Electrospray Ionization Mass Spectrometry: Analysis of the Ca2+-Binding Properties of Human Recombinant α-Parvalbumin and Nine Mutant Proteins

Analytical Biochemistry 268, 64 –71 (1999) Article ID abio.1998.3015, available online at http://www.idealibrary.com on

Electrospray Ionization Mass Spectrometry: Analysis of the Ca 21-Binding Properties of Human Recombinant a-Parvalbumin and Nine Mutant Proteins H. Troxler,* T. Kuster,* J. A. Rhyner,* P. Gehrig,† and C. W. Heizmann* *Department of Pediatrics, Division of Clinical Chemistry and Biochemistry, University of Zurich, Steinwiesstrasse 75, CH-8032 Zurich, Switzerland; and †Institute of Biochemistry, University of Zurich, CH-8057 Zurich, Switzerland

Received August 10, 1998

A set of 10 different recombinant human parvalbumins was used to establish a method for the investigation of the Ca 21-binding properties of proteins by electrospray ionization mass spectrometry (ESI-MS). Human PV WT was found to bind 2 mol Ca 21 ions/mol of protein, whereas its mutants (PV E101V, PV D90A , PV E62V, PV D51A , PV D90A,E101V , PV E62V,E101V , PV D51A,D90A , PV D51A,E62V , PV D51A,E62V,D90A,E101V) containing inactivating substitutions in the Ca 21-binding loops bind 0 or 1 Ca 21 ion per protein molecule, depending on the degree of inactivation. These findings fully agree with previously reported results obtained by flow dialysis experiments. The RPHPLC desalted metal-free proteins were analyzed in 10 mM ammonium acetate at pH 7.0. The experimental conditions were optimized with the recombinant parvalbumin test system before analyzing the Ca 21-binding properties of rat and murine parvalbumins in muscle tissue extracts. ESI-MS revealed that (i) rat and murine a-parvalbumins each bind specifically two Ca 21 ions per protein molecule and (ii) both extracted parvalbumins were found to be posttranslationally modified; each protein is acetylated at the N-terminus. Finally, during our investigations of the murine parvalbumin a sequencing error was detected at the C-terminus where the amino acid at position 109 is Ser and not Thr as mentioned in the SwissProt data base (Accession No. P32848). This work demonstrates the great potential of the ESI-MS technique as a sensitive, specific, and rapid method for direct identification and determination of the stoichiometry of Ca 21-binding proteins and other metalloproteins.

lent review, J. A. Loo (1) recently summarized the developments in the investigation of noncovalent protein complexes analyzed with electrospray ionization mass spectrometry (ESI-MS). 1 Noncovalent interactions have been studied by various physical methods such as flow dialysis, radioactivity measurements, circular dichroism and light scattering. Each of these methods has its strengths and weaknesses. Drawbacks are inaccuracy in molecular weight determination, requirement of large amounts of material, and slowness. Mass spectrometry, on the other hand, is a widely used technique for the determination of molecular masses of organic and bioorganic molecules. The coupling of the method with the soft electrospray ionization technique extended the mass range up to 200,000 Da and made thermally labile compounds of low volatility amenable to the accurate determination of their respective molecular weights. Furthermore, the mild process of the electrospray ionization allows direct detection of intact noncovalent complexes of biomolecules with salts, membranes, and metals. The stoichiometry of such complexes can be deduced directly from the resulting mass spectrum. The method is highly specific, sensitive, and fast. It must be kept in mind, however, that mass spectrometry is performed in the gas phase and, therefore, the results do not necessarily reflect the situation in the condensed phase. A survey of the published literature revealed in general consistency with the expected results and it seems that the

© 1999 Academic Press

Proteins interact noncovalently with target proteins, peptides, oligonucleotides, and metal ions. In an excel64

1 Abbreviations used: ESI-MS, electrospray ionization mass spectrometry; PV, parvalbumin; RP-HPLC, reversed-phase high-performance liquid chromatography; LSIMS/MS, liquid secondary ion tandem mass spectrometry; IPTG, isopropyl b-D-thiogalactopyranoside; PMSF, phenylmethylsulfonyl fluoride; MW, molecular weight.

0003-2697/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.

SPECTROMETRIC ANALYSIS OF Ca 21-BINDING PROPERTIES

gas-phase ions indeed reflect the condensed-phase structure (2). To further support these findings we searched for a test system with high Ca 21-binding affinity and various binding locations that could be investigated under physiological conditions (viz. pH 7). Parvalbumins (PVs) are Ca 21-binding proteins characterized by the EF-hand structure (3, 4). For our studies, we expressed a purified human recombinant a-PV and nine mutants with partially or totally inactivated Ca 21-binding sites, as described earlier (5). One or both functional Ca 21-binding sites were modified by replacing the first, the last, or both amino acids, Asp and Glu, in the Ca 21-binding loops by the two nonpolar amino acids, Ala and Val, respectively. This set of modified compounds with well-known metal-binding stoichiometry is an ideal test model for the critical evaluation of the mass spectrometric approach to compare the gasphase association results with the condensed-phase binding characteristics. Another important goal of this work was the evaluation of the best conditions for determining the complex stoichiometry such as concentration, composition, and pH of the buffer system, ion source potentials, and for establishing methods of desalting and metal displacing for further studies of various metal–protein interactions. The most crucial point hereby is the method of desalting and metal removing. We obtained essential inputs from previous publications dealing with metal-binding stoichiometry of other proteins (6 –11). The binding characteristics of rabbit parvalbumin and bovine calmodulin (6, 7), rat calbindin D 28K (8), human matrilysin (9), of a 26residue peptide (10), and of synthetic calmodulin (11) were analyzed by electrospray mass spectrometry. Desalting and metal removing procedures in the above-mentioned papers vary from the microcentrifuge filters with a molecular weight cutoff of 5000 Da (6, 7) to precipitation with trichloroacetic acid (11) and dialysis (9). All the described techniques are quite sophisticated, time-consuming, and demand skillful experimentation. We obtained very good results by a simple and rapid procedure, namely, using RP-HPLC with a short C 8 column and dissolving the lyophilized proteins in 10 mM ammonium acetate buffer at pH 7.0 prior to ESI-MS experiments. No organic solvent, which would stabilize and sensitize the ESI process by reducing the surface tension and aiding the aerosol formation (12, 13), was added to the buffer. Here, we report on the Ca 21-binding stoichiometry of the test compounds, of rat and of murine a-PV, extracted from tissues and determined by ESI-MS, and on the comparisons with the flow dialysis results.

65

MATERIALS AND METHODS

Materials. HPLC-grade acetonitrile and water were purchased from Merck and Burdick & Jackson, respectively. Trifluoroacetic acid (TFA) was purchased from Perkin–Elmer. Ammonium acetate and water were from Fluka; calcium acetate was purchased from Sigma. Labware. We exclusively used polystyrene vials and plastic vessels. Expression and purification of mutant proteins. The expression and purification of mutant proteins is described elsewhere (5). Briefly, site-directed mutagenesis was carried out to modify the human PV cDNA. The Ca 21-binding sites were inactivated by replacing the first, the last, or both amino acids, Asp and Glu, in the Ca 21-binding loops by the two nonpolar amino acids, Ala and Val, respectively. The wild-type and nine mutant hPV cDNA fragments were cloned into the bacterial expression vector pGEMEX-2 (Promega, Madison, WI). Bacteria from the Escherichia coli strain BL21(DE3)pLysE carrying the expression plasmid were grown to an optical density of 0.4 at 600 nm and the expression was induced by adding 0.5 mM IPTG. Bacteria were harvested, sonificated, and resuspended in 10 mM Tris (pH 8.5). The lysate was heated to 80°C and centrifuged. The supernatant was applied to a Q Sepharose column and the protein was eluted by addition of 50 mM NaCl. Fractions containing homogenous protein were identified by SDS–PAGE and Coomassie brilliant blue staining. Isolation of rat and murine parvalbumin. One gram each of rat and murine muscle tissue was harvested and resuspended in 3 ml of 20 mM Tris (pH 8.5), 0.1 M KCl, 1 mM PMSF. After sonification the lysate was centrifuged at 17,000 rpm for 30 min at 4°C. The supernatant was heated to 80°C for 30 min, cooled on ice, and centrifuged at 17,000 rpm for 30 min at 4°C. The protein was finally purified by RPHPLC with an Applied Biosystems 140B solvent delivery system on a Brownlee RP-300 (C 8 ) Aquapore column (2.1 mm 3 10 cm). Solvents were 0.1% TFA (A) and 0.072% TFA in 80% acetonitrile (B); the flow rate was 50 ml/min. The proteins were eluted with a linear gradient of 40 –70% solvent B over 45 min. Column effluent was monitored with an Applied Biosystems 759A absorbance detector at 220 nm, and fractions were collected manually. Sample preparation. For metal removal the proteins were desalted by RP-HPLC with the system described above, except that a Brownlee RP-300 (C 8) Aquapore column (2.1 mm 3 3 cm) was used. The flow rate was 200 ml/min. The proteins were eluted with a linear gradient of 0 –75% solvent B in 13 min. Column effluent was monitored at 220 nm. The proteins were

66

TROXLER ET AL.

lyophilized at room temperature (SpeedVac, Savant), then relyophilized after addition of 50 ml of 10% methanol. This procedure was repeated twice. For mass spectrometric analysis, the proteins were dissolved in 10 mM ammonium acetate buffer (pH 7.0). The pH of the buffer solution was adjusted by the addition of 1% ammonia and controlled with indicator strips (Fluka, Pehanon). For the analysis of the metal-free protein at pH 3.5 the samples were dissolved in acetonitrile/water (50/50, v/v) containing 0.1% acetic acid. For determining the Ca 21-binding stoichiometry at neutral pH, a 1 mM calcium acetate solution was added to the protein solution to a final concentration of 100 mM Ca 21. Mass spectrometry. The electrospray ionization spectra were acquired on a Perkin–Elmer SCIEX API 365 LC/MS/MS System mass spectrometer with an atmospheric pressure electrospray ion source and a quadrupole mass analyzer with a maximum mass range of 3000 Da. The mass spectrometer was scanned from m/z 1000 to 2500 in 7 s (potentials: IonSpray 5200 V, orifice 30 V, and ring 180 V). Fifteen to 20 scans were averaged to obtain the final spectrum. The flow rate was 10 ml/min. Sample solution (20 ml) was injected with a Rheodyne 8125 injector valve equipped with a 20-ml sample loop. Typical sample quantities used were 800 pmol per analysis. The electrospray carrier solvent used was either acetonitrile/water (50/50, v/v) with 0.1% acetic acid for analysis at pH 3.5 or 10 mM ammonium acetate for analysis at pH 7.0. Before injection of proteins, the capillary was washed with formic acid and then rinsed with the carrier solvent for 4 h to remove traces of acid. Proteolytic digestion. Murine parvalbumin (25 mg) was dissolved in 100 ml freshly prepared 100 mM ammonium bicarbonate (pH 8.5) containing 1.6 mM ethylene glycol bis(b-aminoethyl)-N,N,N9,N9-tetraacetic acid (EGTA), and digested overnight with 2.5 mg of bovine trypsin (Boehringer Mannheim) at 37°C. The resulting mixture was fractionated by RP-HPLC with the same system as described above. We used a Brownlee RP-300 (C 8) Aquapore column (2.1 mm 3 10 cm) for separating the peptides. The peptides were eluted with a linear gradient of 0 – 65% solvent B over 60 min. The flow rate was 100 ml/min. Column effluent was monitored at 220 nm, and the fractions were collected manually. Amino acid sequencing by mass spectrometry. Peptides were mass analyzed by ESI-MS and sequenced by LSIMS/MS. The procedure is described elsewhere (14, 15). Mass spectra were recorded on a TSQ-70 triple stage quadrupole instrument (Finnigan-MAT, San Jose, CA) equipped with a cesium-ion gun (Antek, Palo Alto, CA). The number and location of the acidic Glu

TABLE 1

Molecular Weights of Human Parvalbumin and Its Nine Mutants (Apoproteins)

Protein

Calculated molecular weight (Da)

Determined molecular weight (Da)

PV WT PV E101V PV D90A PV E62V PV D51A PV D90A,E101V PV E62V,E101V PV D51A,D90A PV D51A,E62V PV D51A,E62V,D90A,E101V

11,927.52 11.897.54 11,883.51 11,897.54 11,883.51 11,853.53 11,867.56 11,839.51 11,853.53 11,779.54

11,927.34 6 0.76 11,897.10 6 0.91 11,883.65 6 0.77 11.897.59 6 1.13 11,883.77 6 0.98 11.853.89 6 1.50 11,867.31 6 0.87 11,839.15 6 0.71 11,853.54 6 0.90 11,779.40 6 1.06

and Asp were determined by additional methylation of the peptide fragments (16). RESULTS AND DISCUSSION

Determination of the Molecular Weights and Ca 21Binding Properties of Human PV WT and Nine Mutant Parvalbumins To be suitable for positive ESI-MS, the proteins must carry multiple charges, effected by protonation. To reach that goal, the solvent of the sample must be brought to an acidic pH ('3); under these conditions proteins are denatured and metal-free. To confirm the correct sequence of the PV WT and of its nine mutants, the molecular weights of the denatured species were determined by electrospray mass spectrometry. The sequence of PV WT is found in the SwissProt database (Accession No. P20472). Human PV WT consists of 109 amino acids and exhibits a molecular weight of 11,927 Da. This sequence was loaded onto the GPMAW-program (General Protein/Mass Analysis for Windows, Lighthouse data, 1997) for calculating the molecular weights of the nine mutants. The molecular weights of all recombinant proteins were found to be in good agreement with the calculated values (Table 1). In all but one (PV WT) spectrum, an additional peak with a mass difference of 142 Da was found, corresponding to the acetylated N-terminus. Protonation and deprotonation of proteins in solution modifies their three-dimensional structures and influences the noncovalent binding of ligands such as metal ions. Since the amino acids associated with the binding of such ligands are in close vicinity, a noticeable pH shift that influences the tertiary and/or secondary protein structure may disrupt the metal-binding pockets of the protein, resulting in a different metal stoichiometry.

SPECTROMETRIC ANALYSIS OF Ca 21-BINDING PROPERTIES

Human parvalbumin and the nine mutant proteins were acid-denatured at pH 3.5, conditions that are typical and common for electrospray ionization. Since there is no possibility of investigating the association stoichiometry of the proteins in such an acidic environment, the electrospray mass spectra reveal the molecular weights of the metal-free proteins (Table 1) without any bound ligands. An aqueous solution at a near neutral pH, without excess of organic solvent that would stabilize the electrospray process, is highly recommended for metal ion and protein interaction studies (12). Water is the native environment where proteins associate with other ligands under physiological conditions. Human parvalbumin binds 2 mol of Ca 21 ions/mol protein with high affinity (5). The nine mutant proteins, however, contain inactivating substitutions at positions essential for Ca 21 binding in the CD Ca 21binding site (PV E62V, PV D51A, PV D51A,E62V), the EF site (PV E101V, PV D90A, PV D90A,E101V), or in both (PV E62V, E101V, PV D51A,D90A, PV D51A,E62V,D90A,E101V), and therefore can bind only one or no Ca 21 ion. All of these proteins served as a test system to establish a method to detect and analyze different metal ion binding proteins in tissue and cell extracts. The aim of the work was to introduce a fast and reliable method for the determination of metal ionbinding stoichiometry of proteins. The steps hereby are (i) extraction of the proteins from tissues, (ii) separation of the protein mixture, (iii) metal ion removal and desalting, which makes the proteins amenable to electrospray ionization mass spectrometric analysis, and (iv) dissolving the protein in an appropriately lowconcentration buffer with neutral pH and acquisition of an ESI mass spectrum. Finally, a fixed concentration of the desired metal ion is added to the protein solution and a second ESI mass spectrum is acquired. The mass difference between the two molecular weights, measured with and without the addition of metal ions, reveals the number of specifically bound metal ions. We tested several low buffer concentrations of ammonium acetate and several different pH values; the best results were obtained with 10 mM ammonium acetate at pH 7.0. Hence, this buffer system was used as the solvent. The ion source parameters were identical to those used for usual protein analysis at low pH and set to values such that the ions bear low kinetic energies only. Figure 1 shows the ESI mass spectra of human PV and two different mutant parvalbumins that bind two (PV WT), one (PV D51A), and no Ca 21 ion (PV D51A,E62V,D90A,E101V). The upper part (Figs. 1A, 1C, and 1E) represents the deconvoluted mass spectra of the proteins at pH 7.0 in the absence of Ca 21 ions. All the spectra show the metal-free species (apoproteins) without Ca 21 ion-binding because

67

no such ions are available in the solution. The lower part (Figs. 1B, 1D, and 1F) shows the respective spectra of the proteins measured in presence of Ca 21 ions, effected by the addition of 100 mM calcium acetate (holoproteins). Figure 1A shows the Ca 21-free PV WT, whereas Fig. 1B shows the holoprotein, which binds two Ca 21 ions per molecule. The mass difference of the two main peaks is found to be 176 Da, which corresponds to the addition of two Ca 21 ions minus the loss of four protons replaced by the metal ions. Figures 1C and 1D represent the transformed spectra of PV D51A in which the CD Ca 21-binding loop is inactivated by the substitution of the amino acid Asp by Ala. Therefore, this protein can bind only one Ca 21 ion. In each spectrum of the mutant parvalbumin an additional prominent peak with a mass difference of 142 Da is observed. The peak with the higher molecular mass corresponds to the acetylated species of the protein. The mass difference between the molecular weight of the Ca 21-bound species (Fig. 1D) and the apoprotein (Fig. 1C) is 138 Da, revealing the addition of exactly one Ca 21 ion and the loss of two protons. These results fully confirm those obtained by flow dialysis (5). Figures 1E and 1F represent the ESI mass spectra of PV D51A,E62V,D90A,E101V. Here, both Ca 21-binding sites are inactivated by substitution of two amino acids in the CD binding site and two amino acids in the EF binding site (5). This protein cannot bind any Ca 21 ions, as is confirmed by our results. In both ESI mass spectra the same two main peaks are observed, revealing that the mutant PV and its acetylated species do not bind Ca 21 ions. It is important to consider here that in ESI-mass spectra nonspecific, noncovalent complexes are routinely observed (17, 18). Therefore, the experiment with PV D51A,E62V,D90A,E101V serves as an ideal negative control. All the results prove that in our experiments the Ca 21 binding is specific and that it is not an artifact associated with the soft electrospray process. All other six mutant parvalbumins were successfully analyzed in the same way for their Ca 21-binding stoichiometry (see Table 2, mass spectra not shown). Application of the Method to Tissue Samples After successfully testing the purified parvalbumins, the method was then applied to rat and murine parvalbumins extracted from the skeletal muscle where they are found in high concentrations (19). We analyzed these proteins using the method discussed above. Figures 2A and 2B represent the ESI mass spectra of murine parvalbumin without (Fig. 2A) and with 100 mM Ca 21 ions added (Fig. 2B). Clearly, the shift of the molecular mass of 176 Da indicates two bound Ca 21 ions replacing four protons.

FIG. 1. Deconvoluted electrospray mass spectra of PV WT (A and B), PV D51A(C and D), and PV D51A,E62V,D90A,E101V (E and F). The proteins were dissolved in 10 mM ammonium acetate (pH 7.0). (A, C, and E) without and (B, D, and F) with the addition of 100 mM calcium acetate. Peaks marked with asterisks correspond to Na 1, K 1, and NH 41 adducts of the protein. The mass spectra of both mutant proteins reveal two main peaks corresponding to the native and the acetylated proteins.

68 TROXLER ET AL.

69

SPECTROMETRIC ANALYSIS OF Ca 21-BINDING PROPERTIES TABLE 2

Molecular Weights of Human Parvalbumin and Its Nine Mutants (Holoproteins) Protein

Calculated M r with 2, 1, or 0 Ca 21 ions bound (Da)

Determined M r with 2, 1, or 0 Ca 21 ions bound (Da)

Number of bound Ca 21 ions/molecule

PV WT PV E101V PV D90A PV E62V PV D51A PV D90A,E101V PV E62V,E101V PV D51A,D90A PV D51A,E62V PV D51A,E62V,D90A,E101V

12,003.41 11,935.49 11,921.46 11,935.49 11,921.46 11,891.48 11,867.56 11,839.51 11,891.48 11,779.54

12,003.93 6 0.47 11,935.89 6 1.73 11,922.29 6 1.17 11,935.72 6 0.64 11,922.54 6 0.74 11,891.19 6 0.38 11,867.56 6 1.79 11,839.54 6 1.02 11,892.74 6 2.15 11,779.74 6 0.17

2 1 1 1 1 1 0 0 1 0

Figures 2C and 2D show the spectra of rat parvalbumin with (Fig. 2D) and without Ca 21 addition (Fig. 2C) under neutral conditions. A mass shift is observed between the holoprotein and the apoprotein of 176 Da, which corresponds to the binding of two Ca 21 ions. These two analyses of calcium-binding proteins show that this method can be applied to the analysis of metal ion stoichiometry of naturally occurring proteins. Posttranslational Modifications of Rat and Murine Parvalbumin The ESI-MS analysis of rat a-parvalbumin reveals a measured molecular weight of 11,836.88 6 1.57 Da. The expected molecular mass, which can be found in the SwissProt database (Accession No. P02625), is 11,794 Da. The mass difference between the measured and the expected mass is 42 Da. This value corresponds well with the posttranslational modification, the acetylation of the N-terminal amino acid, and is a wellknown modification of rat parvalbumin. The molecular weight of murine a-parvalbumin, on the other hand, was determined to be 11,841.21 6 0.77 Da. The expected molecular weight from the SwissProt database (Accession No. P32848) was found to be 11,813 Da. For reasons of similarity between rat and murine parvalbumin the same N-terminal posttranslational modification would be expected; the mass difference between the measured and expected molecular mass is, however, 28 Da instead of 42 Da for the supposed acetylation. This difference suggests formylation of the N-terminus instead of acetylation. To verify this hypothesis we digested the murine parvalbumin with trypsin and analyzed the resulting peptides with ESI-MS and LSIMS/MS. The digestion of murine parvalbumin with trypsin yielded several peptides, which were separated by RPHPLC. In addition to fractions corresponding to the expected sequence we detected a peptide with a molec-

ular weight of 1478 Da. Amino acid sequencing by LSIMS/MS revealed the structure of the acetylated N-terminal peptide. A second peptide with a molecular weight of 1381 Da was sequenced and found to correspond to the C-terminus of murine parvalbumin. However, the last amino acid of this peptide was identified as Ser instead of the Thr expected from the published mRNA sequence (20). The N- and C-terminal peptide sequences were confirmed by methylation of the acidic COOH functions. The observed mass shifts further support the proposed structures. Therefore, the published mRNA sequence coding for murine parvalbumin contains a sequencing error previously not recognized. With the correction of the C-terminal amino acid from Thr to Ser and the acetylation of the N-terminal amino acid, the measured molecular weight of the extracted murine a-parvalbumin (11,841.21 Da) corresponds exactly to the calculated value of 11,841.30 Da. CONCLUSION

The analysis of the high affinity Ca 21-binding stoichiometry of proteins by electrospray ionization mass spectrometry is a fast and reliable method. Using PV WT and mutant parvalbumins, a method was established to identify, analyze, and quantify the Ca 21-binding behavior of PVs and other Ca 21-binding proteins from tissue extracts. Desalting and metal removal is performed by a simple and fast RP-HPLC procedure. The proteins were analyzed in aqueous low buffered solutions without any organic solvent under physiological conditions. For each of the parvalbumins, the expected number of bound Ca 21 ions was determined by measuring the molecular weights of the Ca 21-binding and Ca 21-free species. The method was applied to native rat and murine a-parvalbumins. The number of bound Ca 21 ions was determined unambiguously. This reliable and rapid approach can be applied in clinical chemistry to de-

70

TROXLER ET AL.

FIG. 2. Deconvoluted electrospray mass spectra of murine parvalbumin (A and B) and rat parvalbumin (C and D). The proteins were dissolved in 10 mM ammonium acetate (pH 7.0). (A and C) without and (B and D) with the addition of 100 mM calcium acetate. Peaks marked with asterisks correspond to Na 1, K 1, and NH 41 adducts of the protein. Peaks marked with circles correspond to metal-bound species that result from incomplete metal removal.

tect Ca 21 -binding and other metal-binding proteins in various samples. The presence or absence of metal-binding proteins could be an indication for genetic diseases. Both murine and rat parvalbumin were found to be posttranslationally modified. The modifications were detected with ESI–MS by determination of their precise weights. As expected, rat and murine parvalbumins were found to be N-terminally acetylated.

Posttranslational modifications at the N-terminus play an important role in the attachment of proteins to cell membranes and, therefore, mass spectrometry is an ideal tool for such studies. Finally, mass spectrometric sequencing is also of great potential for detecting sequence errors in cDNA or gene sequences. In this work, a sequencing error of murine parvalbumin mRNA was detected by mass spectrometry.

SPECTROMETRIC ANALYSIS OF Ca 21-BINDING PROPERTIES

ACKNOWLEDGMENTS We are grateful to Dr. P. Hunziker for valuable discussions, and to M. Killen and I. Benz for critical reading of the manuscript. This work was supported by the EMDO Foundation, UBS (on behalf of an anonymous client), the Julius Klaus Foundation, and by Biomed 2 (European Union Grant BMH4CT950319/BBW Grant 95.0215-1, Switzerland).

REFERENCES 1. 2. 3. 4. 5.

6. 7. 8.

Loo, J. A. (1997) Mass Spectrom. Rev. 16, 1–23. Gaskell, S. J. (1997) J. Mass Spectrom. 32, 677– 688. Heizmann, C. W. (1984) Experientia 40, 910 –921. Heizmann, C. W., and Hunziker, W. (1991) Trends Biochem. Sci. 16, 98 –103. Rhyner, J. A., Durussel, I., Cox, J. A., Ilg, E. C., Scha¨fer, B. W., and Heizmann, C. W. (1996) Biochim. Biophys. Acta 1313, 179 – 186. Hu, P., and Loo, J. A. (1995) J. Mass Spectrom. 30, 1076 –1082. Hu, P., Ye, Q. Z., and Loo, J. A. (1994) Anal. Chem. 66, 4190 – 4194. Veenstra, T. D., Johnson, K. L., Tomlinson, A. J., Naylor, S., and Kumar, R. (1997) Biochemistry 36, 3535–3542.

71

9. Feng, R., Castelhano, A. L., Billedeau, R., and Yuan, Z. (1995) J. Am. Soc. Mass Spectrom. 6, 1105–1111. 10. Hutchens, T. W., Nelson, R. W., Allen, M. H., Li, C. M., and Yip, T. T. (1992) Biol. Mass Spectrom. 21, 151–159. 11. Lafitte, D., Capony, J. P., Grassy, G., Haiech, J., and Calas, B. (1995) Biochemistry 34, 13825–13832. 12. Veenstra, T. D., Tomlinson, A. J., Benson, L., Kumar, R., and Naylor, S. (1998) J. Am. Soc. Mass Spectrom. 9, 580 –584. 13. Nemirovskiy, O. V., Ramanathan, R., and Gross, M. L. (1997) J. Am. Soc. Mass Spectrom. 8, 809 – 812. 14. Hunt, D. F., Yates, J. R., Shabanowitz, J., Bruns, M. E., and Bruns, D. E. (1989) J. Biol. Chem. 264, 6580 – 6586. 15. Kuster, T., Staudenmann, W., Hughes, G. J., and Heizmann, C. W. (1991) Biochemistry 30, 8812– 8816. 16. Gabrielides, C., McCormack, A. L., Hunt, D. F., and Christakos, S. (1991) Biochemistry 30, 656 – 662. 17. Aplin, R. D., Robinson, C. V., Schofield, C. J., and Westwood, N. J. (1994) J. Chem. Soc., Chem. Commun., 2415–2417. 18. Ding, J., and Anderegg, R. J. (1994) ASMS Proc. 42, 913. 19. Heizmann, C. W., Berchtold, M. W., and Rowlerson, A. M. (1982) Proc. Natl. Acad. Sci. USA 79, 7243–7247. 20. Schleff, M., Zu¨hlke, C., Scho¨ffl, F., and Jockusch, H. (1994) Neuromusc. Disord. 4, 205–217.