- Email: [email protected]
Effect of the mobile phase composition on the separation and detection of intact proteins by reversed-phase liquid chromatography–electrospray mass spectrometry
Journal of Chromatography A, 957 (2002) 187–199 www.elsevier.com / locate / chroma
Effect of the mobile phase composition on the separation and detection of intact proteins by reversed-phase liquid chromatography–electrospray mass spectrometry ´ a,b , *, A.C. Hogenboom a , H. Zappey a , H. Irth a M.C. Garcıa a
Department of Analytical Chemistry and Applied Spectroscopy, Vrije Universiteit, De Boelelaan 1083, 1081 HV Amsterdam, The Netherlands b ´ ´ , Facultad de Quımica ´ , Universidad de Alcala´ , 28871 Alcala´ de Henares, Madrid, Spain Departamento de Quımica Analıtica Received 11 October 2001; received in revised form 27 February 2002; accepted 7 March 2002
Abstract Various buffers (ammonium acetate, ammonium formate, and ammonium hydrogencarbonate), acids (formic acid, acetic acid, heptafluorobutyric acid, and trifluoroacetic acid), and bases (ammonium hydroxide and morpholine) covering the range from 2 to 11.5 have been investigated for their performance in the separation of proteins by reversed-phase liquid chromatography (RPLC) and in their detection by electrospray mass spectrometry (ESI-MS). These additives were first tested for the detection of standard proteins by ESI-MS by flow-injection analysis (FIA). Those additives yielding the highest signals were employed for the separation of standard proteins by using three different reversed-phase columns: two C 18 columns (4.6 mm I.D. and 2.1 mm I.D.) and one perfusion column (2 mm I.D.). The sensitivity of the LC–MS system was evaluated with the column giving the best results and with those LC eluents enabling the LC separation of the proteins and also yielding the highest MS signals. For that purpose, calibration curves were compared for both LC–MS and FIA–MS. Formic acid was the additive yielding the highest responses in FIA–MS and trifluoroacetic acid (TFA) gave the best separation and recovery of the proteins. However, problems related to poor recovery of the proteins in the column when formic acid was used and the significant signal suppression observed in MS when TFA was employed, made neither of them suitable for the sensitive detection of the proteins in LC–MS. 2002 Elsevier Science B.V. All rights reserved. Keywords: Mobile phase composition; Proteins
1. Introduction The biological interest in proteins and peptides has increased the demands in instrumental tools. Electrospray mass spectrometry (ESI-MS) enables the ad* Corresponding author. Departamento de Quımica ´ ´ Analıtica, ´ ´ 28871 Alcala´ de Facultad de Quımica, Universidad de Alcala, Henares, Madrid, Spain. Fax: 134-918-854-915. ´ E-mail address: [email protected] (M.C. Garcıa).
vanced detection of these biological molecules due to its ability to form multiply charged ions and thus, reducing the m /z values to levels that can be detected [1,2]. However, when coupling this powerful detection technique with a so widely employed technique for the separation of proteins such as reversed-phase liquid chromatography (RPLC), incompatibilities that can reduce the performance of the LC–MS system can appear. Indeed, the separation of proteins by RPLC requires mobile phases
0021-9673 / 02 / $ – see front matter 2002 Elsevier Science B.V. All rights reserved. PII: S0021-9673( 02 )00345-X
188
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
which usually are not suitable for ESI-MS detection. In addition, flow-rates normally used for LC are higher than those used for optimum ESI-MS detection. These two factors contribute to the decrease in sensitivity in ESI-MS detection. However, the increasing necessity of detecting very low concentration levels, especially in certain areas of the pharmaceutical industry, has contributed to overcome these limitations. The efforts carried out so far have led to the improvement of the sensitivity by the reduction of the flow-rate. In fact, ESI-MS is not a mass-sensitive detector but a concentration-sensitive one, which means that the signal intensities are not dependent on the flow-rate, at least in the range from 100 nl / min up to 1 ml / min. Consequently, higher sensitivities could be obtained by the employment of capillary columns enabling the use of lower flowrates [3–7]. Another trial to fit LC flow-rates to ESI-MS has been addressed by Banks Jr. et al. [8] who developed an ultrasonic nebulizer enabling the right operation of the ESI-MS system at flow-rates up to several hundred microliters per minute. Despite these efforts to increase the sensitivity of the LC–MS system by the adjustment of the flowrate, not too much has been done regarding the mobile phase incompatibilities between RPLC and ESI-MS. These incompatibilities are due to the fact that separation of proteins by RPLC requires the addition of an additive working as ion-pairing agent. These components are added to increase the hydrophobicity of the proteins by forming ionic pairs with the charged groups and, in addition, enable the defolding of the protein making possible that the buried hydrophobic groups of the proteins become accessible [9]. As a consequence, interaction of the proteins with the hydrophobic stationary phase is possible and, therefore, also their separation. Different chromatographic supports have been used for the separation of proteins with on-line ESI-MS detection. From conventional silica columns [4,5,8,9] to more sophisticated supports such as nonporous [7,10–12], perfusion [3,6,13] or monolithic [14] stationary phases have been used for this purpose. In most of these works, 0.1% trifluoroacetic acid (TFA) has been the additive used in the separation of proteins by RPLC. Unfortunately, although TFA is a suitable ion-pairing agent for the separation of proteins by RPLC, it is a signal supressor in ESI-MS
[15]. In fact, TFA forms very strong ion pairs with proteins (TFA–protein) that are not broken apart by the conditions in the electrospray ionization interface and, consequently, it prevents the ionization of the protein [15]. Therefore, other additives enabling the separation of the proteins by RPLC but at the same time the sensitive detection of proteins by ESI-MS should be explored. In this work, different LC eluents will be evaluated for both the separation by RPLC and the detection by ESI-MS of proteins trying to reach a compromise between the LC and the MS performances.
2. Experimental
2.1. Chemicals and reagents HPLC-grade acetonitrile (ACN) (J.T. Baker, Deventer, The Netherlands), and water (Milli-Q system, Millipore, Bedford, MA, USA) were used in the preparation of the mobile phases. HPLC-grade TFA, heptafluorobutyric acid, and morpholine were from Sigma (St. Louis, MO, USA) and formic acid, acetic acid, ammonium hydroxide, ammonium acetate, ammonium formate, and ammonium hydrogencar¨ (Seelze, bonate were purchased from Riedel-de Haen Germany). Protein samples [cytochrome c, myoglobin, and bovine serum albumin (BSA)] were from Sigma and eggs were purchased in a local market in Amsterdam. Egg-white preparation was performed by the method described by Awade and Efstathiou [16]. The sample was filtered through 0.2-mm PTFE filters prior to analysis. Nitrogen (99.999% purity) used as sheath (5 bar) and auxiliary gas (0.7 bar) was from Praxair (Oevel, Belgium).
2.2. High-performance liquid chromatography A Hewlett-Packard (HP) 1100 Series liquid chromatograph (Hewlett-Packard, Waldbronn, Germany) equipped with a quaternary delivery solvent system and a diode-array detector was used. This system was coupled to a computer with a HP
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
Chemstation data acquisition system. The system was used for both flow injection analysis (FIA) experiments and LC separations. For FIA experiments, the mobile phases consisted of water–ACN (50:50, v / v) and contained a certain concentration of an additive. These mobile phases were filtered using 0.45-mm filters of PTFE (for organic solvents) or a mixture of cellulose esters (for aqueous solutions) (Millipore) and degassed with helium before use. All FIA experiments were performed at 25 8C. For LC experiments, three different reversedphase columns were used: a Luna C 18 column (2503 4.6 mm I.D.) and a Jupiter C 18 column (15032.0 mm I.D.) packed with 5 mm particles (Phenomenex, Torrance, CA, USA) and a Poros R2 / H perfusion column (10032.1 mm I.D.) packed with 10 mm particles (Perseptive Biosystems, Framingham, MA, USA). The separations were carried out with singlestep gradients at flow-rates of 1 ml / min (for the Luna and the Poros columns) or at 0.2 ml / min (for the Jupiter column) followed by reversed gradients to equilibrate the columns between runs to the initial conditions. The injected volume was 5 ml and mobile phases consisted of water with a certain concentration of an additive (mobile phase A) and ACN with the same concentration of the same additive (mobile phase B). The UV detection was performed at 214 nm and the columns were thermostated at 25 8C.
2.3. Mass spectrometry Mass spectrometry detection was performed on a Finnigan LCQ ion trap mass spectrometer equipped with an electrospray ionization interface (San Jose, CA, USA). Xcalibur software was used for data acquisition and data analysis. The instrument was firstly tuned to get the MS conditions enabling the highest responses. These MS conditions were used in all the experiments and were as follows: capillary voltage, 26 or 46 V; spray voltage, 5.0 kV; capillary temperature, 350 8C. Mass spectra were registered in the full-scan mode (m /z, 600–2000; scan time, 2 scans / s, maximum injection time, 100 ms). The MS system was connected to the LC system by a postcolumn splitter to maintain the flow-rate to 100
189
ml / min. All determinations were performed, at least, three times.
3. Results and discussion One of the problems in the hyphenation of RPLC and ESI-MS is the non-compatibility of some of the additives used in the mobile phases for the separation of proteins with the ESI-MS system. For example, as was already discussed in the Introduction, a decrease in response was observed in ESI-MS when TFA was used. In addition, some salts such as phosphate, perchlorate, alkylsulfates or alkylsulfonates, that are used for the separation of proteins by RPLC, are also not suitable for the ESI-MS as they are not volatile [17]. In this work, the additives listed in Table 1 were tested in order to obtain a compromise between the optimum separation and recovery of proteins by RPLC and the optimum detection by ESI-MS. In this list TFA was also included in order to compare the performance of this excellent ion-pairing agent for the separation of proteins by RPLC with the performance with the other additives. Although less volatile, heptafluorobutyric acid is another perfluorinated carboxylic acid that has also been used for the separation of proteins by RPLC. Formic acid can also act as ion-pairing agent for the separation of proteins by RPLC, although the resolution and recoveries obtained are lower than with TFA. Furthermore, it is less acidic and less volatile than TFA. Acetic acid is also suitable for the ESI-MS detection and, in addition, it is safer to handle and cheaper than TFA. Ammonium acetate and ammonium formate have also been employed in some occasions to control the pH of the LC mobile phases in the separation of some peptides and proteins such as hormones [18]. Hydrogencarbonate buffer, morpholine, and ammonium hydroxide can be used as cation-pairing agents for the separation of proteins when it is possible to work at these basic pH values. In fact, protein separations by RPLC are mostly carried out at acidic pH values at which protein exist in the fully protonated form and behave mainly as a cation solute. However, due to the aliphatic character of proteins, they can also be separated by RPLC at neutral or higher pH values [17,19]. Nevertheless,
190
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
Table 1 Relative response (%) of three standard proteins with different mobile phase additives in ESI-MS a,b Additive
Heptafluorobutyric acid Formic acid Trifluoroacetic acid Acetic acid Ammonium formate Ammonium acetate Ammonium formate Ammonium acetate Ammonium hydrogencarbonate Morpholine Ammonium hydroxide
Concentration
0.2% (v / v) 0.2% (v / v) 0.025% (v / v) 0.3% (v / v) 10 mM 10 mM 10 mM 10 mM 50 mM 10 mM 20 mM
pH
Relative response (%)
1.8 2.5 2.5 3.0 3.0 4.5 6.0 6.0 9.0 10.5 11.5
Myoglobin
Cytochrome c
BSA
,1 100 3 46 25 1 ,1 ,1 34 12 24
,1 100 3 56 37 2 ,1 ,1 11 ,1 ,1
3 98 9 100 47 2 ,1 ,1 33 ,1 8
a FIA at 0.1 ml / min; mobile phase, water–ACN (50:50); MS conditions: capillary voltage, 46 V; spray voltage, 5.0 kV; capillary temperature, 350 8C. b Response related to the highest signal observed in every protein.
these conditions are not compatible with conventional silica-based LC columns.
3.1. Effect of different mobile phases on the response in ESI-MS Before studying the performance of the additives listed in Table 1 in LC–MS, their compatibility with the ESI-MS was studied in terms of response with three standard proteins: myoglobin, cytochrome c, and BSA. For that purpose, FIA experiments with mobile phases containing a certain concentration of these additives were performed. The concentrations of these additives in the LC mobile phases are also shown in Table 1 and corresponded to those concentrations normally employed when these mobile phases are used for the separation of proteins by RPLC with the exception of TFA, in which case a lower concentration (0.025%, v / v) was employed [9]. These mobile phases were pumped into the chromatographic system at a flow-rate of 1 ml / min and split to 0.1 ml / min before reaching the ESI-MS system. In these FIA experiments, 2 mM solution of each protein was injected in triplicate. Table 1 groups the average relative responses obtained in these experiments. As can be observed, the variation of the responses with the different additives tested was similar regardless of the protein used. In all cases the highest signals were obtained with formic acid and acetic acid and the lowest responses with
ammonium acetate or ammonium formate at pH 6.0. Furthermore, an increase in the signal was observed for all three proteins when ammonium formate was used at more acidic pH. As regards basic pH values, the additive yielding the highest responses in all cases was ammonium hydrogencarbonate at pH 9.0. On the other hand, when using heptafluorobutyric acid, TFA, ammonium acetate at pH 4.5 or morpholine, the responses were very small (#12%) and in some occasions no signal was obtained. From the results obtained in this study, a group of additives yielding the highest signals for the three proteins [formic acid, acetic acid, ammonium formate (pH 3.0), and ammonium hydrogencarbonate (pH 9.0)] was selected. Moreover, a study on the effect of the variation of the concentration of such compounds on the signal observed by ESI-MS was carried out. In this study, myoglobin and cytochrome c were used and the results obtained are shown in Fig. 1. As can be observed, both proteins gave a similar variation of the relative responses with the concentration of every additive. The highest responses were always found with formic acid in which case the signal increased up to a concentration around 50 mM. Higher concentrations of formic acid resulted in a reduction of the signals of both proteins. As a consequence, formic acid at a concentration of 50 mM (0.2%, v / v) was selected for the separation of the proteins by LC. As regards acetic acid, the responses obtained for
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
191
Fig. 1. Variation of the signal (expressed as response related to the highest signal observed) obtained in ESI-MS for 2 mM solutions of myoglobin and cytochrome c with the concentration of acetic acid, formic acid, TFA, ammonium formate (pH 3) and ammonium hydrogencarbonate (pH 9). FIA at 0.1% ml / min. Mobile phase, water–ACN (50:50). MS conditions: capillary voltage, 46 V; spray voltage, 5.0 kV; capillary temperature, 350 8C.
both proteins were lower than those with formic acid and the highest signals were observed at a concentration of 8 mM. Higher concentrations of acetic acid resulted in a decrease in the response. In this case, two concentrations were chosen, 8 mM (0.05%, v / v) because it was the concentration at which the signal was maximum and 50 mM (0.3%, v / v) since it is the concentration of acetic acid normally employed for the separation of proteins. No significant change in response was observed for myoglobin and cytochrome c when ammonium formate at pH 3.0 was used at concentrations in the range from 5 to 50 mM. In addition, these responses were lower than those observed with formic acid and acetic acid. In this case, a concentration of 10 mM of this buffer was chosen since this concentration has already been used for the separation of proteins [9]. The responses obtained with ammonium hydrogencarbonate at pH 9.0 were even lower than those observed previously up to a concentration of 30 mM. At higher concentrations, the responses significantly increased. In both cases (ammonium formate and hydrogencarbonate), higher concentrations than 70 mM were also studied although problems related to the back-pressure of the system were observed. As was expected, the signal yielded with TFA decreased with increasing concentration with almost complete disappearance at 40 mM (0.4%, v / v). From these results, 0.2%
(v / v) formic acid, 0.05 and 0.3% (v / v) acetic acid, 10 mM acetate buffer, and 50 mM hydrogencarbonate buffer were selected for the MS detection of proteins and were tested for its performance in the separation of proteins by LC.
3.2. Effect of the different mobile phases on the ESI-MS spectra of the proteins Fig. 2 shows the spectra obtained for myoglobin and cytochrome c with the mobile phases previously used. In the case of myoglobin, the spectra obtained with most of the mobile phases tested (formic acid, TFA, ammonium formate, ammonium acetate, morpholine and ammonium hydroxide) presented a single envelope ranging from 1060 to 1884 m /z. In some cases, as with ammonium acetate and ammonium hydroxide, this envelope was shifted to higher charge states. On other hand, when acetic acid and, especially, ammonium hydrogencarbonate were used, the spectra showed a bigger number of peaks and it seems as if different envelopes are overlapping. The non-covalently bound heme group appearing at 616 m /z significantly increased relative to the envelope at basic pH values. In the case of cytochrome c, no signal was obtained with some of the mobile phases tried [ammonium acetate (pH 4.5), morpholine, and ammonium hydroxide]. As with
192
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
Fig. 2. Spectra obtained with different mobile phases for 2 mM solutions of myoglobin and cytochrome c. Experimental conditions as in Fig. 1.
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
myoglobin, these spectra showed a single envelope from 1125 to 1766 m /z with formic acid, TFA, and ammonium formate. However, when acetic acid and ammonium hydrogen carbonate were used, more than one envelope could be observed due to the fact the protein was probably present in different configurations. Moreover, the spectra with acetic acid and ammonium hydrogencarbonate showed the peak corresponding to the heme group while it was not observed with the other mobile phases. The spectra of myoglobin and cytochrome c showed no variations in shape when the concentration of additive in the mobile phase was changed and the only difference was the intensity of the signal observed.
3.3. Chromatographic separations A mixture containing three standard proteins (myoglobin, cytochrome c, and BSA) was separated on three different reversed-phase columns with mobile phases containing any of the four additives selected [0.2% (v / v) formic acid, 0.05 and 0.3% (v / v) acetic acid, 10 mM acetate buffer, and 50 mM hydrogencarbonate buffer] and with 0.025% (v / v) TFA. For this purpose, elution gradients in one step and ranging from 25 to 90% of the mobile phase B (ACN1certain concentration of additive) were optimized. Firstly, the separations were performed with a Luna C 18 column (4.6 mm I.D. and 5 mm particle size) at a flow-rate of 1 ml / min. Secondly, a Jupiter C 18 column (2 mm I.D. and 5 mm particle size) at a flow-rate of 0.2 ml / min was also tested. The problem with narrow columns is that flow-rates employed are low and, as a consequence, the total analysis times and, especially, the equilibration times, are relatively high. When gradient elution is used, LC– MS method development is focused on the reduction of total analysis time, e.g., cycle times of less than 10 min, for optimum use of this expensive MS instrument [20]. In order to obtain high sensitivities and as short analysis times as possible, a narrow perfusion column (2.0 mm I.D. and 10 mm particle size) at a flow-rate of 1 ml / min was also tested. Perfusion columns present a special internal structure enabling the use of higher flow-rates than those normally employed in conventional chromatography. Furthermore, these columns are made of a polymeric
193
material enabling the use of mobile phases with basic pH values, such as hydrogencarbonate buffer at pH 9.0 [21]. The chromatograms obtained with the three columns used are shown in Fig. 3. When ammonium formate and ammonium hydrogencarbonate were used, no peaks were observed in the chromatogram and higher column back-pressures were observed. On the other hand, the chromatograms obtained with 0.05% (v / v) acetic acid are not shown since they were quite similar to those observed with 0.3% (v / v) acetic acid. Regardless the ion-pairing agent used, the same order of elution was obtained on all three columns. Comparing the different additives, TFA enabled the best separation of the proteins with all three columns. Indeed, when formic or acetic acid were used, myoglobin and BSA eluted together. As regards the analysis times, the column allowing the shorter analysis time was the perfusion column with a total analysis time of around 10 min, while the Jupiter column yielded the highest total analysis time (about 1 h). Regarding efficiencies, it is important to take into account that efficiencies are not as good as they could be if using the widely employed 0.1% TFA instead of acetic or formic acid or a concentration four times lower of TFA (0.025%). Nevertheless, it may be interesting the exploration of other kind of stationary phases such as nonporous or monolithic columns in order to try to improve such efficiencies [7,10–12,14]. From these results, the perfusion column was chosen to carry out the separation of the proteins and the evaluation of the performance of the LC–MS system with the two additives enabling the detection and separation of the proteins (formic acid and acetic acid) as well as with 0.025% (v / v) TFA.
3.4. Performance of the LC–MS system The final purpose of this work was to couple the RPLC separation and the ESI-MS detection in order to determine the effect of the ion-pairing agents selected at the concentrations chosen [0.2% (v / v) formic acid, 0.05 and 0.3% (v / v) acetic acid, and 0.025% (v / v) TFA] on the performance of the LC– MS system. These experiments were carried out with the perfusion column and using myoglobin and cytochrome c. For that purpose, the calibration
194
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
Fig. 3. Chromatographic separations of three standard proteins [myoglobin (M), cytochrome c (C), and BSA (B)] obtained with three different reversed-phase columns and with three different acids. Experimental conditions: flow-rate, 1 ml / min for the Luna and the Poros columns and 0.2 ml / min for the Jupiter column; injected volume, 10 ml for the Luna and 5 ml for the Jupiter and Poros columns; mobile phases: A, water1acid, B, ACN1acid; detection, 214 nm; temperature, 25 8C. Concentration of sample injected: 0.5–0.7 mg / ml (Luna), 0.4–0.7 mg / ml (Jupiter), and 0.12–0.21 mg / ml (Poros R2 / H).
curves corresponding to each protein when using the selected ion-pairing agents were obtained. Fig. 4 shows the calibration curves in the range of 0 to 100 pmol / ml. Each calibration curve consisted of at least 15 points and every standard solution was injected in triplicate. In both cases (myoglobin and cytochrome c), the highest sensitivity was obtained with acetic acid and the lowest with TFA. In the case of acetic acid, the response obtained with 0.05% (v / v) was much higher that the observed at 0.3% (v / v). Although this could be expected from the FIA experiments of Fig. 1, the difference in the responses between these two concentrations was much higher that the observed in Fig. 1. Furthermore, the highest response should be observed with 0.2% (v / v) formic acid and not with acetic acid. All these observations
can be explained taking into account the recoveries of the proteins with the different mobile phases. Indeed, the interior structure of the pores in the stationary phase can trap proteins and unfolding of the adsorbed proteins on surfaces of strongly hydrophobic character could cause insufficient recovery when they are eluted with non-suitable eluents [19,22]. When comparing the signals obtained by UV detection for cytochrome c and myoglobin with the different additives selected and with 0.1% (v / v) TFA (Table 2), it was clearly observed that the recovery of proteins decreased with formic acid and acetic acid. In the case of acetic acid, especially at concentration 0.05% (v / v), the recovery was higher than with formic acid. Thus, it is possible to think that the increase in the response observed when using acetic
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
195
Fig. 4. Calibration curves obtained by LC–MS for myoglobin and cytochrome c with the perfusion column and by using 0.2% (v / v) formic acid (m), 0.05% (v / v) acetic acid (d), 0.3% (v / v) acetic acid (j), and 0.025% (v / v) TFA (♦) as additives. LC conditions as in Fig. 3 and MS conditions as in Fig. 1 with the exception of the capillary voltage that was 46 V. Responses (%) related to the highest signal obtained.
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
196
Table 2 Variation of the response (%) obtained in UV for myoglobin and cytochrome c with the POROS R2 / H perfusion column and with different mobile phases additives a,b Response (%)
0.1% (v / v) TFA 0.025% (v / v) TFA 0.2% (v / v) Formic acid 0.3% (v / v) Acetic acid 0.05% (v / v) Acetic acid
Cytochrome c
Myoglobin
96 97 55 79 100
100 86 42 51 57
a Experimental conditions as in Fig. 3. The concentration of the proteins was 5 mM. b Response related to the highest signal observed in each protein.
acid in ESI-MS instead of TFA compensates the decrease in the recovery observed when using acetic acid instead of TFA by RPLC. For formic acid, the increase in response does not compensate the decrease in recovery resulting in almost no increase in the signal by LC–MS. An explanation for the fact that with 0.05% (v / v) acetic acid the response observed was much higher than with 0.3% (v / v) acetic acid (especially for cytochrome c) could be because the recovery in LC was higher. Thus, the combination of this higher recovery (Table 2) and the higher response in MS (Fig. 1) resulted in a higher response in LC–MS with 0.05% (v / v) acetic acid. Nevertheless, this behaviour depends on the protein. While for cytochrome c with 0.05% (v / v) acetic acid the recovery was similar to TFA, for myoglobin the recovery was always lower than with TFA. However, in both cases, the use of 0.05% (v / v) acetic acid as additive in RPLC–ESI-MS significantly increased the response obtained. In order to test the applicability of the use of this additive with other proteins, the RPLC–ESI-MS was used to carry out the separation and detection of the proteins present in egg-white. The egg-white was prepared as described by Awade and Efstathiou [16]. The separation was performed with 0.05% (v / v) acetic acid and with 0.1% (v / v) TFA and the chromatograms were recorded by ESI-MS and by UV at 214 nm. These chromatograms are shown in Fig. 5. As is observed, the UV chromatograms obtained with TFA and with acetic acid presented a similar number of peaks, although their relative responses were differ-
ent. Moreover, the signals obtained with TFA were much higher than the obtained with acetic acid. On the other hand, the MS chromatograms obtained with TFA showed only two big peaks that according to their retention times could be peaks d9 and e9 in the corresponding UV chromatogram. These peaks presented very complex spectra that could correspond to large proteins. When acetic acid was used, the LC– MS chromatogram presented one additional peak in which peaks a, b, and c could overlap. From this peak was possible to extract a spectrum (see Fig. 5) corresponding to a protein whose molecular mass would be around 14 300 and that could be attributed to lysozyme. The LC–MS chromatogram obtained with TFA presented not only less peaks than those observed with acetic acid, but also the LC–MS responses obtained with TFA were about 10% of the response observed with acetic acid. These results would probe that in the case of the egg-white proteins, the use of 0.05% (v / v) acetic acid also improved the response in comparison with 0.1% (v / v) TFA.
3.5. Comparison with FIA In order to reveal the effect of the column on the LC–MS response, similar calibration curves were obtained by FIA (substituting the column by a connector). For that purpose, the mobile phases used consisting of water–ACN (50:50, v / v) and contained the same concentration of the acids previously employed to obtain the calibration curves with the column. As an example, Fig. 6 shows the calibration curves obtained for myoglobin. As was expected taking into account the results shown in Table 1, the highest responses were obtained with formic acid and the lowest with TFA. In the case of acetic acid, the signal observed when the concentration was 0.05% (v / v) was higher than when the concentration was 0.3% (v / v), although the difference in response was not as large as that observed with the column. These results were different from those obtained with the column and confirm that a 0.2% (v / v) of formic acid is not a suitable additive for LC–MS, although it is the most suitable with ESI-MS. Furthermore, the responses obtained with the column were higher than the responses observed by FIA, with the exception of those mobile phases in which the recovery was very
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa Fig. 5. Comparison of the chromatograms corresponding to the separation of proteins from a solution of 10% (v / v) of egg-white with two different additives [0.05% (v / v) acetic acid and 0.1% (v / v) TFA] and with UV and MS detection. Chromatographic conditions: flow-rate, 1 ml / min split to 0.1 ml / min which went to the MS detector and the rest to the UV detector; mobile phases: A, water1acid, B, ACN1acid; UV detection, 214 nm; temperature, 25 8C; injected volume, 5 ml. MS conditions as in Fig. 1. 197
198
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
Fig. 6. Calibration curves obtained by FIA for myoglobin with four different mobile phases. Experimental conditions as in Fig. 1. Responses (%) related to the highest signal obtained.
low. This suppression effect observed in ESI-MS could be attributed to a competition between ions or neutrals for the limited amount of charge and, even, physical space on the surface of a droplet [11,23,24]. Thus, using the column, possible interferences that can compete with the analyte of interest for the charge or space in the droplet are eliminated resulting in higher responses. Finally, Table 3 shows the detection limits obtained with and without column for myoglobin and cytochrome c with the four mobile phases tested. As is observed, the lowest detection limits were obtained when 0.05% (v / v) acetic acid was used and
were 50 fmol / ml (0.25 pmol) for cytochrome c and 100 fmol / ml (0.50 pmol) for myoglobin. If this detection limit is compared with the given by other authors, e.g., for cytochrome c, for a 1.0 mm I.D. column at 50 ml / min (4 pmol) [3] or even for a 320 mm I.D. perfusion column at 125 ml / min (1.6 pmol) [8], we can say that, at least in the cases studied, it is possible to obtain a detection limit quite low (16 times lower than the observed with a 1 mm I.D. column and more than six times lower than the observed with a 320 mm I.D. column) by working with 0.05% (v / v) acetic acid. Thus, if the use of more suitable mobile phases compromising the per-
Table 3 Detection limits (fmol / ml) obtained for myoglobin and cytochrome c with the column and without the column with four different mobile phases Myoglobin
0.2% (v / v) Formic acid 0.05% (v / v) Acetic acid 0.3% (v / v) Acetic acid 0.025% (v / v) TFA
Cytochrome c
Column
No column
Column
No column
500 50–100 500 250–500
250 500 500–1000 1100
250 50 1000 1000
100–250 200 500 1050
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
formance of the LC and the MS system is combined with the use of narrow columns and low flow-rates, the sensitivity of the LC–MS system could improve even more.
4. Conclusions Mobile phases normally used for the separation of proteins by RPLC are not compatible with ESI-MS detection and contribute to a decrease in signal. Alternative mobile phases have been tested for the sensitive detection of proteins. The mobile phases yielding the highest MS signals were formic acid (0.2%, v / v), acetic acid (0.3%, v / v), ammonium acetate (10 mM, pH 3.0), and ammonium hydrogencarbonate (50 mM, pH 9.0). However, only formic and acetic acid enabled the separation of proteins by RPLC. Calibration curves obtained with these two acids by FIA and LC revealed that despite formic acid yielded the highest signal and also enabled the separation of the proteins, recoveries were poor, resulting in higher responses when acetic acid was employed. Furthermore, the comparison of calibration curves obtained by FIA with those observed when the separation took place showed an increase in the sensitivity with the column.
Acknowledgements This research has been supported by a Marie Curie Fellowship of the European Union programme ‘‘Improving Human Research Potential and the Socio– Economic Knowledge Base’’ under contract No. HPMFCT-2000-00640.
References [1] D.N. Nguyen, G.W. Becker, R.M. Riggin, J. Chromatogr. A 705 (1995) 21.
199
[2] O.A. Mirgorodskaya, A.A. Schevchenko, I.V. Chernushevich, A.F. Dodonov, A.I. Miroshnikov, Anal. Chem. 66 (1994) 99. [3] J.F. Banks, J. Chromatogr. A 691 (1995) 325. [4] J.F. Banks, J. Chromatogr. A 743 (1996) 99. [5] J.F. Banks, J. Chromatogr. A 786 (1997) 67. [6] D.B. Kassel, B. Shushan, T. Sakuma, J.P. Salzmann, Anal. Chem. 66 (1994) 236. [7] J.E. MacNair, G.J. Opiteck, J.W. Jorgenson, M.A. Moseley, Rapid Commun. Mass Spectrom. 11 (1997) 1279. [8] J.F. Banks Jr., J.P. Quinn, C.M. Whitehouse, Anal. Chem. 66 (1994) 3688. ¨ [9] H. Schluter, in: M. Kastner (Ed.), Protein Liquid Chromatography, Journal of Chromatography Library, Vol. 61, Elsevier, Amsterdam, 2000, p. 147. [10] D. Corradini, C.G. Huber, A.M. Timperio, L. Zolla, J. Chromatogr. A 886 (2000) 111. [11] J.F. Banks, E.E. Gulcicek, Anal. Chem. 69 (1997) 3973. [12] D.B. Wall, M.T. Kachman, S.S. Gong, S.J. Parus, M.W. Long, D.M. Lubman, Rapid Commun. Mass Spectrom. 15 (2001) 1649. [13] G.J. Opiteck, K.C. Lewis, J.W. Jorgenson, Anal. Chem. 69 (1997) 1518. [14] A. Premstaleer, H. Oberacher, W. Walcher, A.M. Timperio, L. Zolla, J.P. Chervet, N. Cavusoglu, A. Dorsselaer, C.G. Huber, Anal. Chem. 73 (2001) 2390. [15] F. Klink, LC?GC Eur. June (2000) 396. [16] A.C. Awade, T. Efstathiou, J. Chromatogr. A 723 (1999) 69. [17] M.T.W. Hearn, in: M.T.W. Hearn (Ed.), Ion-Pair Chromatography—Theory and Biological and Pharmaceutical Applications, Chromatographic Science Series, Vol. 31, Marcel Dekker, New York, 1985, p. 207. [18] G. Teshina, E. Canova-Davis, J. Chromatogr. 625 (1992) 207. ´ [19] J. Frenz, W.S. Hancock, W.J. Henzel, C. Horvath, in: K.M. Gooding, F.E. Regnier (Eds.), HPLC of Biological Macromolecules—Methods and Applications, Chromatographic Science Series, Vol. 51, Marcel Dekker, New York, 1990, p. 145. [20] J.J. Merdink, J.W. Dolan, LC?GC 18 (2000) 1034. [21] M.C. Garcia, M.L. Marina, M. Torre, J. Chromatogr. A 880 (2000) 169. ˇ in: High-Performance Liquid Chromatography of [22] O. Mikes, Biopolymers and Biooligomers, Part B, Separation of Individual Compounds Classes, Journal of Chromatography Library, Vol. 41B, Elsevier, Amsterdam, 1988, p. B1. [23] D.L. Buhrman, P.I. Price, P.J. Rudewicz, J. Am. Soc. Mass Spectrom. 7 (1996) 1099. [24] M.G. Ikonomou, A.T. Blades, P. Kebarle, Anal. Chem. 62 (1990) 957.
Effect of the mobile phase composition on the separation and detection of intact proteins by reversed-phase liquid chromatography–electrospray mass spectrometry ´ a,b , *, A.C. Hogenboom a , H. Zappey a , H. Irth a M.C. Garcıa a
Department of Analytical Chemistry and Applied Spectroscopy, Vrije Universiteit, De Boelelaan 1083, 1081 HV Amsterdam, The Netherlands b ´ ´ , Facultad de Quımica ´ , Universidad de Alcala´ , 28871 Alcala´ de Henares, Madrid, Spain Departamento de Quımica Analıtica Received 11 October 2001; received in revised form 27 February 2002; accepted 7 March 2002
Abstract Various buffers (ammonium acetate, ammonium formate, and ammonium hydrogencarbonate), acids (formic acid, acetic acid, heptafluorobutyric acid, and trifluoroacetic acid), and bases (ammonium hydroxide and morpholine) covering the range from 2 to 11.5 have been investigated for their performance in the separation of proteins by reversed-phase liquid chromatography (RPLC) and in their detection by electrospray mass spectrometry (ESI-MS). These additives were first tested for the detection of standard proteins by ESI-MS by flow-injection analysis (FIA). Those additives yielding the highest signals were employed for the separation of standard proteins by using three different reversed-phase columns: two C 18 columns (4.6 mm I.D. and 2.1 mm I.D.) and one perfusion column (2 mm I.D.). The sensitivity of the LC–MS system was evaluated with the column giving the best results and with those LC eluents enabling the LC separation of the proteins and also yielding the highest MS signals. For that purpose, calibration curves were compared for both LC–MS and FIA–MS. Formic acid was the additive yielding the highest responses in FIA–MS and trifluoroacetic acid (TFA) gave the best separation and recovery of the proteins. However, problems related to poor recovery of the proteins in the column when formic acid was used and the significant signal suppression observed in MS when TFA was employed, made neither of them suitable for the sensitive detection of the proteins in LC–MS. 2002 Elsevier Science B.V. All rights reserved. Keywords: Mobile phase composition; Proteins
1. Introduction The biological interest in proteins and peptides has increased the demands in instrumental tools. Electrospray mass spectrometry (ESI-MS) enables the ad* Corresponding author. Departamento de Quımica ´ ´ Analıtica, ´ ´ 28871 Alcala´ de Facultad de Quımica, Universidad de Alcala, Henares, Madrid, Spain. Fax: 134-918-854-915. ´ E-mail address: [email protected] (M.C. Garcıa).
vanced detection of these biological molecules due to its ability to form multiply charged ions and thus, reducing the m /z values to levels that can be detected [1,2]. However, when coupling this powerful detection technique with a so widely employed technique for the separation of proteins such as reversed-phase liquid chromatography (RPLC), incompatibilities that can reduce the performance of the LC–MS system can appear. Indeed, the separation of proteins by RPLC requires mobile phases
0021-9673 / 02 / $ – see front matter 2002 Elsevier Science B.V. All rights reserved. PII: S0021-9673( 02 )00345-X
188
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
which usually are not suitable for ESI-MS detection. In addition, flow-rates normally used for LC are higher than those used for optimum ESI-MS detection. These two factors contribute to the decrease in sensitivity in ESI-MS detection. However, the increasing necessity of detecting very low concentration levels, especially in certain areas of the pharmaceutical industry, has contributed to overcome these limitations. The efforts carried out so far have led to the improvement of the sensitivity by the reduction of the flow-rate. In fact, ESI-MS is not a mass-sensitive detector but a concentration-sensitive one, which means that the signal intensities are not dependent on the flow-rate, at least in the range from 100 nl / min up to 1 ml / min. Consequently, higher sensitivities could be obtained by the employment of capillary columns enabling the use of lower flowrates [3–7]. Another trial to fit LC flow-rates to ESI-MS has been addressed by Banks Jr. et al. [8] who developed an ultrasonic nebulizer enabling the right operation of the ESI-MS system at flow-rates up to several hundred microliters per minute. Despite these efforts to increase the sensitivity of the LC–MS system by the adjustment of the flowrate, not too much has been done regarding the mobile phase incompatibilities between RPLC and ESI-MS. These incompatibilities are due to the fact that separation of proteins by RPLC requires the addition of an additive working as ion-pairing agent. These components are added to increase the hydrophobicity of the proteins by forming ionic pairs with the charged groups and, in addition, enable the defolding of the protein making possible that the buried hydrophobic groups of the proteins become accessible [9]. As a consequence, interaction of the proteins with the hydrophobic stationary phase is possible and, therefore, also their separation. Different chromatographic supports have been used for the separation of proteins with on-line ESI-MS detection. From conventional silica columns [4,5,8,9] to more sophisticated supports such as nonporous [7,10–12], perfusion [3,6,13] or monolithic [14] stationary phases have been used for this purpose. In most of these works, 0.1% trifluoroacetic acid (TFA) has been the additive used in the separation of proteins by RPLC. Unfortunately, although TFA is a suitable ion-pairing agent for the separation of proteins by RPLC, it is a signal supressor in ESI-MS
[15]. In fact, TFA forms very strong ion pairs with proteins (TFA–protein) that are not broken apart by the conditions in the electrospray ionization interface and, consequently, it prevents the ionization of the protein [15]. Therefore, other additives enabling the separation of the proteins by RPLC but at the same time the sensitive detection of proteins by ESI-MS should be explored. In this work, different LC eluents will be evaluated for both the separation by RPLC and the detection by ESI-MS of proteins trying to reach a compromise between the LC and the MS performances.
2. Experimental
2.1. Chemicals and reagents HPLC-grade acetonitrile (ACN) (J.T. Baker, Deventer, The Netherlands), and water (Milli-Q system, Millipore, Bedford, MA, USA) were used in the preparation of the mobile phases. HPLC-grade TFA, heptafluorobutyric acid, and morpholine were from Sigma (St. Louis, MO, USA) and formic acid, acetic acid, ammonium hydroxide, ammonium acetate, ammonium formate, and ammonium hydrogencar¨ (Seelze, bonate were purchased from Riedel-de Haen Germany). Protein samples [cytochrome c, myoglobin, and bovine serum albumin (BSA)] were from Sigma and eggs were purchased in a local market in Amsterdam. Egg-white preparation was performed by the method described by Awade and Efstathiou [16]. The sample was filtered through 0.2-mm PTFE filters prior to analysis. Nitrogen (99.999% purity) used as sheath (5 bar) and auxiliary gas (0.7 bar) was from Praxair (Oevel, Belgium).
2.2. High-performance liquid chromatography A Hewlett-Packard (HP) 1100 Series liquid chromatograph (Hewlett-Packard, Waldbronn, Germany) equipped with a quaternary delivery solvent system and a diode-array detector was used. This system was coupled to a computer with a HP
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
Chemstation data acquisition system. The system was used for both flow injection analysis (FIA) experiments and LC separations. For FIA experiments, the mobile phases consisted of water–ACN (50:50, v / v) and contained a certain concentration of an additive. These mobile phases were filtered using 0.45-mm filters of PTFE (for organic solvents) or a mixture of cellulose esters (for aqueous solutions) (Millipore) and degassed with helium before use. All FIA experiments were performed at 25 8C. For LC experiments, three different reversedphase columns were used: a Luna C 18 column (2503 4.6 mm I.D.) and a Jupiter C 18 column (15032.0 mm I.D.) packed with 5 mm particles (Phenomenex, Torrance, CA, USA) and a Poros R2 / H perfusion column (10032.1 mm I.D.) packed with 10 mm particles (Perseptive Biosystems, Framingham, MA, USA). The separations were carried out with singlestep gradients at flow-rates of 1 ml / min (for the Luna and the Poros columns) or at 0.2 ml / min (for the Jupiter column) followed by reversed gradients to equilibrate the columns between runs to the initial conditions. The injected volume was 5 ml and mobile phases consisted of water with a certain concentration of an additive (mobile phase A) and ACN with the same concentration of the same additive (mobile phase B). The UV detection was performed at 214 nm and the columns were thermostated at 25 8C.
2.3. Mass spectrometry Mass spectrometry detection was performed on a Finnigan LCQ ion trap mass spectrometer equipped with an electrospray ionization interface (San Jose, CA, USA). Xcalibur software was used for data acquisition and data analysis. The instrument was firstly tuned to get the MS conditions enabling the highest responses. These MS conditions were used in all the experiments and were as follows: capillary voltage, 26 or 46 V; spray voltage, 5.0 kV; capillary temperature, 350 8C. Mass spectra were registered in the full-scan mode (m /z, 600–2000; scan time, 2 scans / s, maximum injection time, 100 ms). The MS system was connected to the LC system by a postcolumn splitter to maintain the flow-rate to 100
189
ml / min. All determinations were performed, at least, three times.
3. Results and discussion One of the problems in the hyphenation of RPLC and ESI-MS is the non-compatibility of some of the additives used in the mobile phases for the separation of proteins with the ESI-MS system. For example, as was already discussed in the Introduction, a decrease in response was observed in ESI-MS when TFA was used. In addition, some salts such as phosphate, perchlorate, alkylsulfates or alkylsulfonates, that are used for the separation of proteins by RPLC, are also not suitable for the ESI-MS as they are not volatile [17]. In this work, the additives listed in Table 1 were tested in order to obtain a compromise between the optimum separation and recovery of proteins by RPLC and the optimum detection by ESI-MS. In this list TFA was also included in order to compare the performance of this excellent ion-pairing agent for the separation of proteins by RPLC with the performance with the other additives. Although less volatile, heptafluorobutyric acid is another perfluorinated carboxylic acid that has also been used for the separation of proteins by RPLC. Formic acid can also act as ion-pairing agent for the separation of proteins by RPLC, although the resolution and recoveries obtained are lower than with TFA. Furthermore, it is less acidic and less volatile than TFA. Acetic acid is also suitable for the ESI-MS detection and, in addition, it is safer to handle and cheaper than TFA. Ammonium acetate and ammonium formate have also been employed in some occasions to control the pH of the LC mobile phases in the separation of some peptides and proteins such as hormones [18]. Hydrogencarbonate buffer, morpholine, and ammonium hydroxide can be used as cation-pairing agents for the separation of proteins when it is possible to work at these basic pH values. In fact, protein separations by RPLC are mostly carried out at acidic pH values at which protein exist in the fully protonated form and behave mainly as a cation solute. However, due to the aliphatic character of proteins, they can also be separated by RPLC at neutral or higher pH values [17,19]. Nevertheless,
190
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
Table 1 Relative response (%) of three standard proteins with different mobile phase additives in ESI-MS a,b Additive
Heptafluorobutyric acid Formic acid Trifluoroacetic acid Acetic acid Ammonium formate Ammonium acetate Ammonium formate Ammonium acetate Ammonium hydrogencarbonate Morpholine Ammonium hydroxide
Concentration
0.2% (v / v) 0.2% (v / v) 0.025% (v / v) 0.3% (v / v) 10 mM 10 mM 10 mM 10 mM 50 mM 10 mM 20 mM
pH
Relative response (%)
1.8 2.5 2.5 3.0 3.0 4.5 6.0 6.0 9.0 10.5 11.5
Myoglobin
Cytochrome c
BSA
,1 100 3 46 25 1 ,1 ,1 34 12 24
,1 100 3 56 37 2 ,1 ,1 11 ,1 ,1
3 98 9 100 47 2 ,1 ,1 33 ,1 8
a FIA at 0.1 ml / min; mobile phase, water–ACN (50:50); MS conditions: capillary voltage, 46 V; spray voltage, 5.0 kV; capillary temperature, 350 8C. b Response related to the highest signal observed in every protein.
these conditions are not compatible with conventional silica-based LC columns.
3.1. Effect of different mobile phases on the response in ESI-MS Before studying the performance of the additives listed in Table 1 in LC–MS, their compatibility with the ESI-MS was studied in terms of response with three standard proteins: myoglobin, cytochrome c, and BSA. For that purpose, FIA experiments with mobile phases containing a certain concentration of these additives were performed. The concentrations of these additives in the LC mobile phases are also shown in Table 1 and corresponded to those concentrations normally employed when these mobile phases are used for the separation of proteins by RPLC with the exception of TFA, in which case a lower concentration (0.025%, v / v) was employed [9]. These mobile phases were pumped into the chromatographic system at a flow-rate of 1 ml / min and split to 0.1 ml / min before reaching the ESI-MS system. In these FIA experiments, 2 mM solution of each protein was injected in triplicate. Table 1 groups the average relative responses obtained in these experiments. As can be observed, the variation of the responses with the different additives tested was similar regardless of the protein used. In all cases the highest signals were obtained with formic acid and acetic acid and the lowest responses with
ammonium acetate or ammonium formate at pH 6.0. Furthermore, an increase in the signal was observed for all three proteins when ammonium formate was used at more acidic pH. As regards basic pH values, the additive yielding the highest responses in all cases was ammonium hydrogencarbonate at pH 9.0. On the other hand, when using heptafluorobutyric acid, TFA, ammonium acetate at pH 4.5 or morpholine, the responses were very small (#12%) and in some occasions no signal was obtained. From the results obtained in this study, a group of additives yielding the highest signals for the three proteins [formic acid, acetic acid, ammonium formate (pH 3.0), and ammonium hydrogencarbonate (pH 9.0)] was selected. Moreover, a study on the effect of the variation of the concentration of such compounds on the signal observed by ESI-MS was carried out. In this study, myoglobin and cytochrome c were used and the results obtained are shown in Fig. 1. As can be observed, both proteins gave a similar variation of the relative responses with the concentration of every additive. The highest responses were always found with formic acid in which case the signal increased up to a concentration around 50 mM. Higher concentrations of formic acid resulted in a reduction of the signals of both proteins. As a consequence, formic acid at a concentration of 50 mM (0.2%, v / v) was selected for the separation of the proteins by LC. As regards acetic acid, the responses obtained for
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
191
Fig. 1. Variation of the signal (expressed as response related to the highest signal observed) obtained in ESI-MS for 2 mM solutions of myoglobin and cytochrome c with the concentration of acetic acid, formic acid, TFA, ammonium formate (pH 3) and ammonium hydrogencarbonate (pH 9). FIA at 0.1% ml / min. Mobile phase, water–ACN (50:50). MS conditions: capillary voltage, 46 V; spray voltage, 5.0 kV; capillary temperature, 350 8C.
both proteins were lower than those with formic acid and the highest signals were observed at a concentration of 8 mM. Higher concentrations of acetic acid resulted in a decrease in the response. In this case, two concentrations were chosen, 8 mM (0.05%, v / v) because it was the concentration at which the signal was maximum and 50 mM (0.3%, v / v) since it is the concentration of acetic acid normally employed for the separation of proteins. No significant change in response was observed for myoglobin and cytochrome c when ammonium formate at pH 3.0 was used at concentrations in the range from 5 to 50 mM. In addition, these responses were lower than those observed with formic acid and acetic acid. In this case, a concentration of 10 mM of this buffer was chosen since this concentration has already been used for the separation of proteins [9]. The responses obtained with ammonium hydrogencarbonate at pH 9.0 were even lower than those observed previously up to a concentration of 30 mM. At higher concentrations, the responses significantly increased. In both cases (ammonium formate and hydrogencarbonate), higher concentrations than 70 mM were also studied although problems related to the back-pressure of the system were observed. As was expected, the signal yielded with TFA decreased with increasing concentration with almost complete disappearance at 40 mM (0.4%, v / v). From these results, 0.2%
(v / v) formic acid, 0.05 and 0.3% (v / v) acetic acid, 10 mM acetate buffer, and 50 mM hydrogencarbonate buffer were selected for the MS detection of proteins and were tested for its performance in the separation of proteins by LC.
3.2. Effect of the different mobile phases on the ESI-MS spectra of the proteins Fig. 2 shows the spectra obtained for myoglobin and cytochrome c with the mobile phases previously used. In the case of myoglobin, the spectra obtained with most of the mobile phases tested (formic acid, TFA, ammonium formate, ammonium acetate, morpholine and ammonium hydroxide) presented a single envelope ranging from 1060 to 1884 m /z. In some cases, as with ammonium acetate and ammonium hydroxide, this envelope was shifted to higher charge states. On other hand, when acetic acid and, especially, ammonium hydrogencarbonate were used, the spectra showed a bigger number of peaks and it seems as if different envelopes are overlapping. The non-covalently bound heme group appearing at 616 m /z significantly increased relative to the envelope at basic pH values. In the case of cytochrome c, no signal was obtained with some of the mobile phases tried [ammonium acetate (pH 4.5), morpholine, and ammonium hydroxide]. As with
192
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
Fig. 2. Spectra obtained with different mobile phases for 2 mM solutions of myoglobin and cytochrome c. Experimental conditions as in Fig. 1.
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
myoglobin, these spectra showed a single envelope from 1125 to 1766 m /z with formic acid, TFA, and ammonium formate. However, when acetic acid and ammonium hydrogen carbonate were used, more than one envelope could be observed due to the fact the protein was probably present in different configurations. Moreover, the spectra with acetic acid and ammonium hydrogencarbonate showed the peak corresponding to the heme group while it was not observed with the other mobile phases. The spectra of myoglobin and cytochrome c showed no variations in shape when the concentration of additive in the mobile phase was changed and the only difference was the intensity of the signal observed.
3.3. Chromatographic separations A mixture containing three standard proteins (myoglobin, cytochrome c, and BSA) was separated on three different reversed-phase columns with mobile phases containing any of the four additives selected [0.2% (v / v) formic acid, 0.05 and 0.3% (v / v) acetic acid, 10 mM acetate buffer, and 50 mM hydrogencarbonate buffer] and with 0.025% (v / v) TFA. For this purpose, elution gradients in one step and ranging from 25 to 90% of the mobile phase B (ACN1certain concentration of additive) were optimized. Firstly, the separations were performed with a Luna C 18 column (4.6 mm I.D. and 5 mm particle size) at a flow-rate of 1 ml / min. Secondly, a Jupiter C 18 column (2 mm I.D. and 5 mm particle size) at a flow-rate of 0.2 ml / min was also tested. The problem with narrow columns is that flow-rates employed are low and, as a consequence, the total analysis times and, especially, the equilibration times, are relatively high. When gradient elution is used, LC– MS method development is focused on the reduction of total analysis time, e.g., cycle times of less than 10 min, for optimum use of this expensive MS instrument [20]. In order to obtain high sensitivities and as short analysis times as possible, a narrow perfusion column (2.0 mm I.D. and 10 mm particle size) at a flow-rate of 1 ml / min was also tested. Perfusion columns present a special internal structure enabling the use of higher flow-rates than those normally employed in conventional chromatography. Furthermore, these columns are made of a polymeric
193
material enabling the use of mobile phases with basic pH values, such as hydrogencarbonate buffer at pH 9.0 [21]. The chromatograms obtained with the three columns used are shown in Fig. 3. When ammonium formate and ammonium hydrogencarbonate were used, no peaks were observed in the chromatogram and higher column back-pressures were observed. On the other hand, the chromatograms obtained with 0.05% (v / v) acetic acid are not shown since they were quite similar to those observed with 0.3% (v / v) acetic acid. Regardless the ion-pairing agent used, the same order of elution was obtained on all three columns. Comparing the different additives, TFA enabled the best separation of the proteins with all three columns. Indeed, when formic or acetic acid were used, myoglobin and BSA eluted together. As regards the analysis times, the column allowing the shorter analysis time was the perfusion column with a total analysis time of around 10 min, while the Jupiter column yielded the highest total analysis time (about 1 h). Regarding efficiencies, it is important to take into account that efficiencies are not as good as they could be if using the widely employed 0.1% TFA instead of acetic or formic acid or a concentration four times lower of TFA (0.025%). Nevertheless, it may be interesting the exploration of other kind of stationary phases such as nonporous or monolithic columns in order to try to improve such efficiencies [7,10–12,14]. From these results, the perfusion column was chosen to carry out the separation of the proteins and the evaluation of the performance of the LC–MS system with the two additives enabling the detection and separation of the proteins (formic acid and acetic acid) as well as with 0.025% (v / v) TFA.
3.4. Performance of the LC–MS system The final purpose of this work was to couple the RPLC separation and the ESI-MS detection in order to determine the effect of the ion-pairing agents selected at the concentrations chosen [0.2% (v / v) formic acid, 0.05 and 0.3% (v / v) acetic acid, and 0.025% (v / v) TFA] on the performance of the LC– MS system. These experiments were carried out with the perfusion column and using myoglobin and cytochrome c. For that purpose, the calibration
194
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
Fig. 3. Chromatographic separations of three standard proteins [myoglobin (M), cytochrome c (C), and BSA (B)] obtained with three different reversed-phase columns and with three different acids. Experimental conditions: flow-rate, 1 ml / min for the Luna and the Poros columns and 0.2 ml / min for the Jupiter column; injected volume, 10 ml for the Luna and 5 ml for the Jupiter and Poros columns; mobile phases: A, water1acid, B, ACN1acid; detection, 214 nm; temperature, 25 8C. Concentration of sample injected: 0.5–0.7 mg / ml (Luna), 0.4–0.7 mg / ml (Jupiter), and 0.12–0.21 mg / ml (Poros R2 / H).
curves corresponding to each protein when using the selected ion-pairing agents were obtained. Fig. 4 shows the calibration curves in the range of 0 to 100 pmol / ml. Each calibration curve consisted of at least 15 points and every standard solution was injected in triplicate. In both cases (myoglobin and cytochrome c), the highest sensitivity was obtained with acetic acid and the lowest with TFA. In the case of acetic acid, the response obtained with 0.05% (v / v) was much higher that the observed at 0.3% (v / v). Although this could be expected from the FIA experiments of Fig. 1, the difference in the responses between these two concentrations was much higher that the observed in Fig. 1. Furthermore, the highest response should be observed with 0.2% (v / v) formic acid and not with acetic acid. All these observations
can be explained taking into account the recoveries of the proteins with the different mobile phases. Indeed, the interior structure of the pores in the stationary phase can trap proteins and unfolding of the adsorbed proteins on surfaces of strongly hydrophobic character could cause insufficient recovery when they are eluted with non-suitable eluents [19,22]. When comparing the signals obtained by UV detection for cytochrome c and myoglobin with the different additives selected and with 0.1% (v / v) TFA (Table 2), it was clearly observed that the recovery of proteins decreased with formic acid and acetic acid. In the case of acetic acid, especially at concentration 0.05% (v / v), the recovery was higher than with formic acid. Thus, it is possible to think that the increase in the response observed when using acetic
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
195
Fig. 4. Calibration curves obtained by LC–MS for myoglobin and cytochrome c with the perfusion column and by using 0.2% (v / v) formic acid (m), 0.05% (v / v) acetic acid (d), 0.3% (v / v) acetic acid (j), and 0.025% (v / v) TFA (♦) as additives. LC conditions as in Fig. 3 and MS conditions as in Fig. 1 with the exception of the capillary voltage that was 46 V. Responses (%) related to the highest signal obtained.
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
196
Table 2 Variation of the response (%) obtained in UV for myoglobin and cytochrome c with the POROS R2 / H perfusion column and with different mobile phases additives a,b Response (%)
0.1% (v / v) TFA 0.025% (v / v) TFA 0.2% (v / v) Formic acid 0.3% (v / v) Acetic acid 0.05% (v / v) Acetic acid
Cytochrome c
Myoglobin
96 97 55 79 100
100 86 42 51 57
a Experimental conditions as in Fig. 3. The concentration of the proteins was 5 mM. b Response related to the highest signal observed in each protein.
acid in ESI-MS instead of TFA compensates the decrease in the recovery observed when using acetic acid instead of TFA by RPLC. For formic acid, the increase in response does not compensate the decrease in recovery resulting in almost no increase in the signal by LC–MS. An explanation for the fact that with 0.05% (v / v) acetic acid the response observed was much higher than with 0.3% (v / v) acetic acid (especially for cytochrome c) could be because the recovery in LC was higher. Thus, the combination of this higher recovery (Table 2) and the higher response in MS (Fig. 1) resulted in a higher response in LC–MS with 0.05% (v / v) acetic acid. Nevertheless, this behaviour depends on the protein. While for cytochrome c with 0.05% (v / v) acetic acid the recovery was similar to TFA, for myoglobin the recovery was always lower than with TFA. However, in both cases, the use of 0.05% (v / v) acetic acid as additive in RPLC–ESI-MS significantly increased the response obtained. In order to test the applicability of the use of this additive with other proteins, the RPLC–ESI-MS was used to carry out the separation and detection of the proteins present in egg-white. The egg-white was prepared as described by Awade and Efstathiou [16]. The separation was performed with 0.05% (v / v) acetic acid and with 0.1% (v / v) TFA and the chromatograms were recorded by ESI-MS and by UV at 214 nm. These chromatograms are shown in Fig. 5. As is observed, the UV chromatograms obtained with TFA and with acetic acid presented a similar number of peaks, although their relative responses were differ-
ent. Moreover, the signals obtained with TFA were much higher than the obtained with acetic acid. On the other hand, the MS chromatograms obtained with TFA showed only two big peaks that according to their retention times could be peaks d9 and e9 in the corresponding UV chromatogram. These peaks presented very complex spectra that could correspond to large proteins. When acetic acid was used, the LC– MS chromatogram presented one additional peak in which peaks a, b, and c could overlap. From this peak was possible to extract a spectrum (see Fig. 5) corresponding to a protein whose molecular mass would be around 14 300 and that could be attributed to lysozyme. The LC–MS chromatogram obtained with TFA presented not only less peaks than those observed with acetic acid, but also the LC–MS responses obtained with TFA were about 10% of the response observed with acetic acid. These results would probe that in the case of the egg-white proteins, the use of 0.05% (v / v) acetic acid also improved the response in comparison with 0.1% (v / v) TFA.
3.5. Comparison with FIA In order to reveal the effect of the column on the LC–MS response, similar calibration curves were obtained by FIA (substituting the column by a connector). For that purpose, the mobile phases used consisting of water–ACN (50:50, v / v) and contained the same concentration of the acids previously employed to obtain the calibration curves with the column. As an example, Fig. 6 shows the calibration curves obtained for myoglobin. As was expected taking into account the results shown in Table 1, the highest responses were obtained with formic acid and the lowest with TFA. In the case of acetic acid, the signal observed when the concentration was 0.05% (v / v) was higher than when the concentration was 0.3% (v / v), although the difference in response was not as large as that observed with the column. These results were different from those obtained with the column and confirm that a 0.2% (v / v) of formic acid is not a suitable additive for LC–MS, although it is the most suitable with ESI-MS. Furthermore, the responses obtained with the column were higher than the responses observed by FIA, with the exception of those mobile phases in which the recovery was very
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa Fig. 5. Comparison of the chromatograms corresponding to the separation of proteins from a solution of 10% (v / v) of egg-white with two different additives [0.05% (v / v) acetic acid and 0.1% (v / v) TFA] and with UV and MS detection. Chromatographic conditions: flow-rate, 1 ml / min split to 0.1 ml / min which went to the MS detector and the rest to the UV detector; mobile phases: A, water1acid, B, ACN1acid; UV detection, 214 nm; temperature, 25 8C; injected volume, 5 ml. MS conditions as in Fig. 1. 197
198
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
Fig. 6. Calibration curves obtained by FIA for myoglobin with four different mobile phases. Experimental conditions as in Fig. 1. Responses (%) related to the highest signal obtained.
low. This suppression effect observed in ESI-MS could be attributed to a competition between ions or neutrals for the limited amount of charge and, even, physical space on the surface of a droplet [11,23,24]. Thus, using the column, possible interferences that can compete with the analyte of interest for the charge or space in the droplet are eliminated resulting in higher responses. Finally, Table 3 shows the detection limits obtained with and without column for myoglobin and cytochrome c with the four mobile phases tested. As is observed, the lowest detection limits were obtained when 0.05% (v / v) acetic acid was used and
were 50 fmol / ml (0.25 pmol) for cytochrome c and 100 fmol / ml (0.50 pmol) for myoglobin. If this detection limit is compared with the given by other authors, e.g., for cytochrome c, for a 1.0 mm I.D. column at 50 ml / min (4 pmol) [3] or even for a 320 mm I.D. perfusion column at 125 ml / min (1.6 pmol) [8], we can say that, at least in the cases studied, it is possible to obtain a detection limit quite low (16 times lower than the observed with a 1 mm I.D. column and more than six times lower than the observed with a 320 mm I.D. column) by working with 0.05% (v / v) acetic acid. Thus, if the use of more suitable mobile phases compromising the per-
Table 3 Detection limits (fmol / ml) obtained for myoglobin and cytochrome c with the column and without the column with four different mobile phases Myoglobin
0.2% (v / v) Formic acid 0.05% (v / v) Acetic acid 0.3% (v / v) Acetic acid 0.025% (v / v) TFA
Cytochrome c
Column
No column
Column
No column
500 50–100 500 250–500
250 500 500–1000 1100
250 50 1000 1000
100–250 200 500 1050
´ et al. / J. Chromatogr. A 957 (2002) 187–199 M.C. Garcıa
formance of the LC and the MS system is combined with the use of narrow columns and low flow-rates, the sensitivity of the LC–MS system could improve even more.
4. Conclusions Mobile phases normally used for the separation of proteins by RPLC are not compatible with ESI-MS detection and contribute to a decrease in signal. Alternative mobile phases have been tested for the sensitive detection of proteins. The mobile phases yielding the highest MS signals were formic acid (0.2%, v / v), acetic acid (0.3%, v / v), ammonium acetate (10 mM, pH 3.0), and ammonium hydrogencarbonate (50 mM, pH 9.0). However, only formic and acetic acid enabled the separation of proteins by RPLC. Calibration curves obtained with these two acids by FIA and LC revealed that despite formic acid yielded the highest signal and also enabled the separation of the proteins, recoveries were poor, resulting in higher responses when acetic acid was employed. Furthermore, the comparison of calibration curves obtained by FIA with those observed when the separation took place showed an increase in the sensitivity with the column.
Acknowledgements This research has been supported by a Marie Curie Fellowship of the European Union programme ‘‘Improving Human Research Potential and the Socio– Economic Knowledge Base’’ under contract No. HPMFCT-2000-00640.
References [1] D.N. Nguyen, G.W. Becker, R.M. Riggin, J. Chromatogr. A 705 (1995) 21.
199
[2] O.A. Mirgorodskaya, A.A. Schevchenko, I.V. Chernushevich, A.F. Dodonov, A.I. Miroshnikov, Anal. Chem. 66 (1994) 99. [3] J.F. Banks, J. Chromatogr. A 691 (1995) 325. [4] J.F. Banks, J. Chromatogr. A 743 (1996) 99. [5] J.F. Banks, J. Chromatogr. A 786 (1997) 67. [6] D.B. Kassel, B. Shushan, T. Sakuma, J.P. Salzmann, Anal. Chem. 66 (1994) 236. [7] J.E. MacNair, G.J. Opiteck, J.W. Jorgenson, M.A. Moseley, Rapid Commun. Mass Spectrom. 11 (1997) 1279. [8] J.F. Banks Jr., J.P. Quinn, C.M. Whitehouse, Anal. Chem. 66 (1994) 3688. ¨ [9] H. Schluter, in: M. Kastner (Ed.), Protein Liquid Chromatography, Journal of Chromatography Library, Vol. 61, Elsevier, Amsterdam, 2000, p. 147. [10] D. Corradini, C.G. Huber, A.M. Timperio, L. Zolla, J. Chromatogr. A 886 (2000) 111. [11] J.F. Banks, E.E. Gulcicek, Anal. Chem. 69 (1997) 3973. [12] D.B. Wall, M.T. Kachman, S.S. Gong, S.J. Parus, M.W. Long, D.M. Lubman, Rapid Commun. Mass Spectrom. 15 (2001) 1649. [13] G.J. Opiteck, K.C. Lewis, J.W. Jorgenson, Anal. Chem. 69 (1997) 1518. [14] A. Premstaleer, H. Oberacher, W. Walcher, A.M. Timperio, L. Zolla, J.P. Chervet, N. Cavusoglu, A. Dorsselaer, C.G. Huber, Anal. Chem. 73 (2001) 2390. [15] F. Klink, LC?GC Eur. June (2000) 396. [16] A.C. Awade, T. Efstathiou, J. Chromatogr. A 723 (1999) 69. [17] M.T.W. Hearn, in: M.T.W. Hearn (Ed.), Ion-Pair Chromatography—Theory and Biological and Pharmaceutical Applications, Chromatographic Science Series, Vol. 31, Marcel Dekker, New York, 1985, p. 207. [18] G. Teshina, E. Canova-Davis, J. Chromatogr. 625 (1992) 207. ´ [19] J. Frenz, W.S. Hancock, W.J. Henzel, C. Horvath, in: K.M. Gooding, F.E. Regnier (Eds.), HPLC of Biological Macromolecules—Methods and Applications, Chromatographic Science Series, Vol. 51, Marcel Dekker, New York, 1990, p. 145. [20] J.J. Merdink, J.W. Dolan, LC?GC 18 (2000) 1034. [21] M.C. Garcia, M.L. Marina, M. Torre, J. Chromatogr. A 880 (2000) 169. ˇ in: High-Performance Liquid Chromatography of [22] O. Mikes, Biopolymers and Biooligomers, Part B, Separation of Individual Compounds Classes, Journal of Chromatography Library, Vol. 41B, Elsevier, Amsterdam, 1988, p. B1. [23] D.L. Buhrman, P.I. Price, P.J. Rudewicz, J. Am. Soc. Mass Spectrom. 7 (1996) 1099. [24] M.G. Ikonomou, A.T. Blades, P. Kebarle, Anal. Chem. 62 (1990) 957.