Recent Trends of Capillary Electrophoresis-Mass Spectrometry in Proteomics Research

Abstract

Progress in proteomics research has led to a demand for powerful analytical tools with high separation efficiency and sensitivity for confident identification and quantification of proteins, posttranslational modifications (PTMs), and protein complexes expressed in cells and tissues. This demand has significantly increased interest in capillary electrophoresis-mass spectrometry (CE-MS) in the past few years. This review provides highlights of recent advances in CE-MS for proteomics research, including a short introduction to top-down mass spectrometry (TDMS) and native mass spectrometry (native MS), as well as a detailed overview of CE methods. Both the potential and limitations of these methods for the analysis of proteins and peptides in synthetic and biological samples and the challenges of CE methods are discussed, along with perspectives about the future direction of CE-MS.

I. Introduction

Proteomics is a constantly progressing field that endeavors to comprehensively interrogate biological processes by employing a systematic analysis of the proteins expressed in cells and tissues. Mass spectrometry (MS)–based methods continue to be the most widely used methods for high-throughput proteomics research^1–4, and liquid chromatography-mass spectrometry (LC-MS) is the prevailing platform. Confident identification and quantification of proteins, posttranslational modifications (PTMs), and peptides in biological samples rely on rigorous separation of peptides prior to introduction into the mass spectrometer^1,5,6. Capillary electrophoresis is a mature separation technology that is effectively coupled with mass spectrometry (CE-MS) and is well suited for this purpose^7,8. Recent advances in CE-MS platforms have attracted considerable interest from the proteomics community as an alternative to LC-MS^9–11. CE offers advantages over LC, including efficient and fast separations paired with low sample consumption. In addition, solvent consumption (aqueous or organic) in CE-MS is very low, making it environmentally advantageous⁹. The analytical limitations and advantages of CE and LC and some of their contrasting characteristics are presented in Table 1.

Table 1.

The analytical limitations and advantages of CE and LC.

CE - Advantages	CE - Limitations	LC - Advantages	LC- Limitations
Automation	Sample loading	Automation	High organic solvent consumption
Sample size (nL)	Sensible flow rate (buffer, capillary wall)	Sample size (μL)	----
Sensitivity (fmol-nM)	----	Sensitivity (μM)	----
Speed (seconds – minutes)	----	Speed (minutes – hours)	----
High resolution	----	Fast gradient	----
Aqueous solvent	----	Good reproducibility	----
Low flow	----	Stable flow	----

The most common CE-MS technique used in the proteomics field is capillary zone electrophoresis (CZE) coupled with electrospray ionization (ESI). Capillary isoelectric focusing (cIEF)¹² is another useful, but less commonly used CE technique.

In proteomics research, proteins are often identified using a well-established bottom-up approach. This strategy is based on indirect identification of proteins through digestion (enzymatic or chemical) to provide peptides that are further analyzed by either LC-MS or CE-MS^13,14. Proteins can also be investigated using specific or limited digestion to produce large peptides via middle-down strategies¹⁵, or with top-down mass spectrometry (TDMS) analysis of intact proteins. TDMS is an emerging approach that provides detailed structural information on intact proteins, including characterization of PTMs¹⁶. PTM patterns can provide insight into potentially significant biological processes^14,16,17 but can often be lost after the protein digestion^14,18 used in bottom-up approaches. Since CE separations can be performed in aqueous buffers, protein complexes can be separated in their native structures, preserving non-covalent interactions within a protein and between subunits of protein assemblies, prior to MS detection^19–21. Antibodies and their related glycosylated forms are often analyzed using CE-MS via TDMS²² and CE–MS has been used to determine the purity and the stability of proteins in drug discovery²³.

A growing body of research has demonstrated the efficiency of CE-MS and its successful application to proteomics. Numerous reviews have provided an overview of the technical developments of CE-MS^9,10,24 and its application for peptides,^13,25–27 intact protein analysis,^28,29 and biomarker discovery^7,30. In this review, we focus on applications of CE-MS in proteomics research that have been reported since 2008. Parameters for maximal selectivity and sensitivity of a CE-MS platform are discussed, as well as a brief introduction to TDMS and native MS.

II. Top-Down Mass Spectrometry and Native Mass Spectrometry

The proteomics field has been largely dominated by bottom up-based techniques, where proteins are digested prior to MS analysis¹. This approach is well-known for its high-throughput performance; however, critical information about PTMs and subunit protein assemblies (via non-covalent interactions) is generally lost after proteolysis^14,31. Consequently, peptide-based methods may limit our understanding of the intricate and nuanced control of diverse biological activities by PTMs¹⁴. It is commonly accepted that PTMs have significant effects on the fates and functions of cellular proteins¹⁷. Thus, characterization of PTMs is of key importance for understanding cell signaling processes related to clinical disorders¹⁶. Top-down mass spectrometry (TDMS) is an emerging proteomics approach that interrogates proteins as intact entities without prior digestion. Top-down analysis can provide unique insights into a protein’s structure by providing precise molecular weight measurements, the identification of multiple PTMs^5,16,28 and in-depth proteoform characterization (different forms of a single protein generated from a gene)³². In addition, as demonstrated by Gomes and co-workers^33,34, TDMS is well-suited for identification and characterization of branched systems consisting of two proteins. Recent improvements in mass spectrometry instrumentation and bioinformatics have enabled TDMS to be used to characterize and quantify hundreds to thousands of proteoforms in a single analysis¹⁶. TDMS is now being used in a wide range of applications, including the production of therapeutic proteins, such as antibody drug conjugates³⁴. Most TDMS studies are carried out under denaturing conditions, leading to the disruption of the biologically relevant non-covalent interactions, such as protein-protein and protein-ligand complexes³⁵.

While the use of appropriate terminologies to precisely describe the different types of TDMS experiments is still debated, Lermyte and co-workers have proposed a standard lexicon for TDMS³⁶, and we have adopted their recommendations for this review. The schematic of a TDMS workflow is shown in Figure 1.

Figure 1.

Schematic of a TDMS workflow. Reproduced from reference 37, with permission from The Royal Society of Chemistry³⁷.

Most proteins are assembled into protein complexes to achieve functional activity, but it is still unclear how these macromolecular complexes are formed. Native MS provides a proteomics tool that enables identification and characterization of these biomolecular structures and advances the understanding of their biological pathways at the molecular level^19,20,35. Native MS can be used to interrogate the structure of intact proteins in their near native state, with an emphasis on preserving noncovalent protein-protein and protein-ligand complexes²⁰. The success of native MS approaches relies on ESI of proteins in a volatile aqueous buffer that preserves native quaternary and tertiary states in the gas phase^19,20. Native MS provides information about ligand binding, dynamics, stoichiometry of individual subunits, stability, topology, and assembly structural properties³⁵. Recently, the Kelleher group reported a standard operating procedure for native MS³⁸. This valuable tutorial was designed to guide non- and experienced scientists to obtain reproducible and high-quality native MS data for proteins ranging from 30–300 kDa. Figure 2 illustrates a schematic of the information that can be obtained from a native MS experiment.

Figure 2.

Schematic of the information obtained from native MS experiment. Reprinted from reference 39, with permission³⁹.

III. Capillary Electrophoresis-Mass Spectrometry

A. Interface

Since the introduction of CE-MS for analysis of macromolecules, a few soft ionization techniques have been evaluated for CE-MS platforms, but as ESI is the most commonly used technique, we will limit our review to CE-MS applications utilizing ESI. Interfacing CE with ESI requires transferring a voltage to liquid to create the electrospray, which has been done with sheath-liquid or sheathless interfaces. Several thorough reviews have chronicled the considerable progress made in interfacing CE with ESI-MS^9,10,40,41.

The sheath-liquid interface has proven to be versatile and robust and is currently the more commonly used configuration in CE-MS for proteomics research. This interface uses a coaxial sheath-liquid to produce electrical contact between the CE and ESI source. Conversely, the sheathless configuration is based on the direct application of voltage to the background electrolyte (BGE). The Dovichi group has made significant advances in sheath-liquid interface designs, including the electrokinetically pumped low sheath-flow nanospray interface. This sheath-liquid prototype is commercially known as EMASS-II ion source. It has no mechanical flow and a closed electrical circuit that is driven exclusively by electrokinetic flow^42,43. Its application has been successfully demonstrated for both bottom-up^44–48 and top-down analysis^49–52. A potential limitation of the sheath-liquid interface is that the high flow rate of the sheath-liquid dilutes the analyte, thereby causing a potential loss in sensitivity^5,13. Also, suction effects that are caused by both the sheath-liquid and nebulizing gas can be detrimental for separation efficiency⁵³. However, this interface adds flexibility to the system since the spray conditions can be optimized (e.g. organic content) and thus, the ionization process is less dependent on the BGE composition. Although the sheath-liquid usually contains organic solvents that can adversely affect native protein conformations, Shen and co-workers successfully applied an aqueous solution of 25mM ammonium acetate as a sheath-liquid for the analysis of protein complexes from Escherichia coli²¹.

In contrast, the sheathless interface can significantly enhance signal/noise due to decreased analyte dilution and background noise, as the BGE is the only liquid in the system^5,54. In this configuration, the main challenge is maintaining CE flow while generating a closed electrical circuit between the CE separation and the ESI source⁵. This has been accomplished in two different ways. The first method includes an ESI tip coated with a highly conductive metal emitter (gold) or detachable porous emitter to transfer the high voltage to the liquid^54,55. A second strategy includes a porous region close to the ESI tip for electrical connection through an electrolyte⁵. The porous tip sheathless interface was introduced by the Moini group and is the most popular sheathless prototype in proteomics research⁵⁵. The sheathless interface has been widely used in proteomics research to provide ultrasensitive analysis^56–58, although it is limited by potential blocking of the pores. In a recent paper, Belov and co-workers reported a successful application of an etched porous-tip sheathless interface for analysis of complex-forming protein standards, a monoclonal antibody, and a ribosomal proteome of the Escherichia coli in native conditions³⁵. This interface enabled characterization of proteins without dilution prior to nano-ESI flow rates (10–20 nL/min). The ultra-low flow rates enabled higher sensitivity from this interface than from sheath-liquid designs. Previously, the Yates group demonstrated the power of a porous tip sheathless interface for highly sensitive top-down analysis^59,60. In two studies, the porous tip interface provided the sensitivity needed for front-end separation and top-down analysis. Nguyen and co-workers achieved high sensitivity analysis of tryptic peptides and proteins at low nanoliter flow-rates using a sheathless interface that is based on a sub-micrometer fracture located directly in the capillary⁶¹. In this study, a detection limit of 0.045 pmol/μL was achieved for a model peptide (leucine enkephalin) with gentle ionization, which suggests its suitability to probe proteins in their tertiary and quaternary states. A detailed experimental examination of a sheathless transient capillary isotachophoresis (cITP)/CZE−MS interface was provided by Guo and co-workers⁶². In this design, the electric contact between the BGE and ESI source enabled the use of a conductive liquid that is in contact with the metal-coated surface of the ESI emitter. The metal-coating on the ESI emitter was found to be highly sensitive and robust for approximately 100 analyses of a mixture of 9 peptide standards. The most common types of sheath-liquid and sheathless interfaces are presented in Figure 3.

Figure 3.

Schematic representations of (A) porous ESI tip sheathless interface and (B) electrokinetically pumped sheath-flow nanospray. Reprinted from references 63 and 45, respectively, with permission^45,63.

B. Separation

CE separations rely on the electrophoretic mobility of the analytes in an electric field. While many CE-based techniques have been reported for the analysis of proteins and peptides, CZE is the most commonly used in proteomics research^10,64–66. In CZE separation is based on the analyte migration (according to size/charge) in different zones at different velocities⁶⁷. The ionic strength and pH of the BGE drive the solute’s electrophoretic mobility. In CZE, band broadening is only caused by longitudinal diffusion. This is particularly attractive for protein analysis because they have small diffusion coefficients^28,29. Notable examples of CZE-MS include the intact mass analysis of biopharmaceutical proteins¹¹ and monoclonal antibodies^22,68 where complete baseline separation is often achieved. Figure 4 shows a CZE separation for top-down proteome analysis of Escherichia coli that resulted in high peak capacity (~ 280, calculated using the average peak width at 50% peak height for the Escherichia coli proteome sample)⁶⁹.

Figure 4. and Figure 5.

Base peak electropherogram of a CZE separation for top-down analysis in Escherichia coli proteome. Reprinted from reference 69, with permission⁶⁹.

Intensity of the proteins from the CZE-MS method with different BGEs and concentrations. Reprinted from reference 69, with permission⁶⁹.

Zhao and co-workers demonstrated the value of combining CZE separations with electron dissociation techniques for top-down analysis⁷⁰. In this study, a baseline separation was first obtained for a mixture of four standard proteins and then the CZE strategy was extended to the analysis of a biologically relevant protein mixture derived from the Mycobacterium marinum secretome. More recently, McCool and co-workers reported a CZE separation in conjunction with activated ion electron transfer dissociation to interrogate the proteome of Escherichia coli via TDMS where 328 proteins and 3028 proteoforms were identified using 1% spectrum level and 5% proteoform-level false discovery rates⁷¹. In another study, the precision and robustness of CZE separations were confirmed for the analysis of monoclonal antibody Fc/2 charge variants using top-down and middle-down strategies¹⁸. CZE has also proven to be a valuable tool for protein identification via bottom-up proteomics where ultrasensitive and fast analysis can be obtained^72–74. Using a bottom-up approach, Zhang and co-workers identified 27,000 peptides and 4,400 proteins with CZE⁴⁶. Despite the high number of identifications, some limitations were observed, including marginal identifications during the first 20 min of the separation, and a dramatic reduction in identifications after 85 min. Additional optimization could improve the methodology.

There have been several theoretical models reported to predict the electrophoretic mobility of a peptide. Since it has been generally accepted that residue shape and size have little effect on the electrophoretic mobility of a peptide, these models are based on the use of BGE pH, side group pKa, N-terminal amine, and C-terminal acid to determine the charge of the peptide. The Dovichi group reported a sequence-specific model to predict peptide electrophoretic mobility via CZE separations⁷⁵. The authors attributed their success to the selection of major sequence-specific features that alter the peptide charge in acidic conditions. Acidic residues (Asp and Glu) next to the N-terminus were observed to most influence the deviation of experimental mobility values from values predicted via classical models. The accuracy of this model suggests a priori selection of separation conditions for targeted analysis. More recently, Chen and co-workers reported a model to predict the electrophoretic mobility of phosphopeptides via CZE separation⁸⁶. According to this modeling data, phosphorylation reduces the charge of a peptide, thereby its electrophoretic mobility.

CE-MS has been used to study PTMs such as glycosylation. Glycosylation is an important and common PTM in biological systems, but its detailed characterization is challenging because of its heterogeneity in localization and forms^76–79. CZE separation has frequently been used to analyze glycosylated peptides and glycans^49,57,80,81 and CZE separation via TDMS was used for glycosylation characterization⁸². Post-translational phosphorylation plays critical roles in biological systems⁸³ and it can also influence the electrophoretic mobility of peptides. Because phosphate groups have 2 pK_as (acid and neutral) they can add one or two negative charges to the protein or peptide depending on the pH of the BGE. Faserl and co-workers demonstrated the value of CZE for the separation of 70 synthetic peptides that were post-translationally modified including phosphorylation, acetylation, methylation, and nitration⁸⁴. These authors observed that multi-phosphorylated peptides were poorly retained on the stationary phase of a reversed-phase LC-MS method, resulting in a decreased number of peptide identification relative to the CE-MS analysis. Ludwig and co-workers reported a CZE-MS approach that was able to identify over 2,300 phosphorylated peptides in a single analysis⁸⁵, thus illustrating the advantages of CZE separation over LC for PTM analysis. Many studies have suggested that CZE separations are more capable of identifying hydrophilic peptides than LC^48,87. This may be due to the LC gradient, which increases the proportion of the organic modifier throughout the analysis to detect peptides with different hydrophobicity, whereas CE conditions are basically continuous. Since there is a significant difference in size/charge between phosphorylated and unmodified peptides, CZE can achieve outstanding resolution.

Despite the high separation efficiency of CZE for proteins and peptides, separation in only one dimension is not enough to effectively reduce the complexity of proteomic samples prior to MS detection. Multidimensional separation can enhance peak capacity and lead to better separation of proteins or peptides before being introduced into the mass spectrometer. Better separation leads to enhanced ionization efficiency and reduced ion suppression. Qu and co-workers reported a LC-CZE strategy for the characterization of site-specific glycan heterogeneity where the use of CZE for analysis of reversed-phase LC glycopeptide fractions reduced the complexity of the sample and enabled glycan structural elucidation⁸⁸. The enhanced sensitivity of CZE in the second dimension is an added benefit for the analysis. Recently, Yang et al. reported the use of a reversed-phase LC-CZE approach for the identification of proteins in MCF7 proteome via bottom-up proteomics⁴⁷. The authors evaluated C18 ZipTip and nano reversed-phase LC fractionations prior to CZE-MS analysis. Analysis using nano reversed-phase LC combined CZE-MS resulted in the identification of 7,500 proteins and 60,000 peptides. These results demonstrated an increase in sensitivity over analysis after separation using C18 ZipTip only, and highlighted the advantages of using a LC-CZE-MS bottom-up approach. Jooss and co-workers demonstrated the value of on-line coupling of nano reversed-phase LC and CZE for analysis of intact proteins⁸⁹. In this study, a mechanical valve was used to couple LC with CE which was used to separate a protein mixture (RNAse B, cytochrome c, lysozyme c, and myoglobin).

McCool and co-workers demonstrated the value of orthogonal multidimensional separation for a large-scale top-down study that identified 850 proteins and 5700 proteoforms from Escherichia coli⁵¹. In this study, size exclusion chromatography (SEC) and reversed-phase LC were used to reduce the sample complexity prior to CZE-MS analysis. This platform resulted in a peak capacity of ~4,000 for separation of intact proteins. High peak capacity and resolving power improved the data acquisition, which resulted in an enhanced representation of the proteins in the sample.

Jooss and co-workers demonstrated the power of CZE-CZE-MS for the characterization of intact monoclonal antibody charge variants²³. In another report, Jooss et al. showed, for the first time, a successful application of CZE in conjunction with drift tube ion mobility-mass spectrometry (DTIM-MS) for the separation and characterization of native and APTS-labeled N-glycans released from different protein sources⁹⁰. This platform seems to have the potential to resolve isomeric forms of glycans. LC and CZE can also independently generate complementary identification of peptides, proteins, and proteoforms, thus expanding proteomic coverage. For instance, Li and co-workers reported a bottom-up strategy that combines the data from CZE-MS and reversed-phase LC-MS⁹¹. This combination resulted in an improved number of identified unique peptides where hydrophilic and basic peptides with low molecular weight were predominantly identified via CZE-MS.

Further advances in CE-MS technologies can enable the identification of many more cellular proteins and add to our understanding of protein interactions with other cellular species. ESI, in conjunction with high resolution mass spectrometers, enables the identification of protein non-covalent interactions with other proteins or small ligands in protein complexes. The analysis of proteins in complexes by CE–MS is poorly explored, with only a few publications in recent years. An early report for analysis of protein complexes via CZE-MS was described by Belov and co-workers³⁴. More recently, Shen and co-workers demonstrated the use of SEC-CZE for analysis of protein complexes and proteoforms under near native conditions²¹, resulting in the identification of 144 proteins, 672 proteoforms, and 23 protein complexes from the Escherichia coli proteome.

Another common CE separation technique used in proteomics research is cIEF. Although it offers high resolution separations, cIEF has yet to gain widespread use in the proteomics field. In cIEF separation is based on the isoeletric point of the analyte, which is mixed with an ampholyte before being placed in the capillary. A pH gradient in the capillary is initiated by the ampholyte in conjunction with an anolyte and a catholyte. The analytes then migrate through the capillary until they reach their isoelectric points. A limitation of this technique that the catholyte is sprayed into the MS during the ESI process, which can affect the background signal and cause ion suppression. Also, it is not compatible with a sheathless interface.

cIEF-MS with a sheath-liquid interface has been used for the analysis of monoclonal antibodies, with Dai et al. reporting the separation of intact monoclonal antibody⁹². Resolving intact charge variant peaks is a noted challenge, but using this approach, nine charge variants were resolved. In another report, Dai and Zhang used cIEF with a middle-down strategy to reduce sample complexity to achieve characterization of the charge heterogeneity for the same monoclonal antibody⁹³. cIEF has been also reported for the analysis of other intact proteins⁹⁴.

C. Background Electrolyte

BGE selection is critical to ensure successful CE separation and MS sensitivity. In addition to volatility, an effective BGE varies one (±) pH unit around the pK_a value. In CE-MS applications, the type of interface should also be taken into consideration while choosing a BGE. BGE selection is most critical for a sheathless interface because the BGE is the only liquid present during the ESI process and acidic pH is recommended to facilitate protein or peptide ionization efficiency. Conversely, sheath-liquid interface is more flexible and an acidic pH can be obtained from the sheath-liquid reservoir, so the BGE pH range can vary between 1 and 8 units. Compatibility between the BGE and sheath-liquid is of key importance to avoid moving boundaries in the separation medium that can adversely affect the separation efficiency. Table 2 shows the most commonly used BGEs for CE-MS applications in proteomics research^{46,47,95–98}.

Table 2.

The most commonly used chemicals for BGE preparation in proteomics research.

Chemicals	pKa	Buffer range	Formula	pH adjustment (acid/base)
Acetic Acid	4.80	----	CH3COOH	----
Formic Acid	3.80	----	HCOOH	----
Ammonium acetate pKa 1	4.76	3.76 – 5.76	CH3COONH4	CH3COOH or NH4OH
Ammonium acetate pKa 2	9.20	8.20 – 10.20	CH3COONH4	CH3COOH or NH4OH
Ammonium bicarbonate pKa 1	6.35	5.35 – 7.35	NH4HCO3	HCOOH or NH4OH
Ammonium bicarbonate pKa 2	9.20	8.20 – 10.20	NH4HCO3	HCOOH or NH4OH
Ammonium bicarbonate pKa 3	10.30	9.30 – 11.30	NH4HCO3	HCOOH or NH4OH
Ammonium formate pKa 1	3.80	2.80 – 4.80	NH4COOH	HCOOH or NH4OH
Ammonium formate pKa 2	9.20	8.20 – 10.20	NH4COOH	HCOOH or NH4OH
Ammonium hydroxide	9.20	----	NH4OH	----

The electrophoretic mobility of peptides is influenced by the BGE pH. For instance, in a pH lower than ~3.5, migration is predominantly influenced by the basic residues. Acidic pH is especially useful in peptide and protein analysis using uncoated fused silica capillaries. At low pH the free silanols in the capillary wall are protonated and the separation of proteins and peptides is enhanced due to a reduction of the electroosmotic flow (EOF) and reduction of electrostatic interactions. A strategy for large scale top-down analysis using lysate from Escherichia coli was developed by Lubeckyj and co-workers⁶⁹. In this approach, the sheath-liquid was composed of 0.2% formic acid and 10% methanol in water. The authors compared different BGE compositions (acetic acid (5−10%)) in water and formic acid ((0.1−0.5%) in water). A BGE composed of 0.1% formic acid (pH 2.8) resulted in peak intensities equivalent to the acetic acid BGEs, as illustrated in the Figure 5.

Nevertheless, acetic acid BGEs (pH ~ 2.4 and ~2.2) resulted in a wider separation window. The use of acetic acid also increased the migration time of the proteins, which might be attributed to the lower pH of acetic acid that resulted in an increase in hydrodynamics in protein radii, exacerbation of protein unfolding, and reduction of EOF.

Interestingly, addition of organic solvents (e.g., propan-2-ol) to an acidic BGE enhances the separation efficiency and MS detection, as demonstrated by Yates group in a top-down analysis of Pyrococcus furiosus proteins using a sheathless interface⁵⁹. The addition of organic solvent to the BGE reduces the EOF, which results in improved resolution.

Dada and co-workers evaluated a BGE composed of aqueous−aprotic dipolar solvent (N,N-dimethylacetamide (DMA) and N,N-dimethylformamide (DMF)) for peptide mapping⁹⁹. The use of this BGE dramatically enhanced the separation efficiency over conventional BGEs composed of acetic acid diluted in water. The improvement in separation efficiency might be attributed to the influence of DMA on the electrophoretic mobility of the peptides and EOF.

Ammonia salts (acetate and bicarbonate) have been used in CZE native separations. As can be observed in Table 2, these salts (acetate, bicarbonate, and formate) have more than one pK_a, which allow their use in more than one pH range. Ammonia salts can achieve a pH compatible with physiological conditions (pH 6.8–7.4), which is of key importance to probe structures of cellular proteins in complexes. In native ESI-MS, protein complexes are transferred from solution to the gas phase and environmentally induced changes in pH can perturb their tertiary and quaternary conformations while in solution. Ammonium acetate can achieve a pH between 6.8 and 7.4 when dissolved in water and despite a low buffering capacity (ammonium acetate buffers in the pH ranges of 3.8–5.8 and 8.2–10.20, see Table 2), it is often used in native MS. Since ammonium acetate does not constitute a buffer system that can effectively maintain macromolecular complexes in solution at physiological pH conditions^81,100, it suggests that an ammonium acetate solution in the pH range of 6.8 to 7.4 becomes acidic during the ESI process. Ammonium bicarbonate has been evaluated as an alternative to ammonium acetate for native MS applications due to its ideal pH stabilization at physiological conditions. Hedges and co-workers investigated the behavior of ammonium bicarbonate in protein ESI-MS using holomyoglobin¹⁰¹. This protein was chosen as a model due to its rapid unfolding in solvent environmentally induced changes. As illustrated in Figure 6, high charge states (as in denaturing conditions) were formed in native conditions and their signal intensities increased proportionally to the increase in buffer ionic strength.

Figure 6.

ESI mass spectra of myoglobin in aqueous solution at pH 7. The salt concentrations (ammonium acetate and ammonium bicarbonate) increase from A to C and from D to F. Reprinted from reference 101, with permission¹⁰¹.

This phenomenon might be attributed to the destabilization effects that are induced by the anion bicarbonate, exacerbating the propensity of the protein to unfold during ESI process.

Interestingly, Bertoletti and co-workers reported a CE-ESI-MS sheathless interface strategy for the separation and characterization of the intact protein folding conformers of Beta 2-microglobulin using ammonium bicarbonate¹⁰². Using this set up, Bertoletti et al. were able to preserve protein structure in native conditions and they did not observe the high charge states that are promoted by ammonium bicarbonate, which might be attributed to the lower molar concentration of ammonium bicarbonate in the BGE. Although a BGE with high ionic strength is useful to avoid band broadening, high ionic strength can increase the diffusion of the sample zone and lead to high current and temperature in the separation capillary. The compositional and site-specific assessment of multiple peptide-deamidation was investigated by Dominguez-Vega and co-workers¹⁰³. In this approach, a mixture of 150mM ammonium formate (pH 6.0)-propan-2-ol -acetonitrile was used as a BGE. The addition of 40% of acetonitrile-propan-2-ol (87.5:12.5) to 60% 150mM ammonium formate (pH 6.0) significantly improved the separation of deamidated and deacetylated TRI-1144 species and positional isomers. Acetonitrile might account for the fast separation and propan-2-ol may provide better resolution of the degradation products.

D. Capillary

Theoretically, since proteins have low diffusion coefficients, they should produce very high separation efficiency in CE. However, hydrophobic and/or ionic interactions between protein and capillary wall surfaces can lead to increased protein retention in the capillary, or to a lesser extent, peak tailing. Coated capillaries can be used to limit protein adsorption on the capillary wall. Although non-permanent coatings such as UltraTrol LN can be used, permanent coatings are preferred, particularly for high throughput analysis and when interfacing CE with a mass spectrometer. A capillary coating can be achieved using polyacrylamide, aryl pentafluoryl groups, polysaccharides, or polyamines. Examples of coatings are described in Table 3.

Table 3.

Most commonly used coatings in proteomics research.

Coating	Charge
Linear polyacrylamide (LPA)	Neutral
Fluorocarbon (FC)	Neutral
Polyvinyl alcohol (PVA)	Neutral
Polyethylenimine (PEI)	Cation
Polybrene-dextran sulfate-polybrene (PB-DS-PB)	Cation
Bovine serum albumin (BSA)	Cation
Poly(arginine) (PA)	Cation

EOF suppression or reversal can result in higher separation efficiency, and the EOF can be regulated by the capillary coating. Neutral coating with polyacrylamide is widely used in proteomics research due to its ability to eliminate EOF and maintain effective CE separations. The electropherograms in Figure 7 illustrate a separation of proteins using uncoated and neutral (LPA) coated capillaries.

Figure 7.

Elution profiles of (1) cytochrome c, (2) lysozyme, (3) ribonuclease A, and (4) chymotrypsinogen A on uncoated and LPA coated capillaries at the same electrophoretic conditions. Reprinted from reference 104, with permission¹⁰⁴.

Belov and co-workers investigated both bare fused silica and neutral coated capillaries under native conditions. The bare fused silica capillary produced a marginal separation efficiency for intact proteins and protein complexes as a result of protein adsorption to the capillary wall, whereas the neutral coated capillary generated high separation efficiency by preventing protein adsorption³⁵. Haselberg et al. reported almost zero EOF with the use of a neutral capillary for the analysis of intact proteins¹⁰⁵. In another report, a neutral coating was successfully used for analysis of conformers and dimers of antithrombin¹⁰⁶. Acidic N-linked glycans were also successfully analyzed using a neutral coated capillary¹⁰⁷.

The Dovichi group has described a method using polyacrylamide to produce a neutral coated capillary¹⁰⁸ which was evaluated for the analysis of four standard proteins (cytochrome c, myoglobin, beta-lactoglobulin, and carbonic anhydrase), intact antibody, and tryptic digested peptides. These studies showed that high reproducibility with a relative standard deviation for migration time in separation window lower than 1% was achieved.

A capillary with a polyethylenimine (PEI) positively charged coating has also been used for both protein and peptide separations. This coating generally results in high separation efficiency with high number of plates due to its ability to create a positively charged surface on the capillary and reverse the EOF direction. The use of positively charged coatings requires the application of negative voltage on the inlet (reversed polarity); as a result, proteins and peptides will migrate in the opposite direction of the EOF. A potential limitation of this coating is the generation of a very high flow-rate in the separation capillary, which becomes more problematic as BGE pH is reduced. Han et al. demonstrated the high separation efficiency of PEI coated capillaries for the analysis of protein complexes and Pyrococcus furiosus proteins via TDMS^59,60.

Other cationic polymers such as PA and polybrene have also been investigated for CE-MS applications¹⁰⁹. The Kelleher group demonstrated the use of a PA coated capillary for the separation of intact proteins with molecular masses ranging from 30 to 80 kDa. In this study, 30 proteins were identified from P. aeruginosa PA01 whole cell lysate¹¹⁰. A procedure for the production of polybrene coating has been reported by Dominguez-Vega and co-workers¹⁰³. The capillary was evaluated for the analysis of compositional and site-specific assessment of multiple peptide-deamidation where fast and reproducible separations were achieved. A method to maximize the elimination of SDS interference in antibody separations was reported by Sanchez-Hernandez and co-workers¹¹¹. In this approach, effective removal of SDS was demonstrated in both neutral and positively charged coated capillaries. A summary of recent CZE-ESI-MS based methods in proteomics research is shown in Table 4.

Table 4.

Summary of recent CZE-ESI-MS based methods in proteomics research.

Sample	BGE	Capillary	Separation	Interface	Peptide	Protein	Proteoform	Protein Complex	MS-based Method	Ref.
Escherichia coli	5−10% acetic acid	LPA	CZE	Sheath-liquid	----	200	600	----	Top-down	69
Escherichia coli	10% acetic acid	LPA	SEC-LC-CZE	Sheath-liquid	----	850	5700	----	Top-down	51
Pyrococcus furiosus	0.1% acetic acid- 20% IPA	PEI	CZE	Sheathless	----	134	291	----	Top-down	60
Escherichia coli	40 mM ammonium acetate, pH 7.5	LPA	CZE	Sheathless	----	42	144	8	Native MS	35
Escherichia coli	50 mM ammonium acetate, pH 6.9	LPA	SEC-CZE	Sheath-liquid	----	144	672	23	Native MS	21
Yeast	10 mM ammonium acetate (pH 7	Uncoated	CZE	Sheath-liquid	----	49	----	----	Bottom-up	112
MCF7 proteome	5% acetic acid	LPA	LC-CZE	Sheath-liquid	60,000	7500	----	----	Bottom-up	47
Mouse brain proteome	5% acetic acid	LPA	SCX-LC-CZE	Sheath-liquid	65,000	8200	----	----	Bottom-up	113
Xenopus laevis	5% acetic acid	LPA	LC-CZE	Sheath-liquid	22,000	4100	----	----	Bottom-up	114
Monoclonal antibody charge variants	2M acetic acid	PVA	CZE-CZE	Sheath-liquid	----	----	----	----	Intact analysis	23
Therapeutic proteins	15−20% DMA or DMF in 20%acetic acid or 2% formic acid	LPA	CZE	Sheath-liquid	----	----	----	----	Peptide mapping	99
Therapeutic antibodies	10% acetic acid	PVA	CZE	Sheath-liquid	----	----	----	----	Peptide mapping	115
Monoclonal antibody	40 mM Ammonium acetate, pH 7.5 (native MS) 50% methanol, 1% formic acid (middle-down)	LPA	CZE	Sheathless	----	----	----	----	Native MS, middle-down	116
Monoclonal antibodies	0.1% formic acid	LPA	CZE	Sheath-liquid	----	----	----	----	Top–down	68
Monoclonal antibodies	5–30% acetic acid	LPA	CZE	Sheath-liquid	----	----	----	----	Intact analysis	22
Trastuzumab	10% acetic acid	Uncoated	CZE	Sheathless	----	----	----	----	Bottom-up	117
Biopharmaceutical proteins	3% acetic acid	PEI	CZE	Sheathless	----	----	----	----	Intact analysis and top-down	11
Beta2-microglobulin	(50 mM ammonium bicarbonate, pH 7.4	Uncoated	CZE	Sheathless	----	----	----	----	Intact analysis	102
Mixture (9 peptides)	0.1 M acetic acid	Uncoated	CZE	Sheathless	----	----	----	----	Intact analysis	62

E. Preconcentration

The loading capacity of CE is very limited, which can adversely affect the sensitivity of the method and reduce peptide and protein identifications. Many preconcentration techniques to enhance injection volumes in CE-MS applications have been described, based on either electrophoretic (stacking) or chromatographic (solid-phase extraction (SPE)) principles. Recently, a dynamic pH junction has been used in top-down and bottom-up proteomics. This stacking technique relies on the preparation of samples in a basic buffer and injection (with pressure) into the separation capillary. Under application of an electric field, negatively charged proteins or peptides migrate toward the proximal tip of the capillary and are neutralized by an acidic BGE. A focused plug is formed and target analytes become positively charged, migrating towards the ESI source^25,118.

McCool and co-workers described a protocol for top-down analysis of complex samples using a dynamic pH junction where a microliter-scale sample-loading volume was achieved⁵². Zhu and co-workers described the use of dynamic pH junction in bottom-up proteomics¹¹⁹. In this study, a pH junction injection of 130 nL of BSA resulted in the identification of 40 peptides and produced 70% coverage. A 400 nL pH junction injection of an Escherichia coli sample enabled the identification of 527 peptides and 179 proteins. The authors found that reproducibility of the migration times between the pH junction and normal injection was good. The pH junction was further extended to three intact standard proteins where the peak intensity increased more than 10-fold compared to conventional injection.

Transient isotachophoresis (tITP) is a less common stacking approach for preconcentration of peptides and proteins in proteomics research. In this strategy, the sample is mixed with two electrolytes, one with very high mobility (leading) and one with very low mobility (terminating). These electrolytes generate a highly conductive zone when separation voltage is applied. Proteins or peptides are then concentrated into zones between the leading and terminating electrolytes. tITP requires an optimization of both electrolyte concentrations and loading volume to enable efficient stacking of the target analytes^{13,27,65,120–122}.

Sample preconcentration using a chromatographic principle such as SPE was used by Wang et al. where large sample volumes can be trapped on reversed-phase media and eluted using an elution buffer prior to CE separation¹²³. This technology has the advantage that it can be automated and the multistep elution CE can be used as a semi-two-dimensional separation method. Wang et al. used reversed-phase SPE in conjunction with tITP for sample concentration of a Pyrococcus furiosus tryptic digest, which resulted in low background noise and high sensitivity⁸⁷. The Dovichi group synthetized a sulfonate–silica hybrid strong cation-exchange (SCX) monolith at the inlet of the separation capillary and used it for on-line solid-phase extraction (SPE) sample preconcentration in conjunction with a dynamic pH junction for the analysis of tryptic digest peptides¹²⁴. This approach enabled injections at μL levels with a separation efficiency of over 25,000 theoretical plates. In another example, Zhang and co-workers synthetized a SCX monolith in a fused silica capillary and coupled it to a neutral coated capillary. This was used for on-line SPE preconcentration in conjunction with dynamic pH separation for Escherichia coli samples via bottom-up proteomics and the results were compared to a conventional LC-MS treatment of the same sample¹²⁵. There was no significant difference in the number of peptides and proteins identified by CZE-MS and LC-MS methods. The same group further eliminated the zero dead volume union by preparing SCX and the neutral coated capillary sequentially in a single fused silica capillary, which resulted in an enhanced protein and peptide identifications when compared to previous reports¹²⁶.

F. Quantitative Analysis

In proteomics research, quantitative information is critical to gain insights into clinical disorders. Quantification via CE-MS is particularly valuable for clinical studies where samples are often obtained in limited quantities. Several bottom-up strategies are available for protein quantification, including stable isotope labeling by amino acids in cell culture (SILAC), isobaric tag for relative and absolute quantitation (iTRAQ), label-free, and isotope-coded affinity tags (ICAT). Faserl and co-workers reported a SILAC approach that resulted in a quantification of 1371 phosphopeptides by CE-MS, without using any enrichment strategy¹²⁷. Varadi and co-workers described a two-plex method for quantitative analysis of N-glycans via CE-MS¹²⁸. The CE-MS method was found to be linear and precise. It was further applied for analysis of a therapeutic protein where the authors observed differences in N-glycan structures from two different lots. The low sample consumption of CE enabled them to perform quantitative studies of the same sample via LC-MS. In terms of quantitative performance, CE-MS was found to be similar to LC-MS. However, the LC-MS platform failed to resolve N-glycan isomers that were resolved via CE-MS. In another report, Mittermayr and co-workers described a protocol for robust and reproducible quantification of N-glycans labeled using ^12/13C₆ aniline or 2-aminobenzoic acid by CE-MS¹²⁹. Lombard-Banek et al. reported a CE-MS approach for the quantification of proteins in single embryonic cells via bottom-up proteomics¹³⁰. In this study, the quantification of ~150 non-redundant protein groups revealed significant translational cell heterogeneity along multiple axes of the embryo at very early stage of development. More recently, the same research group reported the use of CE-MS in conjunction with label-free single-cell proteomics for quantification ~ 750 protein groups in 16-cell Xenopus laevis (frog) embryo with about 5ng of protein digest¹³¹.

Guo et al. reported quantification of a peptide mixture by CE-MS⁶². In this approach, selected reaction monitoring was used to target the analytes. The method was highly reproducible and robust and provided reliable quantification of a peptide mixture with a coefficient of variation (CV) between 0.2 and 1.8% for the migration time. Wang et al. described the use of cITP/CZE for quantification of targeted peptides of a tryptic digest BSA. They observed a quantitation limit of 10 pM with a CV lower than 22%¹²¹. Sun and co-workers described a quantitative parallel reaction monitoring of peptide abundance where the relative standard deviation (RSD) of the migration time was lower than 4% and the RSD of the fragment ion was approximately 20%¹³². Zhong and co-workers reported a quantitative analysis of glycans labeled with multiplex parbonyl-reactive tandem mass tags¹³³.

Top-down analysis provides quantitative information on diverse proteoforms that are often missed in bottom-up approaches. In quantification using CE-MS via TDMS, the intensities of the peaks corresponding to unmodified and modified proteins can be compared to obtain the abundance of a proteoform. Bush et al. characterized and quantitated individual molecular proteoforms of interferon-β1¹¹. Haselberg and co-workers used CE-MS for analysis of intact glycoforms¹³⁴. Quantitative information was based on the relative intensities of the detected glycoforms where glycoform concentrations were estimated to range between 0.35 to 630 nM. In a TDMS study, Lubeckyj et al. demonstrated a label-free strategy for the quantification of thousands of proteoforms from zebrafish brain using CZE-MS¹³⁵. This quantitative approach revealed that the abundance of proteoforms differs significantly in different regions of the zebrafish brain.

IV. Conclusions and Future Directions

In the past few years advances in MS technologies, along with the need for separation techniques with high speed and resolving power, have renewed interest in the use of CE-MS for proteomics research. Improvements in interfaces have enhanced the robustness and reproducibility of CE-MS methods, resulting in highly confident identification and characterization of peptides, cellular proteins, PTMs, and protein complexes. These improvements have also enhanced sensitivity, enabling the application of CE to complex and low concentrated samples. Despite susceptibility to more technical difficulties and issues than LC-MS, with improvements in instrumentation CE-MS has, in some applications, outperformed conventional LC-based methods (e.g., separation of intact proteins and protein complexes). Highly sensitive ionization efficiency can be achieved using both sheathless and sheath-liquid interfaces at lower flow-rate (in the separation capillary) than LC, which reduces background noise and increases MS detectability. We suggest the use of sheathless interface for TDMS and native MS due to its unparallel sensitivity. The state-of-the-art single shot CZE-ESI-MS, along with neutrally coated capillaries (~1 meter), have substantially enhanced the peak capacity of CE-MS methods in proteomics research. No single dimension, however, can provide complete reduction of sample complexity. Rather, utilization of multidimensional separations is the most effective strategy for improving peak capacity and reducing sample complexity. Sample loading is the major drawback of CE, which can be addressed using simple protocols, as demonstrated in top-down and bottom-up applications^52,87. The use of CE-MS for quantitative proteomics, TDMS, and native MS is exceedingly low, with only a few publications over the past few years. Cross-linking and MS have been successfully used to interrogate protein structures; however, cross-linking reports by CE-MS are scarce. The Dovichi group is leading the technological advances to realize the hope that CE-MS can be as routinely used in proteomics research as LC-MS.

Given that CE is well-suited for native separations, applications in native MS will expand to interrogate protein dynamics and conformation. In addition to a detailed proteoform characterization, TDMS can provide structural information of proteins covalently attached to other proteins while native MS provides a detailed information on protein subunit arrangements via non-covalent interactions. These individual features suggest that TDMS and native MS can be complementary at several levels and can be merged to provide comprehensive protein structural information. The combination of these MS-based methods will yield fruitful benefits for both structural biology and proteomics research.

The future technological developments in CE-MS applications will continue, influenced in part by the proteomics field. For instance, microchip CE-MS analysis of a bovine hemoglobin pepsin digest has demonstrated high peak capacity and speed when compared to a conventional LC separation of the same sample¹³⁶. The use of microchip CE-MS has also been demonstrated for the separation of intact monoclonal antibody variants where fast analysis was achieved (~ 4 min)¹³⁷. Another example is capillary electrochromatography that has been successfully applied for the separation of peptides^138,139. Although further development of these CE-based systems is still required, it is clear they are promising for separation of proteins and peptides and will be a step forward in proteomics research.

Acronyms

BGE

Background electrolyte

BSA

Bovine serum albumin

CE-MS

Capillary electrophoresis-mass spectrometry

cIEF

Capillary isoelectric focusing

cITP

Transient capillary isotachophoresis

CV

Coefficient of variation

CZE

Capillary zone electrophoresis

DMA

N,N-dimethylacetamide

DMF

N,N-dimethylformamide

DTIM-MS

Drift tube ion mobility-mass spectrometry

ESI

Electrospray ionization

EOF

Electroosmotic flow

FC

Fluorocarbon

ID

Identification

LC-MS

Liquid chromatography-mass spectrometry

LPA

Linear polyacrylamide

MS

Mass spectrometry

native MS

Native mass spectrometry

PA

Poly(arginine)

PB-DS-PB

Polybrene-dextran sulfate-polybrene

PEI

Polyethylenimine

PTMs

Posttranslational modifications

PVA

Polyvinyl alcohol

RSD

Relative standard deviation

SCX

Strong cation-exchange

SEC

Size exclusion chromatography

SPE

Solid-phase extraction

TDMS

Top-down mass spectrometry

tITP

Transient isotachophoresis