sole analyte. This is analyte suppression, or the normal analyte suppression effect, to be rigorous. Mathematically, normal matrix suppression is the manifestation of quasiequilibrium in one proton transfer reaction, while normal analyte suppression is that of two or more quasi-equilibria. Even when a sample contains more than one peptide, a particular peptide in the sample can be quantified by using a calibration curve constructed with that peptide only, if all the proton transfer reactions are in quasi-equilibrium. We checked this for several equimolar binary combinations of peptides. The results for the pairs (Y5K, Y5R) and (YLYEIAR, Y5R) are listed in Table 1. In both cases, adequate quantification (within ±30%

from Sigma (St. Louis, MO). Peptides Y5K, Y5R, YLYEIAR, and DLGEEHFK were purchased from Peptron (Daejeon, Korea). Tryptic digest of cytochrome c was purchased from Waters Corp. (Taunton, MA). An aqueous solution of peptide(s) and other analyte(s) was mixed with a water/acetonitrile solution of CHCA. One microliter of a solution containing analytes and 25 nmol of CHCA was loaded and vacuum-dried.

■

RESULTS AND DISCUSSION Calibration Curves. Earlier, we mentioned that I([P + H]+)/I([M + H]+) was directly proportional to c(P) when the matrix-to-peptide proton transfer was in quasi-equilibrium. To cover the data over a wide dynamic range, let us take the logarithms of both quantities. log{I([P + H]+ )/I([M + H]+ )} = log c(P) + log Q

(4)

The ion abundance ratio versus concentration represented in a log−log plot becomes linear with a slope of 1.0. For several peptides in CHCA, we obtained MALDI spectra by TIC control.23 Each sample consisted of 0.01−250 pmol of a peptide in 25 nmol of CHCA. TIC of 3000 particles per shot was adopted as the preset value, which corresponded to a Tearly of 890 K. Calibration curves for two peptides, Y5K and YLYEIAR, are shown in panels a and c of Figure 1, respectively. As expected, both curves are linear with slopes close to 1.0 and deviate from linearity when the amounts of the peptides are larger than 30 pmol. For comparison, we represented the same data as the log−log plot of I([P + H]+) versus c(P) in panels b and d of Figure 1, respectively. Even though the plots were approximately linear at low concentrations, they began to deviate from linearity at 2.0 pmol (i.e., at a lower concentration than in the former plot). This resulted in a narrower dynamic range. When quantification was attempted for 100 pmol of a peptide in 25 nmol of CHCA, a factor of 2 error was observed when the I([P + H]+)/I([M + H]+) versus c(P) plot was used, whereas an order of magnitude error was observed when the I([P + H]+) versus c(P) plot was used. We acquired the MALDI spectrum of pure CHCA under the same experimental condition and measured the abundance of the matrix ion, I0([M + H]+). We also measured the same abundances, I([M + H]+), from MALDI spectra of matrix− peptide mixtures. Then, the extent of matrix suppression, S, defined as follows was calculated. S = 1 − I([M + H]+ )/I0([M + H]+ )

Table 1. Quantification Results for Equimolar Peptide Pairs peptide pair Y5K, Y5R

Y5R,YLYEIAR

a

amounta of each peptide loaded 1.0 1.0 3.0 3.0 10 10 30 30 1.0 1.0 3.0 3.0 10 10 30 30

amounta determined Y5K Y5R Y5K Y5R Y5K Y5R Y5K Y5R Y5R YLYEIAR Y5R YLYEIAR Y5R YLYEIAR Y5R YLYEIAR

1.2 1.2 2.8 3.3 9.9 12 17 43 0.8 0.8 2.7 3.6 11 12 43 65

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

matrix suppression (%)

0.2 0.3 0.3 0.2 2.0 4 3 18 0.1 0.3 0.2 0.7 2 4 5 11

15 15 43 43 71 71 87 87 30 30 41 41 73 73 93 93

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

3 3 2 2 7 7 4 4 4 4 8 8 6 6 2 2

Number of picomoles of peptide in 25 nmol of CHCA.

of the correct value or less) was possible when matrix suppression was around 70% or lower. Here again, the 70% matrix suppression rule for reliable quantification held. At the moment, we do not know why the rule holds. We just speculate that any factor affecting the quantification accuracy does so by affecting matrix suppression. A factor of 2 quantification error was observed at around 90% suppression in these cases. The error increased rapidly as the suppression further increased. Using a Peptide as an Internal Standard. Suppose that the matrix-to-peptide proton transfer reactions for two peptides at a selected Tearly are in quasi-equilibrium. Then, (eq 3) holding for each peptide results in the following equation.

(5)

To be rigorous, the abundances of all the matrix-derived ions must be added to evaluate I and I0. This is unnecessary, however, because both relative and absolute abundances of all the ions are fixed when Tearly is fixed. In the case of CHCA, we added the abundances of [CHCA + H]+, [CHCA + H − H2O]+, and [CHCA + H − CO2]+ in the estimation of I and I0. This was simply to improve the signal-to-noise ratio. The percentage of matrix suppression thus obtained is shown in each calibration curve in Figure 1. Deviation from linearity starts at around 75% in the I([P + H]+)/I([M + H]+) versus c(P) plots, while the corresponding value is around 30% in the I([P + H]+) versus c(P) plot. The fact that the I([P + H]+)/ I([M + H]+) versus c(P) plot displays a wider linear dynamic range is additional evidence that the matrix-to-peptide proton transfer is in quasi-equilibrium. Peptide Mixtures. When there are two peptides in a sample, our model for analyte ion formation will postulate two

I([P1 + H]+ )/I([P2 + H]+ ) = (Q 1/Q 2){c(P1)/c(P2)}

(6)

With Q1/Q2 being virtually constant, eq 6 suggests a direct proportionality between the peptide ion abundance ratio and the concentration ratio. The quantification method utilizing a peptide as an internal standardwithout temperature controlhas been widely used.13 However, this is the first time that the theoretical basis for the methodwith temperature controlhas been elucidated. As a test of this method, CHCAMALDI spectra were acquired for samples containing 10 pmol 8798

dx.doi.org/10.1021/ac401967n | Anal. Chem. 2013, 85, 8796−8801

Analytical Chemistry

Article

nmol of CHCA. We could not further increase the sucrose concentration because the sample became quite inhomogeneous. The MALDI-TOF spectrum acquired for the above sample is shown in Figure 3. The sodium adduct ion, [sucrose

of Y5R and 0.03−250 pmol of YLYEIAR. A calibration curve depicted as the I([YLYEIAR + H]+)/I([Y5R + H]+) versus c(YLYEIAR)/c(Y5R) plot is shown in Figure 2. In this method, the maximum matrix suppression that allows reliable quantification tends to be somewhat lower than in our method.

Figure 3. MALDI-TOF spectrum for a sample containing 1.0 pmol each of Y5K, Y5R, YLYEIAR, and DLGEEHFK and 1.0 nmol of sucrose in 25 nmol of CHCA. TIC of 3000 particles per shot was used for temperature control.

Figure 2. Calibration curve (a solid line with filled circles) for 0.03− 250 pmol of YLYEIAR obtained by using 10 pmol of Y5R as the internal standard. Percentages of matrix suppression are shown as open circles. TIC of 3000 particles per shot was used for temperature control.

+ Na]+, was the main ionic species from sucrose. Even though sucrose comprised as much as 3.8 mol % of the sample and outnumbered each peptide by 3 orders of magnitude, the abundance of [sucrose + Na]+ was weak and matrix suppression was only 34%. Quantification results for the four peptides are listed in Table 2, which were adequate as expected from low matrix suppression. We also acquired the MALDI spectrum of a sample containing 1.0 pmol of Y5R and 10 pmol of insulin in 25 nmol of CHCA. Here again, we could not further increase the amount of insulin because of the sample inhomogeneity. The MALDI spectrum acquired for this sample is shown in Figure 4. The abundance of the insulin ion is very low in this spectrum. Since matrix suppression was only 15%, the peptide quantification result was good, as shown in Table 2. We also acquired the MALDI spectrum of a sample containing 1.0 pmol each of Y5K, Y5R, YLYEIAR, and DLGEEHFK and 250 pmol of ATP in 25 nmol of CHCA. As can be seen from the results listed in Table 2, reliable quantification of the peptides was possible because matrix suppression was 53%. The results presented so far suggest that a peptide in a sample can be quantified with good accuracy even when the sample is heavily contaminated by materials that are weaker bases than the peptide such as carbohydrates, nucleic acids, among others, and the matrix suppression is the guideline to check the reliability of the quantification. Dilution of a Sample with High Analyte Concentrations. When a good quantification result is not expected because matrix suppression is too high, one can improve the accuracy by diluting the sample. Even though this can be done by trial and error, a simple formula relating the extent of dilution to the matrix suppression factors can be used. We will treat the situation as if ISD21 of matrix and analyte ions does not occur because the final formula is the same regardless of the occurrence of ISD, as long as Tearly is fixed. By rearranging eq 5, we obtain the following expression for I([M + H]+).

In our method, the matrix ion abundance is needed to evaluate the peptide-to-matrix ion abundance ratio. On the other hand, in many spectral acquisitions with commercial MALDI-TOF instruments, low m/z ions are routinely deflected away so that detector saturation30 can be avoided. Then, using a peptide as an internal standard may look technically more convenient than our method. However, the former method suffers from a few weaknesses, as described in the following discussion. One of the key features of our work is the improvement in spectral reproducibility through temperature selection or control, which requires measurement of ion abundances at low m/z. When ISD of a peptide ion is used to select spectra with a particular Tearly, our experience has been that it is much better to use the relative abundance of an ion at low m/z (e.g., immonium ions) rather than that of b- or y-type ions at high m/z. Then, measurement of ion abundances at low m/z becomes imperative. In this paper, we are claiming that matrix suppression is a useful guideline for reliable quantification, which is another reason why we must acquire spectral data at low m/z. Once we decide to acquire data at low m/z, there is no reason to add a peptide as an internal standard, which is entirely unnecessary. Peptides in Complex Mixtures. Experimentally, two factors may determine the upper limit for the amounts of contaminants in a sample that allow reliable quantification. One is matrix suppression (studied so far), and the other is the requirement that it should be possible to prepare a homogeneous sample even when the contaminants are present. In the case of peptide mixtures, homogeneous samples that show high matrix suppression could be prepared . The 70% rule was established through measurements for such samples. For most biological molecules other than peptides, the concentration limit imposed by the need to prepare homogeneous samples came earlier than that by matrix suppression. As an example, we prepared a sample containing 1 pmol each of Y5K, Y5R, YLYEIAR, and DLGEEHFK and 1 nmol of sucrose in 25

I([M + H]+ ) = (1 − S)I0([M + H]+ ) 8799

(7)

dx.doi.org/10.1021/ac401967n | Anal. Chem. 2013, 85, 8796−8801

Analytical Chemistry

Article

Table 2. Quantification Results for Some Peptides in the Presence of Contaminants contaminanta

analyte

amountb loaded

sucrose (1.0 nmol)

YLYEIAR Y5K DLGEEHFK Y5R Y5R YLYEIAR Y5K DLGEEHFK Y5R

1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0

insulin (10 pmol) ATP (250 pmol)

a

amountb determined 1.2 0.9 1.2 0.9 0.8 0.9 0.9 0.8 0.8

± ± ± ± ± ± ± ± ±

matrix suppression (%)

0.2 0.1 0.1 0.2 0.1 0.2 0.1 0.2 0.1

34 34 34 34 15 53 53 53 53

± ± ± ± ± ± ± ± ±

1 1 1 1 5 2 2 2 2

Amount of each contaminant in 25 nmol of CHCA. bNumber of picomoles of each peptide in 25 nmol of CHCA.

Table 3. Dilution and Quantification of a Sample with Y5K in a Peptide Mixturea matrix suppression (%) after dilution dilution factor original sample 2.0 3.3 10

calculated

measured c

62 53 31

83 66 ± 2 55 ± 5 26 ± 1

quantification resultsb 6.4 24 ± 3 29 ± 1 32 ± 3

a Conditions: 30 pmol of Y5K and a tryptic digest of 6.5 pmol of cytochrome c in 25 nmol of CHCA. bNumber of picomoles of Y5K in 25 nmol of CHCA. cBefore dilution.

Figure 4. MALDI spectrum of a sample containing 1.0 pmol of Y5R and 10 pmol of insulin in 25 nmol of CHCA. A Tearly of ≈890 K was selected by selecting spectra with the I([CHCA + H − H2O]+)/ I([CHCA + H]+) ratio lying in the range of 3.7−4.4.

improved. For example, when the sample is diluted by a factor of 2, eq 11 predicts the matrix suppression of 0.62 versus 0.66 that is actually measured. Then, the experimentally determined amount of Y5K became 12 pmol, corresponding to 24 pmol in the original sample versus the correct value of 30 pmol. As demonstrated in this example, sample dilution can be especially useful for quantifying peptides in protein digestion mixtures.

In a previous study, we found that the total ion abundances in MALDI spectra do not change with the analyte concentration or are equal to those in MALDI of pure matrix.22,27 I([M + H]+ ) + I([P + H]+ ) = I0([M + H]+ )

■

(8)

CONCLUSION In quantification of analytes by MALDI, matrix suppression has been regarded as a nuisance. In this work, we divided matrix suppression into two parts, normal and anomalous. The normal matrix suppression is associated with matrix-to-analyte proton transfer in quasi-equilibrium, and it is taken care of when the data are plotted in the form of analyte-to-matrix ion abundance ratio versus analyte concentration. The anomalous effect that prevents reliable quantification becomes noticeable when matrix suppression becomes larger than 70%. It is unknown what causes the anomalous effect and why the 70% rule holds. The rule suggests that quantification of an analyte in a contaminated sample is not affected directly by the amounts of contaminants but indirectly through their influence on matrix suppression. This allows quantification of small amounts of peptides in samples containing large amounts of less basic materials such as carbohydrates and nucleic acids. In actual analysis, the difficulty in preparing a homogeneous sample rather than matrix suppression often dictated the upper limit for the amounts of contaminants allowed for reliable quantification.

Inserting eq 7 into eq 8, we obtained the following relation. I([P + H]+ ) = S I0([M + H]+ )

(9)

Inserting eqs 7 and 9 into eq 3 and replacing Q with the equilibrium constant K, we obtain the relation between K and S.

Kc = S /(1 − S)

(10)

Defining S1 and S2 as the matrix suppressions at the analyte concentrations of c1 and c2 and further manipulating eq 10 holding at the two concentrations, we obtained the following relation. c 2/c1 = (S2/S1)[(1 − S1)/(1 − S2)]

(11)

That is, when the matrix suppression for the current sample and the desired value for a diluted sample are provided, the extent of dilution needed for the current sample solution can be calculated. As an example, quantification data for a sample containing 30 pmol of Y5K and a tryptic digest of 6.5 pmol of cytochrome c in 25 nmol of CHCA obtained by MALDI with TIC of 3000 are shown in Table 3. The matrix suppression for this sample was 83%, higher than our guideline of 70%. Quantification for the peptide resulted in 6.4 pmol, much smaller than the correct value. Upon dilution of the solution, matrix suppression decreased as predicted by eq 11, and the quantification result

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest. 8800

dx.doi.org/10.1021/ac401967n | Anal. Chem. 2013, 85, 8796−8801

Analytical Chemistry

■

Article

ACKNOWLEDGMENTS This work was supported by the National Research Foundation of Korea (NRF) Grant funded by the Korean government (MEST) (2012054350). S.H.A. thanks the Ministry of Education, Republic of Korea, for a Brain Korea 21 Fellowship.

■

REFERENCES

(1) Elliott, M. H.; Smith, D. S.; Parker, C. E.; Borchers, C. J. Mass Spectrom. 2009, 44, 1637−1660. (2) Ong, S.; Mann, M. Nat. Chem. Biol. 2005, 1, 252−262. (3) Bucknall, M.; Fung, K. Y. C.; Duncan, M. W. J. Am. Soc. Mass Spectrom. 2002, 13, 1015−1027. (4) Weston, A. D.; Hood, L. J. Proteome Res. 2004, 3, 179−196. (5) Standing, K. G. Curr. Opin. Struct. Biol. 2003, 13, 595−601. (6) Lill, J. Mass Spectrom. Rev. 2003, 22, 182−194. (7) Gutierrez, J. A.; Dorocke, J. A.; Knierman, M. D.; Gelfanova, V.; Higgs, R. E.; Koh, N. L.; Hale, J. E. BioTechniques 2005, 38, S13−S17. (8) Ju, S.; Yeo, W.-S. Nanotechnology 2012, 23, 135701. (9) Hustoft, H. K.; Reubsaet, L.; Greibrokk, T.; Lundanes, E.; Malerod, H. J. Pharm. Biomed. Anal. 2011, 56, 1069−1078. (10) Olsen, J. V.; Ong, S.; Mann, M. Mol. Cell. Proteomics 2004, 3, 608−614. (11) Gerber, S. A.; Rush, J.; Stemman, O.; Kirschner, M. W.; Gygi, S. P. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 6940−6945. (12) Orlando, R.; Lim, J. M.; Atwood, J. A., III; Angel, P. M.; Fang, M.; Aoki, K.; Alvareaz-Manilla, G.; Moremen, K. W.; York, W. S.; Tiemeyer, M.; Pierce, M.; Dalton, S.; Wells, L. J. Proteome Res. 2009, 8, 3816−3823. (13) Wilkinson, W. R.; Gusev, A. I.; Proctor, A.; Houalla, M.; Hercules, D. M. Fresenius’ J. Anal. Chem. 1997, 357, 241−248. (14) Wilkins, M. R.; Sanchez, J.; Gooley, A. A.; Appel, R. D.; Humphery-Smith, I.; Hochstrasser, D. F.; Williams, K. L. Biotechnol. Genet. Eng. Rev. 1996, 13, 19−50. (15) Claverol, S.; Schiltz, O. B.; Gairin, J. E.; Monsarrat, B. Mol. Cell. Proteomics 2003, 2, 483−493. (16) Cohen, L. H.; Gusev, A. I. Anal. Bioanal. Chem. 2002, 373, 571− 586. (17) Glish, G. L.; Vachet, R. W. Nat. Rev. Drug Discovery 2003, 2, 140−150. (18) Szájli, E.; Fehér, T.; Medzihradszky, K. F. Mol. Cell. Proteomics 2008, 7, 2410−2418. (19) Duncan, M. W.; Roder, H.; Hunsucker, S. W. Briefings Funct. Genomics Proteomics 2008, 7, 355−370. (20) Bae, Y. J.; Park, K. M.; Kim, M. S. Anal. Chem. 2012, 84, 7107− 7111. (21) Brown, R. S.; Carr, B. L.; Lennon, J. J. J. Am. Soc. Mass Spectrom. 1996, 7, 225−232. (22) Ahn, S. H.; Park, K. M.; Bae, Y. J.; Kim, M. S. J. Mass Spectrom. 2013, 48, 299−305. (23) Ahn, S. H.; Park, K. M.; Bae, Y. J.; Kim, M. S. J. Am. Soc. Mass Spectrom. 2013, 24, 868−876. (24) Chang, W. C.; Huang, L. C. L.; Wang, Y.-S.; Peng, W.-P.; Chang, H. C.; Hsu, N. Y.; Yang, W. B.; Chen, C. H. Anal. Chim. Acta 2007, 582, 1−9. (25) Hillenkamp, F.; Peter-Katalinić, J. MALDI MS. A Practical Guide to Instrumentation, Methods and Applications; Wiley-VCH: Weinheim, Germany, 2007. (26) Howard, K. L.; Boyer, G. L. Anal. Chem. 2007, 79, 5980−5986. (27) Bae, Y. J.; Shin, Y. S.; Moon, J. H.; Kim, M. S. J. Am. Soc. Mass Spectrom. 2012, 23, 1326−1335. (28) Park, K. M.; Bae, Y. J.; Ahn, S. H.; Kim, M. S. Anal. Chem. 2012, 84, 10332−10337. (29) Yoon, S. H.; Moon, J. H.; Kim, M. S. J. Am. Soc. Mass Spectrom. 2010, 21, 1876−1883. (30) Westman, A.; Brinkmalm, G.; Barofsky, D. F. Int. J. Mass Spectrom. Ion Processes 1997, 169/170, 79−87.

8801

dx.doi.org/10.1021/ac401967n | Anal. Chem. 2013, 85, 8796−8801