RAPID COMMUNICATIONS IN MASS SPECTROMETRY Rapid Commun. Mass Spectrom. 2003; 17: 672–677 Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/rcm.958

The shielding effect of glycerol against protein ionization in electrospray mass spectrometry Maria Anita Mendes1, Jocelei Maria Chies2, Ana Christina de Oliveira Dias2, Spartacos Astofi Filho3 and Mario Sergio Palma1* 1

Laboratory of Structural Biology and Zoochemistry, CEIS/Dept. Biology, Institute of Biosciences, Sa˜o Paulo State University (UNESP), Rio Claro, SP-Brazil 2 Laboratory of Development and Production, Cenbiotenzimas, Federal University of Rio Grande do Sul, Porto Alegre RS-Brazil 3 CAM Federal University of Manaus, AM-Brazil Received 17 July 2002; Revised 22 January 2003; Accepted 27 January 2003

Most commercial recombinant proteins used as molecular biology tools, as well as many academically made preparations, are generally maintained in the presence of high glycerol concentrations after purification to maintain their biological activity. The present study shows that larger proteins containing high concentrations of glycerol are not amenable to analysis using conventional electrospray ionization mass spectrometry (ESI-MS) interfaces. In this investigation the presence of 25% (v/v) glycerol suppressed the signals of Taq DNA polymerase molecules, while 1% (v/v) glycerol suppressed the signal of horse heart myoglobin. The signal suppression was probably caused by the interaction of glycerol molecules with the proteins to create a shielding effect that prevents the ionization of the basic and/or acidic groups in the amino acid side chains. To overcome this difficulty the glycerol concentration was decreased to 5% (v/v) by dialyzing the Taq polymerase solution against water, and the cone voltage in the ESI triple-quadrupole mass spectrometer was set at 80–130 V. This permitted observation of a mass spectrum that contained ions corresponding to protonation of up to 50% of the ionizable basic groups. In the absence of glycerol up to 85% of the basic groups of Taq polymerase became ionized, as observed in the mass spectrum at relatively low cone voltages. An explanation of these and other observations is proposed, based on strong interactions between the protein molecules and glycerol. For purposes of comparison similar experiments were performed on myoglobin, a small protein with 21 basic groups, whose ionization was apparently suppressed in the presence of 1% (v/v) glycerol, since no mass spectrum could be obtained even at high cone voltages. Copyright # 2003 John Wiley & Sons, Ltd.

In recent years the advances in the technology of the manipulation of recombinant DNA have permitted the use of many proteins as tools in molecular biology. Methods for cloning, direct sequencing, clinical diagnosis and many other uses1 have proliferated, with the current ability to produce from microgram to milligram quantities of particular proteins. Thus, commercial insulin, growth hormones, cytokines,2 and DNA polymerases,3,4 among other recombinant proteins, have been used as therapeutic proteins and/or commercial biochemicals in contemporary biotechnological processes. The proteins resulting from these protocols are generally submitted to sodium dodecyl sulfate/polyacrylamide gel electrophoresis (SDS-PAGE) as the only criterion to check the homogeneity of the preparations. However, some of these proteins may present different molecular forms4,5 caused by post-translational modifications, or may even suffer artifactual proteolytic cleavage.5,6 Taking into account the standard *Correspondence to: M. S. Palma, Laboratory of Structural Biology and Zoochemistry, CEIS/IBRC-UNESP, Avenue 24A, 1515 Bela Vista, Rio Claro, SP-Brazil, CEP 13506-900. E-mail: [email protected]

error of the electrophoretic methods, small changes in molecular weight (MW) may not be revealed. Electrospray ionization (ESI) mass spectrometry (MS) is a rapid and precise method for determining molecular masses of proteins and can be used to validate protein sequences;7 in addition, it may be used as an important technique to evaluate the protein purity/homogeneity. The mass accuracy achievable using ESI-MS is generally within the range 0.01– 0.05% of the calculated masses,7,8 and it has been used to characterize many recombinant proteins.9,10 No mutant or post-translationally modified protein has been identified only by comparison between theoretical and experimental masses of the intact protein. However, ESI-MS/MS may be used to characterize post-translational modifications,11 and also to identify errors in cDNA sequences.12 Most commercial recombinant proteins used as molecular biology tools, and also even many of those prepared in the course of academic research, are generally maintained in the presence of high glycerol concentrations after purification to maintain the stability of their biological activity. The effects of many salts, detergents and chaotropic agents are relatively well documented to have a significant effect in decreasing the Copyright # 2003 John Wiley & Sons, Ltd.

Glycerol shielding effect in ESI-MS of proteins 13,14

performance of ESI-MS protocols. To our knowledge there is only a single study in the literature characterizing the effect of protein stabilization by glycerol leading in some circumstances to the suppression of the protein signal in EIS-MS.15 However, no detailed investigation appears to have been performed to characterize this suppressive effect. We describe here the suppression of protein signals in ESIMS analysis of both recombinant Taq DNA polymerase and horse heart myoglobin in the presence of high concentrations of glycerol, and also the use of an appropriate optimization of mass spectrometric conditions to overcome this difficulty.

EXPERIMENTAL Material and methods All solvents used (HPLC quality) were purchased from Mallinckrodt. Horse heart myoglobin and glycerol were acquired from Sigma Chemical Co. Bidistilled and ultra-purified water used in all experiments was prepared using a Barnsted system.

Protein cloning, purification and treatment Taq DNA polymerase was cloned from Thermus aquaticus, expressed in Escherichia coli, as described by Lawyer et al.4 and purified according to the protocol of Engelke et al.3 After purification a part of the protein preparation was maintained in the presence of 25% (v/v) glycerol, and part was maintained soluble in the presence of bidistilled water and the absence of glycerol. When necessary glycerol was added in known concentrations to the Taq polymerase preparation. To purify the Taq polymerase domain fragment (MW 46 270 Da) from the intact protein, part of the enzyme preparation (250 mg) maintained in the absence of glycerol was initially filtered through an AMICON-30 filter; the protein material retained by the filter was dissolved in 5 mM ammonium acetate (pH 6.8) and then submitted to gel filtration chromatography on a Sephadex G-75 column (30.0  1.5 cm), previously equilibrated with the same solvent. Elution was performed with 10 mM ammonium acetate (pH 6.8) at a flow rate of 12 mL/h, and 3-mL fractions were collected. Protein elution was monitored by measuring the absorbance at 280 nm; the protein fractions were pooled and concentrated by lyophilization.

Electrophoresis SDS-PAGE was performed on a 10% gel at 30 mA for 2 h. Proteins were stained with Coomassie Brilliant Blue. The MW markers used for SDS-PAGE were: phosphorylase B (97.4 kDa), bovine serum albumin (66.2 kDa), glutamate dehydrogenase (55.0 kDa), ovalbumin (42.7 kDa), aldolase (40.0 kDa) and carbonic anhydrase (31.0 kDa), all purchased from Promega.

N-Terminal amino acid sequencing When necessary the primary sequence of the N-terminal region of a specific protein was submitted to automated Edman degradation sequencing using a gas-phase sequencer PPSQ 21A (Shimadzu, Kyoto, Japan).

Determination of glycerol concentration The protein solutions containing glycerol, after and before dialysis, were collected and submitted to glycerol analysis Copyright # 2003 John Wiley & Sons, Ltd.

673

by using an enzymatic assay based on the conversion of glycerol to dihydroxyacetone in presence of NAD. The reaction was catalyzed by the enzyme glycerol dehydrogenase. The NADH, which is formed in stoichiometric quantities by this reaction, was estimated by UV absorption at 340 nm.

Mass spectrometric analysis The determination of the homogeneity of the protein preparation and of molecular mass was performed by mass spectrometry using some adaptations to the system described elsewhere.16 Samples were dissolved in 50% (v/v) acetonitrile containing 0.1% (v/v) trifluoroacetic acid (TFA), and analyzed using a QUATTRO II triple-quadrupole mass spectrometer (Micromass, Altrincham, UK) equipped with a standard electrospray probe; solutions were infused at ca. 5 mL/min. During all experiments the source temperature was maintained at 808C and the needle voltage at 3.6 kV, the drying gas flow (nitrogen) was 200 L/h, and the nebulizer gas flow was 20 L/h. The mass spectrometer was calibrated with intact horse heart myoglobin and its typical cone-voltage induced fragments. The cone-to-skimmer voltage controlling the ion transfer to the mass analyzer was manually varied from 30 to 130 V. About 50 pmol of each sample were continuously infused into the electrospray transport solvent using a microinfusion pump (KD Scientific), connected to a 500-mL microsyringe (Hamilton) which in turn was connected to the ESI probe using a silica microcapillary. The ESI spectra were obtained in the continuous acquisition mode, scanning from m/z 500– 2500 with a scan time of 7 s. The mass spectrometer data acquisition and processing system was equipped with MassLynx and Transform software for handling and deconvoluting spectra.

RESULTS AND DISCUSSION The purified recombinant Taq DNA polymerase was apparently homogeneous in the presence of 25% (v/v) glycerol, since it was recorded as a single protein band of MW 94 kDa in SDS-PAGE (Fig. 1(a)). When 50 pmol of the fresh enzyme preparation were analyzed in the presence of 25% (v/v) glycerol no ESI-MS spectrum could be obtained for the enzyme, probably suppressed by the high glycerol concentration (results not shown). However, in the presence of 5% (v/v) glycerol and using 50% (v/v) acetonitrile containing 0.1% (v/v) TFA as solvent at a cone voltage of 32 V, ESI-MS analysis produced a bimodal envelope of peaks from m/z 590 to 1921 (Fig. 2(a)); deconvolution of this spectrum resulted in two apparent MW values of 22 580  17and 24 330  12 Da. The molecular form with MW 22 580 Da resulted from a series of peaks in this envelope corresponding to ionized protein molecule populations containing from 20 to 31 positive charges (designated ‘A’ in Fig. 2(a)), and the form with MW 24 420 Da resulted from an envelope of peaks corresponding to molecules containing from 13 to 19 positive charges (designated ‘B’ in Fig. 2(a)). When the sample cone voltage was increased to 80 V, a second bimodal envelope of peaks appeared in the mass spectrum from m/z 634 to 1972 (Fig. 2(b)). The deconvolution Rapid Commun. Mass Spectrom. 2003; 17: 672–677

674

M. A. Mendes et al.

Figure 1. SD-PAGE of purified recombinant Taq DNA polymerase (a) in the presence of 25% (v/v) glycerol and (b) in the absence of glycerol; standard protein MW markers appear in the left lane. yielded two molecular forms with MW 46 880  34 and 93 324  57 Da. The form with MW 46 880 Da results from an envelope of peaks corresponding to protein molecules containing from 40 to 68 positive charges (designated ‘C’ in Fig. 2(b)); the form with MW 93 324 Da, which seems to represent the intact Taq polymerase, results from a population of molecules containing from 47 to 67 positive charges (designated ‘D’ in Fig. 2(b)). However, this bimodal envelope was still very complex, and did not provide a reliable deconvolution result for MW determination. Therefore, the cone voltage was increased to 130 V, resulting in another bimodal envelope of peaks from m/z 764 to 2059 (Fig. 2(c)); under these experimental conditions, the first envelope of peaks was deconvoluted as the molecular form ‘A’, already observed in Fig. 2(a). The deconvolution of the second envelope of peaks resulted in a molecular mass of 93 320 Da, already designated as molecular form ‘D’ in Fig. 2(b). Therefore, high glycerol concentrations (25% v/v) seem to prevent ionization of Taq polymerase, since no spectrum was obtained under these conditions. However, on decreasing the glycerol concentration to 5% (v/v), different MW forms were detected at different cone voltages. Thus, at 32 V, two forms of MW 22 580 and 24 420 Da were detected, while, at 80 V, two other MW forms were identified at 46 880 and 93 324 Da. When 130 V was applied to the sampling cone, both the smallest and the largest forms were detected. These results suggest that an intact form of Taq polymerase plus three smaller fragments of this enzyme co-exist in the preparation. The fragments may result from proteolytic action after gene expression and/or during the purification protocol. Probably, glycerol interacts strongly with Taq polymerase and all its fragments, promoting strong hydrogen bonding between glycerol molecules and the surface of each protein form in a way that keeps these fragments tightly united in a larger cluster similar to the intact Taq polymerase (form ‘D’); at least in SDS-PAGE these fragments were not detected in the presence of glycerol (Fig. 1(a)). Glycerol at a concentration of 5% (v/v) seems to prevent the detection of the forms with MW 46 880 and 93 320 Da Copyright # 2003 John Wiley & Sons, Ltd.

Figure 2. ESI-MS spectra of Taq DNA polymerase in the presence of 5% (v/v) glycerol at different cone voltages: (a) 32 V; (b) 80 V; and (c) 130 V. under low cone-voltage conditions. Probably, the intact protein and its largest fragment interact more strongly with glycerol, partially preventing the ionization of the basic amino acid residues in these molecular forms. Taq polymerase was prepared in the absence of glycerol, and 50 pmol of this enzyme were analyzed in presence of 50% (v/v) acetonitrile containing 0.1% (v/v) TFA. A rather complex ESI-MS spectrum was obtained when the cone voltage was 45 V (Fig. 3(a)). Since different envelopes of peaks overlapped one another, the deconvolution and the assignment of each series of peaks, followed by their respective charge determinations, were performed using centroided data (Figs. 3(b)–3(e)). The insert in Fig. 3(a) reveals that four different MW forms were detected in the absence of glycerol at a relatively low cone voltage (45 V), corresponding to the same forms already detected in the presence of 5% (v/v) glycerol at high cone voltages (Figs. 2(a)–2(c)), i.e., 22 066  18, 24 080  14, 46 290  36 Da, and the intact Taq polymerase with 93 416  61 Da. SDS-PAGE of this glycerolfree Taq polymerase preparation revealed the presence of a band of MW 46 kDa, while the smaller forms (22–24 kDa) were not detected by this method (Fig. 1(b)). The form presenting MW 46 kDa was isolated from the Taq polymerase preparation in the absence of glycerol (results not shown), as described in the Experimental section. The N-terminal region of this Taq fragment was sequenced by Edman degradation, yielding the sequence EGERLL. Thus, taking into account this N-terminal sequence and the molecular mass value obtained Rapid Commun. Mass Spectrom. 2003; 17: 672–677

Glycerol shielding effect in ESI-MS of proteins

675

Figure 3. (a) ESI-MS spectra of Taq DNA polymerase in the absence of glycerol, acquired in the continuum mode, with the cone voltage adjusted to 45 V; (b)–(e) decomposition of the ESI-MS spectrum in (a) into different envelopes of peaks, represented in centroid mode. after purification (46 290 Da), this protein was identified as the DNA polymerase domain of the Taq enzyme, which corresponds to the Taq polymerase fragment 420–835 (with expected MW 46 270.76 Da); this fragment is still active (results not shown). These results also corroborate the proposal that somehow glycerol interacts strongly with protein fragments, keeping them so tightly bound (probably through hydrogen bonds between glycerol molecules and protein fragment surfaces) that they cannot be separated/ detected from each other by using mild experimental conditions. The centroided representation permitted individual deconvolution of the ESI-MS spectral envelopes for each form of different MW. Figure 3(b) shows that the envelope of peaks corresponding to the form of MW 22 066 Da was produced by a population of molecules containing from 23 to Copyright # 2003 John Wiley & Sons, Ltd.

33 positive charges (designated the ‘A’ series of centroid peaks in Fig. 3(b)), i.e., in the same range of charges as that observed in the presence of glycerol. Figure 3(c) shows that the envelope of peaks corresponding to the Taq fragment of MW 24 083 Da was produced by molecules containing from 18 to 33 positive charges centered around 25 charges (designated the ‘B’ series of centroided peaks in Fig. 3(c)). In the presence of glycerol this fragment produced an envelope of peaks corresponding to molecular species containing from 12 to 19 positive charges (Fig. 2(a)). Thus, the presence of glycerol seems to be partially shielding some basic residues of this protein fragment. In the absence of glycerol the form corresponding to Taq fragment 420–835 yielded an envelope of peaks produced by protein molecules with from 49 to 58 positive charges (designated the ‘C’ series of centroided peaks in Fig. 3(d)) Rapid Commun. Mass Spectrom. 2003; 17: 672–677

676

M. A. Mendes et al.

when a cone voltage of 45 V was used. In contrast, in the presence of 5% (v/v) glycerol, no signal was detected for this protein form at the same cone voltage, although an envelope of peaks interpreted as arising from this form was detected using 80 V cone voltage. This envelope was produced by protein molecules containing from 43 to 68 positive charges. By setting the sample cone voltage to 45 V in the absence of glycerol, the intact Taq polymerase produced an envelope of peaks which corresponded to molecules containing from 47 to 117 positive charges (designated the ‘D’ series of centroided peaks in Fig. 3(e)), while, in the presence of 5% (v/v) glycerol, the envelope of peaks for this protein was detected only at sample cone settings 80 V. Under these conditions the envelope was produced by protein molecules containing from 47 to 69 positive charges, centered around 59 charges. Taking into account the total number of the basic amino acid residues present in the intact Taq polymerase molecule (76 arginines, 42 lysines, 18 histidines) in addition to the amino terminal residue, one might expect that a maximum of 137 positive charges is possible for this protein. This means that, in the absence of glycerol, up to 85% of the basic groups were ionized at relatively low cone voltage, while, in the presence of 5% (v/v) glycerol, only up to 50% of these groups were ionized in the mass spectrum, even at high cone voltage. Thus, it is very clear that the shielding effect caused by glycerol also required higher cone potentials to allow detection of the protein signal. Of course the cone potential has no effect on the electrospray process itself, i.e., the nature and distribution of the ions emerging from the electrospray source and presented to the API interface are entirely independent of the potentials within the interface. A possible explanation for this and related observations described above is that glycerolbound dimers (or even higher oligomers) can be formed by electrospray in the presence of glycerol for Taq DNA polymerase and its larger fragment, but not for the smaller (20 kDa) fragments. At lower cone voltages these glycerolbound oligomers survive the collision conditions in the API interface, but possess m/z values too high to be detected by the instrument used in this work. At higher cone voltages the collision conditions are more energetic and protein monomers are formed and detected, although with a lower chargestate distribution than that formed in the absence of glycerol due to the glycerol shielding effect that operates in the electrospray process itself. In order to corroborate these observations, a similar experiment was performed with a smaller protein, horse heart myoglobin. In the absence of glycerol the ESI-MS spectrum of this protein, obtained using a cone voltage of 35 V, revealed a single (unimodal) envelope of peaks from m/z 645 to 1540, corresponding to protein molecules containing from 11 to 21 positive charges. The deconvolution of this spectrum resulted in a MW value of 16 924  12 Da (Fig. 4(a)). The presence of 0.1% (v/v) glycerol did not change significantly the spectrum observed in its absence; however, it was necessary to set the cone voltage to 45 V in order to obtain a useful protein signal (Fig. 4(b)). In the presence of 0.5 % (v/v) glycerol the ESI-MS spectrum became partially suppressed, presenting a small envelope of peaks from m/z 942 to 1210, corresponding to protein molecules containing Copyright # 2003 John Wiley & Sons, Ltd.

Figure 4. ESI-MS spectrum of horse heart myoglobin under different experimental conditions with inserts representing the deconvolution of each spectrum: (a) in the absence of glycerol and with the cone voltage set to 35 V; (b) in the presence of 0.1% (v/v) glycerol with the cone voltage set to 45 V; (c) in the presence of 0.5% (v/v) glycerol with the cone voltage set to 60 V; and (d) in the presence of 1% (v/v) glycerol with the cone voltage set to 95 V. from 14 to 19 positive charges (Fig. 4(c)). However, it must be emphasized that with 0.5% glycerol present, no signal was observed at cone voltages lower than 60 V. Nonetheless, in spite of this effect, the deconvolution of these data resulted in a consistent MW value (16 918  14 Da). When the mass spectrometric analysis of myoglobin was performed in the presence of 1% (v/v) glycerol, no ESI-MS signal was observed, even by setting high cone voltages (up to 95 V; Fig. 4(d)). Once more it seems that glycerol is interacting with the protein, preventing the ionization of the basic side chains of arginine and lysine residues. The present observations for the effect of glycerol on the charge-state distribution of myoglobin are qualitatively similar to those reported previously15 for other small proteins (lysozyme and cytochrome c).

CONCLUSIONS The addition of glycerol after the final purification step, for both natural and recombinant proteins, has the purpose of preventing these molecules from denaturation.15 However, Rapid Commun. Mass Spectrom. 2003; 17: 672–677

Glycerol shielding effect in ESI-MS of proteins

in addition to the stabilizing effect on the protein molecules, high glycerol concentrations strongly influence the electrospray mass spectrometric performance, since they suppress the protein signal resulting in a flat and noisy base line. Other authors have also reported that protein samples containing high glycerol concentrations are not amenable to ESI interfaces, requiring the use of high nozzle potential in an ESITOFMS instrument.15 An explanation for this effect is proposed, based on strong interactions between glycerol molecules and the side chains of the basic amino acid residues in the proteins, thus interfering with protonation of some of these residues. For some larger proteins these interactions with glycerol are sufficiently large that glycerol-bound dimers (or higher oligomers) are formed by the electrospray process, and these species have m/z values too high to be detected by the mass spectrometer. However, by increasing the voltage in the API interface, it is possible to create collision conditions sufficiently violent that some protein ions without any clustered glycerol molecules are formed and detected, although with a lower charge-state distribution than that observed in the absence of glycerol. This same proposal can also account for the fact that our Taq polymerase preparation, when analyzed by SDS-PAGE in the presence of glycerol, appeared to contain only the intact protein although both mass spectrometric and SDS-PAGE analysis in the absence of glycerol clearly indicated the presence of a fragment of the native enzyme with about half the correct MW. It is thought that glycerol could bind the fragments together to form dimers that cannot be resolved from the intact protein by SDS-PAGE. Finally, similar experiments on a much smaller protein (myoglobin) yielded evidence for the effects of added glycerol on both suppression of ionization and a shift in the charge-state distribution that were qualitatively similar to those reported previously15 for other small proteins. Thus, the present observations on our Taq DNA polymerase

Copyright # 2003 John Wiley & Sons, Ltd.

677

preparation cannot be attributed to instrumental effects, but reflect different phenomena that become apparent only for larger proteins.

Acknowledgements This work was supported by grants from FAPESP (CAT/ CEPID and SMOLBNET) and Instituto do Mileˆnio (CNPq/ MCT). MAM is a post-doctoral fellow of FAPESP; MSP is a researcher of the National Research Council (CNPq, 500079/90–0).

REFERENCES 1. Innis MA, Gelfeld DH, Sniski JJ, White TJ. PCR Protocols: A Guide to Methods and Applications. Academic Press: San Diego, 1990. 2. Cleland JL. Impact of protein folding on biotechnology. In Protein Folding In Vivo and In Vitro, Vol. 256, Cleland JL (ed.). American Chemical Society, Washington, DC, 1993; 1–21. 3. Engelke DR, Krikos A, Bruck ME, Ginsburg D. Anal. Biochem. 1990; 191: 396. 4. Lawyer FC, Stoffel S, Saiki RK, Myambo K, Drummond R, Gelfand DH. J. Biol. Chem. 1989; 264: 6427. 5. Chien A, Edgar DB, Trela JM. J. Bacteriol. 1976; 127: 1550. 6. Kaledin AS, Slyusarenko AG, Gorodetskii SI. Biokymiya 1980; 454: 644. 7. Ashton DS, Beddell CR, Green BN, Oliver RWA. FEBS Lett. 1994; 342: 1. 8. Chait BT, Kent BH. Science 1992; 257: 1885. 9. Pedrocchi M, Schafer BW, Durussel I, Cox JA, Heizmann CW. Biochemistry 1994; 33: 6732. 10. Fohr UG, Heizmann CW, Engelkamp D, Schafer BW, Cox JA. J. Biol. Chem. 1995; 270: 21056. 11. Raftery MJ, Harrison CA, Alewood P, Jones A, Geczy CL. Biochem. J. 1996; 316: 285. 12. Raftery MJ, Geczy CL. Anal. Biochem. 1998; 258: 285. 13. Canarelli S, Fisch I, Freitag R. J. Chromatogr. A 2002; 948: 139. 14. Liu CL, Hofstadler SA, Bresson JA, Udseth HR, Tsukuda T, Smith RD, Snyder AP. Anal. Chem. 1998; 70: 1797. 15. Grandori R, Matecko I, Mayr P, Mu¨ller N. J. Mass Spectrom. 2001; 36: 918. 16. Chassaigne H, Lobinski R. Analyst 1998; 123: 2125.

Rapid Commun. Mass Spectrom. 2003; 17: 672–677