key: cord-0003958-9ahn4q88
authors: He, Muyi; Luo, Pan; Hong, Jie; Wang, Xiaofeng; Wu, Haimei; Zhang, Rongkai; Qu, Feng; Xiang, Ye; Xu, Wei
title: Structural Analysis of Biomolecules through a Combination of Mobility Capillary Electrophoresis and Mass Spectrometry
date: 2019-01-31
journal: ACS Omega
DOI: 10.1021/acsomega.8b03224
sha: 1ebb99d6c23999bab025feddb7a13b81b36d1947
doc_id: 3958
cord_uid: 9ahn4q88

[Image: see text] The 3D structures of biomolecules determine their biological function. Established methods in biomolecule structure determination typically require purification, crystallization, or modification of target molecules, which limits their applications for analyzing trace amounts of biomolecules in complex matrices. Here, we developed instruments and methods of mobility capillary electrophoresis (MCE) and its coupling with MS for the 3D structural analysis of biomolecules in the liquid phase. Biomolecules in complex matrices could be separated by MCE and sequentially detected by MS. The effective radius and the aspect ratio of each separated biomolecule were simultaneously determined through the separation by MCE, which were then used as restraints in determining biomolecule conformations through modeling. Feasibility of this method was verified by analyzing a mixture of somatostatin and bradykinin, two peptides with known liquid-phase structures. Proteins could also be structurally analyzed using this method, which was demonstrated for lysozyme. The combination of MCE and MS for complex sample analysis was also demonstrated. MCE and MCE–MS would allow us to analyze trace amounts of biomolecules in complex matrices, which has the potential to be an alternative and powerful biomolecule structure analysis technique.

The activities and functions of a biomolecule in physiological and pathological processes closely correlate to its 3D structure. 1−3 Therefore, biomolecule structure characterization is essential in understanding biological processes and developing targeted drugs. To date, many techniques have been developed and applied in resolving biomolecule structures, including X-ray crystallography, 4−6 NMR, 7−10 cryo-electron microscopy, 11−13 and SAXS. 14, 15 Although relatively high-resolution biomolecule structure information could be obtained through these techniques, their applications in analyzing trace amounts of biomolecules, especially under in situ or in vivo conditions, are limited by the factor that target biomolecules need to be purified to certain homogeneity and in certain amounts. 5 With high sensitivity and specificity, MS is capable of identifying trace amounts of biomolecules in complex matrices. 16−21 Not only the molecular weight information but also the primary sequence could be resolved through highresolution MS and tandem MS, respectively. 22−24 As a result, MS has been widely applied in studying biological systems. It is specifically attractive to use MS-based techniques for the structural and functional investigation of the trace amount of biomolecules in vivo. However, significant challenges exist when using conventional MS for 3D structure studies of biomolecules. To solve this problem, ion mobility spectrometry (IMS) 25−30 and capillary electrophoresis (CE) 31−38 have been increasingly coupled with MS to enhance its structure analysis capability. As a gas-phase technique, analytes are vaporized and ionized before their collision cross sections are measured in IMS, 39, 40 thus raising debates whether structures of biomolecules could be fully maintained. 41 Ivory et al. introduced a series of method utilizing counterflows to concentrate analyses in electrophoresis. 42−45 Although CE is a solution-phase technique that separates analytes based on their charge states and sizes, the electroosmotic flow (EOF) in CE is sensitive to its operating conditions, which affects its reproducibility and also its application in biomolecule structural determination. 46, 47 Through chemical modifications of charged groups on a protein surface, the protein ladder method could determine the effective charge and size of a protein by capillary electrophoresis. 48−51 Here, we developed a liquid-phase biomolecule separation and 3D structure analysis technique, named mobility capillary electrophoresis (MCE). MCE is a new type of CE method in which a constant liquid flow was used to replace EOF. Precise control of the liquid flow is beneficial to ion structure analysis. MCE has the capability of not only separating complex samples but also acquiring the sizes of biomolecules. Comparing with the protein ladder method, MCE does not require chemical modifications and has the capability of mixture analyses. The use of MCE in combination with MD simulations as a structural determination method was demonstrated here for peptide mixtures and a protein sample. A home-built MCE system was successfully coupled with a mass spectrometer for complex sample analysis. The MCE−MS system showed high stability, reproducibility, and capabilities for biomolecule separation, identification, and structure determination.

MCE separates the analytes based on their charge states, hydrodynamic sizes, and geometries. With minimized EOF and well-controlled operating parameters, MCE has high stability and reproducibility in the first place. More importantly, the correlation between ion elution time and ion effective size could be obtained and used to calculate the effective radius of a biomolecule in solution. To better reconstruct their 3D structures, biomolecules were assumed to have ellipsoid shapes. With this ellipsoid approximation, a restrain curve was acquired for a biomolecule, which defines its aspect ratio versus its radius. This restrain curve was then used to evaluate or estimate the occurrence probability of each conformation acquired from MD simulations. The coupling of MD simulations with MCE experiments was first demonstrated for angiotensin I in an MS friendly solvent, and its conformations with top probabilities were picked out based on this scoring methodology. This method was then applied for the analysis of a peptide mixture in which somatostatin and bradykinin in a pH neutral solution were separated and structurally analyzed. The predicted structures of bradykinin and somatostatin obtained from this method were compared with those obtained from NMR measurements. Finally, performances of the home-built MCE−MS system were characterized through the analyses of biomolecule mixtures and BSA using a typical bottom-up proteomic strategy.

Ion Motion Modeling. In MCE experiments shown in Figure 1 , a constant liquid flow was used to provide a driving force in a liquid channel (fused silica capillary in this case), and an electric field was also applied to separate different ions in the liquid channel. When dissolved in solutions, peptides, proteins, and other biomolecules are typically ionized, if pH of the solution is different from their isoelectric points. These ions will have directional motions when an electric field is applied, which is known as electrophoresis. On the other hand, ions will also experience a viscous drag force (F f ) when their motion speed is different from the liquid flow. Besides the amount of charge an ion possesses, ion migration time also depends on the size and geometry of the ion. In general, ions with larger effective sizes will experience stronger viscous drag force. Therefore, ion migration time (t) in MCE could be used to calculate the ion effective radius (R).

To accurately calculate the ion effective radius, ion migration in MCE needs to be precisely controlled and characterized. Mathematical models were first established to study this ion migration process. The electric field force (F e ) acting on an ion could be expressed as

in which q is the amount of charge possessed by the ion, E is the separation electric field intensity, U is the dc potential applied across the separation capillary, and L is effective length of the separation capillary. Different from neutral molecules, the velocity difference between ions and liquid flow causes a viscous drag force.

where v r is the relative speed of an ion, η is the viscosity coefficient of the solution, and ξ = 6πηR. Under equilibrium, F e would be equal to F f . The absolute migration speed (v absolute ) could then be expressed as v v v absolute carrier r = + (3) in which v carrier is the flow rate of the liquid flow. EOF was minimized in MCE so that liquid flow rate could be precisely controlled. To minimize any residual EOF effects on the liquid flow rate, neutral markers (phenol or DMSO) were used to accurately measure the velocity of the liquid flow. Ion displacement (S) within the separation capillary could be determined by 

Article in which m is the mass of the ion. After solving eq 4, ion displacement with respect to time is

According to eq 5, the measured ion migration time can be used to calculate the ion equivalent radius. MCE Parameter Optimization. Theoretical analysis of eq 5 shows that capillary length, solvent viscosity, separation voltage, liquid driving pressure, and liquid flow rate would all affect ion migration time. As shown in Figure 2 , there is an inverse relationship between ion effective radius and ion migration time. A 50/40 cm long (total/effective length) fused silica capillary with an i.d. of 75 μm was used in mobility capillary electrophoretic theoretical calculation. A cation with a molecular weight of 1048 Da and +2 charge was selected as a model ion. As shown in Figure 2 , there are three segments in a characteristic curve: the front end, the turning point, and the back end. In the front end, there would be a very small difference in ion migration time even for ions with a large effective radius difference. Meanwhile, the ion migration time would no longer reflect the ion radius difference in the back end. Therefore, to improve measurement accuracy, the analysis condition should be controlled so that the range of ion migration time is located at the "turning point" of the characteristic curve. As shown in Figure 2 , a higher electric field would typically shift the characteristic curve up, while a higher driving pressure (or liquid flow rate) could shift the characteristic curve down. While electric field and liquid flow rate are the two most important parameters, capillary length, charge state of ions, and buffer viscosity also have impacts on ion migration time. Details about these effects could be found in the Supporting Information ( Figure S2 ).

To find the optimal running conditions, a series of experiments were carried out. An MS friendly solvent (methanol/water buffer v/v 1:1) was first used as the buffer solution. Electrophoretic diagrams of three peptides under different working conditions are given in Figure S3 . It is found that the calculated equivalent ion radius showed consistency under different working conditions ( Figure S4 ), suggesting that this method is not sensitive to running conditions. It should be noticed that decreasing the liquid flow rate will increase the analysis time. Higher separation voltage can lead to higher Joule heat, which will affect stability of the system. Finally, −20 kV and 60 mbar were selected as the optimized operating conditions in MCE experiments.

Ellipsoid Approximation. An ellipsoid approximation was used to approximate the steric geometry of biomolecules, 52 which was further simplified by spheroid approximation (b = c). Under the current working conditions, the liquid flow has a very low Reynolds number (∼0.075), and the fluid direction is assumed to be parallel to the ellipsoid axis of symmetry. A length-to-diameter ratio was defined as 

Since the ion effective radius (R) could be obtained from MCE experiments using eq 5, eq 6 sets a restrictive condition for the 3D structure of a biomolecule. 52 Figure 3a −c shows a typical process of using the MCE experiment to obtain the restrictive condition. The peptide angiotensin I was used to demonstrate this process. As plotted in Figure 3a , ion migration time (t) of angiotensin I could be obtained by performing MCE analysis, while the neutral marker peak in front was used to calibrate the liquid flow rate. Repeated experiments ( Figure 3a) show that MCE has great repeatability, and a CV (coefficient of variance) value of 0.385% was achieved for angiotensin I. The peak width and tailing effect of angiotensin I could be attributed to two factors: (1) To achieve precise control of the liquid flow rate, laminar flow was used in MCE, and the EOF was minimized. (2) To make sure the MCE solvent is compatible with the MS solvent, a low-sodium or no-sodium solvent was used in this study. With a low-sodium solvent, conductivity of the running buffer is much lower than that of the sample bands, which would cause a tailing effect in the peak due to the electromigration dispersion effect. 53, 54 Another possible reason for peak broadening may be due to the weak buffer capacity; the peptides could cause local pH effects that affected the peak shape. After solving eq 5, it is found that angiotensin I has an equivalent radius of 9.7966 ± 0.0385 Å. By substituting the ion equivalent radius (R) in eq 6, a one-to-one relationship could be obtained between the length-to-diameter ratio (Φ) of angiotensin I and its semi-axis radius (c) as shown in Figure 3c , thus setting a structural restraint for angiotensin I in the liquid solution (named the restrain curve).

Coupling MCE with MD Simulation. As an experimental method, MCE provides a geometry restraint on biomolecules under analysis. By coupling MCE with MD simulations, a scoring method was proposed to predict the most likely conformations present in the solvent condition. As shown in Figure 3d , possible structures of peptides and proteins could be obtained using different types of MD simulation methods. The MCE measurement results were used as a restraint on these simulation results and to help pick out structures with higher probabilities. To do so, ellipsoid approximation was also applied on simulated structures. The radii of gyrations (Rg) of each conformation were first calculated (Figure 3e ), 55 

After further approximating biomolecules using the spheroid approximation, the effective diameter (d) would be

As a result, the length, diameter, and of course length-todiameter ratio (Φ) could be obtained for each simulated conformation.

As a demonstration, MD simulation was carried out for angiotensin I in solution. A 100 ns MD simulation under the MCE experimental conditions produced 10000 conformations, which were then used as the conformation library, as shown in Figure 3d . The root-mean-square deviations (RMSD) of angiotensin I were extracted and are shown in Figure S5 . Following the data processing procedure shown in Figure 3d ,f, the gyration radii of angiotensin I in x-, y-, and z-directions were calculated and are plotted in Figure 3e . Ellipsoid approximation was then performed in which the principal semi-axes of the ellipsoid (a, b, c) were calculated (Figure 3f) .

MD simulation and MCE experimental results were then matched by plotting the length-to-diameter ratios of the 10000 simulated conformations of angiotensin I into the corresponding restrain curve figure. Figure 3g shows the matching result in which each simulated conformation was shown as a data point. 

The score of each conformation is defined as the distance of the corresponding data point from the restrain curve.

The smaller this score is, the higher probability of the corresponding biomolecule conformation is believed to exhibit in solution. The inset in Figure 3g shows the conformation of angiotensin I (in licorice representation) with the minimum score (0.7623 Å). The top five conformations in terms of closer distances from the restrain curve are also listed in Table 1 . Results show that the rod-like angiotensin I shows a random coil structure. Although detailed structures of these five conformations have differences, they have similar geometries. The small distances of simulated conformations from the restrain curve suggest that this MCE-based method has reasonably good agreement with MD simulations.

Mixture Analysis. Many high-resolution structural analysis methods need high-purity samples or even crystal samples, which restricts the application of these methods in highthroughput or complex sample analyses. With the capability of ion separation, MCE could also be applied in the structure analysis of biomolecules in mixtures. As a proof-of-concept demonstration, the separation and structure analysis of the somatostatin and bradykinin mixture were performed using MCE. Figure 4a plots a typical electrophoretic diagram of the mixture in which the neutral marker (DMSO), somatostatin, and bradykinin were separated in MCE under the conditions of 25 kV and 60 mbar. Repeated MCE experiments were also carried out ( Figure S6 ), and CV values of 1.740 and 0.824% were achieved for somatostatin and bradykinin, respectively. Following the procedure in Figure 3a −c, the effective radii of somatostatin and bradykinin were calculated as 9.9940 ± 0.1739 and 8.5183 ± 0.0702 Å, respectively. After the ellipsoid approximation, restrain curves of angiotensin I and bradykinin were obtained and are shown in Figure 4b ,c. To predict steric conformations of these two peptides, MD simulations were then performed under neutral pH conditions as described in the Methods section. A 100 ns MD simulation produced 2000 conformations, which were then used as the conformation library. Following the data processing procedure in Figure 3d −f, the gyration radii and principal ellipsoid semi-axes of somatostatin and bradykinin in the x-, y-, and z-directions were obtained ( Figure S7 ). Figure 4b ,c plots the matching results between simulated conformations and restrain curves obtained from MCE measurements. After applying the scoring method proposed earlier, conformations of somatostatin and bradykinin with top probabilities were obtained (Figure 4b,c) . These two conformations have scores of 1.998 × 10 −4 and 2.085 × 10 −5 Å, respectively. To validate the feasibility of this method, 

Article the top conformations found using our method were also compared with those presented in the NMR data bank. The figures in the right in Figure 4b ,c are the comparison results in which new cartoon representations are presented for clarification. Conformations of somatostatin and bradykinin found using our scoring method (plotted in yellow) agree well with those downloaded from the NMR data bank (plotted in green and white). 57, 58 The RMSD for bradykinin between the best simulation conformation and the NMR conformation with aligned Cα atoms is 2.478 Å, while the corresponding RMSD for somatostatin is 1.655 Å.

Structure analysis of a protein, lysozyme, was also demonstrated using this MCE method. Based on MCE elution time, the effective radius of lysozyme was first calculated as 20.7020 ± 0.3223 Å. The ellipsoid approximation restrain curve of lysozyme was obtained and is shown in Figure 4d . To predict the steric conformations of this protein, MD simulations were then carried out under pH ∼6 conditions as described in the Methods section. A 10 ns MD simulation produced 1000 conformations, which were then used as the conformation library. The gyration radii and principal ellipsoid semi-axes of lysozyme in the x-, y-, and z-directions were obtained ( Figure  S7 ). Figure 4d plots the matching results between simulated conformations and restrain curves obtained from MCE measurements. After applying the scoring method proposed earlier, the conformation of lysozyme with top probability was obtained ( Figure 4d ) with a score of 0.2611 Å.

The top conformation acquired using our method was also compared with that presented in the NMR data bank, which is shown in the right of Figure 4d . The conformation found using our scoring method (plotted in yellow) agrees well with that downloaded from the NMR data bank (plotted in green). 59 The RMSD for lysozyme between the best simulation conformation and the NMR conformation with aligned Cα atoms is 1.333 Å.

MCE−MS for Mixture Analyses. The combination of MCE and MS would be powerful in terms of complex sample analysis. The home-built MCE−MS system as shown in Figure 5a was first optimized (details could be found in the Supporting Information) and then characterized using a mixture of six compounds, which includes an amino acid: phenylalanine (3.7 mg/L); two peptides: angiotensin II (10 mg/L) and bradykinin (16 mg/L); three proteins: insulin (80 mg/L), lysozyme (160 mg/L), and cytochrome c (160 mg/L). Under the optimized working conditions, these six compounds could be well separated in two domains: the migration time domain and the m/z ratio domain. Figure 5b plots the three-dimensional MCE− MS spectrum, and Figure 5c is the corresponding twodimensional heat map. As noted in Figure 5c , these six biomolecules show up at different locations of the heat map due to their property differences. In the m/z ratio domain, the sequence of ions is determined by the m/z ratios of ions in the gas phase after nanoESI. In the migration time domain, the sequence of ions is mainly determined by physical sizes, charge, and folding states of ions in the liquid phase. Since the drag force is proportional to the ion effective radius (R) and the electric field force is proportional to ion charge (q), ion migration time would be proportional to q and inversely proportional to R. As shown in Figure S1 , larger ions with less charge would show up earlier in the chromatogram, and the number of charge would have bigger impacts on ion migration time than the ion effective radius. Results in Figure 5 show that phenylalanine shows up before angiotensin II and angiotensin II shows up before proteins, which might be due to the fact that larger biomolecules would possess more charges in solution. However, insulin has a very compact three-dimensional structure with disulfide bonds and having less available charging sites, which might result in a smaller q/R value than bradykinin. As a result, insulin travels faster than bradykinin in the MCE capillary. Figure 5d plots the total ion chromatogram (TIC) of MCE and the corresponding mass spectra of these six compounds.

The analysis of this six-compound mixture was also performed on a conventional CE instrument (Lumex Capel-105M capillary electrophoresis system, Russia). In the capillary zone electrophoresis mode, five peaks were observed for these six compounds in which insulin and bradykinin were not separated. Furthermore, the elution sequence of these compounds is also different from MCE ( Figure S11) , which is cytochrome c, lysozyme, insulin/bradykinin, angiotensin II, and phenylalanine. Results suggest that this MCE method has a different separation mechanism and it is complementary not only to conventional liquid chromatography (LC) but also complementary to conventional CE. It should also be noticed that MCE separates analytes before ionization, which is different from MCE. As a result, ions from the same analyte would elute and show up simultaneously in MCE as shown in Figure 5 , while ions from the same analyte but at different charge states would spread across a much broader range in MCE. For example, cytochrome c ions would have the same charge state (at least very close to each other) in solution before ionization, and they would have a similar migration time in MCE. The different charge states observed in the mass spectra are due to the ESI process, which would also unfold cytochrome c ions. On the other hand, in a typical MCE experiment, cytochrome c would be ionized first 

Article and then subject to the MCE separation. Hence, cytochrome c ions at different charge states would be separated and have different migration times in MCE.

Liquid-phase chromatography coupled with MS has been widely used in protein analysis. 60 The use of MCE−MS for BSA hydrolysate analysis and its comparison with conventional LC− MS method were performed ( Figure S12) . A 51% peptide coverage rate was obtained using MCE−MS, compared to the 55% peptide coverage rate using LC−MS. With much shorter analysis time and similar peptide coverage rate, this MCE−MS method could benefit high-throughput omics studies. The use of MCE for isomer analysis was also shown and is plotted in Figure  S13 . With complementary separation mechanisms, MCE could also be used together with LC or CE techniques so that multidimensional chromatography methods could be achieved. Since the current MCE separation still takes about 10−20 mins, offline coupling methods could be possible.

In this study, MCE was introduced and characterized as a liquidphase biomolecule 3D structure analysis technique and an MS compatible chromatography method. By assuming biomolecules have ellipsoid shapes, the approximate shape information of biomolecules could be promptly obtained using MCE and used as a restraint for biomolecule structure determination. Separation and structure analysis of a peptide mixture were demonstrated, suggesting that MCE could be used as a fast and convenient method for structure analysis of biomolecules in complex matrices. Charge states of peptides and proteins were calculated from theoretical and empirical equations, which are derived based on the proton affinities of amino acids. Therefore, the current method is limited to peptide and protein analyses under low ionic strength working conditions. In the future, techniques such as Taylor dispersion 61 and ion current measurement could be included so that biomolecule charge state and shape could be obtained simultaneously. The peak width/shape in MCE experiments could also be improved in the future, by using thinner capillaries and more conductive solvents. The coupling of MCE with MS could further extend its application for large-scale in vitro and in vivo biomolecule structural and function investigations. Any new fused silica capillary was cleaned by rinsing it with methanol/water running buffer for 30 min before use. In this MCE system, no EOF is needed, so activation of the capillary was not required and not performed before use. EOF in the capillary was minimized by either using unactivated capillary at low pH conditions (pH ∼3) or using coated capillary. Low-salt or no-salt solvent systems were used to further help lowering the EOF and to make the MCE compatible with MS. Of course, other methods, such as coating and applying external fields, could also be applied to minimize the EOF. 47, 62, 63 Between runs, the capillary was rinsed with running buffer for 3 min. When analyzing angiotensin I, an MS compatible solvent, methanol/ water (v/v 1:1 with 0.1% FA, η = 1.62 mPa·S) was used as the running buffer. Angiotensin I was diluted to 1 mg/mL, and 1‰ phenol (1 mol/L) was also added in the sample solution. Phenol was selected and used as a neutral marker in acidic conditions to calibrate the velocity of the liquid flow. 63 In the following experiments, neutral coated capillaries were used under neutral pH conditions. When analyzing the mixture of somatostatin and bradykinin, a pH neutral solution containing 2 mM NaCl in water (pH = ∼7, η = ∼0.89 mPa·S) was applied as the running buffer. Somatostatin and bradykinin were diluted to 1 mg/mL, and 1‰ DMSO was added as the neutral marker. Lysozyme was analyzed using water as the running buffer (η = ∼0.89 mPa·S), and its pH was adjusted to 6 by FA. Lysozyme was diluted to 2 mg/mL, and 5‰ phenol (1 mol/L) was added as the neutral marker in acidic conditions.

Working conditions of the Lumex system were modified as follows to perform MCE experiments (otherwise specified): samples were injected by applying a 50 mbar pressure for 5 s; a −20 kV running voltage was applied over a 50 cm long fused silica capillary, which has an effective length of 40 cm (distance from the sample injection point to the UV detector); a 60 mbar pressure was applied to drive the liquid flow; an environment temperature of 25°C was maintained for the capillary; detection wavelength of the UV detector was 214 nm. Working conditions of −15 kV voltage and 90 mbar pressure were applied during the lysozyme analysis.

Home-built MCE−MS Instrument Setup. Figure S1 (as well as in Figure 5a ) shows a schematic illustration of the homebuilt MCE−MS system, in which the MCE system was coupled with the MS system through a nanoESI source. The MCE system mainly consists of four parts: a sample injection part, a running buffer pumping system, a separation capillary, and a DC separation voltage module. To operate the MCE system, samples were first injected into the fluid channel using a syringe (Hamilton) driven by a syringe pump (KD Scientific Syringe Pump Company), and a microvalve (Agilent 1200 Series 2 Position/ 6 Port, VICI, No. G1162A) was used to accurately control the sample injection amount. 64 During the sample injection period, the microvalve was switched to the quantitative loop, and samples were injected from the syringe to the quantitative loop with a pumping speed of 20 μL/h for 0.1 min.

After sample injection, the microvalve was switched to the quantitative loop, and another syringe pump was connected and used to pump the running buffer (methanol/water v/v 1:1 with 0.1% FA) into the flow channel for sample separation. The same

Article fused silica capillary (unactivated) with a length of 40 cm was used as the separation channel. During MCE−MS experiments, the injected sample was flushed into the separation channel by the running buffer at a constant flow rate. At the same time, a constant dc separation voltage was applied across the separation capillary as shown in Figure S1 . A negative high voltage was applied at the back end of the capillary (close to the sample injection side), and the front end of the capillary was grounded (close to the nanoESI side). After separation within the capillary, analytes would be ionized by the nanoESI source (FS360-50-15-D, New Objective, Woburn, MA), and MS analyzed by the consequent Q-TOF mass spectrometer (Agilent G6520B Accurate-Mass Q-TOF). A negative high voltage from 1300 to 1600 V was applied on the MS inlet, which provides ionization voltage for the nanoESI source. The following MS parameters were used in experiments: gas temperature 300°C, gas flow rate Tryptic Digestion of BSA. One microgram of BSA was denatured and reduced by a 200 μL solution containing 8 M urea and 10 mM DTT in 50 mM NH 4 HCO 3 buffer (pH 8.3) at 37°C for 2 h. Alkylation was performed in a 50 mM IAA solution at room temperature for 1 h in the dark. After alkylation, the sample was diluted using a 50 mM NH 4 HCO 3 buffer to give a final urea concentration of 1 M. Tryptic digestion was then performed at a protein-to-trypsin concentration ratio of 50:1 (w/w) for 16 h at 37°C. 65 After digestion, residual trypsin activity was quenched by adding 3% FA (v/v) and boiling the sample for 10 min. Next, the digested peptide samples were desalted by reversed-phase C18 SPE column and then lyophilized to complete dryness. Dry powder was dissolved in running buffer and centrifuged at 17000g for 5 min before MCE−MS analysis.

MD Simulation. MD simulations were conducted with the AMBER99SB-ildn force field 66 in GROMACS 2016.1. 67 The AmberTools 16 was used to create peptides that were not supported in the data bank. 68 Charge states of peptides in the experiments were calculated based on the literature (refer to the Supporting Information for details). 69, 70 It is found that angiotensin I has a charge state of +3 at a pH of 3. For angiotensin I, MD simulation was performed on a methanol/ water solvent box (1:1 volume ratio), which has a dimension of ∼65 Å. The formic anion was the counterion. Under the native experimental conditions (pH = 7 in water solution), both somatostatin and bradykinin show the +2 charge state. The chloride anion was the counterion. In this case, MD simulations were performed on a sodium chloride water solvent box, which has a dimension of ∼52.5 Å. In all simulations, energy minimization was performed and followed by 1 ns equilibration at 300 K. The overall simulation window was 100 ns. Lysozyme MD simulation was conducted with the GROMOS96 54a7 force field. Charge states of lysozyme in the experiments were calculated by propka-3.1. 71, 72 It is found that lysozyme has a charge state of +9 at a pH of 6. The starting system consisted of a cubic simulation box of X = Y = Z = ∼76.9 Å, containing one lysozyme molecule and 14405 water molecules. The formic anion was the counterion. The overall simulation window was 10 ns.

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10 

The authors declare no competing financial interest.

Protein activity regulation by conformational entropy

Dynamic personalities of proteins

Coupling of receptor conformation and ligand orientation determine graded activity

Cool' crystals: macromolecular cryocrystallography and radiation damage

Accessing protein conformational ensembles using room-temperature X-ray crystallography

Crystal Cryocooling Distorts Conformational Heterogeneity in a Model Michaelis Complex of DHFR. Structure

NMR spectroscopy brings invisible protein states into focus

An NMR perspective on enzyme dynamics

Studying excited states of proteins by NMR spectroscopy

Open-to-closed transition in apo maltose-binding protein observed by paramagnetic NMR

Mature HIV-1 capsid structure by cryo-electron microscopy and allatom molecular dynamics

Cryo-electron microscopy structures of the SARS-CoV spike glycoprotein reveal a prerequisite conformational state for receptor binding

The bacteriophage ϕ29 tail possesses a pore-forming loop for cell membrane penetration

Structural characterization of flexible proteins using small-angle X-ray scattering

Accurate small and wide angle x-ray scattering profiles from atomic models of proteins and nucleic acids

Direct Ionization of Large Proteins and Protein Complexes by Desorption Electrospray Ionization-Mass Spectrometry

Mass spectrometry in bioinorganic analytical chemistry

Expanding the structural analysis capabilities on an Orbitrap-based mass spectrometer for large macromolecular complexes

The diverse and expanding role of mass spectrometry in structural and molecular biology

Mass spectrometry captures structural intermediates in protein fiber self-assembly

Foldability of a Natural De Novo Evolved Protein

Quantitative analysis of phospholipids in functionally important membrane domains from RBL-2H3 mast cells using tandem high-resolution mass spectrometry

Top-down highresolution mass spectrometry of cardiac myosin binding protein C revealed that truncation alters protein phosphorylation state

High resolution, high-throughput amide deuterium exchange-mass spectrometry (DXMS) determination of protein binding site structure and dynamics: utility in pharmaceutical design

Anhydrous protein ions

Biomolecule Analysis by Ion Mobility Spectrometry

The power of ion mobility-mass spectrometry for structural characterization and the study of conformational dynamics

Structural Modeling of Heteromeric Protein Complexes from Disassembly Pathways and Ion Mobility-Mass Spectrometry

Uncovering the Stoichiometry of Pyrococcus furiosus RNaseP, a Multi-Subunit Catalytic Ribonucleoprotein Complex, by Surface-Induced Dissociation and Ion Mobility Mass Spectrometry

Complementary Glycomic Analyses of Sera Derived from Colorectal Cancer Patients by MALDI-TOF-MS and Microchip Electrophoresis

Protein unfolding and conformational studies by capillary electrophoresis

Real-Time Monitoring of Protein Conformational Dynamics in Solution Using Kinetic Capillary Electrophoresis

Detection of coexisting protein conformations in capillary zone electrophoresis subsequent to transient contact with sodium dodecyl sulfate solutions

Surface-Confined Aqueous Reversible Addition-Fragmentation Chain Transfer (SCAR-AFT) Polymerization Method for Preparation of Coated Capillary Leads to over 10 000 Peptides Identified from 25 ng HeLa Digest by Using Capillary Zone Electrophoresis-Tandem Mass Spectrometry

Predicting Electrophoretic Mobility of Tryptic Peptides for High-Throughput CZE-MS Analysis

Qualitative and quantitative metabolomic investigation of single neurons by capillary electrophoresis electrospray ionization mass spectrometry

MALDI MS Guided Liquid Microjunction Extraction for Capillary Electrophoresis-Electrospray Ionization MS Analysis of Single Pancreatic Islet Cells

Number of Solution States of Bradykinin from Ion Mobility and Mass Spectrometry Measurements

Collision Cross Sections of Proteins and Their Complexes: A Calibration Framework and Database for Gas-Phase Structural Biology

Gaseous conformational structures of cytochrome c

Simultaneous separation of negatively and positively charged species in dynamic field gradient focusing using a dual polarity electric field

Development of a membrane-less dynamic field gradient focusing device for the separation of low-molecular-weight molecules

Counterflow isotachophoresis in a monolithic column

Stacking in a continuous sample flow interface in capillary electrophoresis

EOF measurement by detection of a sampling zone with endchannel amperometry in microchip CE

A Two-Step Method for Rapid Characterization of Electroosmotic Flows in Capillary Electrophoresis

Using Charge Ladders and Capillary Electrophoresis to Measure the Charge, Size, and Electrostatic Interactions of Proteins

Hydrodynamic radius ladders of proteins

Determination of effective protein charge by capillary electrophoresis: effects of charge regulation in the analysis of charge ladders

Measuring the size and charge of proteins using protein charge ladders, capillary electrophoresis, and electrokinetic models of colloids

Low Reynolds number hydrodynamics with special applications to particulate media

Concentration distributions in free zone electrophoresis

Predicting Peak Symmetry in Capillary Zone Electrophoresis: The Concept of the Peak Shape Diagram

and the GROMACS development team. GROMACS User Manual Version

The molecular basis of subtype selectivity of human kinin G-protein-coupled receptors

Elucidating the Role of Disulfide Bond on Amyloid Formation and Fibril Reversibility of Somatostatin-14 RELEVANCE TO ITS STORAGE AND SECRETION

Pressuredependent changes in the solution structure of hen egg-white lysozyme

Quantitative mass spectrometry in proteomics

Simultaneous evaluation of ligand binding properties and protein size by electrophoresis and Taylor dispersion in capillaries

Stable cationic capillary coating with successive multiple ionic polymer layers for capillary electrophoresis

Capillary zone electrophoresis with electroosmotic flow controlled by external radial electric field

Identification of Novel Biomarkers for Sepsis Prognosis via Urinary Proteomic Analysis Using iTRAQ Labeling and 2D-LC-MS/MS

Automatic assignment of metal-containing peptides in proteomic LC-MS and MS/MS data sets

Improved side-chain torsion potentials for the Amber ff99SB protein force field

High performance molecular simulations through multi-level parallelism from laptops to supercomputers

The Amber biomolecular simulation programs

Calculation of the isoelectric point of tryptic peptides in the pH 3.5-4.5 range based on adjacent amino acid effects

The Focusing Positions of Polypeptides in Immobilized pH Gradients Can Be Predicted from Their Amino Acid Sequences

Very fast empirical prediction and rationalization of protein pKa values

PROPKA3: Consistent Treatment of Internal and Surface Residues in Empirical pKa Predictions

Article