A simple and efficient approach for preparing cationic coating with tunable electroosmotic flow for capillary zone electrophoresis-mass spectrometry-based top-down proteomics

Abstract

Background

Capillary zone electrophoresis-tandem mass spectrometry (CZE-MS/MS) has become a valuable analytical technique in top-down proteomics (TDP). CZE-MS/MS-based TDP typically employs separation capillaries with neutral coatings (i.e., linear polyacrylamide, LPA). However, issues related to separation resolution and reproducibility remain with the LPA-coated capillaries due to the unavoidable non-specific protein adsorption onto the capillary wall. Cationic coatings can be critical alternatives to LPA coating for CZE-MS/MS-based TDP due to the electrostatic repulsion between the positively charged capillary inner wall and proteoform molecules in the acidic separation buffer. Unfortunately, there are only very few studies using cationic coating-based CZE-MS/MS for TDP studies.

Results

In this work, we aimed to develop a simple and efficient approach for preparing separation capillaries with cationic coating, i.e., poly(acrylamide-co-(3-acrylamidopropyl) trimethylammonium chloride [PAMAPTAC]) for CZE-MS/MS-based TDP. The PAMAPTAC coating-based CZE-MS produced significantly better separation resolution of proteoforms compared to the traditionally used LPA-coated approach. It achieved reproducible separation and measurement of a simple proteoform mixture and a complex proteome sample (i.e., a yeast cell lysate) regarding migration time, proteoform intensity, and the number of proteoform identifications. The PAMAPTAC coating-based CZE-MS enabled the detection of large proteoforms (≥30 kDa) from the yeast cell lysate reproducibly without any size-based prefractionation. Interestingly, the mobility of proteoforms using the PAMAPTAC coating can be predicted accurately using a simple semi-empirical model.

Significance

The results render the PAMAPTAC coating as a valuable alternative to the LPA coating to advance CZE-MS-based TDP towards high-resolution separation and highly reproducible measurement of proteoforms in complex samples.

Graphical Abstract

Introduction

Top-down proteomics (TDP) aims to characterize diverse intact proteoforms resulting from alternative splicing, post-translational modifications (PTMs), and proteolytic cleavages in complex biological samples and to discover proteoform biomarkers for various diseases [1–4]. It has been noted that the effective separation of proteoforms prior to mass spectrometry (MS) is critical to reduce the complexity and capture the lower abundant species [5,6]. Capillary zone electrophoresis-tandem mass spectrometry (CZE-MS/MS) has been more and more recognized as a valuable method for TDP because of its low sample consumption, high peak capacity, and high sensitivity for various proteoform samples [4,7–10].

Coatings on the capillary inner wall are crucial for CZE-MS/MS-based TDP to maintain high separation efficiency and wide separation windows. Neutral coatings (e.g., linear polyacrylamide, LPA) and positively charged coatings (e.g., polyethyleneimine, PEI) have been utilized in CZE-MS-based analyses of large biomolecules (i.e., proteoforms) [10–14]. The silanol group on the inner wall of the bare fused silica capillary carries a negative charge at pH above 2, causing a flat electroosmotic flow (EOF) due to the electrical double layer forming on the surface [15]. Commonly used linear polyacrylamide (LPA) neutral and hydrophilic coating minimizes the EOF and reduces the static adsorption, thereby enabling a high separation efficiency and wide separation windows [10, 16–18]. However, in our most recent study, we discovered that LPA coating still has significant protein adsorption, which leads to the reproducibility issue in long-term TDP measurements [19]. There is an urgent need to investigate new capillary coatings to improve the separation reproducibility and separation resolution of proteoforms by CZE. The positively charged coating could generate a countercurrent EOF while analytes are separated by their charge-to-size ratios under a strong electric field [20]. The cationic coating can benefit the separation resolution and prevent the non-specific adsorption of positively charged proteoforms by Coulombic repulsion force [21]. Therefore, positively charged coatings can be valuable alternatives to the LPA coating to further advance CZE-MS/MS-based TDP.

Positively charged coatings can be dynamic or covalent [21–23]. The dynamic coating is quick implementation and easy to remove. However, the additives in the background electrolyte (BGE) usually have a lower compatibility coupling with MS [23–25]. Permanent coatings and semi-permanent, such as polyethyleneimine (PEI), polybrene (PB), and successive multiple ionic-polymer layer (SMIL), maintain reproducibility of migration time and a stable EOF [26–31]. However, very few MS-based intact protein studies have employed positively charged coating-based CZE-MS [32–37]. Much more effort is needed to investigate the positively charged coatings (i.e., SMIL)^27,31 for CZE-MS-based TDP of complex biological samples. Here, we systematically studied the performance of one positively charged coating for CZE-MS/MS-based TDP. The positive charge coating is based on one monomer carrying a permanent positive charge, i.e., (3-acrylamidopropyl) trimethylammonium chloride (APTAC). The monomer APTAC has been used in one recent study to prepare positive charge coating for CZE-UV and CZE-MS analyses of basic small molecular drugs [38]. In that study, it has been demonstrated that co-polymerization of acrylamide and APTAC can make a tunable EOF for CZE separation, and the EOF is stable in pH 2~10. In this study, we refined the procedure of preparing the co-polymer of poly(acrylamide-co-(3-acrylamidopropyl)trimethylammonium chloride) (PAMAPTAC) and implemented it in CZE-MS/MS-based TDP. The stability and reproducibility were well-evaluated in the study. The enhanced separation resolution of the capillary indicates its strong potential for the identification of low-abundance, large, and hydrophobic proteoforms. The variable charge density of the cationic coating provides the possibility for broad biological applications.

Experimental Section

Materials and Chemicals

Bare fused silica capillaries (50-μm i.d., 360-μm o.d.) were purchased from Polymicro Technologies (Phoenix, AZ). Ammonium persulfate, ammonium bicarbonate (NH₄HCO₃), Dulbecco’s phosphate buffered saline, 3-(Trimethoxysilyl)propyl methacrylate and Amicon Ultra centrifugal filter units (0.5 mL, 30 kDa cut-off) were purchased from Sigma-Aldrich (St. Louis, MO). Hydrofluoric acid, acrylamide, and LC/MS grade water, methanol, acetic acid, and formic acid were purchased from Fisher Scientific (Pittsburgh, PA). Urea was bought from Alfa Aesar. Protease inhibitor (cOmplete ULTRA Tables, Roche) and phosphatase inhibitor (PhosSTOP, Roche) was bought through Fisher Scientific.

Sample preparation

A standard protein mixture (accumulative concentration 1.25 mg mL⁻¹: 0.05 mg mL⁻¹ ubiquitin, 0.2 mg mL⁻¹ myoglobin, 0.3 mg mL⁻¹ carbonic anhydrous, and 0.7 mg mL⁻¹ bovine serum albumin) was prepared in the BGE solution of 5% acetic acid (pH 2.4).

Around 50 g of baker’s yeast (Saccharomyces cerevisiae, strain ATCC 204508/S288c) was added in 1 L of yeast extract peptone dextrose (YPD) medium (autoclaved) and cultured at 37 °C (300 rpm shaking) overnight in an incubator shaker (Thermo Scientific MaxQ 4000). The yeast was harvested by centrifugation (3,000 g, 5 min), followed by washing with Dulbecco’s phosphate buffered saline (dPBS) for three times. Two grams of the yeast pellet was resuspended in 10 mL of lysis buffer containing 8 M urea, 100 mM ammonium bicarbonate (ABC, pH 8.0), complete protease inhibitor and phosphatase. The cells were lysed (3 min, 3 times) using a homogenizer 150 for (Fisher Scientific) and sonicated on ice (3 min, 5 times) with Branson Sonifier 250 (VWR Scienfic). The concentration was determined using a bicinchoninic acid (BCA) kit.

Around 200 μg of yeast lysates was centrifuged on an Amicon 30 kDa molecular cut-off filter at 14, 000 g for 20 min. The proteins retain inside of the filter was washed two times with 100 mM ABC, two times with water, and two times with 5% AA. A small amount of precipitation was noticed during acidifying. Around 30 μL solution was collected after buffer exchange (1.4 mg mL⁻¹). The protein concentration was measured by Bradford assay. Prior to CE separation, the sample was centrifuged at 10,000 g for 3 min to remove potential precipitates.

Capillary preparation

To prepare the capillary for coating, the bare fused silica capillaries (1 m length, 50 μm i.d., 360 μm o.d.) were flushed with 300 μL of 1 M sodium hydroxide (NaOH), water, 1 M hydrochloric acid (HCl), water and methanol, successively. After degassing with nitrogen, 50% (v/v) 3-(trimethoxysiyl) propyl methacrylate in methanol was introduced into the capillary. Both ends of the capillary were sealed and the capillaries were incubated at room temperature for three days grafting double bonds. The capillaries were then rinsed by methanol and dried under nitrogen.

The linear poly acrylamide (LPA) coating was prepared based on our previous procedure with minor modifications [17]. Briefly, 500 μL of 4% (w/v) acrylamide was mixed with 3.5 μL of 5% ammonium persulfate (APS). The solution was degassed with nitrogen for 15 minutes and then infused into the capillary using a vacuum. Both ends of the capillaries were sealed. The capillaries were incubated in 50 °C water bath for 1 hour. After incubation, the solution was pushed out using 200 μL of water. Note: the polymer pushed out of the capillary is gel-like sticky.

Three 50% PAMAPTAC-coated capillaries were made. The experimental condition is based on the literature [38], while the initiator was changed to APS. For copolymer preparation, 1 mL of APTAC/acrylamide monomer containing 25 or 50 mol.% APTAC was prepared. For the 50% APTAC, a 500 μL of 0.7 mol/L acrylamide was mixed with 500 μL of 0.7 mol/L APTAC and 7 μL of 5% (w/v) APS in a 1.5 capped vial. By drilling a hole on the cap, the solution was degassed with nitrogen for 15 minutes, and then infused into the capillary using a vacuum. Both ends of the capillaries were sealed, and the capillaries were incubated in a 66 °C water bath for 1.5 hours. To note: APS generates the radical at a high temperature and in a basic condition [39,40]. Given the weak basic of the ammonium ion hydrolysis (pH 8.6 for the coating solution), it is suggested to conduct the coating step in a time-sensitive manner and in a completed degas condition. After incubation, the solution was pushed out using 200 μL of water. Note: the polymer pushed out of the capillary is gel-like sticky.

One end of the coated capillary was etched with hydrofluoric acid (HF) for 85 minutes to reduce the outer diameter to around 70 μm. The reduction of capillary outer diameter reduced the distance between the separation capillary end and the spray emitter orifice of CE-MS interface used in this study, minimizing the sample diffusion in the spray emitter and improving the sensitivity [41]. A detailed video procedure can be accessed from our published work [42].

CZE-ESI-MS/MS

A Beckman CESI 8000 capillary electrophoresis autosampler (Sciex) was coupled to the mass spectrometer by an in-house-constructed electrokinetically pumped sheath-flow CE-MS nanospray interface [43,44]. The electrospray emitter was a borosilicate glass capillary (1.0 mm o.d., 0.75 mm i.d.) pulled with a Sutter instrument P-1000 flaming/brown micropipette puller with an orifice size 20 ~30 μm and a length of 5 cm. The sheath liquid consists of 0.2% (v/v) formic acid and 10% (v/v) methanol. The spray voltage is around + 2.0 kV and the distance between the spray emitter orifice and the mass spectrometer entrance was ~2.8 mm.

The inlet of the capillary was inserted to the background electrolyte of CZE, 5% (v/v) acetic acid (pH 2.4). For all experiment, a 50 nL injection was carried by applying 5 psi pressure for 9.5 s, based on the Poiseuille’s law. For the separation of standard protein mixtures, + 30 kV was applied at the injection end of the LPA coated capillary for 30 minutes. And – 30 kV was applied at the injection end of PAMAPTAC coated capillary for 40 minutes. A 10-psi pressure was also applied during the last ten minutes of the runs. For the yeast lysate, the separation time was extended to 60 minutes. A 10-minute flush with 10 psi and +/− 30 kV was used to clean up the capillary.

Most of the experiments were conducted using an Orbitrap Exploris 480 mass spectrometer except the cross-capillary of yeast lysate, which was conducted on the Q-Exactive HF mass spectrometer. The MS parameters are listed in Table 1. The ion transfer tube temperature was 320 °C. The intact protein mode was turned on and the low-pressure mode was selected. Dynamic exclusion was enabled with the following settings: repeat count as 3, exclusion duration as 30 s and the exclusion of isotopes was enabled. For Low-high condition, an additional 5V source fragmentation energy was enabled.

Table 1.

Summary of the MS parameters used in this study.

Instrument	Scan	Resolution	m/z range	AGC target	Maxi Injection time	Microscan	Data-dependent acquisition
Exploris 480 (High-high)	Full MS	480k	500–3200	3E6	Auto	1	Top 6, intensity threshold >1E4, charge state 5–60, 2 m/z isolation window, normalized HCD energy of 25%
	MS/MS	120k	200–2000	1E5	Auto	3
Exploris 480 (Low-high)	Full MS	7.5k	500–2500	3E6	Auto	10	Stepped normalized HCD energies of 25, 35, and 45%. Others were the same as “High-high”.
	MS/MS	120k	200–2000	1E5	Auto	2
Q-Exactive HF	Full MS	120k	600–2000	3E6	100 ms	3	Top 5, normalized HCD energy of 20%. Others were the same as the Exploris 480.
	MS/MS	60k	200–2000	1E6	200 ms	1

Data Analysis

The standard protein mixture was analyzed using Xcalibur software (Thermo Fisher Scientific) for the intensity, migration time, and full width at half maximum of proteoforms. For complex samples analyzed by low-high mode, the RAW data was deconvoluted using UniDec to detect the large proteoforms [45].

For complex samples analyzed by high-high mode, proteoform identification and quantification from yeast lysate was performed using TopPIC (Top-down mass spectrometry-based Proteoform Identification and Characterization) pipelines [46]. Briefly, the raw file was converted to mzML files by MSConvert using peak-picking algorithm [47]. Then, the spectral deconvolution was performed using Top-down mass spectrometry Feature Detection (TopFD, version 1.7.0) with default parameter [48]. The database search was performed using TopPIC (1.7.0) against UniProt proteome database of Yeast (UP000002311, 6735 entries, accessed on 03/24/2023). The parameters were set as follows: mass error tolerance of matching masses of 25 ppm, maximum number of variable modifications (including acetylation, phosphorylation, oxidation and methylation) of 3, maximum number of mass shifts of unknown modification of 2 ranging from - 500 Da to 500 Da. The false discovery rates (FDRs) were estimated using the target-decoy approach. The spectrum level FDR cutoff was 1%, and the proteoform level FDR cutoff was 5%. The Top-down mass spectrometry-based identification of Differentially expressed proteoforms (TopDiff) was used to perform label-free quantification of the proteoforms. The lists of identified proteoforms from the yeast replicates are shown in Supporting Information I. The MS RAW files were deposited to the ProteomeXchange Consortium via the PRIDE [49] partner repository with the dataset identifier PXD051862.

Results and discussion

Comparison of LPA and PAMAPTAC coatings for CZE-MS-based protein measurement

We first compared LPA-coated and PAMAPTAC-coated capillaries in terms of proteoform separations using a standard protein mixture in a mass range of 8.6–66.7 kDa. An acidic BGE, 5% (v/v) acetic acid (pH 2.4), was chosen based on our previous experience to ensure that all the proteoforms were positively charged and to reduce the potential interactions between proteoforms and the capillary inner wall with the PAMAPTAC coating. We measured the EOF mobility of one bare fused silica capillary, one LPA-coated capillary, and one 50% PAMAPTAC-coated capillary using a neutral marker benzyl alcohol. We determined the EOF mobilities as 2.01E-8, 2.27E-9, and 4.88E-8 m² V⁻¹ s⁻¹, respectively. The EOF mobility of PAMAPTAC-coated capillary is in the same magnitude range (2~5E-8 m² V⁻¹ s⁻¹) as poly(diallyldimethylammonium chloride), poly(allylamine hydrochloride), and PEI in previous reports [50, 51]. We need to point out that the EOF direction of the PAMAPTAC-coated capillary is opposite from the bare fused silica capillary due to the positive charges on the capillary inner wall. Therefore, for the 50% PAMAPTAC capillary, we applied a negative high potential (−30 kV) at the sample injection end for separation. In contrast, for the LPA-coated capillary, +30 kV was applied at the sample injection end for the CZE separation. The neutral LPA coating minimized the EOF, and the migration of positively charged proteoforms was primarily due to their electrophoretic mobilities under the positive electric field. Given the +2 kV at the other end of the separation capillary for ESI, the separation voltage for the LPA-coated capillary and PAMAPTAC-coated capillary were +28 kV and −32 kV, respectively. For the PAMAPTAC-coated capillary, the forward EOF and backward electrophoresis produced a complete inversion in the proteoform migration order compared to the LPA-coated capillary. As depicted in Figure 1, CZE-MS using LPA and PAMAPTAC-coated capillaries produced reproducible separation and detection of the standard protein mixture. The PAMAPTAC-coated capillary provided a much wider separation window compared to the LPA-coated capillary (10 vs. 5 minutes). The migration time of five main peaks in PAMAPTAC-coated capillary are recorded in Table 2, with relative standard deviations (RSDs) of migration time less than 1%, highlighting the nice reproducibility.

Figure 1.

Electropherograms of a standard protein mixture analyzed by CZE-MS with a PAMAPTAC-coated capillary (A) and an LPA-coated capillary (B). Fifty nanoliter of the sample was injected for CZE-MS analysis in one run. The sample contained four standard proteins (carbonic anhydrous, CA; myoglobin, Myo; Ubiquitin, Ubq; bovine albumin serum, BSA) and the impurity superoxide dismutase (SOD) and it was dissolved in the background electrolyte (BGE, 5% acetic acid) of CZE. The dashed lines in (A) and (B) show zoomed-in electropherograms (~100x)).

Table 2.

Summary of the migration time of standard protein peaks using a PAMAPTAC-coated capillary.

	Migration Time (min)
Run	BSA	Ubq	Myo	CA	SOD1
PAMAPTAC-1	35.28	32.99	28.25	25.01	22.73
PAMAPTAC-2	34.99	33.07	28.34	24.92	22.75
PAMAPTAC-3	35.34	33.31	28.58	25.06	22.81
RSD (%)	0.53	0.50	0.60	0.28	0.18

We further compared the separation resolution of the standard proteins by LPA-coated and PAMAPTAC-coated capillaries, as shown in Table 3. CZE-MS using the PAMAPTAC-coated capillary improved the separation resolution between the adjacent protein peaks, on average, by 107%, compared to that using the LPA-coated capillary. Additionally, the PAMAPTAC coating provided much higher separation efficiency compared to the LPA coating, evidenced by the much better average number of theoretical plates from the PAMAPTAC coating (35,183 vs. 12,648). Interestingly, two proteoforms of myoglobin (with or without a phosphorylation) and three proteoforms of ubiquitin (+0, +42, and +45 Da) were separated using the PAMAPTAC-coated capillary but not the LPA-coated capillary. Modifications such as phosphorylation and acetylation altered the net charge carried by the proteoform, thereby affecting the charge-to-size ratio and separation resolution. The data highlights the advantages of PAMAPTAC coating for bettering proteoform separation compared to the LPA coating. We need to point out that the preparation of the PAMAPTAC coating is as simple as the LPA coating. The results demonstrate that the PAMAPTAC coating could be a valuable alternative to the commonly used LPA coating for further advancing TDP.

Table 3.

Summary of the separation resolution between adjacent protein peaks of a standard protein mixture measured by CZE-MS using an LPA-coated capillary and a PAMAPTAC-coated capillary.

Run	BSA - Ubq	Ubq - Myo	Myo - CA	CA-SOD1
LPA-1	0.82	2.66	1.96	6.2
LPA-2	0.84	2.4	2.08	5.52
LPA-3	1.12	2.42	1.97	5.54
Mean ± SD	0.93 ± 0.17	2.49 ± 0.14	2.00 ± 0.07	5.75 ± 0.38
PAMAPTAC-1	2.21	7.26	5.31	4.8
PAMAPTAC-2	1.49	7.16	5.17	3.88
PAMAPTAC-3	1.83	7.25	5.47	4.35
Mean ± SD	1.84 ± 0.36	7.22 ± 0.06	5.32 ± 0.15	4.34 ± 0.46

CZE-MS/MS using PAMAPTAC-coated capillaries for TDP of a complex yeast cell lysate

We further evaluated the performance of PAMAPTAC coating for CZE-MS/MS characterization of a complex sample, i.e., a yeast cell lysate. The yeast lysate extracted by 8M urea in 100 mM ammonium bicarbonate (pH 8.0) was buffer exchanged to 5% (v/v) acetic acid (pH 2.4). A fifty nanoliter sample (~70 ng) was injected into the same one-meter capillary under identical separation conditions as the standard proteins. The duration was extended to 70 minutes and repeated for six runs, Figure 2A. Our CZE-MS technique is reproducible for the yeast cell lysate analysis across the six runs regarding the number of proteoform identifications (8% RSD), the normalized-level (NL) intensity (15% RSD), and the overall separation profiles. We also randomly selected three proteoforms identified in all six runs in Figure 2A. The RSDs of those three proteoforms across the six runs are less than 2% in migration time and less than 18% in proteoform intensity. The data indicate reproducible measurements of proteoforms using our CZE-MS technique.

Figure 2.

Summary of the yeast cell lysate data from CZE-MS/MS using a PAMAPTAC-coated capillary. (A) The successive six electropherograms of CZE-MS/MS analyses of a yeast cell lysate. The normalized level (NL) refers to the highest signal intensity of the base peak electropherograms. The number of identified proteoforms is labeled for each run. (B) Two examples of proteoforms identified across six runs. (C) Proteoform overlap between runs. (D) The box plot of Pearson’s correlation coefficients of proteoform intensity between any two runs. (E) The histogram of molecular weight distribution of the proteoforms identified in Run 1.

On average, approximately 154 proteoforms were identified per run, with 69 proteoforms consistently identified across all runs. Although the total number of identified proteoforms is limited, the ratio of identified proteoforms larger than 10 kDa is 43.70 ± 1.92 %. This ratio is higher than those reported in previous studies [19,52]. A histogram of molecular weight distribution is shown in Figure 2E. The observed distribution shift can be attributed to a different sample preparation technique. In this study, a 30 kDa MWCO filter was used with 6 washes, rather than the 10 kDa MWCO filter with three washes. Efficient washing in sample preparation resulted in fewer identifications and relatively larger proteoforms. Another reason for the limited numbers of proteoform identification was the smaller sample injection amount due to the lack of an online concentration method. In future studies, various online sample stacking techniques will be investigated, including large-volume sample stacking, dynamic pH junction, transient isotachophoresis, and field-amplified sample stacking [53–56]. Figure 2B presents two proteoform examples. The large ribosomal subunit protein uL22A (RPL17A) is the largest proteoform consistently identified in each run. The intact proteoform was identified with 38 matched fragment ions and a low E-value of 3.77E-08. The average migration time of RPL17A is 43.36 min with a remarkably low relative standard deviation of 1.24%, underscoring the high reproducibility of our CZE-MS/MS system. The second proteoform, the elongation factor 3A (YEF3) with dual phosphorylation sites at pSer 1039 and pSer 1040, was also identified in all six runs. This elongation factor is required for the yeast ribosome and plays a role as a negative regulator of nonderepressible 2 (GCN2) kinase activity [57,58]. The two phosphorylation sites were reported earlier in large-scale phosphorylation analysis [59,60]. Additionally, two proteoforms of nascent polypeptide-associated complex subunit alpha (EGD2) covering Gly89-Lys174 were identified. One proteoform has a phosphorylation (+80.0010 Da) on Ser 101 and the other doesn’t. The migration time of the phosphorylated form and unphosphorylated form was 17.74 ± 0.09 and 16.61 ± 0.04 min, respectively. The single phosphorylation altered the net charge of the proteoform, thereby altering its electrophoretic mobility. The distinct difference in migration time further demonstrates the high resolution of our CZE-MS method for proteoform separation. We also assessed the proteoform identification between every two runs, noting around 60% of proteoforms overlapped between pairs of runs, Figure 2C. The box plot of Pearson’s correlation coefficients of the intensity of overlapped proteoforms, Figure 2D, indicates that the CZE-MS system is quantitatively reproducible with a mean and a medium value of 0.88 and 0.91, respectively.

We further evaluated the capillary-to-capillary reproducibility of the PAMAPTAC coating. We prepared another two PAMAPTAC capillaries (capillary 2 and 3) following the same procedure. Both standard protein mixtures and complex yeast samples were separated under identical separation conditions as used in PAMAPTAC capillary 1. As illustrated in Figures 3A and 3B, the separation profiles of the standard protein mixture remained consistent with the previous capillary, which suggests high capillary-to-capillary reproducibility. Moreover, Figure 3C shows the separation of the same yeast sample using three PAMAPTAC capillaries on a QE-HF instrument. Despite minor shifts in migration times, the separation profile was well-consistent, further confirming the high reproducibility across capillaries for complex samples.

Figure 3.

The reproducibility across capillaries. (A) and (B) are the electropherograms of standard protein mixture using different PAMAPTAC capillaries. (C). The electropherograms of the same yeast lysate using three different PAMAPTAC capillaries.

The identified proteoforms predominantly range between 5–20 kDa, as shown in Figure 2E. Large proteoform detection from complex cell lysates is usually restricted in global TDP due to their wide charge state distributions and complicated isotopic peaks, resulting in low sensitivity [61]. In a typical TDP study, the mass of identified proteoforms usually is lower than 30 kDa [4,19,49]. Here we investigated the potential of our CZE-MS/MS technique with PAMAPTAC coating for detecting large proteoforms in a yeast cell lysate. In low-resolution MS1, the isotopic peaks of each charge state were merged into a single peak, simplifying the mass spectra, improving the measurement quality, and enabling the determination of large proteoform mass. We applied low-resolution MS1 (resolution 7,500 at m/z of 200) with 10 microscans and deconvoluted based on charge-state distribution by UniDec [45]. The CZE separation condition remains the same.

Without any size-based prefractionation method, two large proteoforms, 44.95 kDa and 45.14 kDa co-migrated, Figure 4B, and they shared similar charge state distributions. Figures 4C and 4D illustrate the detection of additional large proteoforms in the 25–35 kDa range. The data demonstrate that the cationic coating-based CZE-MS has a high potential to advance TDP toward the identification of large proteoforms. It is important to note that the limited backbone cleavages further hinder the identification of the large proteoforms using the commonly used HCD. This insufficient gas-phase fragmentation of large proteoforms is a critical challenge in the field of TDP. Coupling cationic coating-based CZE-MS with electron- or photon-based fragmentation techniques for the characterization of large proteoforms will be performed in our future studies.

Figure 4.

Low-resolution MS1 of the yeast cell lysate. (A). The successive three electropherograms of the yeast lysate at the low-resolution MS1. (B), (C) and (D) are the mass spectra and deconvoluted masses of large proteoform examples detected. The deconvolution is performed using UniDec software [45].

Proteoforms’ electrophoretic mobility prediction from CZE-MS measurement using PAMAPTAC-coated capillary

Previous studies have demonstrated that the tunable EOF rate of PAMAPTAC capillaries, adjustable from 0 to 4E-8 m²V⁻¹s⁻¹, depending on the charged monomer ratio, could enhance the electrophoretic separation resolution [38,62]. In this study, we examined the separation of standard protein mixtures using a PAMAPTAC capillary with 50% APTAC, observing a significantly higher separation resolution compared to an LPA capillary. We also tested the same standard protein mixture using a PAMAPTAC capillary with 25% APTAC, Figure 5A. The lower percentage of positively charged monomer led to a reduced EOF and slower migration velocity, increasing the average separation resolution between CA and myoglobin from 5.32 ± 0.15 to 6.38 ± 0.66. We noted that reducing EOF using 25% APTAC decreased the EOF-driven proteoform dispersion [63], which is one important reason for improved separation resolution. Consequently, this adjustment delayed the migration of ubiquitin and BSA, requiring the pressure application after 60 minutes of separation. These findings state the impact of charged monomer concentration on both migration time and separation resolution of proteoforms.

Figure 5.

The predictability of cationic coated capillary electrophoresis. (A) The electropherogram of standard protein mixture by 25% charged monomer. The dashed line shows a magnified view of the data (~100x). (B) The predicted and experimental mobility of the unmodified proteoforms in the yeast lysate using 50% PAMAPTAC capillary.

To predict the migration time of proteoforms under EOF and electrophoretic migration, we adopted a semi-empirical model of the peptides and proteoforms in the open-tubular CZE separation [64–66]. Briefly, the theoretical electrophoretic mobility μ_EP is calculated based on the net charge of proteoform, Q, and the proteoform mass, M. The net charge is determined by accounting for positively charged residues (Ks, Rs, Hs, and the N-terminal amino group) in an acidic background electrolyte (BGE), and the proteoform mass is represented by the precursor mass. We proposed the predicted mobility μ, combining the measured EOF mobility μ_EOF with reverse electrophoretic mobility and an optimized coefficient:

$$\mathit{\text{Predicted}}\mu ={\mu }_{EOF}-{\mu }_{EP}=0.488-5*\frac{\text{ln}\left(1+0.35Q\right)}{M^{0.411}}$$ (1)

The experimental mobility (μ) is calculated by dividing the experimental velocity of the solute (v) by the applied electrical field (E):

$$\mathit{\text{Experimental}}\mu =\frac{v_{ep}}{E}=\frac{\frac{L}{t_m}}{\frac{V}{L}}=\frac{L^2}{V*t_m}$$ (2)

where L is the capillary length in meters, t_m is the migration time in seconds, and V is the applied potential in volts.

Validation using unmodified proteoforms identified in yeast lysate showed a good linearity (R²=0.94), as depicted in Figure 5B. Based on the correlation, the migration time of proteoforms can be predicted using this model. To accurately predict the mobility, the coefficients in equation (1) need to be optimized for each specific case. Integrating this predictive model with the correlation between EOF rate and charged monomer percentage enables a robust framework for optimizing separation conditions tailored to diverse complex samples.

Conclusions

We developed a simple and efficient approach for preparing cationic capillary coating for CE-MS-based TDP. The cationic coating, PAMAPTAC, significantly enhanced the separation resolution, with its reproducibility confirmed through intra- and inter-capillary replicates. Importantly, our approach facilitated the detection of large proteoforms (greater than 30 kDa) without prefractionation, demonstrating its strong potential in identifying large and hydrophobic proteoforms while minimizing adsorption. Additionally, this tunable EOF rate and predictable mobility offer promising potential for broader biological applications.

The current method prepared samples in the BGE solution, limiting the sample injection amount as well as the number of identified proteoforms. Future studies should focus on optimizing the online concentration method and reducing the sample complexity to enhance proteoform identification. Field-amplified sample stacking could be beneficial while maintaining proteins in a positively charged form [67]. Further investigations are required to explore the long-term durability of the coating.

• A cationic coating was developed for CZE-MS/MS-based top-down proteomics.

• The cationic coating improved the separation resolution of proteoforms.

• The cationic coating-based CZE-MS achieved reproducible proteoform measurement.

• Large proteoforms were detected from a yeast cell lysate using CZE-MS.

• Electrophoretic mobility of proteoforms from CZE-MS can be predicted accurately.