Charge Ratio Analysis Method: Approach for the Deconvolution of

A new method to interpret electrospray mass spectral data based on calculating the ratio of mass-to-charge (m/z) values of multiply charged ions is de...
0 downloads 0 Views 399KB Size
Anal. Chem. 2005, 77, 111-119

Charge Ratio Analysis Method: Approach for the Deconvolution of Electrospray Mass Spectra Simin D. Maleknia*,†,‡ and Kevin M. Downard§

School of Science, Griffith University, Australia, Sciformatics, Sydney, Australia, and School of Molecular and Microbial Biosciences, The University of Sydney, Australia

A new method to interpret electrospray mass spectral data based on calculating the ratio of mass-to-charge (m/z) values of multiply charged ions is described. The massto-charge ratios of any two multiply charged ions corresponding to a single compound are unique numbers that enable the charge states for each ion to be unequivocally identified. The multiply charged ions in electrospray mass spectra originate from the addition or abstraction of protons, cations, or anions to and from a compound under analysis. In contrast to existing deconvolution processes, the charge ratio analysis method (CRAM), identifies the charge states of multiply charged ions without any prior knowledge of the nature of the charge-carrying species. In the case of high-resolution electrospray mass spectral data, in which multiply charged ions are resolved to their isotopic components, the CRAM is capable of correlating the isotope peaks of different multiply charged ions that share the same isotopic composition. This relative ratio method is illustrated here for electrospray mass spectral data of lysozyme and oxidized ubiquitin recorded at lowto high-mass resolution on quadrupole ion trap and Fourier transform ion cyclotron mass spectrometers, and theoretical data for the protein calmodulin based upon a reported spectrum recorded on the latter. Electrospray ionization mass spectrometry1-3 (ESI-MS) has revolutionized the analysis of biopolymers directly from solution and is easily coupled to chromatographic separation techniques.4 A unique feature of the ionization process is that a distribution of multiply charged ions is produced, where each ion contains more than one charge-carrying species. When mass spectra are recorded on instruments at relatively low to moderate mass resolution, in which the isotopes are unresolved, deconvolution algorithms5,6 are typically applied to * To whom correspondence should be addressed. School of Science, Griffith University, Nathan Campus, Brisbane, QLD 4111, Australia. Phone: + 61 7 3875 7894. E-mails: [email protected], [email protected]. † Griffith University. ‡ Sciformatics. § The University of Sydney. (1) Dole, M.; Mack, L. L.; Hines, R. L. J. Chem. Phys. 1968, 49, 2240-2249. (2) Yamashita, M.; Fenn, J. B. J. Phys. Chem. 1984, 88, 4451-4459. (3) Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, G. M. Science 1989, 246, 64-71. (4) Whitehouse, C. M.; Dreyer, R. N.; Yamashita, M.; Fenn, J. B. Anal. Chem. 1985, 57, 675. (5) Mann, M.; Meng, C. K.; Fenn, J. B. Anal. Chem. 1989, 61, 1702-1708. 10.1021/ac048961+ CCC: $30.25 Published on Web 12/03/2004

© 2005 American Chemical Society

first calculate the charge states of the ions and, subsequently, the molecular weight of the compound. These algorithms work by assuming that adjacent ions of a distribution differ by one charge integer and that all ions of the distribution support the same charge-carrying species. It is common for such algorithms to measure a molecular weight on the basis of the m/z ratio for each ion and plot this value in a molecular weight spectrum (deconvolution spectrum). Most deconvolution algorithms developed to interpret ESI mass spectra initially assume the identity of the charge-carrying species associated with the multiply charged ion in order to derive a molecular weight value. The following expression (eq 1) is used to calculate the molecular weight (M) of a compound from the mass-to-charge ratios (Rz) of multiply charged ions (of charge z) that originate from the addition or abstraction of charge-carrying species (m).

Rz ) (M ( zm)/z

(1)

From the mass-to-charge values for any two ions (Rz)a and (Rz)b within the electrospray mass spectrum and their corresponding charge difference (zb ) za + integer) together with a prior assumption of the nature of the charge-carrying species (m), two simultaneous equations can be constructed from which a molecular weight (M) is calculated. One of the limitations of the above algorithm is that the charge-carrying species (m) is assumed to be uniform for all ions. To overcome this limitation, an algorithm7 was developed that considers the charge-carrying species to also be a variable. In this case, a three-dimensional molecular weight spectrum is produced in which the mass of the charge-carrying species is determined as the value at which the intensity of the molecular weight peak is a maximum. Entropy-based calculations have also been used to interpret electrospray mass spectra.8,9 In these approaches, a molecular weight range is first predicted and hypothetical multiply charged ions across a charge state distribution are fit to the mass spectral data. The entropy-based procedures also include noise reduction algorithms that are effective for the processing of mixtures and low signal-to-noise spectra.9 (6) Hagen, J.; Monnig, C. A. Anal. Chem. 1994, 66, 1877-1883. (7) Labowsky, M.; Whitehouse, C.; Fenn, J. B. Rapid Commun. Mass Spectrom. 1993, 7, 71-84. (8) Ferrige, A. G.; Seddon, M. J.; Green, B. N.; Jarvis, S. A.; Skilling, J. Rapid Commun. Mass. Spectrom. 1992, 6, 707-711. (9) Reinhold, B. B.; Reinhold: V. N. J. Am. Soc. Mass Spectrom. 1992, 3, 207215.

Analytical Chemistry, Vol. 77, No. 1, January 1, 2005 111

Here, we illustrate that the mathematical ratio of any two integers is a unique number, and hence, the charges for any two ions of a charge-state distribution can be uniquely assigned from such ratios (Table 1). This approach is applied to representative electrospray mass spectral data for several proteins recorded at low- and high-mass resolution to identify both the charge states of ions and, subsequently, the nature of the charge-carrying species. We also demonstrate how the method is superior for determining the charge states of ions recorded at high resolution in which the isotopes are resolved, in that CRAM can identify charge states with greater accuracy and can also correlate the isotopic peaks of different multiply charged ions to their common isotopic compositions. EXPERIMENTAL SECTION All protein samples were obtained from Sigma Chemicals (St Louis, MO) and were used without further purification. The protein solutions were prepared at a concentration range of 1-5 µM in water and methanol (both 49 vol %) containing 2% acetic acid. The oxidized ubiquitin sample was prepared by mixing an aqueous solution of a 10 µM protein sample with an equal volume of a 30% peroxide solution (Chem-Supply, Gillman, SA, Australia) for 5 min. Electrospray mass spectra were recorded on a quadrupole ion trap (LCQ Classic, ThermoFinnigan Corporation, San Jose, CA) or a 4.7 T FTICR (APEX, Bruker Daltonics, Billerica, MA) mass spectrometer operated in the positive ion mode. All spectra were mass-calibrated with the use of external calibrants. The quadrupole ion trap was calibrated with a mixture containing caffeine, a tetrapeptide with a sequence MRFA, and a perfluorinated polymer (Ultramark 1621). The FTICR instrument was calibrated with the multiply charged ions of angiotensin-1 (2+ and 3+ ions) and ubiquitin (10+ to 12+ ions). All calibrants are well characterized with known elemental compositions and masses calculated to at least five decimal places. Protein solutions were infused at a rate of between 3 and 5 µL/min using a needle voltage in the range of 4.2-4.5 kV. RESULTS AND DISCUSSION Theoretical Basis of the Charge Ratio Analysis Method (CRAM). The theoretical basis of the charge ratio analysis method (CRAM) is that the ratio of two multiply charged ions, a and b, originating from the same compound can be derived from eq 1.

(Rz)a/(Rz)b ) (zb(M ( zam))/(za(M ( zbm))

(2)

For relatively large molecular weight proteins, when M > zam or zbm, the ratio of charge states (eq 2) simplifies to

(Rz)a/(Rz)b ) zb/za

(3)

Therefore, by simply dividing the mass-to-charge ratios (Rz) of two multiply charged ions, the inverse ratio of their two charge states can be calculated (eq 3). A comparison of this calculated ratio to the unique ratio of any two integers (Table 1) then identifies the charge state values for each of the ions. Once the charge states are identified, the mass-to-charge ratios for two ions are multiplied by their charge state values (eqs 4 and 5). For any 112

Analytical Chemistry, Vol. 77, No. 1, January 1, 2005

pair of ions, a simple equation is derived (eq 6), which enables the mass of the charge-carrying species (m) to be determined.

(Rz)aza ) (M ( zam)

(4)

(Rz)bzb ) (M ( zbm)

(5)

((Rz)az - (Rz)bz) ) (m (za - zb)

(6)

Once values of z and m are obtained, a molecular weight (M) measurement is calculated from eq 1. Table 2 illustrates how CRAM is used to calculate the number of charges (z), the mass of charge-carrying species (m), and the molecular weight (M) of a compound from its electrospray ionization mass spectral data. Lysozyme has a theoretical average (i.e., weighted for all isotopes) molecular weight of 14305.1438 Da. The first column of Table 2 lists theoretically generated massto-charge ratios (Rz)n for lysozyme ions with 10-13 charges formed by the addition or abstraction of protons and the addition of sodium ions or potassium ions. From a pair of theoretically generated (Rz)n values, the ratios for two successive ions are calculated. These calculated ratios (Rz)n/(Rz)n+1 are then compared to the relative ratios of two integer numbers listed in Table 1 to identify the number of charges on each ion. For example, the ratio of 1301.47542/1431.52218 is 0.90915, which is in closest agreement with the 0.90909 value derived from dividing 10 by 11 (Table 1). This comparison enables the charges of ions at m/z of 1431.52218 and 1301.47542 to be determined as 10+ and 11+ ions. After identifying the charge numbers (zn), the (Rz)n values are multiplied by their corresponding number of charges listed as [(Rz)nzn] in Table 2. The mass of the charge-carrying species is then identified from the difference of two ((Rz)nzn) values, as shown in the last column of Table 2. As this example illustrates, the identity of charge-carrying species is based on the mass-to-charge ratios and the calculated number of charges for the ions only. From the first set of (Rz)n values (Table 2), the mass of adduct is calculated to be 1.0078, identifying the charge-carrying species to be protons. The mass of the charge-carrying species from the second and third set of (Rz)n values are calculated to be 22.9897 and 39.0984, which correspond to the mass of sodium and potassium, respectively. The last set of numbers in Table 2 correspond to theoretically generated negatively charged ions of lysozyme formed by the removal of between 10 and 13 protons. In this example of negative ions, the charges of ions and, subsequently, the nature of the charge-carrying species (i.e., protons) are also identified by the application of CRAM, as illustrated in Table 2. Table 2 also illustrates that the errors derived by subtracting the calculated ratios from their corresponding ratio of integers (Table 1) increase when the masses of adduct species increase. For example, when the adduct species are protons, the calculated ratios are within 6 × 10-5 of their theoretical values, and when the adduct species are potassium ions, the calculated ratios are within 244 × 10-5 of their theoretical values. This increase is associated with the assumption that M > zam or zbm during the derivation of eq 3. Obviously, as the mass of the adduct species increases, the values of zam or zbm become more significant relative to the mass of the protein (M), and the errors increase

Analytical Chemistry, Vol. 77, No. 1, January 1, 2005

113

a

The ratio of numbers represents a number from 1 to 20 divided by 1 through 20. Note that column two starts with the integer 1 (i.e., 1 divided by 1).

Table 1. Ratio of Two Integer Numbers From 1 to 20 (Calculated to Five Decimal Points) Used for the Application of the CRAMa

Table 2. Illustration of the CRAM Utilizing Theoretical Mass-to-Charge Ratios for the Protein Lysozymea

a An average molecular weight (M ) of 14 305.1438 was used in these calculations. Theoretical mass-to-charge ratios were calculated from eq av 1 by the addition or abstraction of 10-13 protons and the addition of sodium or potassium ions. An average atomic mass was used for potassium.

accordingly. Even with the increasing error limits, the charge numbers are identified with a high confidence, since the ratios of numbers are sufficiently unique (Table 1). In this example, the ratios are calculated from two consecutive numbers. Even in the case of the potassium adduct ion series with the highest associated errors, in which the ratios from consecutive numbers are only considered (bold section in Table 1), the charges on the ions could only be assigned to values between 10+ and 13+. Although the example above describes the CRAM approach for deconvoluting electrospray data from consecutive mass-tocharge values, the method is not limited only to consecutive values. The ratios of two numbers are unique, regardless of their order, as illustrated by the numbers in Table 1 that correspond to the relative ratios for any two integer numbers from 1 to 20. The ratios of two consecutive numbers are listed in the two diagonal rows above and below the center diagonal row. For each consecutive set of numbers listed in Table 1, the ratios increase monotonically starting from a value of 0.5 (i.e. 1 divided by 2 when considering the ratio of the smaller number divided by the next larger number of the set) to 0.66667 for the ratio of 2 divided by 3 up to 0.95000 for the ratio of 19 and 20. Conversely, for each consecutive set of numbers in Table 1, the ratios of two numbers also decrease monotonically when considering the ratio of the larger integer divided by the smaller number of the set. For example, starting from a value of 2 (i.e. 2 divided by 1), the next number along the diagonal is 1.5 (derived from 2 divided by 3) down to 1.05263 for the ratio of 20 divided by 19. As shown in Table 1, the ratios of several numbers are the same. For example, 0.33333 is the ratio for 1/3, and any multiple of this 1/3 ratio such as 2/6 and 3/9. One way of reducing this ambiguity of the charge state assignment is to include one consecutive charge state in the calculations. In this example, if the ratio of 0.33333 was based on two ions with 3 and 9 charges, 114 Analytical Chemistry, Vol. 77, No. 1, January 1, 2005

the inclusion of a ratio derived from a mass-to-charge value for a consecutive multiply charged ion of the series, the charge states can be assigned without ambiguity. Including mass-to-charge values for a multiply charged ion with 4 or 10 charges will result in the CRAM being applied to ion series with 3, 4, and 9 charges or 3, 9 and 10 charges. Therefore, the ratios could now be calculated for 3/4, 3/9, and 4/9 or for 3/9, 3/10, and 9/10. Further, the inverse ratios (eq 3) that would be calculated are represented by 4/3, 9/3 and 9/4 or for 9/3, 10/3 and 10/9. In general, the inclusion of ratios of the mass-to-charge values for more ions in these calculations results in a greater confidence in charge state assignment. Applications of CRAM to Low- and High-Resolution Experimental ESI-MS Data. The electrospray mass spectrum of lysozyme recorded at relatively low resolution on a quadrupole ion trap is shown in Figure 1. Table 3 shows how the charges of each ion, the mass of the charge-carrying species, and the molecular weight of the protein are calculated by the CRAM. The electrospray mass spectrum of lysozyme shows five abundant ions with m/z of 1193.1 to 1789.1. The m/z for these five multiply charged ions are listed in column 1 of Table 3. In this example, two sets of ratios are calculated as [(Rz)n+1/(Rz)n] and [(Rz)n/ (Rz)n+1]. Comparison of the values for these ratios to Table 1 reveals that these ions have a charge state distribution of 8+ to 12+. The calculated ratios listed in columns 2 and 5 of Table 3 are all in very good agreement with the corresponding ratio of integer numbers listed in Table 1. The ratio of 0.91671 derived from 1193.1/1301.5 has the lowest deviation of 4 × 10-5 when compared to the ratio of 11/12 (Table 1), and the ratio of 1.099 88 based on 1431.5/1301.5 has the highest deviation of 12 × 10-5, as compared to the ratio of 11/10. These calculations show that the charge ratio analysis method is able to assign charge numbers with a high accuracy.

Although the ratios in Table 3 are calculated for consecutive multiply charged ions, the charge ratio analysis method can be applied to multiply charged ions without any specific order. In this example, when the two ions of 1301.5 and 1789.5 are selected, two ratios of 0.72729 (1301.5/1789.5) and 1.37495 (1789.5/1301.5) are calculated that relate these ions as 11+ and 8+ (Table 1). Following the identification of the charge states (z), the mass of the charge-carrying species is calculated, and the average value (m ) 1.01) corresponding to a proton is used to calculate the molecular weight of the protein. The molecular weight of lysozyme is calculated by multiplying the (Rz)n values by their charge states zn and subtracting the mzn values accordingly, which in this example provides an average molecular weight of 14 304.9 Da. When the resolution of the mass analyzer is sufficient to provide isotopically resolved peaks for the multiply charged ions, the charge states of ions can be simply calculated from the m/z differences of the resolved isotopic clusters (1/z).10,11 Therefore, the high-resolution electrospray data provides an internal 1-Da mass scale that can be used to calculate the charge states of ions unambiguously. For example, for an ion with 10 charges, the m/z spacing of the isotopes corresponds to 1/10 or 0.10 u. The electrospray mass spectrum of oxidized ubiquitin recorded on a Fourier transform ion cyclotron (FTICR) mass spectrometer (shown in Figure 2) displays four dominant multiply charged ion clusters. The insert shows the expansion of the 12+ ion with a resolution of 54 K based on the peak at m/z 716.01234 (fwhm). The high resolving power of the FTICR analyzer affords isotopic separations of all ions, and their charge states can be quickly determined from their m/z separation, as illustrated in Table 4, which contains the m/z values for the two most abundant ion clusters. The spacing between the isotopic peaks for these two multiply charged ions are listed in column 2 of Table 4. When averaged, these values of 0.08161 and 0.08829 u correspond to multiply charged ions of 12+ and 11+, respectively. Once the charge states of the ions are calculated, their m/z values are multiplied by their charge states to identify the chargecarrying species (using eq 6) and the molecular weight of the protein (using eq 1). In this example, when considering only the most abundant isotopic peaks at m/z 716.01234 and 781.01095, (10) Henry, K. D.; McLafferty, F. W. Org. Mass Spectrom. 1990, 25, 490-492. (11) Senko, M. W.; Beu, S. C.; McLafferty, F. W. J. Am. Soc. Mass Spectrom. 1995, 6, 52-56.

Table 3. Application of the CRAM for Experimental Mass Spectral Data of Lysozyme

Figure 1. Electrospray mass spectrum of lysozyme recorded on a quadrupole ion trap.

Analytical Chemistry, Vol. 77, No. 1, January 1, 2005

115

the charge-carrying species are identified to be protons (average m ) 1.03310). The molecular weight of oxidized ubiquitin is calculated as 8580.04 Da on the basis of the average of the two molecular weights determined from the most abundant isotope peaks. Ubiquitin has one N-terminal methionine residue that is solvent-accessible and highly oxidizable upon reactions with oxygen-containing radicals.12,13 Reactions with hydrogen peroxide result in methionine oxidation,14 and the 16+ u addition in the molecular weight measurement of the oxidized ubiquitin has verified this oxidative reaction. The above example shows that although the measure of the isotopic spacing can assign charge states, greater accuracy can be obtained when the CRAM is applied. The averaged values for the isotopic spacing (1/z) of 12+ and 11+ ions (0.08161 and 0.08829 u) differ by 0.00172 and 0.00261 u from their corresponding to theoretical values of 0.08333 and 0.09090. As shown in column 8 of Table 4, the ratios of m/z values for all corresponding isotope peaks within each multiply charged ion cluster are within 11 × 10-5 of the ratio of 11/12 (Table 1). Therefore, the charge states can be assigned with a 20-fold greater accuracy upon application of the CRAM, as compared with measurements of the spacing between ion peaks within each cluster. An even greater confidence in assigning the charge states of ions is obtained when the CRAM is applied to larger molecular weight species, and this feature of the CRAM is illustrated with calmodulin as a model protein. The theoretical m/z values corresponding to those for the resolved isotopes of three multiply charged ions (denoted A, B, and C) for the protein calmodulin are shown in Table 5. These values were obtained at a mass resolution power of 50 K with relative ion intensities adjusted according to those observed in a reported mass spectrum of the protein recorded on a 9.4-T FTICR instrument.15 The application of the CRAM enables the charge state of the ions to be determined from the ratios of m/z values of the isotopically enriched ions (12) Maleknia, S. D.; Chance, M. R.; Downard, K. M. Rapid Commun. Mass Spectrom. 1999, 13, 2352. (13) Maleknia, S. D., Downard, K. M. Mass. Spectrom. Rev. 2001, 20, 388. (14) Maleknia, S. D.; Brenowitz, M.; Chance, M. R. Anal. Chem. 1999, 71, 3965. (15) Hakansson, K.; Cooper, H. J.; Hudgins, R. R.; Nilsson, C. L. Curr. Org. Chem. 2003, 6, 1503-1525.

116

Analytical Chemistry, Vol. 77, No. 1, January 1, 2005

Table 4. Molecular Weight Measurement for Oxidized Ubiquitin From FTICR Mass Spectral Data

Figure 2. Electrospray mass spectrum of oxidized ubiquitin, with the expansion of the 12+ charge state (insert), recorded on a 4.7-T FTICR.

Analytical Chemistry, Vol. 77, No. 1, January 1, 2005

117

a The ratios with the same isotopic compositions described in the text are shown in bold squares. b Error ) (measured isotopic spacing -1/z). c Error ) (calculated ratios in bold - corresponding ratio of integers from Table 1).

Table 5. Application of the CRAM for the Correlation of Ions of the Protein Calmodulin That Share a Common Isotopic Composition

across the multiply charged clusters with an average error of 0.4 × 10-6 (last column in Table 5). This compares to an average error of 1.6 × 10-4 (column 4 of Table 5) with the isotopic spacing method. Therefore, the charges of the ions are assigned by application of the CRAM with a 400-fold greater confidence. In summary, when comparing the isotope spacing method to the CRAM for the same data sets (Tables 4 and 5) the charge states are assigned with a higher accuracy by the CRAM. Correlation of the Isotope Peaks of Different Multiply Charged Ions That Share a Common Isotopic Composition by Application of the CRAM. A vital step in measuring the molecular weight of a compound using high-resolution mass spectral data is the requirement that a correct isotopic distribution be assigned to ion peaks of a resolved cluster.16 For reasonably large molecules (>10 000 Da), a resolved ion cluster is likely not to contain a detectable peak associated with ions that contain only the lightest isotopes for each element it is composed of. Thus, in the case of the theoretical calmodulin data represented in Table 5, each of the peaks (designated A1 to A10, B1 to B13, and C1 to C9) represents multiply charged ions for protein molecules enriched in heavy isotopes. The absence of a monoisotopic peak (or more correctly, a peak representing ions containing only the lightest isotope of each element) is due to the low probability that a molecule consisting of many atoms (many hundreds) contains not a single heavy isotope (such as 13C). If the correct isotopic composition is not assigned to a particular peak of the resolved ion cluster, a molecular weight measurement based on it will be in substantial error.17 To overcome this, isotopically resolved ion clusters recorded in the mass spectrum have to date been compared with those theoretically derived.18 By definition, this requires that the elemental composition of the compound under analysis be known in advance, a situation that may often not be the case. Where the experimental and theoretical isotopically resolved ion cluster are misaligned, the molecular weight measured for the compound can be brought into error by several Daltons. As will be illustrated for the data shown in Table 5, it is also not possible to correlate multiply charged ions of each cluster that share a common isotopic composition based upon their relative intensity. The theoretical m/z values for the ion peaks of the three multiply charged ions (of charge 14+, 13+, and 12+ denoted A, B, and C respectively) of calmodulin are shown in Table 5 as values A1-A10, B1-B13, and C1-C9. For this data set, only ions with a relative abundance above 17% have been considered as those detected above an artificial noise threshold (broken line in Figure 3). Therefore, the three multiply charged ions display a different number of isotopic peaks associated with them, ranging from 9 for the multiply charged ions of cluster C to 13 for those of cluster B. Furthermore, the relative intensities for ions within each resolved isotopic cluster differ. The most abundant isotope ions within each isotopic cluster correspond to peaks A5, B6, and C5. Yet it will be shown, through the application of the CRAM, that these ions do not share a common isotopic composition. (16) Senko, M. W.; Beu, S. C.; McLafferty, F. W. J. Am. Soc. Mass Spectrom. 1995, 6, 229-233. (17) Beavis, R. C. Anal. Chem. 1993, 65, 496-497. (18) Horn, D. M.; Zubarev, R. A.; McLafferty, F. W. J. Am. Soc. Mass Spectrom. 2000, 11, 320-332.

118 Analytical Chemistry, Vol. 77, No. 1, January 1, 2005

Figure 3. Alignment of resolved isotope peaks for the multiply charged ions (12+ to 14+) of the protein calmodulin that share a common isotopic composition. Artifical threshold set at 17% of the abundance of most intense ion (broken line).

Application of the CRAM allows multiply charge ions of different isotopic clusters to be correlated according to their common isotopic composition. This, in turn, assists with the determination of a more accurate and reliable molecular weight. This feature of the CRAM is illustrated in Table 5 (columns 6-15) when the m/z of one of the isotope peaks for one particular multiply charged ion is divided by the m/z of all the isotope peaks for a second multiply charged ion. For example, when the m/z value of 1200.00038 (A1) is divided by all the m/z values for isotopically enriched ions of the second multiply charged ion (B1-B13), the ratio of values for A1/B3 (i.e. 0.92857) is identical to the ratio of charges of the ions 13 and 14 (13/14 or 0.92857). Therefore, the ions associated with peaks A1 and B3 share the same isotopic composition. Similarly, ratios of m/z values for peaks A2/B4, A3/B5, etc. (shown in bold boxes in Table 5 along the diagonal of columns 6 to 15) within experimental error share the same value and, thus, also share a common isotopic composition. It is noteworthy that the most abundant isotope peak of the 14+ ions (A5) shares a common isotopic composition to ions at m/z value given at B7 that is not the most abundant isotope peak of the 13+ ion cluster. In the case of the 13+ and 12+ charge states of calmodulin, the ratio of values B1/C1 is in the closest agreement with the ratio of charges (12/13 or 0.92307), such that ions at m/z values of B1 and C1 through B9 and C9 have the same isotopic composition, as illustrated in Table 5. To show that the CRAM can be applied to nonconsecutive multiply charged ions, the ratio of m/z values for the isotope peaks of the 14+ and 12+ ions are shown in Table 5.

In this case, peaks A1 and C3, A2 and C4, etc. (shown in bold boxes), share the same ratio value to 12/14 or 0.85714. Figure 3 displays all three multiply charged ions aligned according to those ions that share a common isotopic composition. This isotope peak-matching feature of the CRAM is particularly useful for the analysis of unknown samples and mixtures to distinguish where the resolved isotope peaks of one component overlap with isotope peaks of another component. CONCLUSIONS A powerful new method for the deconvolution of electrospray mass spectral data is described. This charge ratio analysis method (CRAM) is based upon calculating the ratio of m/z values of multiply charged ions in order to determine the charge state of the ions, the nature of the charge-carrying species, and subsequently, the molecular weight of the compound under analysis. The approach substantially differs in its execution from other deconvolution programs that must typically assume the nature of the charge-carrying species in order to derive a correct molecular weight value. In contrast, the CRAM makes no such assumption and derives the mass of the charge-carrying species once the charge states for the ions are determined. We have also demonstrated that the CRAM provides a superior measure of the charge state of ions, as clearly demonstrated for theoretically and

experimentally generated ESI-MS data sets for the proteins lysozyme, oxidized ubiquitin, and calmodulin. The charge ratio analysis method can be easily applied for processing of mass spectral data recorded at either low- or high-mass resolution. An additional feature of CRAM is its ability in high resolution mass spectral data, for which isotope peaks of multiply charged ions are resolved, to identify those ions of each cluster that share a common isotopic composition. This improves the reliability of interpreting such data and measuring the accurate molecular weights of compounds without error. We are presently completing a computer algorithm based on the CRAM to analyze ESI mass spectral data in an automated manner. ACKNOWLEDGMENT High-resolution ESI mass spectra for oxidized ubiquitin were recorded at Griffith University on a Bruker 4.7T APEX FTICR mass spectrometer purchased with funds by the Australian Research Council (through the Linkage Infrastructure, Equipment and Facilities grant LE02/37908) and Griffith University. Received for review July 15, 2004. Accepted October 20, 2004. AC048961+

Analytical Chemistry, Vol. 77, No. 1, January 1, 2005

119