Probing the proton channels in subunit N of Complex I from Escherichia coli through intra-subunit cross-linking

Abstract

Respiratory Complex I appears to have 4 sites for proton translocation, which are coupled to the oxidation of NADH and reduction of coenzyme Q. The proton pathways are thought to be made of offset half-channels that connect to the membrane surfaces, and are connected by a horizontal path through the center of the membrane. In this study of the enzyme from Escherichia coli, subunit N, containing one of the sites, was targeted. Pairs of cysteine residues were introduced into neighboring α-helices along the proposed proton pathways. In an effort to constrain conformational changes that might occur during proton translocation, we attempted to form disulfide bonds or methanethiosulfonate bridges between two engineered cysteine residues. Cysteine modification was inferred by the inability of PEG-maleimide to shift the electrophoretic mobility of subunit N, which will occur upon reaction with free sulfhydryl groups. After the cross-linking treatment, NADH oxidase and NADH-driven proton translocation were measured. Ten different pairs of cysteine residues showed evidence of cross-linking. The most significant loss of enzyme activity was seen for residues near the essential Lys 395. This residue is positioned between the proposed proton half-channel to the periplasm and the horizontal connection through subunit N, and is also near the essential Glu 144 of subunit M. The results suggest important conformational changes in this region for the delivery of protons to the periplasm, or for coupling the actions of subunit N to subunit M.

Graphical abstract

1. Introduction

Complex I plays a key role in oxidative phosphorylation by coupling the citric acid cycle to the generation of proton motive force, which is used for ATP synthesis. This enzyme is typically found embedded in the inner membranes of mitochondria in eukaryotes, or in the plasma membranes of bacteria (for reviews see [1,2]). Structurally, the enzyme is L-shaped, and comprises a peripheral arm that faces the inside of the mitochondrion or bacterial cell, and a membrane arm that is integral to the membrane. In all species examined so far, the membrane arm is formed from seven core subunits. In mammals they are coded for by mitochondrial genes. In Escherichia coli, the subject of this study, the seven subunits are coded by the nuo operon, nuoA, nuoH, and nuoJ-N, along with another six genes that code for the subunits of the peripheral arm [3]. It is thought that there are four sites for translocation of protons, corresponding to the likely ratio of H⁺/NADH of 4. Three sites are proposed to be found in each of the three similar subunits L, M and N. A fourth site was proposed to be formed by subunits A, H, J and K [4]. In each case the proposed proton pathway consists of two half-channels from the membrane surface that are connected by a horizontal channel that runs through the center of the membrane.

The crystal structures of bacterial Complex I showed that the three subunits L, M and N have nearly identical conformations [4,5]. The primary difference is the extra structural component of subunit L, consisting of two TM (transmembrane) spans connected by helical segments that run along the cytoplasmic surface (Fig. 1). Cross-linking studies [6], and the results of several other approaches [7–11], have indicated that this helical element is likely to have a structural role, rather than a dynamic role in proton translocation. Apart from this extra structural component of subunit L, each of these three subunits consists of 14 α-helical TM spans. In each subunit, two of the TM spans are formed by helices with a discontinuous region of 5–8 amino acids near the center of the membrane. Each so-called broken helix is part of a bundle of 5 TM helices, which we refer as a sector. Sector 1 is formed by TM4–8 and sector 2 is formed by TM9–13. Within each subunit the two sectors have pseudo-symmetry, and are related by a two-fold screw axis centered in the membrane [5]. So, within each subunit, the sectors and the broken helices have opposite orientations with respect to the membrane surfaces. Each of the broken helices is associated with a predicted half-channel that runs from the center of the membrane to the cytoplasmic or periplasmic surface. Within each subunit, each broken helix is connected to the other broken helix by a horizontal, water-containing channel. The predicted path of protons within L, M or N subunits during enzyme turnover is through a half-channel from the cytoplasm, followed by horizontal transport through the subunit, and, finally, exit to the periplasm by another half-channel.

Figure 1.

Structure of the bacterial Complex I from Thermus thermophilus. A. The entire Complex I is shown, with the peripheral subunits colored gray. The prosthetic groups, 1 FMN and all Fe-S clusters are shown in space filling, and colored by the CPK scheme. The membrane subunits are colored individually and are named according to the E. coli nomenclature: H-magenta, A-orange, J-green, K-yellow, N-light gray, M-blue, and L-red. The image is developed from PBD file 4hea [4]. B. The membrane arm of bacterial Complex I from Thermus thermophilus. The molecule is rotated 90° from panel A, and the view is from the cytoplasmic surface. The coloring scheme is the same as in panel A, except for subunit N. In subunit N, TM4–8 are colored light blue (sector 1), and TM9–13 are colored pink (sector 2). The remaining helices are colored light gray: TM1–3 and TM14. The image is developed from PDB file 4he8 [4].

Several amino acid residues have been identified in each subunit by mutagenesis that are thought to mediate the transport of protons [12–19]. A strictly conserved Lys residue in broken helix TM7 (Lys 217 in nuoN) is found at the center of the membrane near the end of a half-channel from the cytoplasm. Nearby is a conserved Glu in TM5 (Glu 133 in nuoN), which is also at the interface with an adjacent subunit. In the other broken helix, TM12, a conserved Lys or Glu residue (Lys 395 in nuoN) is found, also at the center of the membrane, and at the end of a half-channel from the periplasm. Several other polar residues, conserved among the individual subunits, are found along the putative proton pathways. The tight coupling of Complex I activity is indicated by the observation that mutagenesis of a single residue in the membrane arm can nearly totally eliminate both NADH oxidation and proton translocation. The basis for the tight coupling of quinone reduction to proton translocation is unknown. For the proton translocation site nearest to the site of quinone reduction, developing negative charge on the quinone during reduction could directly drive local conformational changes, leading to modulation of the pK_a values of key amino acid residues. For proton translocation that occurs at downstream sites there would seem to need to be a network of linked conformational changes between subunits J-K-N-M-L.

In the work described here, based on the proposed proton pathways in subunit N [4,5], and where conformational changes might occur, we engineered cysteine residues that were positioned to form cross-links. Sites were chosen primarily in the helical regions of the TM helices in each of the two sectors. The results showed that some of the cross-links in sector 2, in the vicinity of Lys 395 of TM12, were able to impair activity of Complex I.

2. Materials and Methods

2.1 Materials

All materials used were described in previous publications [6]. MTS (Methanethiosulfonate) cross-linking reagent, M2M (1,2-Ethanediyl bismethanethiosulfonate) was from Toronto Research Chemicals (Toronto, Canada). Methoxypolyethylene glycol maleimide (5,000 daltons) and poly(ethylene glycol) methyl ether maleimide (2,000 daltons), both referred to as PEG-maleimide, were purchased from Sigma-Aldrich (St. Louis, MO). Subunit N was detected by the monoclonal rat anti-HA (high affinity) antibody from Roche (Indianapolis, IN). Oligonucleotides for mutagenesis and sequencing were synthesized by Eurofins (Huntsville, AL). DNA sequencing was performed in Lone Star Labs (Houston, TX).

2.2 Methods

Plasmids pUC19-L’MN [13] or pUC19 L’MN-SpeI [6] were used for construction of mutations in nuoN. Both of the plasmids are 6.5 kb and contain a 3′ truncated gene for L and full length genes for subunits M and N. They differ only in that pUC19 L’MN-SpeI contains a unique SpeI endonuclease site immediately downstream of the nuoN gene. Mutations in N were transferred to the expression vector, pBA400, using two unique restriction sites. Cysteine substitutions were initially introduced into subunit N by QuikChange mutagenesis. The endogenous cysteine, residue 88, was changed to alanine, and all subsequent constructs were made in this background, which is referred to as “wild type”. Previous work had shown that mutagenesis of this residue had no effect on activity of the enzyme [12], and this was confirmed here. Two strategies were used for generating double Cys-substituted mutants. First, plasmids containing single Cys substitutions were digested with two endonucleases and the insert was ligated in the vector containing the second Cys mutation. Or, the plasmids containing the first Cys mutations were used as the templates in QuikChange mutagenesis for creating the second Cys mutations. The double Cys mutants were transferred into plasmid pBA400 or pBA400-SpeI [6], which contains a full size nuo operon and transformed into strain BA14 that carries a chromosomal deletion for the genes of all complex I subunits [20].

Each of the doubly Cys-substituted mutants were grown on M63 minimal salt media agar plates to confirm that all could grow with acetate as sole carbon source. Acetate plates contained 1.36% KH₂PO₄, 0.2% (NH₄)₂SO₄, 0.05% FeSO₄·7H₂O, 1.5% Agar, 0.02% MgSO₄, 0.001% vitamin B1 and 0.2% potassium acetate, pH 7.0. Growth was evaluated by visual inspection after 48–72 hours at 37°C. For membrane preparations all mutants and wild type E. coli cells were grown in rich media (3% tryptone, 1.5% yeast extract, 0.15% NaCl and 1% (v/v) glycerol) at 30° C and the cultures were harvested at A₆₀₀ = ~1.2. Cells were suspended in TMG buffer (50 mM Tris-HCl, 5 mM MgCl₂, 10% glycerol, pH 7.5) or TMG-Acetate buffer (50 mM Tris-acetate, 5 mM magnesium acetate, 10% glycerol, pH 7.5) and passed through a French Press at 14,000 psi. Membrane vesicles were prepared as described previously [6,13,20].

For cross-linking with Cu²⁺ membrane vesicles in TMG buffer were incubated with 1.5 mM CuCl₂ at 4° C for 1 hour. The cross-linking reaction was terminated by the addition of 15 mM Na₂EDTA and 20 mM NEM (N-ethylmaleimide) and was incubated at 4°C for 10 min. Membrane vesicles were mixed with an equal volume sample buffer (350 mM Tris·HCl, 10% SDS, 30% glycerol, 0.12 mg/ml bromophenol, pH 6.8) and incubated at 4° C for 1 hour before analyzing by SDS electrophoresis. For cross-linking with MTS reagents, a 25 mM methanethiosulfonate (MTS) Reagent solution was made in dimethyl sulfoxide immediately before cross-linking. Results are reported for 1,2-ethanediyl bismethanethiosulfonate (M2M), but other lengths were also tested. Membrane vesicles in TMG·acetate buffer were treated with 250 μM MTS reagent and incubated at 4°C for 10 min. The reaction was stopped by the addition of 20 mM Na₂EDTA at 4°C for 15 min. Membrane vesicles were incubated with an equal volume of sample buffer for 1 hour before analyzing by SDS electrophoresis. The procedures were adapted from those described previously, [21,22].

For SDS electrophoresis and immunoblotting, 40 μg of membrane protein were run on 12% acrylamide gels using procedures previously described [6]. PEG-maleimide was added to samples to 1 mM and incubated at room temperature for 1 hour before electrophoresis [23]. Blots shown are representative of 2 or 3 blots in total.

Enzymatic assays were conducted according to the methods described previously, with an average value of 165 ± 48 nmoles NADH/min per mg membrane protein [6,13,20]. In brief, deamino-NADH oxidase activity assays were started with 0.15 mM deamino-NADH (extinction coefficient 6.22 Mm⁻¹ cm⁻¹) and the absorbance was monitored at 340 nm for 2 minutes. In all assays, deamino-NADH was used, rather than NADH, to eliminate contributions from the alternative NADH oxidase found in E. coli membranes [24]. Proton translocation assays were conducted by measuring the fluorescence quenching of the acridine dye ACMA as a ΔpH indicator using excitation and emission wave lengths of 410 and 490 nm, respectively, using a PTI QuantaMaster 40 spectrofluorometer. Both deamino-NADH oxidase activity assays and proton translocation assays were conducted by using 150 μg/ml membrane protein in deamino buffer (50 mM MOPS, 10 mM MgCl₂, pH 7.3, adjusted with KOH) at room temperature. The 1 μM uncoupler FCCP (carbonyl cyanide p-(trifluoromethoxy) phenylhydrazone) from a 1 mM ethanol stock was added to eliminate the buildup of a proton gradient during NADH oxidase assays. For proton translocation assays, ACMA was added to 1 μM, while other concentrations were the same as for oxidase assays.

For structural analysis the Protein Data Bank file 3rko [5] was used for design of mutations and for structural analysis. It includes subunits A, J, K, L, M, and N of the membrane arm of the E. coli enzyme, and lacks subunit H of the membrane arm, and all of the subunits of the peripheral arm, B, CD, E, F, G, and I.

3. Results

To probe possible conformational changes in subunit N, sites for the introduction of cysteine residues were designed to cross-link pairs of TM helices within each sector. Sector 1 (TM4–8) is thought to contain the half-channel to the cytoplasm, and sector 2 (TM9–13) is thought to contain the half-channel to the periplasm. These half-channels are connected by a perpendicular channel through the center of subunit N. Residues to be mutated to cysteine were selected in adjacent helices, and the alpha carbons were measured to be within 4–8 Å [25], so that disulfides might be formed, or that the cysteines could be bridged by shorter bis-methanethiosulfonates [26].

3.1 Cross-linking at the cytoplasmic end of sector 1 had no effect on enzyme activity

The existence of a half-channel to the cytoplasm in sector 1 was supported in recent work [27,28] by molecular dynamics simulations, which showed access of numerous water molecules. Two pairs of cysteines were engineered in this region: L142C (in TM5) + V238C (in TM8) and S152C (in TM6) + A234C (in TM7), as shown in Fig. 2. In Fig. 2A the NADH oxidase activities of the wild type membranes, and the two double mutants, both before and after treatment with CuCl₂ (1.5 mM) for 15 min at room temperature. Before treatment, both of the double mutants had NADH oxidase activity that is about 90% of the wild type rate. After treatment, there is no appreciable loss of activity. Similarly, as shown in Fig. 2B, there is no significant loss of proton translocation after treatment as indicated by the quenching of ACMA fluorescence. Evidence for high yield cross-linking is provided in Fig. 2C, in which PEG-maleimide is used in conjunction with CuCl₂. At the right, when both reagents are omitted, the immunoblot shows the characteristic band of subunit N, which appears just above the 50 kilodalton standard. If PEG-maleimide is added to the SDS-denatured sample before electrophoresis, the band is shifted to a higher apparent molecular weight, consistent with the reaction of 2 free sulfhydryl groups with PEG-maleimide, each carrying an additional molecular mass of 2000 daltons. When the samples are treated with CuCl₂, without the PEG-maleimide, there is no apparent shift in electrophoretic mobility of subunit N. If PEG-maleimide treatment is applied after reaction with CuCl₂, the band for subunit N is not shifted, consistent with the loss of free sulfhydryl groups due to disulfide formation. In some cases it appears that PEG-maleimide modification of subunit N can interfere with antibody recognition, depending upon the site of cysteine substitution. This is likely due to interference of the PEG chain with the antibody, and is dependent upon the location of the mutation. The positions of the mutated residues are depicted in Fig. 2D: 142 (red) + 238 (yellow) and 152 (blue) + 234 (orange). The two sectors are colored pink (1) and cyan (2). The additional helices, TM1, 2, 3, and 14, are colored gray.

Figure 2.

Analysis of cross-linking at the Cytoplasmic side of sector 1 in subunit N (TM4–8). L142C (in TM5) + V238C (in TM8) and S152C (in TM6) + A234C (in TM7) were analyzed. A. The effect of CuCl₂ treatment on NADH oxidase activity of membrane vesicles from wild type, or from the two double mutants. Deamino-NADH is used as the substrate. The means and standard errors are shown for at least 4 measurements from at least 2 independent preparations. B. The effect of CuCl₂ treatment on the rates of NADH-driven proton translocation as indicated by quenching of the fluorescence of ACMA, for the two double mutants. Representative traces of 2–3 preparations are shown. C. Immunoblots of the two double mutants. PM indicates treatment with PEG-maleimide and Cu²⁺ indicated treatment with CuCl₂. The bands for subunit N and the shifted bands after PEG-maleimide-2000 treatment (N-PM) are indicated at the right. Molecular weight markers are shown at the left: 50,000 and 80,000. D. The location of the cysteine substitutions are indicated by space filling views of the original residues: L142 is in red, V238 is in yellow, S152 is in blue, and A234 is in orange. The helices are colored as in Figure 1B: Sector 1 in light blue and sector 2 in pink. At left, the view is from the cytoplasm. At right, the view is rotated 90°, with the cytoplasmic surface at the top. The image is developed from PBD file 3rko [5].

3.2 Cysteines at the periplasmic end of sector 1 could not be cross-linked

Three pairs of cysteine residues were engineered at the periplasmic end of sector 1: L128C (TM5) + L170C (TM6), V177C (TM6) + L203C (TM7), and L128C (TM5) + V253C (TM8), as shown in Figure 3. This end of sector 1 was predicted to be closed to protons or water [5]. The same analysis was carried out for these double mutants, before and after treatment with CuCl₂. No difference was detected in NADH oxidase activity (Fig. 3A) or NADH-driven proton translocation (Fig. 3C), but for these mutants there was no evidence of disulfide formation (Fig. 3D). Other attempts to form disulfides included treatment with Cu-phenanthroline, iodine and MTS bi-functional cross-linkers, but these methods also failed to produce cross-linked products. In Fig. 3B, the residue pairs are colored: 128 (blue) + 170 (red), 177 (yellow) + 203 (green), and 128 (blue) + 253 (orange).

Figure 3.

Analysis of cross-linking at the periplasmic side of sector 1 in subunit N (TM4–8). L128C (in TM5) + L170C (in TM6), V177C (in TM6) + L203C (in TM7) and L128C (in TM5) + V253C (in TM8) were analyzed. A. The effect of CuCl₂ treatment on NADH oxidase activity of membrane vesicles from the three double mutants. Deamino-NADH is used as the substrate. The means and standard errors are shown for at least 4 measurements from at least 2 independent preparations. B. The location of the cysteine substitutions are indicated by space filling views of the original residues: L128 is in blue, L170 is in red, V177 is in yellow, L203 is in green, and V253 is in orange. The helices are colored as in Figure 1B: Sector 1 in light blue and sector 2 in pink. At top, the view is from the periplasm. Below, the view is rotated 90°, with the cytoplasmic surface at the top. The image is developed from PBD file 3rko [5]. C. The effect of CuCl₂ treatment on the rates of NADH-driven proton translocation as indicated by quenching of the fluorescence of ACMA, for the two double mutants. Representative traces of 2–3 preparations are shown. D. Immunoblots of the three double mutants. PM indicates treatment with PEG-maleimide-2000 and Cu²⁺ indicates treatment with CuCl₂. The bands for subunit N and the shifted bands after PEG-maleimide treatment (N-PM) are indicated at the right. Molecular weight markers are shown at the left: 50,000 and 80,000.

3.3 Cross-linking in sector 2 had moderate effects on enzyme activity

Three cysteine pairs were constructed at the cytoplasmic end of sector 2: N293C (TM10) + S358C (TM11), V343C (TM11) + L372C (TM12), and T377C (TM12) + A429C (TM13), as shown in Figure 4. This end of sector 2 was predicted to be closed to protons or water [5]. S358 is actually in the connection between TM11 and TM12. It is part of a short helix that interacts with the cytoplasmic end of TM10. The first two double mutants were treated with the MTS reagent, M2M, and the last pair was treated with CuCl₂. In general, cysteines that can form disulfides are also readily cross-linked by shorter MTS reagents such as M2M [29]. All three double mutants had about 80% of the wild type NADH oxidase activity before treatment (Fig. 4A). Only T377C + A429C showed a significant loss of activity after treatment, about 30%. Measurements of proton translocation using ACMA fluorescence quenching were consistent with the NADH oxidase activities for all three mutants (Fig. 4C). In the immunoblots (Fig. 4D), N293C + S358C showed the standard pattern for a cross-linked product: the only shifted band of subunit N was after reaction with PEG-maleimide alone. For the other two mutants, PEG-maleimide did not shift the band for subunit N after treatment with M2M or with CuCl₂, consistent with cross-linked product, however, the situation for reaction with PEG-maleimide alone was not as clear. In the case of V343C + L372C, a second band also appeared below the shifted band, which is likely a nonspecific band that was also shifted. In the case of T377C + A429C, the yield of cross-linked product appeared to be incomplete. The PEG-maleimide treatment caused some of the N band to shift, whether or not prior treatment with CuCl₂ was done. That suggests that a higher yield of cross-linking would have significantly reduced enzyme activity. In Fig. 4B, the residue pairs are colored: 293 (red) + 358 (yellow), 343 (green) + 372 (blue), and 377 (orange) + 429 (magenta).

Figure 4.

Analysis of cross-linking at the cytoplasmic side of sector 2 in subunit N (TM9–13). N293C (in TM10) + S358C (in TM11), V343C (in TM11) + L372C (in TM12) and T377C (in TM12) + A429C (in TM13) were analyzed. A. The effect of M2M or CuCl₂ treatment on NADH oxidase activity of membrane vesicles from the three double mutants. Deamino-NADH is used as the substrate. The means and standard errors are shown for at least 4 measurements from at least 2 independent preparations. B. The location of the cysteine substitutions are indicated by space filling views of the original residues: N293 is in red, S358 is in yellow, V343 is in green, L372 is in blue, T377 is in orange, and A429 is in magenta. The helices are colored as in Figure 1B: Sector 1 in light blue and sector 2 in pink. At top, the view is from the cytoplasm. Below, the view is rotated 90°, with the cytoplasmic surface at the top. The image is developed from PBD file 3rko [5]. C. The effect of M2M or CuCl₂ treatment on the rates of NADH-driven proton translocation as indicated by quenching of the fluorescence of ACMA, for the two double mutants. Representative traces of 2–3 preparations are shown. D. Immunoblots of the three double mutants. PM indicates treatment with PEG-maleimide-2000, M2M indicates treatment with 1,2-ethanediyl bis-methanethiosulfonate, and Cu²⁺ indicates treatment with CuCl₂. The bands for subunit N and the shifted bands after PEG-maleimide treatment (N-PM) are indicated at the right. Molecular weight markers are shown at the left: 50,000 and 80,000.

The periplasmic end of the second sector was predicted to contain a half channel to the periplasm, and Ser 322 in particular was suggested to mediate water transport [5]. Molecular dynamics studies showed water near Ser 322, but not a continuous chain [27]. Three cysteine pairs were constructed in this region: V311C (TM10) + L410C (TM13), A315C (TM10) + V401C (TM12), and S322C (TM11) + V401C (TM12), as shown in Figure 5. The latter two mutants showed reductions in NADH oxidase activity of about 30 and 25 %, respectively, after treatment with CuCl₂, as shown in Fig. 5A. Modest reductions in NADH-driven proton translocation were also seen for these two mutants upon CuCl₂ treatment (Fig. 5C). The effects of CuCl₂ treatment on the first double mutant, V311C + L410C, were not significant (Fig. 5BA,C). The immunoblots were consistent with a high yield of disulfide formation. In the case of A315C + V401C (TM12), and S322C + V401C (Fig. 5D, right), each PEG-maleimide had a mass of 5,000 daltons, and so the total shift appeared to be 10,000 daltons. In Fig. 5B, the residue pairs are colored: 311 (red) + 410 (yellow), 315 (green) + 401 (blue), and 322 (orange) + 401 (blue).

Figure 5.

Analysis of cross-linking at the periplasmic side of sector 2 in subunit N (TM9–13). V311C (in TM10) + L410C (in TM13), A315C (in TM10) + V401C (in TM12) and S322C (in TM11) + V401C (in TM12) were analyzed. A. The effect of CuCl₂ treatment on NADH oxidase activity of membrane vesicles from the three double mutants. Deamino-NADH is used as the substrate. The means and standard errors are shown for at least 4 measurements from at least 2 independent preparations. B. The location of the cysteine substitutions are indicated by space filling views of the original residues: V311 is in red, L410 is in yellow, A315 is in green, V401 is in blue, and S322 is in orange. The helices are colored as in Figure 1B: Sector 1 in light blue and sector 2 in pink. At top, the view is from the periplasm. Below, the view is rotated 90°, with the cytoplasmic surface at the top. The image is developed from PBD file 3rko [5]. C. The effect of CuCl₂ treatment on the rates of NADH-driven proton translocation as indicated by quenching of the fluorescence of ACMA, for the two double mutants. Representative traces of 2–3 preparations are shown. D. Immunoblots of the three double mutants. PM indicates treatment with PEG-maleimide-2000 (left panel), PEG-maleimide-5000 (right panel) and Cu²⁺ indicates treatment with CuCl₂. The bands for subunit N and the shifted bands after PEG-maleimide treatment (N-PM) are indicated at the right. Molecular weight markers for the left panel: 50,000 and 80,000, and for the right panel: 50,000, 65,000 and 80,000.

3.4 Cross-linking in the central cavity near Lys 395 caused a loss of activity

Two additional pairs of cysteine substitutions were constructed in the second sector: S301C (TM10) + L383C (TM12) and S305C (TM10) + L383C (TM12), as shown in Figure 6. This region contains the essential residue Lys 395 (TM12). L383 is located on the cytoplasmic edge of the non-helical segment of TM12, while Lys 395 is located within the helical segment on the periplasmic side of the non-helical segment. The sites of cysteine substitution were chosen to be closer to the middle of the membrane near the horizontal proton pathway; the previous sites in this sector were located much closer to the cytoplasmic and periplasmic surfaces. In contrast to the other mutants analyzed in this report, these two double mutants both showed significant reduction in NADH oxidase before treatment for cross-linking, relative to the wild type. Both retained about 40% of the activity, as shown in Fig. 6A. After treatment with the cross-linking reagent M2M, S301C + L383C lost essentially all of its activity, both NADH oxidase (Fig. 6A) and NADH-driven proton translocation (Fig. 6B). The remaining NADH oxidase activity, about 10% of the untreated rate, is near the background level [13]. The other double mutant S305C + L383C lost little or no activity after treatment with M2M, consistent with a low yield of cross-linked product. Immunoblots were complicated by the fact that the cross-linked products appeared to be detected poorly by the HA-antibody used for subunit N, especially after PEG-maleimide modification. Faint bands can be seen indicating that PEG-maleimide shifted the band for subunit N. After treatment with M2M, especially in the case of S301C + L383C, the PEG-maleimide does not appear to shift the band for subunit N, indicating cross-linked product. In Fig. 6D, the residue pairs are colored: 301 (yellow) + 383 (blue) and 305 (red) + 383 (blue).

Figure 6.

Analysis of Cross-linking at the Center of Sector 2 in Subunit N (TM9–13). S301C (in TM10) + L383C (in TM12) and S305C (in TM10) + L383C (in TM12) were analyzed. A. The effect of M2M treatment on NADH oxidase activity of membrane vesicles from the three double mutants. Deamino-NADH is used as the substrate. The means and standard errors are shown for at least 4 measurements from at least 2 independent preparations. B. The effect of M2M treatment on the rates of NADH-driven proton translocation as indicated by quenching of the fluorescence of ACMA, for the two double mutants. Representative traces of 2–3 preparations are shown. C. Immunoblots of the two double mutants. PM indicates treatment with PEG-maleimide-2000, and M2M indicates treatment with 1,2-ethanediyl bis-methanethiosulfonate. The bands for subunit N and the shifted bands after PEG-maleimide treatment (N-PM) are indicated at the right. Molecular weight markers are shown at the left: 50,000 and 65,000. D. The location of the cysteine substitutions are indicated by space filling views of the original residues: S301 is in yellow, L383 is in blue, and S305 is in red. The helices are colored as in Figure 1B: Sector 1 in light blue and sector 2 in pink. At the left, the view is from the cytoplasm. At right, the view is rotated 90°, with the cytoplasmic surface at the top. The image is developed from PBD file 3rko [5].

4. Discussion

In Complex I the sites of NADH oxidation and ubiquinone reduction are distant from the sites of proton translocation in the membrane arm. Therefore, Complex I must use an indirect coupling mechanism [30], and would be expected to undergo important conformational changes during turnover. Studies of the bovine enzyme have shown that the NADH-reduced enzyme appears to adopt substrate-dependent conformations based on changes in rates of chemical modification [31], trypsin digestion [32], and inter-subunit cross-linking [33]. Similar conclusions have been drawn from studies of bacterial Complex I [34–37]. In previous work from this lab [6,38], a cross-linking approach using engineered cysteine substitutions in Complex I from E. coli was used to constrain the lateral helix of subunit L. It was found that cross-linking it to other subunits, M, N or K, did not significantly impair the activity of the enzyme. In contrast, cross-links between engineered cysteine residues in subunits M and N, within the membrane, in the vicinity of the of key residues Lys 395 of N and Glu 144 of M (see Fig. 7), reduced enzyme activity to a very low level. For example, N subunit M388C (TM12) was cross-linked to M subunit M145C (TM5) and N subunit V430C (TM13) was cross-linked to M subunit A155C (TM5). In the current work we have used a similar cysteine cross-linking approach to probe the proton pathways within the channels of subunit N. In the case of the cysteine pairs near the periplasmic surface (Figure 3), there was no evidence of cross-link formation. It is likely that these residues, as found in inverted membrane vesicles, were not accessible to the reagents used to promote cross-linking.

Figure 7.

Location of deleterious cross-links with respect to key residues of subunit N. Subunit N is colored as previously: Sector 1 in light blue, sector 2 in pink, and remaining TM helices in light gray. Subunit M, at the left, and subunits J and K at the right are colored dark gray. In subunit N, key residues along the central proton pathway are shown in space filling view, with Glu residues in red and Lys residues in blue. Several waters are also visible, colored red. The images are developed from PBD file 3rko [5]. A. The view is from the cytoplasm. At the right, Glu 133 from sector 1 of subunit N is shown in proximity to Glu 72 of subunit K. At the left, Lys 395 from sector 2 of subunit N is shown in proximity to Glu144 of subunit M. B. The view is rotated 90° from that in panel A, with the cytoplasm at the top. Residues from cross-links that cause deleterious effects are highlighted in space filling views: M145 in pink (nuoM) + M388 in green, T377 in orange + A429 in magenta, S301 in yellow + L383 in cyan, A315 in violet + V401 in pink, and S322 in black + V401 in pink.

In sector 1 (TM4–8), a half-channel is expected to connect the cytoplasm to the center of the membrane. We succeeded in making only 2 cross-links in this region, both near the cytoplasmic surface. Neither the substitutions by cysteine, nor the formation of disulfide bonds had any significant effect on the activity of Complex I, indicating that the entry of hydrogen ions was not impeded. One explanation is that there is redundancy in the entry channels. On the basis of the crystal structure [5] it was proposed that a second entry channel exists at the interface of subunits K and N. This was confirmed by molecular dynamics [27], in which water was found to fill both half-channels. It is also possible that the disulfide bonds are unable to prevent the penetration of water in this half-channel. It is clear that the location of the disulfide between L142C in TM5, 7 residues from E133, and V238C in TM8, 9 residues from K243, did not diminish activity. Likewise for the formation of a disulfide between S162C (TM6) and A234C (TM7), in the broken helix. This is in contrast to many ion channels, in which disulfide bond formation restricts important conformational changes, such as for gating ions [39,40].

In sector 2 (TM9–13) the success rate of forming cross-links was much higher, and all positions tested could be cross-linked with at least moderate yield. In this sector the half-channel is expected to run from the center of the membrane to the periplasmic surface. Cysteine residues were substituted into positions at the cytoplasmic surface, the periplasmic surface, and near the center of the membrane. Only cross-links that formed with the broken helix, TM12, had any significant effect on Complex I activity. For example, cross-links between V311C (TM10) + L410C (TM13) near the periplasmic surface, and between N293C (TM10) + S358C (TM11) near the cytoplasmic surface, had no effect on activity. In contrast, 5 different cross-links to various positions in the broken helix TM12 from TM10, TM11 or TM13, had various effects on Complex I activity, summarized in Table 2.

Table 2.

Summary of Deleterious Cysteine Cross-links

Residues in Subunit N	Cross-link	Activity before treatmenta	Activity after treatmenta	Residual activityb	Location in Subunit N
S301C + L383C	M2M	47±3%	11±1%	23%	Central cavity
A305C + L383C	M2M	50±2%	35±4%	70%	Central cavity
A315C + V401C	disulfide	100±6%	67±4%	67%	Open half-channel
S322C + V401C	disulfide	83±4%	64±3%	77%	Open half-channel
T377C + A429C	disulfide	75±5%	52±2%	69%	Closed channel

Activities are the means (± standard errors) of at least 4 measurements

^aDeamino-NADH oxidase activity as a percent of untreated wild type rate, which was 165 ± 48 nmol/min/mg protein (n=15) [10].

^bActivity after treatment as a percentage of activity before treatment.

The most notable cross-link was formed between S301C (TM10) + L383C (TM12), in which nearly all activity was lost. This cross-link, involving M2M, and illustrated in Figure 7, appeared to fill an internal cavity between Lys 247 and Lys 395 in the horizontal proton pathway. Therefore, it is likely to have obstructed the path of protons or water either directly, or indirectly through restriction of conformational changes in that region. The similarly located residues A305C + L383C had a similar deleterious effect before treatment with M2M, but the cross-link did not appear to form efficiently. In the predicted half-channel to the periplasm in sector 2, two cross-links, A315C + V401C and S322C + V401C, had small effects on activity, in contrast to both cross-links in the cytoplasmic half-channel of sector 1, which had no effect. That might be due to the absence of redundancy in the periplasmic half-channel, as was indicated by molecular dynamics [27]. Residue Ser 322 was previously identified as a possible contributor to proton transport through this half-channel to the periplasm [5]. Likewise, Val 401 is one helical turn away from His 405, which was identified to be in the cavity near the link to the periplasm [5]. Finally, the cross-link between T377C + A429C occurs in the predicted closed end of sector 2. Even so, some loss of activity occurred upon formation of a disulfide cross-link. In this case, Ala 429, which is one helical turn away from Tyr 425, also identified to be in the cavity near the path to the periplasm [5]. In addition, since Thr 377 is just 7 residues from the discontinuous section of TM12, the disulfide might restrict conformational changes associated with interactions between subunits N and M. A similarly placed residue in the discontinuous region, M388C, when cross-linked to M145 of subunit M (see Fig. 7B) had profound effects on activity [6].

In summary, four regions of subunit N have been probed by cross-links between engineered cysteine residues. The results support several predicted aspects of the proton pathways through subunit N. There appear to be redundant cytoplasmic half-channels, a single periplasmic half-channel, and that the horizontal proton pathway between key residues Lys 247 and Lys 395 can be blocked by a cross-link.

Table 1.

Summary of Cysteine Substitutions Used in Cross-linking

Location	Sector	Residue 1	TM helixa	Residue 2	TM helixa	Cα-Cα distance (Å)	Cross-linking agentb
Cytoplasmic	1	L142	5	V238	8	6.1	CuCl2
Cytoplasmic	1	S152	6	A234	7*	5.1	CuCl2
Periplasmic	1	L128	5	L170	6	10.0	–
Periplasmic	1	V177	6	L203	7*	4.6	–
Periplasmic	1	L128	5	V253	8	7.4	–
Cytoplasmic	2	N293	10	S358	11c	4.7	M2M
Cytoplasmic	2	V343	11	L372	12*	6.2	M2M
Cytoplasmic	2	T377	12*	A429	13	4.9	M2M
Periplasmic	2	V311	10	L410	13	6.9	CuCl2
Periplasmic	2	A315	10	V401	12*	6.4	CuCl2
Periplasmic	2	S322	11	V401	12*	5.8	CuCl2
Central Cavity	2	S301	10	L383	12*	8.6	M2M
Central Cavity	2	S305	10	L383	12*	8.7	M2M

^a*identifies the broken helices, TM7 and 12.

^bThe results of these cross-linking agents are shown in Figures 2–5. “–” indicates that cross-linking was not observed

^cS358 is located in a short helix that is part of the connection between TM11 and TM12.

Highlights.

• Proton pathways in subunit N of Complex I from E. coli are probed by cross-linking.

• Disulfide cross-links in the cytoplasmic half-channel did not affect enzyme activity.

• Cross-links in the periplasmic half channel had small effects on enzyme activity.

• A cross-link formed in the central cavity near Lys 395 caused loss of activity.

Abbreviations

ACMA

9-amino 3-chloro 2-methoxy acridine

BA14

nuoA-N deletion strain in E. coli

FCCP

carbonyl cyanide p-(trifluoromethoxy) phenylhydrazone

MTS

methanethiosulfonate

M2M

1,2-ethanediyl bismethanethiosulfonate

pBA400

expression vector for Complex I

PEG-maleimide

one of the polyethylene glycol maleimides

TM

transmembrane

TM7

for example, transmembrane helix 7