Protein charge determination and implications for interactions in cell extracts

Abstract

Decades of dilute‐solution studies have revealed the influence of charged residues on protein stability, solubility and stickiness. Similar characterizations are now required in physiological solutions to understand the effect of charge on protein behavior under native conditions. Toward this end, we used free boundary and native gel electrophoresis to explore the charge of cytochrome c in buffer and in Escherichia coli extracts. We find that the charge of cytochrome c was ∼2‐fold lower than predicted from primary structure analysis. Cytochrome c charge was tuned by sulfate binding and was rendered anionic in E. coli extracts due to interactions with macroanions. Mutants in which three or four cationic residues were replaced with glutamate were charge‐neutral and “inert” in extracts. A comparison of the interaction propensities of cytochrome c and the mutants emphasizes the role of negative charge in stabilizing physiological environments. Charge–charge repulsion and preferential hydration appear to prevent aggregation. The implications for molecular organization in vivo are discussed.

Introduction

Protein surfaces are rich in ionizable residues that carry a formal charge within the physiological pH range.1, 2 Noncovalent interactions that involve charged groups act across the greatest distances3, 4, 5 and impact significantly on protein behavior. For instance, there is considerable literature reporting the effects of charged residues on the kinetics6, 7, 8 and thermodynamics9, 10 of protein recognition. Charge–charge attraction is frequently exploited in the design of biosupramolecular systems including DNA origami,11 coacervate‐based protocells,12, 13 and protein assemblies.14

Although the influence of net charge on protein behavior is well‐appreciated, protein charge is rarely measured. And previous charge determination studies have revealed disparities between the experimentally derived and computational values for proteins.15, 16, 17 This disparity emphasizes the complex, co‐operative nature of charge–charge interactions on polyelectrolyte surfaces. Studies of protein charge, a system property, therefore remain crucial for developing models of protein “charge structure”15 and for understanding specific ion effects,18, 19, 20, 21, 22 posttranslational modifications,23 and ligand binding.24, 25

The significance of charge–charge interactions in cellulo is implicit in their long‐range nature and the abundantly Coulombic character of physiological environments.26, 27, 28 For instance, protein charge produces a potential energy ‘halo’ that extends from the van der Waals surface of the protein for up to 10 Å at physiological ionic strengths.3, 29 A similar distance separates macromolecules in the crowded cytoplasm. Thus screened charge–charge interactions are of biological importance.26, 29 Furthermore, the isoelectric point distributions of all proteomes are generally bimodal, demarcating cationic and anionic protein populations.27, 28 While the paucity of proteins with pIs between pH 7.4–7.6 is universal, the relative abundance of cationic and anionic proteins varies according to species and subcellular location.28 Approximately 90% of the most abundant proteins in the E. coli cytoplasm are anionic at physiological pH suggesting that charge–charge repulsion is critical in stabilising the cytoplasm against random aggregation.27 The term “quinary structure”30 has emerged to denote the charge‐directed interactions that transiently link functionally‐related complexes in vivo. Advancements in the fields of NMR spectroscopy, fluorescence methods and electron microscopy have resulted in an improved understanding of charge effects and quinary structure in cells30, 31, 32, 33, 34, 35 or similar crowded environments.5, 36, 37, 38, 39 However, studies of macromolecular charge structure in live cells are currently unattainable and are scarcely performed in native‐like solutions.3, 5, 29

Various free‐solution electrophoretic methods exist for determining the charge of purified proteins.2, 15 Of these, real‐time electrophoretic mobility membrane confined electrophoresis (REM‐MCE) is straightforward, fast, and accurate.15 This technique measures the electrophoretic mobility (μ) of a protein by using moving boundary electrophoresis.40, 41 From μ, the effective valence of the protein (Z _eff; the unitless ratio of the Coulombic charge to the elementary proton charge) can be calculated as [Eq. (1)]:15

$$\mu \hspace{0.17em}=\hspace{0.17em}{Z}_{\text{eff}}{Q}_{\mathrm{p}}/6\pi R_{\mathrm{s}}\eta$$

where Q _p is the proton fundamental charge, R _s is the Stokes radius of the protein, and η is the solvent viscosity. REM‐MCE can be used to explore protein charge in solutions across a range of ionic strengths and pH values to reveal preferential ion interactions with the test protein surface. MCE instrumentation can also be used for steady state electrophoresis (SS‐MCE). In SS‐MCE experiments, the magnitude of the applied electric field is relatively low (i.e., 0.05–0.2 V/cm) and macroion flux in the electric field is countered by macroion flux due to diffusion. When the steady state is reached (10–12 h), the flux of ions due to electrophoresis is balanced by the flux due to diffusion and a stable macroion concentration gradient forms. The macroion concentration gradient is exponential with respect to the x axis (position in cm) and can be fit to the following equation to yield the effective charge [Eq. (2)]:41

$$c\left(x\right)\hspace{0.17em}=\hspace{0.17em}\sigma \hspace{0.17em}+\hspace{0.17em}{c}_{\mathrm{o}}\hspace{0.17em}{\text{exp}}^{[{\mathrm{Z}}_{\text{eff}}[E/(k_{B}T)](x-x{}_{\mathrm{o}})]}$$

where σ is the baseline offset (cm⁻¹), c _o is the protein concentration (mg/mL) at an arbitrary initial reference point x _o (cm), T is the temperature (K), E is the electric field (V/cm), and k _B is Boltzmann's constant (1.3807 × 10⁻²³ J/K). Native gel electrophoresis can also be used to measure protein charge in dilute or complex solutions (i.e., extracts42) and is routinely applied to monitor complex formation.42, 43 This approach is therefore suited to the comparison of test protein charge in simple and physiological samples. Herein, REM‐MCE, SS‐MCE, and native gel electrophoresis were used to probe the charge of cytochrome c at pH 7.0 in dilute, simple solutions, or in concentrated E. coli extracts, which mimic the crowded, heterogeneous cytoplasm.

In buffers containing KCl, KNO₃, K₂SO₄, or MgCl₂ at an ionic strength (I) of 100 mM, the measured formal valence of cytochrome c was ∼2‐fold lower than predicted from its primary structure. REM‐MCE revealed the decrease in cytochrome c valence effected by sulfate binding or charge‐inverted (Arg/Lys → Glu) mutations (4 mutants referred to as single–quadruple). The successive decrease in charge in the mutant series correlated with the declining “stickiness” of each mutant in concentrated E. coli extracts.37 Importantly, cytochrome c and the single and double mutants behaved as anions in E. coli extracts, due to interactions with macroanions. In contrast, the triple and quadruple mutants were charge neutral and apparently “inert”. These data emphasize the prominent role of anionic particles in physiological solutions. The implications of these findings are considered in light of the current models for intracellular molecular organization.

Results and Discussion

Cytochrome c charge is low and salt‐dependent

REM‐MCE was used to determine the electrophoretic mobility of cytochrome c at pH 7.0 in 10 mM BIS–TRIS propane (BTP) plus KCl, KNO₃, MgCl₂, or K₂SO₄ at an ionic strength of 100 mM (Supporting Information Fig. S1). KCl is the standard salt for MCE studies owing to the near identical transport properties of the K⁺ and Cl^‐, which reduces electroosmotic flows.15, 16, 40 Preferential ion–protein interactions can be identified by comparing protein mobility in buffers containing different salts.16, 19 Salts composed of simple, monovalent ions are preferred.15, 16, 21, 40 To eliminate differences in heme iron valence, only the oxidized protein was studied by REM and SS‐MCE. Table 1 lists the electrophoretic mobility (μ), the effective valence (Z _eff), and the Debye–Huckel–Henry valence (Z_DHH) of cytochrome c in each buffer. Z_eff represents the solution‐specific valence of the protein with its associated counterion cloud. Z_DHH corrects for counterion size, the electrophoretic effect and electrostatic screening to provide an estimate of the formal protein valence.19, 21 Z_DHH is therefore a useful indicator of preferential ion interactions.15, 16 The Z_eff of cytochrome c was ∼3 times less than the Z_DHH as predicted from MCE theory.15 For clarity, only the Debye−Huckel−Henry valence will be quoted hereafter.

Table 1.

Average μ, Z_eff, and Z_DHH Values for Cytochrome c Determined by REM‐MCE ^a

Salt (I = 100 mM)	μ × 10−5 cm2/s V	Z eff	Z DHH
KCl	6.15 (±0.03)	1.34 (±0.01)	4.10 (±0.03)
KNO3	5.53 (±0.02)	1.21 (±0.01)	3.71 (±0.02)
MgCl2	5.24 (±0.06)	1.15 (±0.01)	3.49 (±0.08)
K2SO4	∼0	∼0	∼0

^aThe average standard deviation, in parentheses, calculated from two measurements.

In buffers containing KCl, KNO₃ or MgCl₂ the cytochrome c valence ranged from +3.50 to +4.10 (Table 1). The calculated valence of ferric cytochrome c is +7.8 at pH 7.0 assuming model pK _a values.44 However, presently there is no means of calculating an accurate protein valence. The approximate two‐fold difference between the calculated and measured cytochrome c valence highlights the limitations of such “back‐of‐the‐envelope” calculations, which assume; (1) constant pK _a values and (2) that only proton binding modulates protein charge. The sulfate‐bound cytochrome c has a calculated charge of +5.8 (PDB 1YCC, crystals grown from ∼90% sat. ammonium sulfate).45 Interestingly, in K₂SO₄ solutions, cytochrome c did not form a concentration gradient in positive or negative electric fields indicating that it was charge neutral (Supporting Information Fig. S1). Membrane polarization can occur during REM‐MCE and hamper boundary formation.40 SS‐MCE, the most accurate method for protein valence determination,16, 19, 21 employs electric fields tenfold lower than those used in REM‐MCE to eliminate boundary broadening due to membrane polarization. However, concentration gradients never formed during SS‐MCE experiments of cytochrome c in K₂SO₄ solutions (data not shown). Thus, the diminished cytochrome c valence likely arises from sulfate binding. Preferential sulfate binding was identified previously by free solution electrophoresis for other cationic proteins including lysozyme from T4 bacteriophage16 and hen egg white19 as well as RNase.21 The neutralization of cytochrome c by interactions with sulfate is supported by the observation of sulfate binding sites in crystal structures.45 Similarly, cytochrome c interactions with the supramolecular anion sulfonato‐calix[4]arene46 occur via a C^αNN structural motif47 suggesting that this site is optimized for anion binding. Attempts to explore cytochrome c valence changes in calixarene‐containing solutions by REM‐MCE were hampered by interactions between the confining membrane and the calixarene (Supporting Information Fig. S2). The effects of physiologically‐relevant anions (e.g., glutamate, phosphates, and polycarboxylates) cannot be studied for the same reason.21, 40 Owing to the similar geometry, radii, and anionic nature of sulfates and phosphates,48 it is expected that phosphate binding would effect comparable changes in cytochrome c valence. Notably, there is structural evidence of phospholipid (cardiolipin) binding to cytochrome c.49

Specific ion effects on cytochrome c interactions in E. coli extracts

Previously, size exclusion chromatography (SEC) was used to identify specific ion effects on macromolecular assemblies in E. coli extracts.18, 37 Cytochrome c was observed to elute in the high molecular weight fractions when the SEC buffer contained 100 mM potassium glutamate,37 the physiologically abundant salt which is present at concentrations upwards of 100 mM in E. coli.50 This result suggested that cytochrome c was bound to macromolecules in the E. coli extract (i.e., preferentially solvated3 by neighboring macromolecules). The interactions of cytochrome c were partially disrupted when the buffer contained 100 mM sodium chloride.37 Here, SEC was employed to determine whether cytochrome c–sulfate interactions prevail in physiological environments. In buffers containing 35 mM K₂SO₄ (I = 100 mM) cytochrome c eluted from the SEC column in the high molecular weight fractions (50–65 mL) with a minor amount in the low molecular weight fractions (70–80 mL; Fig. 1), where pure cytochrome c typically elutes (Supporting Information Fig. S3). A similar elution profile was obtained in buffers containing 100 mM KCl or KNO₃ although cytochrome c was absent from the 70–80 mL fractions in the latter cases (Fig. 1). Thus, it appears that most of the charged‐based interactions of cytochrome c with E. coli macromolecules were not disrupted in the presence of sulfate, nitrate, or chloride. It is likely that the high effective concentration of negative charge found at macroanionic surface patches19, 26, 51 combined with the high surface‐to‐volume ratios in cell extracts renders cytochrome c interactions with macromolecules more favorable than cytochrome c interactions with simple anions.

Figure 1.

SDS‐PAGE analysis of size exclusion chromatograms of DNase I‐treated E. coli cell extracts containing overexpressed cytochrome c. The SEC buffers were 20 mM Na₂HPO₄ or TRIS–HCl (pH 7.0) and contained K₂SO₄, KCl, KNO₃, or MgCl₂, at an ionic strength of 100 mM. The arrow marks the migration position of cytochrome c. The gel lanes are labeled; MM: molecular weight marker; CE: cell extract; 45–80: fraction volume (mL).

Interestingly, in buffers containing 35 mM MgCl₂ (I = 100 mM) cytochrome c eluted exclusively in the low molecular weight fractions (75 and 80 mL; Fig. 1). These fractions yielded the typical cytochrome c HSQC spectrum confirming the identity and purity of the protein (Supporting Information Fig. S4). Mg²⁺ binds tightly to biological oxyanions such as phosphates and carboxylates.18, 51, 52 The disruptive effect of Mg²⁺ in extracts is expected to arise from its displacement of charge‐based interactions between cationic and anionic patches on macromolecular surfaces.18 Owing to the large number of lysine residues at the cytochrome c surface, it is likely that lysine−phosphate and/or lysine–carboxylate interactions drive cytochrome c complexation with nucleic acids and/or proteins, respectively. To elucidate the origin of its stickiness in extracts, cytochrome c interactions with nucleic acids were probed by pretreating the extracts with nucleases (RNase A or DNase I) prior to SEC in 20 mM Na₂HPO₄, 100 mM KCl, pH 7.0. Cytochrome c eluted in the high molecular weight fractions in both experiments (Fig. 1; DNase treated extracts only). The presence of cytochrome c in high molecular weight complexes, after nucleic acid digestion, indicates that cytochrome c interacts mainly with anionic proteins. Although protein–protein interfaces are typically devoid of lysine residues,53 the role of lysine in driving nonspecific interactions (lysine–carboxylate salt bridges) is implied here.

Charge‐inverted mutants

Protein characterization under conditions of reduced charge can reveal the importance of electrostatics in governing protein behavior.2, 16, 37, 41 For instance, it was demonstrated previously that charge‐inverted (Arg/Lys → Glu) cytochrome c mutants were less sticky in E. coli extracts compared to the wildtype protein.37 REM‐MCE was used to investigate the effective valence of a series of charge‐inverted cytochrome c mutants at pH 7.0 in buffers containing 10 mM BTP, 100 mM KCl. The series included a single (R13E), double (R13E/K73E), triple (R13E/K73E/K87E), and quadruple (R13E/K73E/K87E/K100E) mutant (Fig. 2). Table 2 shows the μ, experimentally‐derived, and calculated valences for the mutants. The Z _DHH and Z _cal for each protein have the same overall pattern: wildtype cytochrome c has the highest valence followed by the single, double and triple/quadruple mutants (Table 2). However, Z _cal is substantially greater than the Z _DHH and overestimates the valence change of each mutation.

Figure 2.

Electrostatic surface representations of wildtype cytochrome c and the charge‐inverted mutants, with positive and negative potentials colored blue and red, respectively. The mutant positions are labeled and the exposed heme edge is black.

Table 2.

Average μ, Z_DHH, and Z_cal values for Cytochrome c Mutants Determined by REM‐MCE in 10 mM BTP, 100 mM KCl at pH 7.0 ^a

Mutant	μ × 10−5 cm2/s V	Z DHH	Z cal
Single (R13E)	3.81 (±0.03)	2.50 (±0.02)	5.8
Double (R13E/K73E)	2.34 (±0.02)	1.56 (±0.02)	3.8
Triple (R13E/K73E/K87E)	∼0	∼0	1.8
Quadruple (R13E/K73E/K87E/K100E)	∼0	∼0	−0.8

^aThe average standard deviation, in parentheses, calculated from two measurements.

The triple and quadruple mutants did not form electrophoretic concentration gradients during REM or SS‐MCE experiments indicating that they have a net charge of ∼0 (see “Materials and Methods”, Supporting Information Figs. S5 and S6). Proteins with a valence ∼0 tend to self‐associate.20 Dynamic complexes and aggregates are unsuited to studies by REM‐MCE.15 Therefore, sedimentation‐velocity analytical ultracentrifugation (SV‐AUC) was used to check the oligomeric state of representative samples. The wildtype, double and quadruple mutant yielded a single species with an average sedimentation coefficient (s) of 1.74, which corresponds to the molecular weight of the monomer (Fig. 3). Therefore, the lack of electrophoretic gradient formation in the triple and quadruple samples was not the result of oligomerization or aggregation. Similarly, SEC analysis of the pure triple and quadruple mutants was consistent with a 13 kDa monomer, even at total protein concentrations of 1 mM (12.7 mg/mL, Fig. 3). Together these data indicate that although the triple and quadruple have a net charge of ∼0 they remain monomeric and water‐soluble.

Figure 3.

(A) A representative sedimentation coefficient distribution obtained from a 0.5 mg/mL sample of the quadruple mutant in 10 mM BTP, 100 mM KCl, pH 7.0. The sedimentation distribution shows a single peak at 1.69 s with no evidence of aggregation at higher s values. (B) SDS‐PAGE analysis of the elution profile of the pure quadruple mutant (12.7 mg/mL) during SEC in 10 mM TRIS–HCl, 100 mM KCl pH 7.0. The gel lanes are labeled; MM: molecular weight marker; pure: pure sample before SEC; 45–80: fraction volume (mL). The arrow marks the migration position of the quadruple mutant.

Electrophoresis reveals preferential interactions of cytochrome c and mutants

To explore the charge and interactions of cytochrome c and the mutants under conditions of greater physiological relevance, native gel electrophoresis was used to compare protein migration in buffer and in E. coli extracts (Fig. 4). Note that the extracts used for native gel electrophoresis were prepared from nonexpressing E. coli cultures grown to saturation in LB (see “Materials and Methods”). This approach was expected to yield extracts with higher macromolecular concentrations and more native‐like compositions compared to those prepared from lower density cultures, overexpressing a test protein. Qubit fluorometry revealed that the total protein, RNA, and DNA concentrations of DNase I‐treated extracts prepared from saturated LB cultures was approximately 80, 7, and 1 mg/mL, respectively. Therefore, these extracts can be considered moderately crowded3 and heterogeneous. Extracts prepared from E. coli cultured on minimal medium and overexpressing 15N‐labeled cytochrome c were approximately half as concentrated. This finding is supported by SDS‐PAGE analysis comparing the total protein content of both extract types (Supporting Information Fig. S7).

Figure 4.

2% agarose gels showing the migration of cytochrome c and the mutants in buffer and in E. coli extracts. The electrophoresis buffer contained 20 mM Na₂HPO₄ or TRIS–HCl at pH 7.0 plus the indicated salt at an ionic strength of 100 mM. Cytochrome c and mutants are red or pink (due to the oxidized or reduced heme, respectively) and can therefore be easily distinguished on the gel. The gel lanes are labeled WT, SM, DM, TM and QM to indicate the presence of cytochrome c and the single, double, triple and quadruple mutants, respectively.

The native gel experiments were performed in four different electrophoresis buffers at pH 7.0 containing 20 mM Na₂HPO₄ or TRIS–HCl ± KNO₃, K₂SO₄, or MgCl₂ at an ionic strength of 100 mM. In keeping with the MCE data, pure cytochrome c always migrated toward the cathode with a migration distance that was short (e.g., 5 mm in 20 mM Na₂HPO₄, pH 7.0) despite the relatively high applied voltage and long run times (see “Materials and Methods”). This result confirmed that cytochrome c is weakly cationic. The migration distance of cytochrome c was shorter in the presence of buffers containing KNO₃ or K₂SO₄ at an ionic strength of 100 mM compared to buffer alone (Fig. 4) which demonstrates the dampening of cytochrome c charge by anions. The reduction in cytochrome c charge effected by charge inversion was also evident. For example, the migration of the triple and quadruple mutants was poor in buffer and in the presence of KNO₃ or K₂SO₄ confirming that they are charge neutral (Fig. 4). In the presence of Mg(NO₃)₂ cytochrome c and the mutants migrated toward the cathode. Interestingly, the migration distance of cytochrome c was greater in buffer containing 35 mM Mg(NO₃)₂ than in buffer alone (Fig. 4). This finding disagrees with the REM‐MCE data which showed that the μ of cytochrome c was lower in buffers containing MgCl₂ than those containing KCl or KNO₃ (Table 1). In addition, the cytochrome c and mutant bands were thin and diffuse in the presence of Mg(NO₃)₂ (Fig. 4). Perhaps such aberrant protein migration patterns occur as a result of Mg²⁺ interactions with agarose‐bound anions.54 We suspect that Mg²⁺, a potent disruptor of protein interactions,18, 51 displaces cytochrome c (and mutant) interactions with the agarose‐bound anions.

When present in E. coli extracts the behavior of cytochrome c was substantially altered, with migration towards the anode (Fig. 4). The diffuse nature of the cytochrome c band in cell extracts suggests that it interacted nonspecifically with numerous anionic proteins or assemblies. The migration distance of the cytochrome c‐containing assemblies was shorter in the presence of salt (KNO₃ or K₂SO₄ at I = 100 mM) compared to buffer alone (no salt) indicating that these assemblies had a lower net charge at physiological ionic strength (i.e., they were charge‐screened; Fig. 4). However, cytochrome c interactions were not disrupted in buffers containing KNO₃ or K₂SO₄. These charge‐screened assemblies are thus reminiscent of the anionic “clusters” expected to pervade the cytoplasm.26, 27 When the electrophoresis buffer contained 35 mM Mg(NO₃)₂, cytochrome c (and the mutants) migrated toward the cathode, demonstrating the disruption of complexes by Mg²⁺ cations (Fig. 4). The extensive precipitation observed near the loading well highlights the deleterious effects of macromolecular charge neutralization in crowded environments, and supports the role of (screened) charge–charge repulsion in stabilizing the cytoplasm.5, 26

In low ionic strength buffer, the single mutant migrated toward the anode as part of a macroanionic complex in the E. coli extract (Fig. 4). When the buffer contained KNO₃ or K₂SO₄, low concentrations of the single mutant migrated toward the cathode also. This result suggests that the interactions between the single mutant and anionic E. coli proteins are weaker than those of cytochrome c. An even greater concentration of the double mutant migrated toward the cathode indicating that its interactions with E. coli proteins were further weakened. The migration of the triple and quadruple mutants were closely similar in both dilute and cell extract samples irrespective of the electrophoresis buffer (Fig. 4), confirming that these mutants did not interact with E. coli macroions.37 The triple mutant contains 20 cationic residues (arginine, lysine, histidine) and 14 anionic residues (aspartate and glutamate). The heterogeneous charge distribution of the triple mutant (Fig. 2) likely renders the protein highly soluble, despite its low net charge. Notably, glutamate and aspartate contribute most favorably to protein solubility,55 possibly by facilitating water clustering and structure formation (i.e., electrostriction) at protein surfaces.51, 55, 56 Crystal structure analyses show that glutamate and aspartate carboxylates are frequently excluded from protein–macromolecule interfaces interacting, instead, with water.57 Differences in the interaction propensities of cytochrome c and the mutants in extracts suggests that anionic side chains play a similar role in physiological environments, that is, disfavoring protein–macromolecule interactions through charge–charge repulsion and preferential hydration.26, 55

Conclusions

Despite the significance of charge on biomolecular interactions in vivo, studies of biomolecular charge in complex, physiological environments are rare. A combination of REM‐MCE, SS‐MCE, native gel electrophoresis, and SEC was used herein to compare the valence and interactions of cytochrome c in simple and native‐like solutions. The experimentally‐derived valence of cytochrome c was ∼2‐fold lower than the calculated value and was tuned by anion binding. Although weakly cationic in buffer, cytochrome c was rendered anionic by interactions with proteins in E. coli extracts. This finding supports the idea that anionic assemblies or clusters crowd the cytoplasm. The drastic difference in cytochrome c valence in buffer versus extracts calls into question the validity of dilute solution studies. The use of protein charge characterization as a means of exploring quinary‐like interactions is also apparent.

Variations in the electrostatic potential of the cytochrome c surface were affected by a series of K/R → E mutations. Electrophoresis revealed that mutants carrying a net positive valence in buffer became anionic in E. coli extracts. The triple and quadruple mutants were charge neutral in buffer and extracts. These findings suggest that the increased concentration of glutamate residues on the surface of the triple and quadruple mutants cause them to interact preferentially with water. This result supports the role of negative charge in maintaining the stability of the E. coli cytoplasm through charge–charge repulsion and preferential hydration.26, 27, 57

Materials and Methods

Protein production

Saccharomyces cerevisiae cytochrome c C102T and its mutants were expressed and purified according to previously described methods.37 After purification, the proteins were oxidized with a 2–3‐fold excess of K₃[Fe(CN)₆] at 4°C.

Membrane confined electrophoresis

Comprehensive descriptions of the MCE instrument have been reported previously.15, 40, 41 All MCE experiments were performed at 20°C after at least 2 h of complete system equilibration. A constant flow (10 mL/h) of buffer (20 mM BIS−TRIS propane plus KCl, KNO₃, MgCl₂, or K₂SO₄ at an ionic strength of 100 mM, pH 7.0) was used to condition the sample‐confining membranes (8 kDa molecular weight cutoff). Oxidized protein samples were studied at a concentration of 0.5 − 1 mg/mL. During REM‐MCE experiments, currents of ±1 mA were applied across the cuvette such that electric fields of ±2–2.5 V/cm were achieved. Initially, both positive and negative electric fields were applied to each sample and 250 absorbance scans (at 230 nm) were acquired for each electric field. Proteins that did not form concentration gradients in response to positive or negative electric fields at magnitudes of either 2 − 2.5 or 20 − 25 V/cm or during SS‐MCE experiments were assumed to have a net charge of zero. SS‐MCE experiments were performed in electric fields of ±187 or 374 mV/cm over 24 h and absorbance scans were acquired every 15 min.

The MCE data were analysed using the Spin Analytical MCE Analysis Software. The electrophoretic mobility was converted to Z _eff and Z _DHH using ZUtilities (http://www.rasmb.org/). A radius (R _s) of 1.65 nm58 was used for cytochrome c and the charge‐inverted mutants. Anionic radii of 0.121, 0.129, and 0.230 nm were used for Cl⁻, ${\text{NO}}_3^{-}$, and SO₄ 2 ⁻, respectively.19

Size exclusion chromatography

SEC was performed at 21°C on an Åkta FPLC using an XK 16/70 column (1.6 cm diameter, 65 cm bed height) packed with Superdex 75 (GE Healthcare).18, 37 Prior to sample injection, the column was equilibrated with 150 mL of SEC buffer using a continuous flow rate of 1.5 mL min⁻¹. Cell extract samples (850 μL) were injected onto the column and 1 mL fractions were collected. Sample elution was monitored at 280 nm. The elution buffer was 20 mM TRIS–HCl or Na₂HPO₄ plus a salt at 100 mM ionic strength. The pH was adjusted to 7.0 and the buffers were filtered and degassed.

Sedimentation‐velocity analytical ultracentrifugation

SV‐AUC experiments were performed using a Beckman Coulter XLA analytical ultracentrifuge. Data were acquired at 50,000 rpm at 20°C in 10 mM BTP, 1 mM ascorbic acid, 100 mM KCl, pH 7.0. Reduced protein samples of 0.05, 0.15, and 0.5 mg/mL were equilibrated at 20°C prior to ultracentrifugation. Sedimentation was monitored at 550 nm (pathlength = 1.2 cm). The concentration of cytochrome c was calculated assuming an extinction coefficient of 27.5 mM ^– 1 cm⁻¹.37 The data were fit using SEDFIT59 and plotted in GUSSI.

E. coli extract preparation and analysis

Cell extracts were prepared from minimal medium cultures expressing 15N‐labeled cytochrome c as previously described18, 37 or from saturated LB cultures (OD₆₀₀ = 6.5–7.0).60 Briefly, cells were harvested from 50 mL of culture, resuspended in 1 mL of 20 mM Na₂HPO₄, pH 7.0, and frozen (−20°C) overnight. After thawing, cell lysis was completed by sonication. The lysate was treated with 10 μg/mL DNase I or RNase A and centrifuged for 30 min at 20,000g at room temperature. Qubit fluorometry was used to determine the total protein, RNA and DNA concentration of the extracts.

Extracts prepared from cultures overexpressing 15N‐labeled cytochrome c were used for SEC analysis and were observed to precipitate after 40 min at 21 or 30°C. The extracts were stable for >6 h when stored on ice. Extracts prepared from nonexpressing cultures were DNase I‐treated and used for agarose gel electrophoresis after the addition of 0.2 mM cytochrome c or the mutants. These extracts were stable for up to 8 h on ice or at room temperature but the stability decreased upon addition of cytochrome c (precipitation occurred after 1.5 h at 30°C).

Native gel electrophoresis

20 μL samples of 0.2 mM cytochrome c or the charge‐inverted mutants were analysed in 2% agarose gels (13.5 cm × 14.0 cm) prepared in 20 mM Na₂HPO₄, or 20 mM TRIS–HCl at pH 7.0. High voltage agarose gel electrophoresis is beset by resistive heating which leads to gel melting. To offset the effects of resistive heating, the native gel electrophoresis experiments were performed at 4°C. Gels were equilibrated in the electrophoresis buffer for 30 min at 4°C prior to running at a constant voltage (100 V) for 30 min at 4°C. The buffer temperature was 6–12°C after electrophoresis, indicating that the gels were not heated substantially during the experiment. The gels were imaged directly after electrophoresis using a flatbed scanner. Each gel was processed identically using Adobe Photoshop. Changes to the color balance, saturation, brightness and contrast were applied uniformly across the images.

NMR spectroscopy

¹H, 15N HSQC spectra were acquired at 30°C with 8 scans and 64 increments on a Varian 600 MHz Spectrometer equipped with a HCN coldprobe. The spectra were processed in NMRPipe61 and analyzed in CCPN.62

Disclosure

T.M.L. and K.J. are the Spin Analytical chief scientific officer and product support specialist, respectively. Spin Analytical produces the membrane confined electrophoresis instrument. The MCE data presented herein were not interpreted nor presented with a view toward financial gain.

Supporting information

Supporting Information