Characterization of the Nqo5 subunit of bacterial complex I in the isolated state

The subunits that comprise bacterial complex I (NADH:ubiquinone oxidoreductase) are also found in more complicated mitochondrial enzymes in eukaryotic organisms. Although the Nqo5 subunit is one of these conserved components and important for the formation of complex, it has been little studied. Here, we report structure analyses of isolated Nqo5 from Thermus thermophilus. Biochemical studies indicated that the C‐terminal region following the 30‐Kd subunit motif is disordered in the isolated state, while the remaining portion is already folded. Crystallographic studies of a trypsin‐resistant fragment revealed detailed structural differences in the folded domain between the isolated and complexed states.

The subunits that comprise bacterial complex I (NADH:ubiquinone oxidoreductase) are also found in more complicated mitochondrial enzymes in eukaryotic organisms. Although the Nqo5 subunit is one of these conserved components and important for the formation of complex, it has been little studied. Here, we report structure analyses of isolated Nqo5 from Thermus thermophilus. Biochemical studies indicated that the C-terminal region following the 30-Kd subunit motif is disordered in the isolated state, while the remaining portion is already folded. Crystallographic studies of a trypsin-resistant fragment revealed detailed structural differences in the folded domain between the isolated and complexed states.
NADH:ubiquinone oxidoreductase (complex I) is one of the respiratory chain components (complexes I-IV and ATP synthase) in the mitochondrial inner membrane of eukaryotes or the plasma membrane of prokaryotes. The enzyme carries out an electron transfer reaction from NADH to ubiquinone, and makes a proton concentration gradient across the membrane by proton pumping. More than 40 subunits compose the complex I of mammalian mitochondria, and the resultant molecular mass is about 1000 kDa. On the other hand, bacterial complex I (~500 kDa) has a simpler composition with 13-15 subunits. Respiratory complexes from the thermophilic bacterium Thermus thermophiles have frequently been used in the structural studies because of their high stability [1][2][3][4]. The whole structure of bacterial complex I, including the transmembrane subunits was first determined at 3.3 A resolution [5], while eukaryotic complexes I from Yarrowia lipolytica and Bos taurus were reported more recently [6,7].
A subunit with a molecular weight of 30 kDa (30 Kd subunit for the bovine enzyme, NDUFS3 for the human enzyme) is encoded in nuclear DNA in the case of most eukaryotes [8]. The subunit has a signature pattern of EREx 2 (D/E)(L/I/V/M/F/Y) 2 x 6 (H/K)x 3 (K/R/P)x(L/I/V/ M)(L/I/V/M/Y/S) in the C-terminal region, which is known as the respiratory-chain NADH dehydrogenase 30-Kd subunit signature. The subunit is contained in all homologous enzymes, and the human homolog (NDUFS3) is involved in diseases [9][10][11]. The bacterial homologous proteins are denoted as Nqo5 (for T. thermophilus and Paracoccus denitrificans) or NuoC (for Escherichia coli). The subunit composes the hydrogenase-like module together with the Nqo4, Nqo6, and Nqo9 subunits. Homologous proteins of the Nqo5 subunit are also contained in membrane-bound multisubunit hydrogenases [12], while soluble-type hydrogenases are not. It has been reported that the subunit has a central role in the assembly of complex I [13][14][15][16][17][18]. Although the structure of the Nqo5 subunit in the complexed state has been elucidated, the details of the assembly process remain unclear.
In this paper, we report structural studies of the conserved subunit Nqo5 of the bacterial complex I from T. thermophilus. Strictly folded and flexibly extended portions of the Nqo5 subunit in the isolated state are distinguished according to biochemical and crystallographic results. The implications for roles in assembly process and conformational changes of the whole complex are discussed based on a comparison of Nqo5 structures in the isolated and complexed states.

Expression and purification
The nqo5 gene from the genome of T. thermophilus HB8 was ligated into pET11a (Novagen, Madison, WI, USA). The transformed Escherichia coli strain BL21 (DE3) was grown at 310 K in LB medium and 0.1 mM of ampicillin. Cultured cells were broken by sonication, and the cell membrane was removed by centrifugation (10 000 g). The supernatant was heated at 343 K. The centrifuged supernatant was successively applied to a Super Q Toyopearl 650M column (Tosoh, Tokyo, Japan), Resource Q column (GE Healthcare, Buckinghamshire, UK), CHT5-I hydroxyapatite column (Bio Rad, Hercules, CA, USA), Superdex200 (GE Healthcare) column, and HiPrep desalting column (GE Healthcare). Selenomethionyl Nqo5 was expressed in B834(DE3), and purified in a manner similar to the native protein.

Protease degradation assays
Purified Nqo5 was digested with trypsin in a buffer solution (50 mM Tris-HCl, pH 8.0). The ratio of the weight of protein to protease was 100 : 1. The solutions were incubated at 277 K, and were sampled at 0, 1, 2, 4, and 24 h. The reaction was terminated by adding PAGE buffer and immediately stored at 253 K. SDS/PAGE analyses were performed with 15% gels. Gels were stained with Coomassie brilliant blue. The trypsin-resistant fragment with a molecular mass of 15.6 kDa was purified with a size exclusion column, Superdex75 (GE Healthcare).

Mass spectrometry
The masses of full-length Nqo5 and the fragment of Nqo5 digested by trypsin at 277 K for 24 h were determined by matrix-assisted laser desorption ionization mass spectrometry (MALDI-MS) (Voyager-DE RP; Applied Biosystems, Waltham, MA, USA). Digested protein solution was desalted with ZipTip C18 (Merck Millipore, Billerica, MA, USA) for MALDI-MS experiments. 10 lL of the desalted solution was mixed with 4.0 lL of a solution containing 10 mgÁmL À1 sinapinic acid, 0.1% trifluoroacetic acid, and 50% acetonitrile.

Circular dichroism spectroscopy
Circular dichroism (CD) spectra were measured using a J-805 spectropolarimeter (Jasco, Tokyo, Japan). The samples of the full-length and trypsin-resistant fragment of Nqo5 (0.1 mgÁmL À1 ) were dissolved in 5.0 mM of potassium phosphate buffer in a 1.0-mm quartz cuvette. All CD spectra were measured at 293 K. Five scans were averaged for each spectrum. The secondary structure contents were determined with the JWSSE-408 program (Jasco) using a reference data set [19].

Crystallization
The 15.6-kDa fragment as well as the full-length protein was used for crystallization experiments with the sitting drop vapor diffusion method at 293 K. High-quality crystals were obtained only from the fragment. Rod-shaped orthorhombic crystals with typical dimensions of 0.3 9 0.1 9 0.1 mm 3 were grown in crystallization solutions which consisted of a protein solution (5 mgÁmL À1 protein, 20 mM Tris-HCl pH 8.0 and 5% (w/v) NDSB-195) and a reservoir solution (200 mM calcium chloride and 25% (w/v) PEG4000). Hexagonal crystals with typical dimensions of 0.1 9 0.1 9 0.1 mm 3 were grown from a crystallization solution that consisted of the protein solution and a reservoir solution (10% (w/v) PEG1000 and 10% (w/v) PEG8000). Selenomethionine-derivative crystals were obtained under the same conditions as the Native I crystal.

X-ray data collection
Native data sets (Native I and II) of the wild-type protein and three wavelength data sets of the selenomethionyl protein were collected at 95 K. Diffraction data sets were collected at SPring-8, and processed and scaled using the HKL2000 package [20]. The space group of the Native I crystal is P2 1 2 1 2 1 with unit cell parameters of a = 41. 3 A A. The selenomethionyl derivative crystal belongs to the space group P2 1 2 1 2 1 and the unit cell parameters are A. The crystallographic statistics are listed in Table 1.

Structure determination and refinement
The phasing was performed by the multi-wavelength anomalous diffraction (MAD) method with the phenix.autosol [21]. electron density map was adequate for the model building (Fig. 1A). Autotracing was carried out with the phenix.autobuild [22]. The output structure was subjected to further manual improvement with the Coot program [23]. The structure was refined against the Native I data set using the phenix.refine [22] (Fig. 1B). The final R work and R free factors were 16.9% and 20.9%, respectively. The structure for the Native II crystal was determined by the molecular replacement method with the MOLREP program [24] in the CCP4 suite [25] using the Native I structure as a search model. The R-factor and the correlation coefficient are 38% and 0.61, respectively. The structure was refined with the phenix.refine (Fig. 1C). The final R work and R free factors were 19.5% and 23.4%, respectively. The structures were validated with the Molprobity program [26]. The refinement statistics for Native I and II are listed in Table 2. Figures were prepared using the PyMOL program [27]. The coordinates and structural factors for the Native I and II crystals have been deposited in the Protein Data Bank under accession numbers PDB: 5B3P and PDB: 5B3Q, respectively.

Characterization of the Nqo5 subunit in the isolated state
The Nqo5 subunit from T. thermophilus was expressed in E. coli, and obtained as a soluble protein. In order to elucidate the conformation in the isolated state, protease accessibility assays and circular dichroism (CD) spectroscopy were performed. A fragment with a molecular mass of 15.6 kDa was remained even after long incubation with trypsin ( Fig. 2A). The trypsinresistant fragment was purified for further characterization. The CD spectrum for the fragment is different from that for the whole molecule (Fig. 2B). The content for random conformation is decreased from 17% to 4% by the truncation. The digestion sites of the fragment were determined to be Arg133 and Lys134 at the C-terminal region according to the mass spectrum (Fig. 2C), while the several potential digestion sites exist throughout the molecule (Fig. 2D).

Structural features of the protease-resistant fragment
The full-length protein and the trypsin-resistant fragment (residue 1-134) were subjected to crystallization trials. High-quality crystals were obtained from the fragment. The structure was solved by the multiple wavelength anomalous dispersion (MAD) method using the selenomethionyl protein (Table 1), and refined against the native data sets at 1.65 (Native I in P2 1 2 1 2 1 ) and 3.0 A (Native II in P6 3 ) resolutions ( Table 2). The two strucutures can be superimposed with a root mean square deviation of 1.7 A for all atoms. This finding indicated that the atoms are almost identical despite their quite different crystal packing (Fig. 3). We modeled almost all the trypsin-resistant fragment into the crystal structure (Fig. 4A). The electron density for the last visible residue, Lys134, is weak ( Fig. 4B), indicating the residue has a low occupancy. These facts are consistent with the results from the mass spectra of the digested protein, in which the C-terminal sides of both Arg133 and Lys134 are considered probable digestion sites for trypsin. The structure of the isolated Nqo5 appears to be similar to the subunit in the complexed state (Fig. 5A,  B). The rmsd value between the complexed and isolated states is 2. 3 A for all atoms. The loops B, E, and F largely protrude from the main body (Fig. 4A).
These regions exhibit significant deviations in the comparison between the isolated and complexed states (Fig. 5B). The loop B has a hydrophobic residue Leu22 at the tip. Leu22 interacts with Leu104 and Val342 of Nqo4 in the complex (Fig. 5C-i). The loop E has three proline residues (Pro61, Pro63, and Pro65) and two tyrosine residues (Tyr57 and Tyr60) (Fig. 5Cii). Tyr60 interacts with both Tyr57 of Nqo5 and Tyr366 of Nqo4 by the aromatic-aromatic interaction in the T-shaped geometry in the complexed state. Tyr57 as well as Arg91 interact with Glu341 of Nqo4 in the complexed state. Tyr57 and Tyr60 do not interact with each other in the isolated state, while the conformation of Tyr57 is almost identical in both states. Therefore, the conformation of Tyr60 is altered upon the complex formation. The loop F contains hydrophobic residues such as Pro78 and Trp80. The side chain of Trp80 undergoes a hydrophobic interaction with Val244 of Nqo4 in the complexed state ( Fig. 5C-iii). Changes of the side chain conformation are observed for other hydrophobic residues such as Phe113 (Fig. 5C-iv) at the interface with Nqo4. On the other hand, the conformations of hydrophilic residues such as Asp51, Arg87, Arg91, and Glu117 ( Fig. 5C-ii, v) are similar in both states.
The Nqo5 subunit from T. thermophilus has a sequence of 115 EREVYDLFGIVFEGHPDLRKIL as the respiratory chain NADH dehydrogenase 30 Kd subunit signature. The structure of the signature motif other than Ile135 and Leu136 could be determined in this study. The conformation of this portion is almost identical between the isolated and complexed states.    Conserved residues in the signature are at the surface of the subunit with the exception of Glu115. The side chain of Glu115 forms hydrogen bonds with the main chain N atom of Val104 and the side chain of Thr103. Consequently, the protruding portion of the signature is attached to the main body of the subunit. It has been reported that negatively charged residues (Glu115, Glu117, and Asp120) located at the signature sequence are important for the stability of the complex [28].

Discussion
The present structure indicates that the C-terminal portion is unfolded or loosely folded in the isolated state, while the N-terminal portion is already folded. The C-terminal portion is fixed by interactions with other subunits such as Nqo3, Nqo4, Nqo6, Nqo9, and Nqo16 in the complexed state [5]. NuoC (Nqo5) of E. coli is fused with NuoD (Nqo4), while these are separately expressed in other species [13], and a large amount of a multimer of NuoCD was observed in the cell lysate of E. coli [29]. Therefore, it is plausible that Nqo5 initially forms a subcomplex with Nqo4 and accumulates in cells. Some hydrophobic residues are located at the loop regions. These residues may be involved in the first stage of the complex formation, because these conformations can adapt with high flexibility. The hydrophobic residues are the apparatus for capturing the neighboring Nqo4 subunit. Further associations on the main body surface may be promoted by hydrophobic interactions.
The interface between Nqo4 also consists of many hydrophilic residues in addition to hydrophobic residues (Fig. 6A). In the complex, hydrophilic residues, such as Asp51, Arg87, and Glu117, interact with Glu352, Lys369, Arg371, and Arg409 of Nqo4 (Fig. 5C-v). On the other hand, hydrophobic residues, such as Ile52, Val53, Val71, and Phe89, undergo no  interaction (Fig. 6A). In the isolated Nqo5, water molecules and calcium ions are located at the positions near the interacting residues of Nqo4 (Fig. 6B). These water molecules bound to the interfacial hydrophilic residues should be removed upon the interaction with the residues of Nqo4. Some water molecules may be left at the interface of the complex, but no such molecules are observed in the complex structure due to the low resolution. Indeed, several cavities are found at the interface between Nqo5 and Nqo4 (Fig. 6C). A high content of hydrophilic residues is unusual at the subunit interface of steady protein complexes. However, some examples can be found in the structure of complexes. The bottom sides of the interfaces of the A and B subunits in the V 1 complex also have such high contents of hydrophilic residues [4]. Each interface of the three AB pairs has a different manner of interaction. This flexibility at the interface is suitable for the open-close motion in the complex. Similarly, the hydrophilic residues at the Nqo4-Nqo5 interface may be involved in the flexibility of complex I reported in various studies [30][31][32][33][34]. In addition, the possibility of conformational fixation by calcium ions has been pointed out [35]. In our experiments, the calcium ions are coordinated with Asp51 and Glu73 in the Native I crystal, which is obtained in the presence of calcium ions (Fig. 6B).