Investigation of Electrospray for a Capillary Electrophoresis-Mass Spectrometry Interface in Reverse Polarity and Negative Ion Mode

Abstract

Capillary zone electrophoresis (CZE) paired with mass spectrometry (MS) is a powerful analytical technique for examining mixtures of ionic analytes such as glycosaminoglycans. This study examines the mechanics of the electrospray process for a sheath flow CZE-MS interface under reverse polarity negative ionization conditions. Liquid flow in a sheath flow nano-electrospray CZE-MS interface is driven by two mechanisms, electroosmotic flow (EOF) and electrospray nebulization. The contribution of these two processes to the overall flow of solution to the electrospray tip is influenced by the surface coatings of the sheath flow emitter tip and by the solvent composition. We have investigated the application of this interface to the reverse polarity separation of glycosaminoglycans and find that the role of EOF is far less than has been reported previously, and the electrospray process itself is the largest contributor to the flow of the sheath liquid.

INTRODUCTION

The online combination of capillary zone electrophoresis and mass spectrometry (CZE-MS) provides powerful capabilities for the analysis of mixtures of ionic species that are not readily examined by more conventional hyphenated approaches such as liquid chromatography and tandem mass spectrometry (LC-MS/MS) (1, 2). Recently our laboratory demonstrated the effectiveness of CZE-MS for analyzing glycosaminoglycans (GAGs), a family of sulfated carbohydrates that are difficult to analyze by LC-MS/MS (3). A commercial sheath flow CZE-MS interface provided the means to implement nano-electrospray with a low dilution factor and minimal sample consumption, while providing ample resolution of closely related isomers.

Sulfated carbohydrates exist as negative ions in solution, and thus require reverse polarity for the CZE separation and negative mode ionization for the mass spectrometry detection. Under these conditions, standard bare fused silica separation capillaries are subject to an electroosmotic flow (EOF) that opposes the migration of negative ions, adversely affecting the separation time and resolution for GAGs. Our prior work showed that by modifying the surface of the separation capillary to make it neutral or cationic, shorter separation times could be achieved while maintaining good peak resolution for this class of compounds (3). In a similar fashion, we modified the surface of the glass capillary that forms the emitter of the sheath interface, with the goal of making a cationic surface that would provide a sufficient EOF to sustain stable electrospray. The sheath flow CZE-MS interface has been reported by others to function by the electrokinetic pumping of the sheath liquid by EOF in the emitter tip (4–6). We were surprised to find that neutral coated, or even unmodified bare glass capillaries (with an anionic surface) would function well with reverse polarity CZE and negative ionization mass spectrometry. The present study is motivated by the desire for a better understanding of the sheath flow interface, and the mechanism of liquid flow necessary for electrospray ionization.

The sheath flow CZE-MS interface is something of a hybrid between a traditional ESI interface and a nano-ESI interface (7–11), as it uses no desolvation gas, has no backing pressure, and has a moderately sized orifice of 30 μm. Liquid from a reservoir is introduced to a bare glass emitter sheath that surrounds the exit of the separation capillary, and which is pulled to a fine tip that functions as an electrospray ionization emitter. As the analyte exits the separation capillary, it mixes with the sheath liquid in a nanoliter volume and then exits through an orifice as it undergoes electrospray, at the entrance of a mass spectrometer (12).

There are two mechanisms that are responsible for the flow of sheath liquid through the interface: electroosmotic flow (EOF) and electrospray-driven liquid flow (7–13). EOF is liquid flow caused by the movement of the solvated mobile ions in the electrical double layer present at the inner surface of a glass capillary, and which acts as a nanoscale pump. The ESI-driven flow is created by the aerosolization of solvent at the tip of the emitter. Here we explore the contributions of these forces in reverse polarity CZE and negative ionization mass spectrometry experiments. We use a highly negatively charged glycosaminoglycan sample for this investigation. Cationic and neutral modifications of the inside surface of the bare glass emitter that surrounds the end of the separation capillary were utilized to ascertain the importance of EOF to the performance of the sheath flow CZE-MS interface under reverse polarity conditions.

MATERIALS AND METHODS

Materials.

Bare fused silica (BFS) capillary for CZE (360um o.d. × 50um i.d.) was purchased from PolyMicro Technologies (Phoenix, AZ), borosilicate glass capillaries (1.0mm OD × 0.75mm ID) and pulled coated electrospray emitters (1.0mm OD × 0.75mm ID) were obtained from CMP scientific (Brooklyn, NY). Coating reagents, dichlorodimethylsilane (DMS, Sigma-Aldrich, St. Louis, MO) and N-(6-aminohexyl)aminomethyltriethoxysilane (AHS, Gelest, Morrisville, PA) were prepared in toluene. Ammonium acetate, water, and methanol were of HPLC grade (Fisher Scientific, Hampton, NH). Sodium hydroxide, acetone, and toluene were purchased from Sigma-Aldrich (St. Louis, MO). Ammonium hydroxide and acetic acid were purchased from Avantor (Allentown, PA)

Heparan Sulfate GAG Sample.

The GAG tetrasaccharide sample was prepared by enzymatic depolymerization of heparan sulfate and purified using strong anion exchange high-pressure liquid chromatography (SAX-HPLC), using methods described previously (14, 15). The sample was desalted with a 3 kDa Amicon Ultra centrifugal filter (Millipore, Temecula, CA) prior to separation and mass spectrometry analysis. Filters were conditioned with water, and the sample was then washed with two filter volumes of water (14,000 × g for 25min each). Before analysis, GAG samples were diluted to 5ug/mL in water. The heparan sulfate sample used in this study, referred to as HS1, has been characterized previously, and its structure has been assigned as ΔHexA(2S)-GlcNS(6S)-HexA-GlcNS(6S) (16). Its structure is shown in Figure 1.

Figure 1.

Chemical structure diagram of heparan sulfate sample 1 (HS1). This heparan sulfate tetrasaccharide has 5 sulfo-modifications, making it highly negatively charged.

Emitter surface modification.

The interior surface of the sheath flow emitter was covalently modified using solutions that were prepared in toluene with 1% concentration of either DMS (neutral surface) or AHS (cationic surface). To clean and prepare the emitter for surface modification, the borosilicate glass capillary was rinsed consecutively with 2 mL each of aqueous 0.1 M NaOH, water, methanol, dry acetone, and dry toluene. The surface of the borosilicate glass capillary was then modified by flushing with 500 μL of 1% DMS or AHS over a 20 min period. The borosilicate glass capillary was then consecutively rinsed with 2mL of each dry toluene, dry acetone, and methanol. The modified borosilicate glass capillaries were then processed into emitters with 30 μm orifices using a commercial capillary puller, as described previously (17). For all experiments, the separation capillary was modified with AHS as described in our previous paper (3).

Instrumentation.

Separations were conducted with an Agilent HP 3D capillary electrophoresis instrument (Wilmington, DE). A long bare fused silica capillary was used for CZE of GAG analyte, and its inner surface was modified with AHS to speed up analyte migration. The total length of the capillary ranged from 58–60 cm, and its inner diameter was 50 μm with an internal volume of approximately 1μL. 25mM ammonium acetate was used as both the background electrolyte (BGE) and the sheath liquid (SL) for reverse polarity experiments. The aqueous GAG sample (HS1) was injected for 3 s at 950 mbar followed by a background electrolyte (BGE) injection for 10 s at 10 mbar. The injected volume was 0.1μL. The ionic strength of the injected sample plug is 2–3 orders of magnitude less than that of the background electrolyte, and sample stacking is expected under these conditions and provides a sharp sample front (2). The entrance end of the capillary was then placed into a vial of BGE prior to separation. A voltage of −30 kV was applied to the capillary to drive the separation for all experiments.

An EMASS-II (CMP Scientific, Brooklyn, NY) CZE-MS interface was employed to couple the CZE with a Thermo Scientific Velos Orbitrap Elite mass spectrometer (Bremen, Germany). A separation capillary with a beveled outlet was nested inside a glass emitter sheath with a 0.75 mm inner diameter and a 30 μm tip orifice (CMP Scientific, Brooklyn, NY). The etched end of the capillary was positioned 0.3–0.5 mm from the tip of the sheath emitter orifice to create a mixing volume of ca. 1–5 nL and the emitter tip was filled with sheath liquid (SL), at a concentration of 25 mM ammonium acetate 50–70% methanol. An external power supply provided a voltage of −1.9 kV to the sheath liquid reservoir via a platinum wire, which produces electrospray at the emitter tip. MS detection was performed in negative ion mode. Prior to CZE-MS experiments, a semi-automatic optimization of source parameters was performed using sucrose octasulfate to improve sensitivity of sulfated GAGs and reduce sulfate loss during MS analysis. The Orbitrap was scanned from m/z 150–2000 for GAG oligosaccharides with a specified resolution of 120,000.

RESULTS AND DISCUSSION

Figure 2 shows a schematic of the nanoflow sheath CZE-MS interface developed by Dovichi and Wojcik, and used for these studies (18). For reverse polarity CZE-MS experiments 25mM ammonium acetate in 70:30 methanol:water was used for both the BGE and the SL. A large negative voltage (−30kV) is applied to the BGE vial at the inlet of the separation capillary and a relatively smaller negative voltage (−1.9kV) is applied to the reservoir of the sheath liquid (SL) near the outlet of the separation capillary. The potential applied to the SL serves two purposes. It completes an electrical circuit through the separation capillary by contact of the SL and the BGE near the emitter tip which drives analyte migration through the separation capillary. It also drives the electrospray process by creating a difference in electric potential between the sheath liquid near the emitter and the small potential (−0.33 kV) at the entrance of the mass spectrometer. A Taylor cone forms and electrospray of the analyte is achieved when the electric potential difference is large enough to overcome the surface tension of the solvent. Dilution is reduced, and sensitivity is maximized by placing the outlet of the separation capillary in close contact to the orifice of the sheath emitter, producing a mixing volume that is less than 1 nL.

Figure 2.

Schematic of the CZE-MS sheath flow interface. Background electrolyte (BGE) and sheath liquid (SL) are 25mM ammonium acetate in 70:30 MeOH:H₂O. The separation capillary makes a snug fit with the emitter sheath and establishes a small mixing volume (ca. 200 pL) where BGE and SL meet and complete the electrical circuit. The electric potential difference between the sheath liquid in the emitter and the detector source forms a Taylor cone and electrospray. Covalent modification of the inner-surface used to control EOF in the emitter sheath is represented by the green, as it is destroyed in the tip by the pulling process.

The flow of the sheath liquid must be sufficient to sustain a stable electrospray, as the flow rate of the liquid exiting the separation capillary is too small to serve this purpose by itself. The inventors of this interface describe the driving force for sheath flow as electrokinetic in nature, resulting from the electroosmotic flow of SL within the emitter (5, 18). Under normal polarity conditions (positive voltages for CZE and for the SL), a bare glass surface, with its negatively charged surface, would be expected to provide an EOF that moves the sheath liquid from its reservoir toward the emitter tip. Under reverse-polarity conditions (negative voltages for both CZE and the SL), the EOF will reverse in direction, and oppose the desired flow necessary for electrospray ionization. Previous work using reverse-polarity for negatively charged analytes has employed a cation-modified surface on the emitter, which reversed the direction of the EOF, so that it moved toward the emitter tip (6). Not considered previously was the flow of the SL due to the electrospray process. In the absence of EOF, liquid is drawn through the emitter as the solution is nebulized by ESI. This may be sufficient to overcome EOF even when it flows in the wrong direction. We have made a series of measurements to better understand the relative magnitudes of these forces (EOF versus ESI-driven flow). We were motivated to examine whether or not it was necessary to alter the surface of the emitter in order to make reverse polarity separations. Modifying the surface adds extra complexity to the experiment. This modification process must be accomplished before the emitter is pulled to a fine tip, and heating during this process is certain to damage the modification in an indeterminate portion of the capillary. Also, delamination of the modified surface can occur, leading to a reduced lifetime from plugging of the emitter.

The interior surface of the borosilicate glass emitters was covalently modified to control EOF. We used the same agents for modifying the surface of the emitter as we had previously used for modifying the separation capillary (3). We use a cationic reagent (AHS) to reverse EOF and a neutral reagent (DMS) to nullify EOF. Our cation reagent (AHS) is similar to material described by others (6), but has a longer linker to make a more stable and longer-lasting covalent modification. The heat generated pulling the emitter tip is thought to destroy the surface modification in its narrow, conical region, but the extent of this region is poorly characterized.

EOF behaves like a nano-scale pump, moving bulk solvent, in this case sheath liquid, along the inner surface of the emitter tip. If EOF contributes significantly to electrospray, then the volume of solvent being consumed should be influenced by the direction of the EOF. Figure 3 shows the measured average rate of consumption of sheath liquid from the reservoir for each type of inner-surface. The volume of the sheath liquid in its reservoir was measured before and after 10 h of use, and the difference in volume over that time is the rate of consumption. Under these reverse polarity conditions, the consumption of SL is found to be similar between all three types of inner-surface. The cation-coated surface shows a modestly larger consumption, suggesting that the EOF contributed fractionally to the overall sheath flow for this surface. The neutral-coated and bare glass surface show nearly identical consumption, suggesting that the EOF of the bare glass surface is small compared to the overall sheath flow under the experimental conditions employed. Furthermore, the flow rate of the interface is low, on the order of 10uL/h, as expected for nanospray conditions.

Figure 3.

Comparison of sheath liquid consumption rates between three types of emitter sheath inner surface; DMS (neutral), AHS (cationic), and bare glass surface (BG) (anionic). DMS (neutral) nullifies EOF, AHS (cationic) provides an EOF toward the emitter tip, and BG (anionic) produces an EOF that opposes the sample flow toward the emitter tip.

Prior studies of heparan sulfate tetramers show that under reverse polarity conditions, with 25mM ammonium acetate in 70:30 methanol:water as the BGE and SL, an EOF that opposes the electrophoretic flow (bare fused silica separation capillary) will double CZE migration times compared to favorable EOF (AHS modified separation capillary) (3). This suggests that EOF is a relevant and influential factor in the separation capillary under these solvent conditions. The emitter is borosilicate glass and presents the same inner surface as the separation capillary, but the EOF is weaker because of the much larger diameter of the emitter compared to the separation capillary (750 μm versus 50 μm). A heparan sulfate tetrasaccharide sample, HS1, structure shown in Figure 1, was used to test the stability of ionization when using emitters with different surface modifications. The data in Figure 4 and Table 1 show that, unlike the separation capillary, modifying the surface of the emitter has no observable effect on the migration time, peak shape, or mass spectrum of HS1. Figure 4a shows the electropherogram and mass spectrum obtained with an AHS modified emitter, while the data in 4b was obtained with the same experiment conditions, but with a bare glass emitter. The magnitude of EOF under these conditions appears negligible compared to the ESI flow and has no influence on the outcome of the experiment.

Figure 4.

CZE-MS measurement of HS1 comparing cationic and anionic emitter inner surfaces. An AHS modified separation capillary was used in tandem with a (a) AHS (cationic surace) coated emitter and (b) uncoated glass (anionic surface) emitter. Mass spectra are shown as insets to the electropherograms, and were acquired for the major feature at 23 minutes migration time.

Table 1.

Comparison of migration times and peak widths for different emitter sheath coated surfaces: AHS (cationic), DMS (neutral), and BFS (anionic)

Emitter Type	Migration Time (min)	FWHM (sec)
AHS	21.41	0.18
	21.33	0.19
	21.16	0.21
	21.19	0.19
	21.11	0.21
	21.04	0.22
AHS avg.	21.21	0.20
DMS	21.76	0.17
	21.33	0.21
	21.00	0.19
	21.08	0.20
	21.16	0.22
	21.25	0.20
DMS avg.	21.26	0.20
BFS	21.28	0.19
	21.20	0.16
	21.20	0.21
	21.04	0.17
	20.93	0.18
	20.72	0.18
BFS avg.	21.06	0.18

While the direction of EOF is controlled by the surface of the emitter, the magnitude of EOF can be manipulated by the content of the solvent. Varying the organic composition of the solvent will influence the degree of ionization of the silanol groups on the surface of a bare glass emitter, and therefore the number of ions in the electrical double layer. Highly organic solvent systems, like the one used here to effectively separate GAGs, reduces the degree of ionization of the silanol groups, thus reduces the mobile ions in the double layer, and reduces the EOF. Increasing the aqueous concentration will have the opposite effect, leading to an increase in the EOF. Figure 5 shows the difference in electrospray between a sheath liquid that has 70:30 methanol:water (5a) and one that has 50:50 methanol:water (5b & 5c). By increasing the ratio of water vs. organic in the solvent, degree of ionization of surface silanol groups increases and the magnitude of EOF is expected to increase. This results in unstable electrospray at lower organic concentration for uncoated bare glass emitters, where the EOF is counter to electrospray flow, exhibiting itself as signal dropout observed across the peak in the mobility plot in Figure 5b. Conversely, AHS emitters, which have EOF with the same direction as the electrospray flow, are found to function normally at higher aqueous concentration, as can be seen in Figure 5c. These data suggest the EOF increases with a 50% organic/50% aqueous solution to a magnitude that is comparable to the electrospray driven flow.

Figure 5.

CZE-MS electropherograms and mass spectra (inset) for HS1, obtained with different solvent organic concentrations. HS1 in (a) 70:30 MeOH:H₂O BGE and SL with a bare glass emitter, (b) 50:50 MeOH:H₂O BGE and SL with a bare glass emitter, and (c) 50:50 MeOH:H₂O BGE and SL with a AHS coated emitter. The separation capillary had an AHS modified surface for all three measurements.

Altering the pH of the solvent also changes the magnitude of EOF through changes in the degree of ionization of the surface silanols for bare glass emitters (2, 19). A more basic sheath liquid, obtained by adding ammonium hydroxide, ionizes a larger proportion of the silanol groups present on the inner surface of an uncoated glass emitter and should increase the EOF that opposes the electrospray flow in a reverse polarity CZE experiment. However, we have examined more alkaline solvents (up to pH 10), and find no disruption of the ESI process, suggesting that the magnitude of the EOF does not change appreciably with changes in pH when the optimal organic content of 70% methanol is used for the sheath liquid.

CONCLUSIONS

Collectively, the data presented here show that the accepted mechanism of EOF-driven flow for the sheath liquid in the CZE-MS interface does not apply under all experimental conditions (4). The separation of GAG oligomers by capillary zone electrophoresis is optimal at relatively high concentration of organic solvent (70%). While EOF is still significant in the separation capillary with a solvent of this composition, in the sheath flow interface, EOF is small compared to the flow driven by the electrospray process itself. At lower organic concentration, the EOF in the sheath interface increases enough to be a concern, but this is not an optimal solvent composition for the separation, and thus is not relevant. Surface modified CZE-MS emitters are expensive, are unstable at high alkaline concentrations (pH 10 or higher), and have more limited lifespans than unmodified emitters. Thus, it is significant that an emitter with an unmodified surface can be implemented for reverse polarity separations. If coated emitters are required, then covalently bound coatings are best because they are stable and effective (3).