Report

Organization of Iron-Sulfur Clusters in Respiratory Complex I

See allHide authors and affiliations

Science  29 Jul 2005:
Vol. 309, Issue 5735, pp. 771-774
DOI: 10.1126/science.1113988

Abstract

Complex I of respiratory chains plays a central role in bioenergetics and is implicated in many human neurodegenerative diseases. An understanding of its mechanism requires a knowledge of the organization of redox centers. The arrangement of iron-sulfur clusters in the hydrophilic domain of complex I from Thermus thermophilus has been determined with the use of x-ray crystallography. One binuclear and six tetranuclear clusters are arranged, maximally 14 angstroms apart, in an 84-angstrom-long electron transfer chain. The binuclear cluster N1a and the tetranuclear cluster N7 are not in this pathway. Cluster N1a may play a role in the prevention of oxidative damage. The structure provides a framework for the interpretation of the large amounts of data accumulated on complex I.

Nicotinamide adenine dinucleotide (NADH): ubiquinone oxidoreductase (complex I, EC 1.6.5.3) is the first enzyme of the mitochondrial and bacterial respiratory chains. It catalyzes the transfer of two electrons from NADH to quinone, coupled to the translocation of about four protons across the membrane (1, 2). This process accounts for about 40% of the transmembrane proton gradient generated in NADH oxidation by the mitochondrial respiratory chain. Complex I is one of the largest membrane protein complexes known to date. The simplest version is the prokaryotic enzyme, which has 14 subunits with a combined molecular mass of about 550 kD. Analogs of all of the subunits of bacterial complex I (also referred to as NDH-1) are found in the more elaborate mitochondrial enzyme, and they both contain equivalent redox components (2). Mitochondrial and bacterial enzymes have a characteristic L-shaped structure with one arm (hydrophobic) embedded in the membrane and the other, the hydrophilic peripheral arm, protruding into the mitochondrial matrix or the bacterial cytoplasm (3-7). These similarities allow NDH-1 to be used as a minimal model for the study of complex I. Its atomic structure is not known, and the mechanisms of electron transfer and proton pumping are not established. The association of mutations in complex I subunits with human neurodegenerative diseases, including Parkinson's disease (8), provides additional impetus to efforts to characterize this enzyme. Complex I has also been suggested to be a major source of reactive oxygen species (ROS) production in mitochondria, which can lead to mitochondrial DNA damage and may be one of the causes of aging (9). Dissociation of complex I with chaotropes and detergents has demonstrated that its NADH-binding site and all of its redox centers [noncovalently bound flavin mononucleotide (FMN) and iron-sulfur clusters] are in the peripheral arm (1, 10, 11), whereas the proton-pumping machinery is probably in the membrane arm. On the basis of such fragmentation studies, we have recently proposed a detailed model of the arrangement of subunits in bacterial complex I (12). It is broadly consistent with models developed from consideration of the evolutionary origins of complex I from smaller preexisting modules, such as soluble and membrane-bound hydrogenases (1, 13). Subunits Nqo1 to Nqo3 (Thermus nomenclature is used throughout, with bovine mitochondrial nomenclature shown in parentheses when needed for clarity) form the dehydrogenase domain, capable of NADH oxidation with artificial electron acceptors. It is at the distal end of the peripheral arm. Subunits Nqo4 to Nqo6 and Nqo9 are proposed to form a connecting domain between the dehydrogenase domain and the membrane arm (14). Iron-sulfur (Fe-S) clusters identified with the use of electron paramagnetic resonance spectroscopy (EPR) include the binuclear clusters N1a and N1b and the tetranuclear clusters N2, N3, N4, and N5 (2, 15). Sequence comparisons suggest that complex I contains two more tetranuclear clusters, N6a and N6b, in subunit Nqo9 (TYKY). So far, they have been detected only in subcomplexes or in the recombinant subunit (16, 17). Additionally, complex I from Escherichia coli, from Thermus thermophilus, and from various other bacteria contains another binding motif for the tetranuclear cluster N7, which has been identified by EPR in recombinant subunit Nqo3 (75 kD) but not in situ (18). Thus, complex I is predicted to contain a total of eight or nine Fe-S clusters, making it the most elaborate iron-sulfur protein assembly known.

We have purified and crystallized the hydrophilic domain (peripheral arm) of complex I from T. thermophilus HB-8 (19). It contains all the redox centers and represents more than half of the molecular mass of the entire complex (280 kD out of 520 kD). Analysis of x-ray data using anomalous Fe signals from intrinsic Fe-S clusters allowed us to identify nine Fe-S clusters in this subcomplex and calculate electron density at about 4 Å resolution (table S1). The overall shape of the hydrophilic domain was clear, and many secondary structure features, such as α helices, were visible (fig. S1). The molecular surface of the peripheral arm of complex I and the electron densities of Fe-S clusters are shown in Fig. 1. The clusters are buried within the protein mass, shielded from the solvent phase. The overall appearance of the peripheral arm is pronouncedly Y-shaped, the feature which is partially visible in electron microscopic models of complex I from Neurospora crassa (4) and from the hyperthermophile Aquifex aeolicus (6). The domain is about 155 Å high with an average width of about 70 to 80 Å. These dimensions and overall shape are consistent with those of the peripheral arm in the enzyme from A. aeolicus, indicating that the lower part of the structure, as shown in Fig. 1, forms the interface with the membrane arm of the enzyme.

Fig. 1.

The electron density of Fe-S clusters shown within the molecular surface of the peripheral arm of complex I from T. thermophilus. To calculate the molecular surface, a mask was produced with the CCP4 program Ncsmask (30), using skeletonized density from one hydrophilic domain and a radius of 5 Å around skeleton atoms. The clusters are shown in red as the electron density contoured at 6σ. The number 1 denotes a possible channel for NADH access, reaching within 10 Å of the first Fe-S cluster in the redox chain (N3); 2 denotes a second possible access channel, approaching within about 15 Å of the same cluster; and 3 denotes a short channel that would allow quinone to approach the last cluster in the chain of Fe-S clusters (N2). (A) The Y-shaped view of the domain, with the scale bar shown. The interface with the membrane domain is at the bottom of the structure. (B) Orthogonal view, rotated clockwise about the vertical compared with (A).

From an examination of the electron density of Fe-S clusters, it is clear that two of the clusters are oblong and smaller than the seven other clusters (Fig. 1). Accordingly, during heavy atom refinement [where the clusters were treated as single superatoms (19)], the occupancies of the two smaller clusters were always about half the occupancies of the seven larger clusters. Thus, the two smaller clusters have been assigned as being binuclear and the seven larger ones as tetranuclear. The numbers of the two cluster types are consistent with the maximum number predicted from EPR experiments and sequence comparisons, indicating that our model for the arrangement of clusters in complex I is complete. Mitochondrial complex I and many bacterial enzymes do not contain binding motifs for cluster N7, and so they will contain the two binuclear clusters and the six tetranuclear clusters.

The detailed arrangement of iron-sulfur clusters is shown in Fig. 2. Seven of them are arranged in a continuous chain (indicated by arrows in Fig. 2) with edge-to-edge spacing within 14 Å, the maximum distance for physiological electron transfer reactions (20). This chain, about 84 Å long, is likely to connect the two catalytic sites of the enzyme. It terminates at the tetranuclear cluster close to the interface with the membrane arm. This cluster is likely to be N2, which has the highest (about –100 mV on average), pH-dependent, midpoint potential. It is in the connecting domain subunit Nqo6 (PSST), and it has been suggested that it reduces the quinone at the interface with the membrane domain (1, 2). Cluster N2 is within about 15 Å of the membrane end of the hydrophilic domain and appears to be next to a short channel in which the electron acceptor could sit within about 10 Å of N2 (Fig. 1 and fig. S1). This is in contrast to the suggestion that cluster N2 is about in the middle of the peripheral arm and that ubiquinone moves inside a long channel to reach this cluster (3). Our interpretation is consistent with the observation that a semi-quinone species forms within 12 Å of cluster N2 (21). Although a complete Q binding site will only be revealed in a structure of intact enzyme, it is clear now that the quinone can reside mostly within the membrane, as expected for such a hydrophobic moiety.

Fig. 2.

Arrangement of Fe-S clusters in complex I. Metal sites are composed of magenta spheres for Fe atoms and yellow spheres for S atoms. The overall orientation is similar to the one shown in Fig. 1A. Most of the Fe-S clusters are roughly in the plane of the Y, with cluster N1b offset toward the reader by about 14 Å. Cluster names have been assigned tentatively as described in the text. Cluster N1a is located within subunit Nqo2; N3 in Nqo1; N1b, N4, N5, and N7 in Nqo3; N6a/b in Nqo9; and N2 in Nqo6. The likely pathway of electron transport is indicated by blue arrows. The distances between the clusters given in Å were calculated both center to center and edge to edge (shown in parentheses). Clusters N3 and N5 are separated by 17.6 Å (about 15 Å edge to edge), and clusters N1b and N4 by 19.2 Å (about 16 Å edge to edge).

All remaining clusters can be assigned tentatively on the basis of current knowledge about the arrangement of subunits in complex I and about the association of particular clusters with specific subunits (2, 12, 15). This assignment can be verified once an atomic model is built and refined. The binuclear cluster at the top left in Fig. 2 and the tetranuclear cluster at the top right are separated from the other clusters by about 19 Å and 21 Å edge-to-edge distances, respectively, and so they are unlikely to participate directly in physiological electron transfer. The two tetranuclear clusters above N2 and immediately preceding it in the electron transfer chain are likely to be N6a and N6b, because they are coordinated by the ferredoxin-like subunit Nqo9, thought to be in the connecting domain, close to the interface with the membrane arm (2, 17). Accordingly, the relatively large cluster-free domain to the right of clusters N2 and N6 (Fig. 1B) is likely to consist of the remaining subunits of the connecting domain, Nqo4 (49 kD) and Nqo5 (30 kD).

Subunit Nqo1 (51 kD) contains the NADH-binding site, the primary electron acceptor FMN (midpoint potential –340 mV), and the tetranuclear cluster N3. Therefore, the first tetranuclear cluster in the chain of Fe-S clusters is likely to be N3 (Fig. 2), because subunits Nqo1 and Nqo2 (24 kD) can form a distinct subcomplex capable of NADH oxidation (2, 22) and Nqo2 contains only the binuclear cluster, N1a. This proposal is consistent with the observed strong spin-spin interaction between the semiflavin and cluster N3 (15). The first cluster in the chain of Fe-S clusters should be close to the NADH-binding site. Accordingly, there appear to be two channels in this region of the electron density map, each of which can connect the bulk of solvent phase to about 10 to 15 Å from the cluster N3 (Fig. 1).

Subunit Nqo3 contains the binuclear cluster N1b as well as tetranuclear clusters N4, N5, and N7 (18). Cluster N7 is not present in many species and so may not be a part of the conserved electron transfer pathway. Therefore, it is the most plausible candidate for the distal tetranuclear cluster (Fig. 2). This location is in contrast to the proposal that cluster N7 has a role in electron transfer to the low-potential bacterial electron acceptor menaquinone (23). Rather, the role of N7 may be to stabilize the fold of the Nqo3 subunit, which has no other clusters in its extensive C-terminal domain. This arrangement indicates that the three remaining clusters between N3 and N7 are N1b, N4, and N5. This suggestion provides a reasonably compact arrangement within subunit Nqo3, because clusters N1b, N4, and N5 are coordinated within a segment of about 220 residues in the N-terminal domain of Nqo3. Thus, the binuclear cluster in this group is proposed to be N1b, whereas cluster N5 rather than cluster N4 is likely to be in its assigned position because of its unusual EPR properties: It exhibits very fast spin relaxation and it appears to exist in a mixed spin ground state (S = 1/2 + 3/2) (24). Unlike the other clusters shown in Fig. 2, N5 in its proposed position can interact with more than two clusters (namely with N3, N1b, and N4), which may account for its unusual EPR signals.

The distance between N3 and N5 is greater than the distance between N3 and N1b. Thus, electron transfer from FMN to quinone probably takes the shortest route along the following path: N3-N1b-N5-N4-N6a/b-N6b/a-N2, as indicated by the arrows in Fig. 2. Clusters N1b, N3, N4, and N5 are isopotential (about –250 mV) in situ (15).

This arrangement leaves the binuclear cluster N1a as the only possible candidate for its assigned position, which is also consistent with the co-localization of subunits Nqo1 and Nqo2. Because of its large distance from N3, it is unlikely to participate directly in the electron transfer from FMN to the quinone. This may explain why cluster N1a is not reduced by NADH in most species and has the lowest midpoint potential (about –370 mV on average) (15). Judging from the positions of possible NADH access channels in the electron density (Fig. 1), it is likely that FMN is located between N1a and N3, within 14 Å from either cluster. In this case, one role for the N1a cluster, because of its low potential, could be to interact with FMN and prevent its excessive reduction by redistribution of electrons further down the redox chain. This mechanism could prevent excessive generation of damaging ROS by the reduced flavin, which is likely to be more exposed to the solvent than the Fe-S clusters. A similar role was suggested for heme b in succinate dehydrogenase (25). It would appear that instead of adding a high-potential cofactor near the end of the chain, as is the case for succinate dehydrogenase, in complex I nature has opted for adding a very low–potential cofactor near the beginning of the chain, with a similar effect on the reduction state of the flavin. It will be of great interest to explore, by genetic modification, how changes to the environment of cluster N1a affect the rates of electron transfer and of ROS generation.

The organization of Fe-S clusters in complex I determined here allows us to start to understand the electron transfer mechanism of this enormously complicated enzyme, and it provides a framework for interpreting the vast amount of data accumulated over the past 40 years. It is clear that the fundamental design of the hydrophilic part of complex I is similar to other electron transfer proteins: two catalytic sites are connected by a redox chain (26). However, the chain does not take the simplest, shortest route between the two catalytic sites. In principle, fast electron transfer could be provided with just one cluster in place of N1b and N5. The reason for this added complexity may be related either to the evolutionary origins of complex I from smaller building blocks or to its mechanism. The coupling between electron transfer and proton translocation may be either direct (e.g., Q cycle) or indirect (conformational) (1, 2, 27-29). The arrangement of clusters described here is compatible with a direct mechanism, but it does not exclude an indirect coupling or a combination of both mechanisms.

Supporting Online Material

www.sciencemag.org/cgi/content/full/309/5735/771/DC1

Materials and Methods

Fig. S1

Table S1

References

References and Notes

View Abstract

Navigate This Article