A Comparison of CEMS and LCMS for Peptide Samples

The authors investigated and compared aspects of capillary electrophoresis and liquid chromatography coupled with electrospray-ionization mass spectrometry for the analysis of peptide samples. Despite their different levels of technological maturity, both techniques can be operated at the same level of automation. However, they differ greatly in the interface design, selectivity, sensitivity, and method development aspects.

iquid chromatographymass spectrometry (LCMS) is progressing toward maturity and is widely perceived as a routine analytical technique. For many years researchers have developed and successfully applied the methods and instrumentation for analyzing a wide range of analytes (1,2). Capillary electrophoresis (CE) with tandem UV and MS detection, although still in its infancy, is being used increasingly for the analysis of complex mixtures (3,4). Its use offers different selectivity, higher efficiency, and shorter analysis times than LC.

In LCMS, the outlet of the UV detector is connected to a single-tube spraying device within the ion source (Figure 1). The LC pump maintains the ow, and the nebulizing and drying gas together with the electrospray high voltage assist in the generation of gas-phase ions. The nebulizing and drying gas and electrospray high voltage also are present in CEMS; however, the interface also requires an additional electrical connection to the run buffer to maintain the electrical eld across the separation capillary. To accommodate these requirements and the smaller ow rate in CE, researchers have

(a) Nebulizing gas Sheath liquid Capillary

(c) Nebulizing gas

CE inlet Mass spectrometer

Sprayer

Maria Serwe and Gordon A. Ross

(b) Nebulizing gas From column

Figure 1: Schematics of orthogonal ow sprayers for (a) CEelectrospray-ionization MS and (b) LCelectrospray-ionization MS and (c) a cross-sectional diagram of the mass-selective detector.

developed various designs, including the sheath-liquid interface (5), the liquidjunction interface described by Lee and co-workers (6), and sheathless devices (7). The most commonly used and perhaps most rugged interface for CEMS is the sheathliquid interface, which was rst described by Smith and co-workers (5). In this triple-tube design, the CE separation capillary itself is the center tube of the sprayer, and it is surrounded by two stainless steel tubes. The inner steel tube delivers a sheath liquid or makeup ow, and the outer tube delivers the nebulizing gas (Figure 1). The sheath liquid mainly provides an electrical connection to ground for the CE outlet; however, it also generates the necessary ow for stable electrospray. In our study we investigated the analysis of peptides by capillary zone electrophoresis (CZE) and capillary reversed-phase LC coupled with electrospray-ionization MS. We chose the capillary format of LCMS because the ow rate delivered into the electrospray source has the same order of magnitude as the flow rate generated by CEelectrospray-ionization MS with the addition of a sheath liquid. Furthermore, capillary reversed-phase LC is the most sensitive LC technique and therefore provides a more valid comparison.
Experimental Instrumentation for LCMS: For capillary

reversed-phase LCMS, we used a model HP 1100 series LCMSD system (Agilent Technologies Deutschland GmbH, Waldbronn, Germany). The system comprised a high-pressure binary gradient pump with a vacuum degasser, an autosampler, a diodearray detector, and a mass-selective detector. The tubing had inner diameters of 50120 m, and tubing lengths were minimized. The pump ow rate of 100 L/min was reduced to 5 L/min using an Acurate microow processor (LC Packings, Amsterdam, The Netherlands) and a flow calibrator matched to the analytical column. The ana-

lytical system was controlled using a PC and Agilent ChemStation software (Agilent Technologies). We used a 25 cm 300 m, 5-m dp Vydac model PIC-25-05-C18P3 C18 analytical column and a 1 cm 300 m, 5-m dp Vydac model MGU-30-C18P3 C18 guard column (both from LC Packings). Parameters for LCMS: Solvent A was 0.025% trifluoroacetic acid in water, and solvent B was 0.02% triuoroacetic acid in acetonitrile. The injection program was as follows: draw, inject, wait 5 min, valve bypass, wait 65 min, and valve main pass. The injection program matches the gradient described below and reduces dead volume. The ow rate was 100 L/min split to 5 L/min. The gradient (unless otherwise stated) was as follows: 131% B in 30 min, 3175% B in 5 min, 75% B for 2 min, 751% B in 5 min, and 1% B for 38 min (total gradient time: 80 min). The temperature was ambient. We performed UV detection using a 500nL detection cell with a 10-mm pathlength. The detection signal was 206 nm (10-nm bandwidth), and the reference wavelength was 450 nm (80-nm bandwidth). MS acquisition was performed in the positive-ion mode at 4000 V. The nebulizing gas was nitrogen at 20 psi. The drying gas was nitrogen with a ow rate of 6 L/min at 150 C. The fragmentor voltage was 96 V. (For more details, please refer to gure captions.) Instrumentation for CEMS: All experiments used an Agilent CE electrophoresis system equipped with an MS adapter kit (both from Agilent Technologies), which enabled connection to an external detector. The sheath liquid was delivered by an Agilent 1100 binary pump (Agilent Technologies) equipped with a 1:100 flow splitter. The splitter and the orthogonal ow sprayer are part of a CEelectrospray-ionization MS sprayer kit, which is designed to interface a CE system with the mass-selective detector. All instrumentation was controlled from a single PC using Agilent ChemStation soft-

ware (8).
Parameters for CEMS: The buffer was 10 mM acetic acid at pH 3.5 (unless otherwise stated). We used a 75 cm 50 m bare fused-silica capillary with a 22-cm effective length. We made injections at 50 mbar for varying times. The voltage was 27 kV, applied with 0.3-min voltage ramp. The operating temperature was 25 C. We performed UV detection using a signal wavelength of 206 nm (10-nm bandwidth) and a reference wavelength of 450 nm (80-nm bandwidth). MS acquisition was performed in the positive-ion mode at 4000 V. The sheath liquid was 0.5% acetic acid in 50% methanol (unless otherwise stated) at a 4-L/min ow rate. The nebulizing gas was nitrogen at 10 psi. The drying gas was nitrogen with a 10-L/min ow rate at 150 C. The fragmentor voltage was 96 V. (For more details, please refer to gure captions.) Chemicals: All chemicals were high performance liquid chromatographygrade or better. We purchased the peptides from Sigma (Deisenhofen, Germany).

Results and Discussion The sheath liquid: The need to add a sheath

Table I: Dilution Ratio of the CE Flow by the Sheath Liquid for Two Extreme pH Values* Parameter CE ow rate (L/min) Sheath ow (L/min) Dilution ratio (LCCE) at pH 2.5 0.042 4 94 at pH 9.0 0.212 4 19

* The electroosmotic ow velocities are given in microliters per minute and were calculated for an applied voltage of 27 kV to a 75 cm 50 m capillary for pH 2.5 and 9.0 using electroosmotic ow mobility values of 1.0 104 cm2/ Vs and 5.0 104 cm2/ Vs, respectively. Information from reference 10.

liquid in CEelectrospray-ionization MS has been considered a disadvantage because it dilutes the ow coming from the CE capillary. The ow rate in CE is dictated largely by the electroosmotic ow, which in turn depends on buffer properties such as pH and concentration as well as temperature and applied voltage (9). Thus for a given sheathliquid ow rate (for example, 4 L/min), the ratio of sheath flow to electroosmotic ow can vary from 10 to 100 depending upon the buffer pH (Table I [10]). This dilution factor also can contribute to reducing sensitivity. Given that a 10100-fold dilution of the buffer by the sheath liquid occurs, users have the opportunity to chemically manipulate the ions after separation. We investigated the effects of sheath liquid pH on the analysis of two peptides. Using the pKa values of the amino-acid side chains, we calculated the peptide charge for different pH environments (Figure 2) (11). Because the CE eluent is mixed with the sheath liquid, analysts should expect the sheath liquid pH to inuence the ionization state. This situation is reected in the mass spectra (Figure 3), which exhibit a shift of the molecular ions from those with 3 4 charges to those with only 12 charges when increasing the pH of the sheath liquid. Users

may also observe a different signal-to-noise ratio (S/N ). In all cases, the CE separation was unaffected. Varying the sheath liquid therefore offers the possibility of chemically optimizing the MS detection without compromising the CE separation. By using a binary or quaternary LC pump, analysts can optimize the sheath liquid composition during method development. Postseparation manipulation of the mass spectras appearance in LCMS is possible by adding a modier through a tee into the ow path. Kuhlmann and co-workers (12) demonstrated this effect for peptide analysis by adding propionic acid or isopropanol to weaken the signal-suppressing effects of triuoroacetic acid anions during the electrospray process. Selectivity and resolution: In CZE, analytes are separated based on their mobility differences. An ions mobility is proportional to its chargetoionic volume ratio. Small analytes with high numbers of charges migrate faster in an electric eld than larger analytes with small numbers of charges. In reversed-phase LC, the separation mechanism relies upon differences in the analytes surface hydrophobicity. This difference in hydrophobicity results in different partitioning between the stationary phase and mobile phase with increasing organic modifier concentration. Given these different separation mechanisms, analysts should expect a change in elution order when comparing the same set of analytes (Figure 4). Figure 4 shows that CEMS and LCMS, with their different selectivities, should be considered complementary techniques. To change selectivity in CEMS, analysts can use complexing agents to increase the ionic volume in chiral analysis or change the buffer pH to manipulate the charge of analytes with different pKa values (see Figure 2). However, adding complexing agents can hinder the efficiency of electrospray ionization. Selectivity changes in CE can be implemented simply by switching between run buffer vials that contain background electrolytes at different pHs (Figure 5). A selectivity change cannot be achieved as quickly in LCMS. For a given stationary phase and solvent system, users cannot alter the elution order. A different degree of resolution can be achieved only with different gradient programming (Figure 6), and the trade-off is longer run times. Changing the solvents and the stationary phase, including lengthy column equilibration, is necessary to make major selectivity changes.

5 4 Net charge (q) 3 2 1 0 2 1 2

pH 2.9

NH2DRVYIHPFHLCOOH Angiotensin l (1296.5) (8.2) (12.6) (6.2) (6.2) (3.2) NH2DRVYIHPFCOOH (8.2) (12.6)(6.2) (3.2) pH 6.9 pH 9.5 Angiotensin ll (1046.2)

4 pH 1297.5 649.3 433.2 325.2 1047.2 524.1 349.7

6 1297.5 649.3 433.2 325.2 1047.2 524.1 349.7

10 1297.5 649.3 433.2 325.2 1047.2 524.1 349.7

12

14

Figure 2: Plots of the calculated charge distribution and expected molecular ions (underlined) for angiotensin I (top curve) and II (bottom curve). Also shown are the amino acid sequences, possible charges, and pKas of the chargeable groups (in brackets) for both peptides.

(a) 100 50 0 Relative abundance (b) 100 50 0 (c) 100 50 0

433.0 3 Max: 66066 4 325.6 2 649.2

3 349.6 523.8 2 Max: 27728

3 433.0 4 325.1

648.8

2 Max: 11558

523.7 2 Max: 50598

433.0 3

648.8 2 Max: 8959 1 1296.8

523.7 2 Max: 50256 1046.4 1 400 600 800 m/z 1000 1200

400

600

800 m/z

1000

1200

Figure 3: Mass spectra of angiotensin I (left) and II (right) showing the effects of CEMS sheathliquid pH on molecular ions. Sheath liquid: (a) 0.5% acetic acid in 50% methanol (pH 2.9), (b) 5 mM ammonium acetate in 50% methanol (pH 6.9), (c) 0.5% ammonium hydroxide in 50% methanol (pH 9.5); sample: 0.05 mg/mL angiotensin I and II; injection: 300 mbars; buffer: 10 mM acetic acid (pH 3.5) for all three runs; scan range: m/z 3001300. Other conditions are described in the text.

Sensitivity: In CE, the sample is loaded directly onto the capillary by applying pressure. As a general rule, the sample zone should be no more than 23% of the total capillary volume to preserve peak efficiency. When using 75 cm 50100 m capillaries, the injection volumes are in the 10200 nL range. These low injection volumes are a major advantage when analyzing small sample volumes because several injection volumes can be taken from each sample; however, low injection volumes can be a severe disadvantage when attempting to analyze low-concentration samples. Large-sample injection volumes can be concentrated in LCMS under noneluting conditions at the top of the column. For 180300 m i.d. columns, the injection volume is in the low microliter range. Therefore, very small sample volumes often permit only one analysis, which restricts the possibility of method development. Comparing the limits of detection for automated CEMS and LCMS: The concentration limit of detection for angiotensin II using CEMS in the selected-ion mode at m/z 524.1 was 1 ng/L. This result means a 1 fmol limit of detection at a S/N of 3 for a 300-mbars injection. This limit of detection possibly could be further reduced by increasing the injection volume. In full-scan mode (m/z 1001600), the concentration limit of detection was 50 ng/L. This result corresponds to a limit of detection of 300 fmol for a 2000 mbars injection. Compared with the total ion electropherogram, the extracted ion trace provided an approximate 30% improvement in S/N. The approximately 1000-fold greater mass loading capability in LC provides a concentration limit of detection that is 1000-fold lower than that of CEMS. We loaded the same 6-pg mass in CE and LCMS and obtained selected-ion mode traces of a similar order of magnitude (data not shown). However, in LC this mass was loaded from a 1000-fold lower concentrated sample solution: the concentration limit of detection in selected-ion mode was 1 pg/L, which corresponded to an extrapolated limit of detection of 2 fmol from a 6-L injection. By calculating the limit of detection from the total ion chromatogram in full-scan mode, we obtained a 3-pmol limit of detection for an 18-L injection from a 1 ng/L sample solution. However, if the molecular weight of the analyte was known, the extracted ion trace revealed a much better S/N, which results in an extrapolated limit of detection of approxi-

(a) 6.0 Abundance (106 ) 5.0 4.0 3.0 2.0 1.0 0.0 6 (b) Abundance (105 ) 1.0 0.8 0.6 0.4 0.2 8 2 3 4

5 7

8 9 6 10 12 14 3 6 8 5

1 10

2 4 7 1 10 9

0.0 10 20 30 40 50 Time (min) Figure 4: Peptide separations obtained using (a) CEMS and (b) LCMS. Sample: 10 peptides at 0.16 mg/mL (CEMS) and 0.0053 mg/mL (LCMS) each; injection: 150 mbars for CEMS, 2 L for LCMS; scan range: m/z 350650 for both CEMS and LCMS. Peakspeptides: 1 KRTLRR (MW 829.0), 2 MKRPPGFSPFR (MW 1319.6), 3 DRVYIHPFHL (MW 1296.5), 4 pyroglutanic acid EHWSYGLRPG-NH2 (MW 1182.3), 5 pyroglutanic acid GLYENKPRRPYIL (MW 1672.9), 6 DRVYIHPF (MW 1046.2), 7 VYV (MW 379.4), 8 YGGFL (MW 555.6), 9 YGGFM (MW 573.7), 10 pyroglutanic acid ELYGNK (MW 776.8). 60 70 80 90

(a) 6.0 2 Abundance (106) 5.0 4.0 3.0 2.0 1.0 1 0.0 6 8 3 4

5 7

(b) 1.8 1.6 Abundance (106) 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0 10 12 14 6 8 10 12 14 16 1 10 3 5 2 6 8 7 9 4

8 9 6

10

Time (min)

Time (min)

Figure 5: Extracted ion electropherograms from CE peptide separations obtained using (a) 10 mM acetic acid (pH 3.5) and (b) 20 mM ammonium formate (pH 4.3) run buffers. See Figure 4 for other conditions and peak identication.

mately 750 fmol for angiotensin II. The data presented in this article reect detection limits that are achievable with fully automated and commercially available instrumentation. Analysts can achieve higher sensitivity with MS detection by using preconcentration techniques and sheathless interfaces. Many groups have taken various approaches to membrane preconcentration and solid-phase extraction to decrease the limits of detection by increasing the amount of sample loaded onto the CZE column (10,13). In this technique, workers prepare a small section of capillary that contains an absorbent material upon which the analytes can be concentrated. The extractor section of the capillary is placed in-line with the separation capillary. A large volume of sample is pushed through the absorbent material, washed with a noneluting buffer, and then eluted by injecting a plug of organic solvent followed by the separation buffer. Herring and Qin (10) reported the full-scan analysis of a 20 nM tryptic digest of myosin I heavy-chain kinase using a preconcentrator made of Poros beads and quartz wool together with a sheathless interface. The total amount of protein loaded was 150 fmol. Sheathless interfaces called micro- or nanosprayers are on-capillary or decoupled (14), tapered fused-silica tips. To make the electrical contact, a conductive coating such as gold is sputtered onto the sprayer surface or a wire made of a metal such as platinum is inserted into the capillary (15). Using these devices leads to a more efficient ionization process and higher sensitivity. Figeys and colleagues (16) reported using a 20-m i.d. gold-coated microsprayer to analyze a synthetic peptide. The detection limit was 300 amol in selected-ion mode (loaded from a 0.21-pmol/L solution). They also performed a full-scan analysis of an eight-peptide component mix with collision-induced dissociation for peptide sequencing. The amount of each peptide on the capillary corresponded to 33 pg (approximately 40 fmol), injected from a 1.67ng/L solution. These devices clearly provide greater sensitivities. However, they require special capillaries with inherently short lifetimes. This complication makes their use impractical in routine laboratories. Electrospray interfaces for these devices are not yet commercially available for on-line CEelectrosprayionization MS. Thompson and colleagues (17) described sensitivity improvements through the use of

Abundance ( 105 )

(a)

2.0 1.5 1.0 0.5 0.0

10 2,3 1 15 20 25 5,6 7,8 9 4

30

35

40

45

50

55

60 10 8 5 67 9

Abundance ( 105 )

(b)

1.0 0.8 0.6 0.4 0.2 0.0 10 1.0 0.8 0.6 0.4 0.2 0.0 10

2 3

20

30

40

50

60

70

80

90

100

Abundance ( 105 )

(c)

10 1 2 4 3 20 30 40 50 5 67 8 9

Time (min) Figure 6: Effects of varying gradients on LCMS separation of a 10-peptide mixture. Gradients: (a) 141% B in 40 min; (b) 141% B in 120 min; and (c) 111% B in 3 min and 1141% B in 90 min. The gure shows the extracted ion traces. See Figure 4 for other conditions and peak identication.

transient capillary isotachophoresis. They found that the concentration limits of detection for proteins in full-scan mode could be reduced by two orders of magnitude to approximately 100 nM. To use this type of system, analysts must be skilled in selecting the appropriate background electrolyte. Method development: The need for volatile buffers in LCMS is well documented. However, MS compatibility of buffers is not the rst priority when developing CE methods. In CE method development, the required separation selectivity dictates the buffer choice, and the buffers frequently are incompatible with electrospray, especially when they contain additives such as sodium dodecyl sulfate, cyclodextrins, or ampholytes. Figure 7 compares the analysis of peptides using a 20 mM phosphate buffer and a more volatile 10 mM acetic acid buffer. We used the same sheath liquid in both cases. With the volatile buffer, we detected signals in the total ion electropherogram (Figure 7a, right); however, we observed only some weak signals (Figure 7b, right)

and an unsteady baseline with high noise levels with the nonvolatile buffer system. Signals are visible in the extracted-ion mode (Figure 7b, right, see arrow), which is useful only when the molecular weights of the analytes are known. During the analysis of protonated peptide ions, the nonvolatile phosphate anions presumably decrease the ion intensity by ion-pairing effects or the prevention of ion evaporation during the electrospray process. Generally, volatile reagents like acetic acid, formic acid, ammonium acetate, and formate have been well suited for CEelectrospray-ionization MS. If analysts must adjust to higher pH values, then they should use ammonium hydroxide.
Conclusions

Interfacing LC and CE separations with electrospray-ionization MS and tandem UV detection is equally facile and achievable with the same degree of automation. Using a sheath liquid in CEelectrospray-ionization MS allows the postseparation manipulation of the analytes ionization. The use of

(a) 60 40 20 0 2 (b) Abundance (106) Abundance (mAU)

Abundance (107)

Abundance (mAU)

1.0 0.5

velopment route for CE is restricted when it is interfaced with electrospray-ionization MS because many CE buffers are nonvolatile and incompatible with electrosprayionization MS.
References

0 4 5.0 4.0 3.0 2.0 1.0 0.0 4 6 8 10 12 14 16 6 8 10 12 14 16

2 Time (min)

Time (min)

Figure 7: CEMS separations of the peptide mixture obtained using (a) 10 mM acetic acid (pH 3.5) and (b) 20 mM phosphate (pH 2.5) buffers with the same sheath liquid. Shown are the UV absorbance and total ion electropherograms (left and right, respectively). The arrow indicates the extracted-ion mode signal. See Figure 4 for other conditions and peak identication.

the sheath liquid as a tool for postseparation manipulation should be investigated further. CZE and reversed-phase LC offer quite different selectivities, which make the two techniques complementary. However, manipulation of selectivity and resolution is much easier in CE than in LC. Furthermore, the overall analytical run time for CEMS is much shorter than for LCMS. When comparing the sensitivity of these techniques, it is clear that the concentration sensitivity available from routine LCMS instrumentation is 1000-fold higher than CEMS instrumentation. In both techniques selected-ion mode offers greater sen-

sitivity, but this process requires a previous knowledge of the analytes molecular weights and charges. The suitability of either technique for peptide analysis is more likely to be dictated by the sample concentration available rather than instrumentation concerns. To achieve greater sensitivity for CE analyses, analysts can use some preconcentration techniques. Methods for reversed-phase LC of peptides are well established and any method development comprises changing the mobile-phase gradient and the stationary phases rather than changing the mobilephase constitution. The normal method de-

(1) W.M.A. Niessen, J. Chromatogr. A 794, 407435 (1998). (2) B. Mehlis and U. Kertscher, Anal. Chim. Acta 352, 7183 (1997). (3) J.F. Banks, Electrophoresis 18, 22552266 (1997). (4) D. Figeys and R. Aebershold, Electrophoresis 19, 885892 (1998). (5) R.D. Smith, C.J. Barinaga, and H.R. Udseth, Anal. Chem. 60, 19481952 (1988). (6) E.D. Lee, W. Muck, J.D. Henion, and T.R. Covey, Biomed. Environ. Mass Spectrom. 18, 844850 (1989). (7) J.H. Wahl, D.C. Gale, and R.D. Smith, J. Chromatogr. A 659, 217222 (1994). (8) An Integrated Solution for CE-ESI-MS, publication number 12-5968-1328E, Agilent Technologies Deutschland GmbH (Waldbronn, Germany, 1999). (9) T. Tsuda, in Handbook of Capillary Electrophoresis, J.P. Landers, Ed. (CRC Press, Boca Raton, Florida, 1994), pp. 563590. (10) C.J. Herring and J. Qin, Rapid Commun. Mass Spectrom. 13, 17 (1999). (11) E.C. Rickard, M.M. Strohl, and R.G. Nielsen, Anal. Biochem. 197, 197207 (1991). (12) F.E. Kuhlmann, A. Apffel, S.M. Fischer, G. Goldberg, and P.C. Goodley, J. Am. Soc. Mass Spectrom. 6, 12211225 (1995). (13) A.J. Tomlinson, N.A. Guzman, and S. Naylor, J. Capillary Electrophor. 2, 247266 (1995). (14) K.P. Bateman, R.L. White, and P. Thibault, Rapid Commun. Mass Spectrom. 11(3), 307315 (1997). (15) P. Cao and M. Moini, J. Am. Soc. Mass Spectrom. 8(5), 561564 (1997). (16) D. Figeys, I. van Oosteveen, A. Ducret, and R. Aebersold, Anal. Chem. 68, 18221828 (1996). (17) T.J. Thompson, F. Foret, P. Vouros, and B.L. Karger, Anal. Chem. 65, 900906 (1993).

Copyright Notice Copyright by Advanstar Communications Inc. Advanstar Communications Inc. retains all rights to this article. This article may only be viewed or printed (1) for personal use. User may not actively save any text or graphics/photos to local hard drives or duplicate this article in whole or in part, in any medium. Advanstar Communications Inc. home page is located at http://www.advanstar.com.

Reprinted from LCGC, January 2000

AN

ADVANSTAR 5 PUBLICATION

Printed in U.S.A.

Agilent Technologies
Publication Number 5980-1304E