Structure of subcomplex Iβ of mammalian respiratory complex I leads to new supernumerary subunit assignments Jiapeng Zhua, Martin S. Kinga, Minmin Yub, Liron Klipcana, Andrew G. W. Leslieb, and Judy Hirsta,1 a Medical Research Council Mitochondrial Biology Unit, Cambridge, CB2 0XY, United Kingdom; and bMedical Research Council Laboratory of Molecular Biology, Cambridge, CB2 0QH, United Kingdom

Mitochondrial complex I (proton-pumping NADH:ubiquinone oxidoreductase) is an essential respiratory enzyme. Mammalian complex I contains 45 subunits: 14 conserved “core” subunits and 31 “supernumerary” subunits. The structure of Bos taurus complex I, determined to 5-Å resolution by electron cryomicroscopy, described the structure of the mammalian core enzyme and allowed the assignment of 14 supernumerary subunits. Here, we describe the 6.8-Å resolution X-ray crystallography structure of subcomplex Iβ, a large portion of the membrane domain of B. taurus complex I that contains two core subunits and a cohort of supernumerary subunits. By comparing the structures and composition of subcomplex Iβ and complex I, supported by comparisons with Yarrowia lipolytica complex I, we propose assignments for eight further supernumerary subunits in the structure. Our new assignments include two CHCH-domain containing subunits that contain disulfide bridges between CX9C motifs; they are processed by the Mia40 oxidative-folding pathway in the intermembrane space and probably stabilize the membrane domain. We also assign subunit B22, an LYR protein, to the matrix face of the membrane domain. We reveal that subunit B22 anchors an acyl carrier protein (ACP) to the complex, replicating the LYR protein–ACP structural module that was identified previously in the hydrophilic domain. Thus, we significantly extend knowledge of how the mammalian supernumerary subunits are arranged around the core enzyme, and provide insights into their roles in biogenesis and regulation.

|

CHCH domain electron transport chain NADH:ubiquinone oxidoreductase

| LYR protein | mitochondria |

involved in regulation, homeostasis, mitochondrial biogenesis, or protection against oxidative damage (2, 5, 7, 8). Work to determine the structure of bacterial complex I culminated in the 3.3-Å resolution structure of intact Thermus thermophilus complex I (9–11). The bacterial structures revealed many catalytically important elements, including the flavin mononucleotide that oxidizes NADH, the iron–sulfur (FeS) clusters that transfer electrons from the flavin to ubiquinone, the ubiquinone-binding site, and elements of the proton transfer apparatus: four antiporter-like structures in the membrane, and a long transverse helix that appears to strap the membrane domain together. These features are closely conserved in the medium resolution structures of the complexes from B. taurus (2) (4UQ8.pdb) and the yeast Yarrowia lipolytica (12) (4WZ7.pdb). Furthermore, in the 5-Å resolution electron cryomicroscopy (cryo-EM) structure of the B. taurus enzyme, 18 supernumerary transmembrane helices (TMHs) were observed, and it was possible to propose assignments and partial models for 14 supernumerary subunits (2). One of these assignments (for the 13-kDa subunit) was confirmed recently in Y. lipolytica complex I using the anomalous diffraction from a bound zinc ion (13). Currently, 17 supernumerary subunits remain unassigned, and 11 supernumerary TMHs in the cryo-EM model have no subunits assigned to them. Fractionation of B. taurus complex I into subcomplexes Iα, Iλ, and Iβ has provided a wealth of information on its protein composition, and enabled subunits to be assigned to its major Significance

I

n mammalian mitochondria, respiratory complex I (NADH: ubiquinone oxidoreductase) (1, 2) contains 45 subunits with a combined mass of 1 MDa (3–6). Fourteen core subunits, conserved from bacteria to humans, constitute the minimal complex I and catalyze the energy transducing reaction: NADH oxidation and ubiquinone reduction coupled to proton translocation across the inner membrane. Complex I is thus critical for mammalian metabolism: it oxidizes the NADH generated by the tricarboxylic acid cycle and β-oxidation, produces ubiquinol for the rest of the respiratory chain, and contributes to the proton-motive force that supports ATP synthesis and transport processes. Seven of the core subunits are hydrophilic proteins that are encoded in the nucleus and imported to mitochondria, and the other seven are hydrophobic, membrane-bound proteins (known as the ND subunits) that are encoded by the mitochondrial genome. These two sets of seven subunits form the two major domains of complex I, in the matrix and inner membrane, which confer its characteristic L shape. The protein composition of complex I from Bos taurus heart mitochondria has been characterized extensively, and is the blueprint for the human enzyme (3–6). In addition to the 14 core subunits it contains 31 supernumerary subunits (including two copies of the acyl-carrier protein, SDAP; ref. 2). Several supernumerary subunits are homologous to proteins of known function and some are required for the assembly/stability of the complex, but the roles of most of them are unknown: they may be

www.pnas.org/cgi/doi/10.1073/pnas.1510577112

Mitochondrial complex I (proton-pumping NADH:ubiquinone oxidoreductase) is the largest respiratory chain enzyme. Mammalian complex I contains 45 subunits: the structures of the 14 “core” subunits (which are sufficient for catalysis and conserved from bacteria to humans) were described in the 5-Å resolution structure of Bos taurus complex I, but only 14 supernumerary subunits could be located. Here, we exploit new structural information from the membrane domain of mammalian complex I to assign eight further supernumerary subunits. We locate two oxidatively-folded CHCH-domain subunits in the intermembrane space, and reveal a second LYR protein–acyl carrier protein module. Thus, we extend knowledge of how the supernumerary subunits are arranged around the core, and provide insights into their roles in biogenesis and regulation. Author contributions: J.Z., M.S.K., and J.H. designed research; J.Z., M.S.K., and M.Y. performed research; J.Z., L.K., A.G.W.L., and J.H. analyzed data; and J.H. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. Data deposition: The coordinates and structure factors for subcomplex Iβ have been deposited with the Protein Data Bank (accession no. 5COD). 1

To whom correspondence should be addressed. Email: [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1510577112/-/DCSupplemental.

PNAS | September 29, 2015 | vol. 112 | no. 39 | 12087–12092

BIOCHEMISTRY

Edited by Hartmut Michel, Max Planck Institute of Biophysics, Frankfurt, Germany, and approved August 14, 2015 (received for review May 29, 2015)

domains (3, 5, 14). Subcomplex Iλ contains all of the redox cofactors and represents the hydrophilic domain; subcomplex Iα contains subcomplex Iλ plus additional subunits from the interdomain region; and subcomplex Iβ constitutes the distal section of the membrane domain. Subcomplex Iβ represents the domain of complex I that is the least-well-characterized structurally because it contains many unassigned supernumerary subunits (2) (Table 1). Here, we present the 6.8-Å resolution structure of subcomplex Iβ from B. taurus complex I, determined by X-ray crystallography. By comparing the structures of subcomplex Iβ and intact complex I, supported by comparisons of structural elements present in the B. taurus and Y. lipolytica complexes, we propose assignments for a further eight mammalian supernumerary subunits. Results The Structure of Membrane–Domain Subcomplex Iβ of Complex I. Fig. 1 presents the structure of B. taurus subcomplex Iβ and the relationship between subcomplex Iβ and complex I. The composition of the crystallized subcomplex was confirmed by mass spectrometry analyses on a sample of washed and solubilized crystals; all of the 15 expected subunits of subcomplex Iβ (5, 14) except core subunit ND4 were detected. Two core subunits, ND4 (orange in Fig. 1B) and the first 15 TMHs of ND5 (magenta in Fig. 1B), were readily identified in the electron density map (shown in Fig. S1) of the subcomplex. The C-terminal region of ND5 [the transverse helix that runs along the membrane domain plus TMH16 (2, 11)] is not observed, indicating that it dissociates from the core when the enzyme fragments and is disordered in the crystals. This observation is consistent with the position of TMH16, anchoring the end of the transverse helix at the interface of subunits ND2 and ND4L that are not present in the subcomplex. On the right side of the subcomplex, adjacent to core subunits ND4 and ND5, are six supernumerary TMHs conserved from the cryo-EM structure of intact complex I (2)

(Fig. 2; note that we refer to unassigned TMHs and helices by their labels in the cryo-EM structure of complex I, 4UQ8.pdb; ref. 2). On the left side, two TMHs were observed on the outside of the transverse helix in the intact enzyme (TMHs l and m) (2) but only half of TMH l (below the transverse helix) is observed in the subcomplex, indicating that TMH m and the upper part of TMH l become detached from the core when the transverse helix dissociates. Two additional TMH-like densities (in red in Fig. 2) are also observed in the subcomplex, and form crystal contacts between molecules in the asymmetric unit (Fig. S2). They were not observed in the density map of complex I (2) so may be disordered in the intact enzyme used for cryo-EM. Alternatively, they may be artifacts from reorganization of the structure around the dissociated transverse helix, or from the cocrystallization of dissociated subunits from outside of subcomplex Iβ. Two additional helices on the matrix face, and four short helices on the intermembrane space (IMS) face, have also been added to the model for subcomplex Iβ; they were not included in the cryo-EM model of complex I but weak density corresponding to them is visible in the density map (2). All of the new structural elements added to the model for subcomplex Iβ are shown in red in Fig. 1B. Assignments for Four TMH-Containing Subunits. The conservation of subunits (Table 1) (5, 8) and TMHs (Fig. 2) between complex I and subcomplex Iβ, supported by comparisons between the B. taurus (2) and Y. lipolytica (12) enzymes (Fig. S3), allows us to propose four new subunit assignments. First, we divide the TMHs into three classes. Class I contains seven TMHs from the cryo-EM structure of complex I that have already been assigned to subunits B9, B14.7, B16.6, and MWFE (2). None of these subunits have been detected in subcomplex Iβ (5), and because they are located in a clearly different region of the complex (Fig. 2, TMHs in cyan) they are not present in its structure. Class II contains four additional TMHs (c, d, e, and m) that are also absent completely from the structure of subcomplex Iβ, but they

Table 1. Supernumerary subunits in the membrane domain of complex I (5, 8) B. taurus 15 kDa MWFE B9 PGIV B14.7 B16.6 SDAP MNLL AGGG B12 B15 SGDH B17 B18 ASHI B22 PDSW ESSS KFYI B14.5b

Length*

H. sapiens

Y. lipolytica†

Length*

Features

Subcomplex

Assigned

105 70 83 171 141 144 88 58 72 97 128 143 127 136 158 178 175 125 49 120

NDUFS5 NDUFA1 NDUFA3 NDUFA8 NDUFA11 NDUFA13 NDUFAB1 NDUFB1 NDUFB2 NDUFB3 NDUFB4 NDUFB5 NDUFB6 NDUFB7 NDUFB8 NDUFB9 NDUFB10 NDUFB11 NDUFC1 NDUFC2

NIPM NIMM NI9M NUPM NUJM NB6M ACPM2

88 86 78 171 197 122 92

59 93

NB8M NIAM NI2M NIDM NESM

98 125 109 91 204

1α 1α 1α 1α 1α 1α and 1λ 1α and 1β 1β 1β 1β (1α and 1β)‡ 1β 1β 1β 1β 1β 1β 1β

This work Previously (2) Previously (2) Previously (2) Previously (2) Previously (2) Previously (2)

NB2M NB5M

2-CX9C motif 1 TMH 1 TMH 2 2-CX9C motifs 4 TMHs 1 TMH Acyl carrier (1 TMH) 1 TMH 1 TMH 1 TMH 1 TMH 1 TMH 2-CX9C motif 1 TMH LYR motif

NUXM NEBM NUNM NUUM

169 73 118 89

1 TMH 1 TMH 1 or 2 TMHs 2 TMHs 1 TMH (1 TMH) 1 TMH

(1β)‡

This work

This work This This This This This

work work work work work

*The lengths are given for the mature proteins, for B. taurus and Y. lipolytica. † Homology relationships between the mammalian and Y. lipolytica subunits are based predominantly on the analysis of Huynen et al. (15), and are in accord with the recent summary by Kmita and Zickermann (8). ‡ Subunits detected at low levels.

12088 | www.pnas.org/cgi/doi/10.1073/pnas.1510577112

Zhu et al.

A

B transverse helix

matrix

toe

heel IMS

are located adjacent to it in intact complex I and thus in the fracture zone that cleaves apart to separate subcomplex Iα from subcomplex Iβ (Fig. 2, complete TMHs in green). Finally, class III contains seven TMHs that are present in the structures of both intact complex I and subcomplex Iβ (TMHs f, g, h, i, j, k, and l, shown in blue in Fig. 2), plus the two additional, putative TMHs (Fig. 2, red) that are observed in the subcomplex. Assignments of TMHs in classes II and III are described in the following sections. Specific assignments for the fracture-zone TMHs c, d, e, and m were formulated as follows. First, three TMH-containing subunits have been detected either in both subcomplex Iα and Iβ (B15, one predicted TMH), in neither (KFYI, one predicted TMH), or in subcomplex Iβ at low levels only (B14.5b, one or two predicted TMHs) (5), suggesting they are from the fracture zone, so we assign them to TMHs c, d, e, and m. Second, TMHs c, d, and e form a group of three TMHs that must contain the two-TMH subunit B14.5b. Therefore, the isolated TMH m is from KFYI or B15; we assign it to B15 because both TMH m and subunit B15 are conserved in Y. lipolytica, whereas no KFYI homolog is known (Fig. S3; Table 1; refs. 8, 15). B15 has been detected in both subcomplexes Iα and Iβ (5, 14) and TMH m is suitably positioned to remain weakly associated with the disordered transverse helix in subcomplex Iβ, and with adjacent subunits such as B14.7 and 42 kDa in subcomplex Iα. Indeed, dissociation of the transverse helix severely disrupts the local structure in subcomplex Iβ (Fig. 1), including the loss of helix q that runs across the matrix face (Fig. 3A). As subunit B15 contains a 25-residue N-terminal helix we may speculate that helix q is part of B15. Third, the connectivity within the three TMHs (c, d, and e) in the cryo-EM density for B. taurus complex I was judged ambiguous (2), whereas TMHs d and e were modeled as a single chain in Y. lipolytica complex I (12), tentatively suggesting TMH c is from KFYI and TMHs d and e are from B14.5b. This assignment is supported by the absence of both TMH c (12) and KFYI (8, 15) in Y. lipolytica, and by the peripheral location of TMH c in B. taurus (2) (it interacts only with TMHs d and e, explaining why KFYI is lost upon fractionation whereas B14.5b is retained in low levels; ref. 5). Finally, models for B15, KFYI, and B14.5b, based on secondary structure predictions, fit well into the cryo-EM density (Fig. S4). For B14.5b the unequal lengths of the two helices define TMH e as the shorter, N-terminal helix. Only the identity of the protein forming TMHs d and e in Y. lipolytica remains unclear because no Y. lipolytica B14.5b homolog is known, and only subunit NUXM (with no known B. taurus homolog) has two predicted TMHs (Table 1). NUXM and B14.5b may be different proteins fulfilling the same structural Zhu et al.

role, or [as there are 18 TMHs observed in Y. lipolytica complex I (12) and only 16 predicted] the homolog of B14.5b may not have been identified yet. The latter suggestion is supported by the existence of a known B14.5b homolog in Neurospora crassa (16). In this case, NUXM can be assigned to the two TMHs that are specific to Y. lipolytica (Fig. S3), consistent with the composition of Y. lipolytica subcomplex Iδ (17), and with the position of these two TMHs adjacent to the three N-terminal TMHs of ND2 that are also absent from B. taurus (18). In subcomplex Iβ seven subunits with single predicted TMHs have been detected in apparently stoichiometric levels (AGGG, ASHI, ESSS, MNLL, SGDH, B12, and B17; Table 1; ref. 5). They must account for the TMHs observed in the structure of subcomplex Iβ, assuming that there are no as-yet-undetected proteins present. If the two new TMHs (β1 and β2) are real, then there are fewer proteins than TMHs so one or more subunits may be present in more than one copy, or have a more complex topology than predicted. Currently, there is little information to distinguish the TMHs from one another, but by comparing the structures from B. taurus (2) and Y. lipolytica (12) (Fig. S3) we propose an assignment for subunit ESSS to TMH g as follows.

(hydrophilic arm)

B14.7

left side

m

*β1

l

*β2 MWFE & B9

heel

j B16.6

g

e

*c d

*f

( )i *h *

( )

right side

k toe

Fig. 2. The transmembrane helices present in B. taurus subcomplex Iβ and complex I. The view is from the matrix face. Subunits ND4 and ND5 in subcomplex Iβ are in orange and magenta, respectively, except that the portion of ND5, including the transverse helix, that is not observed in the subcomplex is in wheat. Core subunits in complex I that are not present in the subcomplex are also shown in wheat. Supernumerary TMHs assigned previously that are not present in subcomplex Iβ are in cyan; additional TMHs that are observed in the structure of complex I but not subcomplex Iβ are in green; TMHs observed in the structures of both complex I and subcomplex Iβ are in blue, and the two putative TMHs observed only in subcomplex Iβ are in red. The unassigned TMHs are labeled according to their names in 4UQ8.pdb with those that are not conserved in the structure of Y. lipolytica complex I marked *, and those that are present in the Y. lipolytica structure but do not superimpose on the TMHs in the B. taurus enzyme marked (*).

PNAS | September 29, 2015 | vol. 112 | no. 39 | 12089

BIOCHEMISTRY

Fig. 1. Comparison of the structures of subcomplex Iβ and complex I from B. taurus. (A) The density from subcomplex Iβ (red) is superimposed on that of intact B. taurus complex I (gray, semitransparent). (B) The model for subcomplex Iβ is superimposed on that for intact complex I (pale blue, 4UQ8.pdb). Core subunits ND4 and ND5 in subcomplex Iβ are in orange and magenta, respectively, supernumerary structures observed in both complex I and subcomplex Iβ are in dark blue, and new structures observed in subcomplex Iβ are in red. The figure is presented in the opposite orientation (with the “toe” pointing to the left) to subsequent figures to show the region around the transverse helix.

A

(matrix face)

B22 (n) *q

*42 kDa

B

(IMS face)

fracture zone

180

*w

B16.6

β5

s

PDSW§ (t)

PGIV

*β6 heel

*β3 *β4

*o *p

(hydrophilic arm)

that it binds SDAP-β to the membrane domain (2, 21). Here, we used the models for SDAP-α and SDAP-β to map the B14SDAP-α module (from 4UQ8.pdb for B. taurus complex I) onto the matrix face of subcomplex Iβ, and found that three helices of B14 superimpose closely on three helices (chain n) in subcomplex Iβ, constituting an equivalent module (Fig. S5). Subunit B22, chain n, and SDAP-β are all conserved in Y. lipolytica (12) and increased density on the matrix face of ND5 in B. taurus, relative to Y. lipolytica (Fig. S3), is probably partly due to the longer length of the B. taurus B22 homolog. Therefore, we assign the three helices (chain n) on the matrix face of subcomplex Iβ (and complex I) to subunit B22 (Figs. 3A and 4).

SDAP

15 kDa (r)

*β7 *β8

B18§ (u,v) toe

Fig. 3. The matrix and IMS faces of B. taurus complex I. (A) The hydrophilic arm has been removed to show the supernumerary subunits present on the matrix face of complex I. (B) The supernumerary subunits present on the IMS face of complex I. For both panels, the structures of the supernumerary subunits on the faces are in color and labeled accordingly. The structures of ND4 and ND5 are in light orange with the other subunits in white and the line along which the complex fractures to form subcomplex Iβ (to the right side) is indicated. The figure was created using 4UQ8.pdb for complex I (2) plus additional elements from the structure of subcomplex Iβ. Asterisks indicate elements that are not observed in the structure of Y. lipolytica complex I (12) and § indicates that an alternative assignment in which the assignments for PDSW and B18 are exchanged for one another is possible.

Four unassigned TMHs in subcomplex Iβ (TMHs g, j, k, and l) are closely conserved in the structures of the complexes from B. taurus and Y. lipolytica (Fig. S3) and three of the unassigned TMH-containing subunits identified in subcomplex Iβ (ASHI, ESSS, and B12; Table 1) have known Y. lipolytica homologs. Therefore, ASHI, ESSS, and B12 are candidates for TMHs g, j, k, and l. A domain is present on the matrix face of Y. lipolytica complex I, on the right-side adjacent to ND4 and conserved TMH g (Fig. S3) that is not present in B. taurus. In the model for Y. lipolytica complex I (12) TMH g is connected to one of the helices modeled into this domain, as well as to a conserved helix (helix β5 in Fig. 3B) in the IMS, which is in a position consistent with the antibody label to Y. lipolytica ESSS observed by EM (19). The assignment of ESSS to TMH g explains the extra density on the matrix face of Y. lipolytica complex I because the Y. lipolytica ESSS N-terminal domain of ∼120 residues is twice as long as in B. taurus. Furthermore, in both homologs the predicted TMH is followed closely by a further helix of ∼20 residues (the C-terminal domain contains 45 residues in B. taurus), which is consistent with helix β5 in the IMS. In consequence, we propose that TMH g is formed by subunit ESSS, with the N terminus on the matrix face. Identification of Subunit B22 on the Matrix Face. Two copies of

SDAP, the mitochondrial acyl carrier protein, were observed in the cryo-EM density for B. taurus complex I (2) and named SDAP-α and SDAP-β as they are in subcomplexes Iα and Iβ, respectively. SDAP-α is bound to the hydrophilic domain by subunit B14, an LYR protein (2, 20, 21). SDAP-β is clearly visible in the map from subcomplex Iβ, on the matrix face of the membrane domain. A second LYR family complex I subunit (B22) is also present in subcomplex Iβ (5), raising the possibility 12090 | www.pnas.org/cgi/doi/10.1073/pnas.1510577112

Assignments for Three Subunits on the IMS Face. Two subcomplex Iβ subunits that do not contain any TMHs have not been discussed or modeled thus far: B18 and PDSW. B18 is known to be located on the IMS face, and to contain a CHCH domain—a double CX9C domain that forms a helix-turn-helix structure with two disulfides between the helices (22, 23). Secondary structure predictions for both the B. taurus and Y. lipolytica homologs indicate it is followed by a helix of more than 30 residues. There are two candidates for B18 in the structure of subcomplex Iβ (Fig. 3B): the two helices in chain t, and chains u and v together. Little is known about subunit PDSW: we assign it to the IMS face because it is predicted to contain three helices of more than 15 residues (including one of 35) that are difficult to accommodate in helices observed on the matrix face of subcomplex Iβ but can be readily accommodated on the IMS side, and because it contains conserved Cys residues that may confer structural stability through disulfides formed in the oxidizing IMS (24). The two helices in chain t, and chains u and v together, are thus also candidates for the assignment of PDSW. Although we are confident that B18 and PDSW together account for the four helices in chains t, u, and v, it is not currently possible to confidently

B8 18 kDa B13 B14 SDAP-α 42 kDa

13 kDa

SDAP-β

39 kDa MWFE & B9 B16.6 B14.7

ESSS B15

B22

15 kDa B18* PGIV

KFYI B14.5b

PDSW*

Fig. 4. Summary of the assignments of supernumerary subunits in B. taurus complex I. A semitransparent surface for the density map for mammalian complex I is shown in pale gray, with the surface from the core subunits in wheat. Structural models for the supernumerary subunits are shown in color and labeled accordingly, and structural elements that cannot be assigned confidently in the current map are in blue. Subunits assignments proposed here are labeled in larger font size and underlined to differentiate them from those proposed previously (2). The asterisks for PDSW and B18 indicate that an alternative assignment in which the two assignments are exchanged for one another is possible. The model shown comprises the cryo-EM model (4UQ8.pdb) plus new elements modeled in subcomplex Iβ.

Zhu et al.

Discussion Fig. 4 summarizes our current picture of the locations and structures of the supernumerary subunits in B. taurus complex I. The enzyme contains 45 subunits, comprising the 14 core subunits, and 31 supernumerary subunits. Structural models for the core subunits, together with assignments and partial models for 14 supernumerary subunits, were described in the 5-Å cryo-EM structure (2), leaving 17 subunits unassigned. Here, we have compared the structures of subcomplex Iβ and intact B. taurus complex I, and the density maps from the B. taurus and Y. lipolytica complexes, to propose assignments for a further eight supernumerary subunits. Therefore, assignments have now been proposed for a total of 36 subunits, and 9 subunits remain unassigned. Little is known about three of our newly assigned subunits: KFYI (human NDUFC1), the smallest mammalian subunit that occupies a peripheral location in the complex, or the B14.5b (NDUFC2) or B15 (NDUFB4) subunits. NDUFC2 has been denoted a site of accelerated amino acid replacement due to its coevolution with mitochondrial-encoded ND4 (25); this is consistent with its assignment to a position adjacent to ND4 in our structure. Perhaps because it is encoded on the X chromosome (26), our newly assigned ESSS subunit (human NDUFB11) has received more attention. Mutations in it cause an X-linked disorder, microphthalmia with linear skin defects syndrome (27), and two NDUFB11 isoforms have been identified, with the alternative form linked to apoptosis and neurodegeneration (28). We predict that the 10-residue insertion of the alternative isoform extends the loop between the end of the TMH g and the start of IMS helix β5. Finally, Ser20 in a canonical motif (KRXS) in the N-terminal matrix-domain of B. taurus ESSS is phosphorylated in complex I in a cAMP-dependent manner by protein kinase A (29). However, Ser20 is not conserved in humans so the wider implications of this observation are unclear. Both complex I LYR proteins (subunits B14 and B22, human NDUFA6 and NDUFB9) have now been assigned in the structure. Of the nine further mitochondrial LYR proteins known (21), one interacts with the L-cysteine desulfurase for FeS cluster and respiratory chain biogenesis (30), and several function as respiratory chain assembly factors (21, 31). The LYR motif may recruit the apparatus for FeS cluster insertion to complex II (mitochondrial succinate:ubiquinone oxidoreductase), and it has been suggested that LYR motifs in general may be used to identify new FeS-containing proteins (32). Neither of the LYRcomplex I subunits are FeS proteins; B14 is bound to the FeScontaining hydrophilic arm of complex I but B22 is very distant from it. Rather, the duplicated LYR–SDAP module may indicate the LYR motif recruits the SDAP protein to the complex. As the mitochondrial acyl-carrier protein, the SDAP subunits are derivatized by phosphopantetheines with acyl chains attached (33). Zhu et al.

However, whether they retain any function in fatty-acid synthesis or biosynthesis once sequestered onto the complex is unknown. All three CHCH domain complex I subunits (PGIV, 15 kDa, and B18, human NDUFA8, NDUFS5, and NDUFB7) have now been assigned in the structure of B. taurus complex I. CHCH domain proteins are imported to the IMS and processed by an oxidative protein folding pathway, mediated by Mia40 and Erv1, that forms disulfides between pairs of cysteines (34). PGIV, 15 kDa, and B18 are canonical substrates for Mia40, but noncanonical substrates are also known (24) and PDSW may be folded by the same pathway. The CHCH domain family also includes several respiratory-chain assembly factors and a subunit of cytochrome c oxidase (24, 35). B18 is the only subunit in complex I that is myristoylated (36) so it is interesting that for another CHCH domain protein (CHCHD3) myristoylation has been proposed to be important for transporting it across the outer membrane (37). Similarly, a CHCH domain protein (CHCHD2) required for the full activity of cytochrome c oxidase is translocated to the nucleus during hypoxia to regulate transcription of an isoform of cytochrome c oxidase subunit 4 (38) suggesting that the CHCH domain subunits of complex I may similarly have regulatory roles, perhaps acting through the redox status of their disulfide linkages. In summary, we have proposed assignments for eight further supernumerary subunits in the structure of mammalian complex I, and thus described the most biochemically complete structural model so far. The supernumerary subunits in mammalian complex I comprise approximately half of its mass, and assigning their locations within the overall structure of the enzyme is an essential step toward defining their roles, as well as a staging post on the way to a complete structural model. The supernumerary subunit cohort in general may act as a scaffold, supporting the core structure, or protecting it against oxidative damage. Some supernumerary subunits may have independent functions, in complex I assembly, mitochondrial biogenesis, or enzyme regulation. Pathological mutations in the supernumerary subunits of complex I have been identified clinically in mitochondrial diseases. Our new assignments significantly advance knowledge of the structure of mammalian complex I, and provide new insights into the roles of its supernumerary subunits. Experimental Methods Protein Preparation. Complex I was purified from B. taurus heart mitochondrial membranes as described previously (39) with minor modifications. Here, 150–200 mg of membranes at ∼10 mg mL−1 were solubilized using DDM (n-dodecyl-β-D-maltoside) but the anion exchange and size-exclusion chromatography steps were at pH 7.8 in 0.1 and 0.04% Cymal 7 (7-cyclohexyl-1-heptyl-β- D -maltoside), respectively, and no phospholipids were added. The peak fractions from the size-exclusion column were pooled and concentrated fivefold to ∼300 μL using a 100-kDa molecular-weight cutoff Vivaspin centrifugal concentrator. Then, to remove the glycerol present in the chromatography buffers, the protein was twice diluted with 5 mL of buffer containing 20 mM Tris·HCl (pH 7.8), 150 mM NaCl and 0.05% Cymal-7, and reconcentrated. Finally, the protein was concentrated to 7 mg mL−1; the final detergent concentration was found to reach ∼2% (wt/vol) (40). Protein Crystallization. Initial crystallization screening was performed at 22 and 4 °C in MRC 96-well plates by sitting-drop vapor diffusion (200-nL complex I + 200-nL reservoir) using the Memgold HT-96 and MemStart + MemSys HT-96 kits (Molecular Dimensions). Crystals appeared in Memgold HT-96 [100 mM NaCl, 10 mM Hepes pH 8, and 11% PEG 1,500 (wt/vol)], but diffracted to only ∼20-Å resolution. Extensive optimization of the conditions was then performed, including screening with the Hampton Research Additive Screen. The best platelike crystals of maximally 150 × 100 × 50 μm appeared at 17 °C after 4 wk, and reached final size after 7–8 wk. They were formed by mixing 2-μL protein solution with 1.5-μL reservoir solution comprising 100 mM Tris-Cl (pH 7.8), 100 mM MgCl2, 100 mM NaCl, 150 mM HCO2Na, 3.2 mM DDM, and 10–13% PEG 3350 (vol/vol). The crystals formed were of subcomplex Iβ, not complex I; the subcomplex must form by fragmentation of the intact complex in the crystallization solution.

PNAS | September 29, 2015 | vol. 112 | no. 39 | 12091

BIOCHEMISTRY

decide which subunit is which. We tentatively propose that B18 is from chains u and v (Figs. 3B and 4) because the helices are more parallel than in chain t, the density more closely resembles that assigned previously to the two CHCH domains in subunit PGIV (chain X) in the cryo-EM density map from B. taurus (2), and the lengths of the observed helices are consistent with their predicted secondary structures. We do not exclude the alternative assignment of B18 to chain t and PDSW to chains u and v. Finally, three subunits in complex I contain double CX9C motifs: PGIV was assigned previously to the heel of the complex (2), B18 has been assigned here in subcomplex Iβ, and the 15-kDa subunit is known to be located in subcomplex Iα and on the IMS face (5, 22). On this basis, we searched for the typical two-helix motif in the density maps for the B. taurus (2) and Y. lipolytica (12) complexes and readily identified it in the expected region. Consequently, we assign and model the 15-kDa subunit as a twohelix CHCH domain joined to helix r (Figs. 3B and 4), consistent with secondary structure predictions.

Data Collection and Processing. Crystals were transferred to a cryoprotectant solution comprising 100 mM Tris·HCl (pH 7.8), 100 mM MgCl2, 100 mM NaCl, 150 mM HCO2Na, 3.2 mM DDM, and 30% PEG 3350, then flash frozen in liquid nitrogen. X-ray data were collected to 6.8-Å resolution from a single crystal at 100 K at beam line ID23-1 of the European Synchrotron Radiation Facility using an ADSC Q315R detector. Images were processed with XDS (41) and Mosflm (42). The crystals belong to space group P212121, with six molecules (complexes) per asymmetric unit and 62% solvent content (Table S1). Phasing, Model Building, and Refinement. The structure of subcomplex Iβ was solved by molecular replacement implemented in Phaser (43) with all of the atomic temperature factors set to 100. Search models comprising the bacterial homologs of ND4 and ND5 [3RKO.pdb (10) or 4HEA.pdb (11)] failed to provide a solution, but successful phase determination was achieved using a polyalanine search model comprising ND4 and the first 14 helices of ND5 from the 5-Å resolution cryo-EM structure of B. taurus complex I (4UQ8.pdb; ref. 2). Well-defined electron density for the supernumerary subunits was revealed around the two core subunits and fitted using corresponding elements from 4UQ8.pdb. Structure refinement was performed by rigid-body refinement using Phenix (44) (with each subunit treated as a single rigid

1. Hirst J (2013) Mitochondrial complex I. Annu Rev Biochem 82:551–575. 2. Vinothkumar KR, Zhu J, Hirst J (2014) Architecture of mammalian respiratory complex I. Nature 515(7525):80–84. 3. Walker JE (1992) The NADH:ubiquinone oxidoreductase (complex I) of respiratory chains. Q Rev Biophys 25(3):253–324. 4. Walker JE, et al. (1992) Sequences of 20 subunits of NADH:ubiquinone oxidoreductase from bovine heart mitochondria. Application of a novel strategy for sequencing proteins using the polymerase chain reaction. J Mol Biol 226(4):1051–1072. 5. Hirst J, Carroll J, Fearnley IM, Shannon RJ, Walker JE (2003) The nuclear encoded subunits of complex I from bovine heart mitochondria. Biochim Biophys Acta 1604(3): 135–150. 6. Carroll J, et al. (2006) Bovine complex I is a complex of 45 different subunits. J Biol Chem 281(43):32724–32727. 7. Hirst J (2011) Why does mitochondrial complex I have so many subunits? Biochem J 437(2):e1–e3. 8. Kmita K, Zickermann V (2013) Accessory subunits of mitochondrial complex I. Biochem Soc Trans 41(5):1272–1279. 9. Sazanov LA, Hinchliffe P (2006) Structure of the hydrophilic domain of respiratory complex I from Thermus thermophilus. Science 311(5766):1430–1436. 10. Efremov RG, Sazanov LA (2011) Structure of the membrane domain of respiratory complex I. Nature 476(7361):414–420. 11. Baradaran R, Berrisford JM, Minhas GS, Sazanov LA (2013) Crystal structure of the entire respiratory complex I. Nature 494(7438):443–448. 12. Zickermann V, et al. (2015) Structural biology. Mechanistic insight from the crystal structure of mitochondrial complex I. Science 347(6217):44–49. 13. Kmita K, et al. (2015) Accessory NUMM (NDUFS6) subunit harbors a Zn-binding site and is essential for biogenesis of mitochondrial complex I. Proc Natl Acad Sci USA 112(18):5685–5690. 14. Finel M, Skehel JM, Albracht SPJ, Fearnley IM, Walker JE (1992) Resolution of NADH: ubiquinone oxidoreductase from bovine heart mitochondria into two subcomplexes, one of which contains the redox centers of the enzyme. Biochemistry 31(46): 11425–11434. 15. Gabaldón T, Rainey D, Huynen MA (2005) Tracing the evolution of a large protein complex in the eukaryotes, NADH:ubiquinone oxidoreductase (Complex I). J Mol Biol 348(4):857–870. 16. Huynen MA, de Hollander M, Szklarczyk R (2009) Mitochondrial proteome evolution and genetic disease. Biochim Biophys Acta 1792(12):1122–1129. 17. Angerer H, et al. (2011) A scaffold of accessory subunits links the peripheral arm and the distal proton-pumping module of mitochondrial complex I. Biochem J 437(2): 279–288. 18. Birrell JA, Hirst J (2010) Truncation of subunit ND2 disrupts the threefold symmetry of the antiporter-like subunits in complex I from higher metazoans. FEBS Lett 584(19): 4247–4252. 19. Abdrakhmanova A, et al. (2004) Subunit composition of mitochondrial complex I from the yeast Yarrowia lipolytica. Biochim Biophys Acta 1658(1-2):148–156. 20. Angerer H, et al. (2014) The LYR protein subunit NB4M/NDUFA6 of mitochondrial complex I anchors an acyl carrier protein and is essential for catalytic activity. Proc Natl Acad Sci USA 111(14):5207–5212. 21. Angerer H (2015) Eukaryotic LYR proteins interact with mitochondrial protein complexes. Biology (Basel) 4(1):133–150. 22. Szklarczyk R, et al. (2011) NDUFB7 and NDUFA8 are located at the intermembrane surface of complex I. FEBS Lett 585(5):737–743. 23. Banci L, et al. (2012) Structural characterization of CHCHD5 and CHCHD7: Two atypical human twin CX9C proteins. J Struct Biol 180(1):190–200. 24. Herrmann JM, Riemer J (2014) Three approaches to one problem: Protein folding in the periplasm, the endoplasmic reticulum, and the intermembrane space. Antioxid Redox Signal 21(3):438–456. 25. Gershoni M, et al. (2014) Disrupting mitochondrial-nuclear coevolution affects OXPHOS complex I integrity and impacts human health. Genome Biol Evol 6(10):2665–2680.

12092 | www.pnas.org/cgi/doi/10.1073/pnas.1510577112

body) and the densities from the six molecules of the asymmetric unit were averaged using the NCS-averaging function in Coot (45). Subsequently, polyalanine helices not present in 4UQ8.pdb were added to the model using Coot (their orientations around the helix axis are not known), to account for further clearly defined density features. The final poly-alanine model was rerefined by rigid-body refinement to Rfree = 43.3%, with all of the atomic temperature factors set manually to 100. TMH and Secondary Structure Predictions. TMHs were predicted using TMHMM2 (46), HMMTOP2 (47), and the TOPCONS suite (48) (seven methods in total) and are presented as consensus values with less represented values in brackets and single outliers discarded. Secondary structures were predicted using PSIPRED (49). ACKNOWLEDGMENTS. We thank Professor Randy Read (Cambridge) for advice on molecular replacement, the European Synchrotron Radiation Facility (ESRF) and the Diamond Light Source for access to synchrotron facilities, and the staff of beam lines ID23-1, ID14-4 (ESRF) and I02, I03, I04, and I24 (Diamond) for assistance. This work was funded by The Medical Research Council Grants U105663141 (to J.H.) and U105184325 (to A.G.W.L.).

26. Carroll J, Shannon RJ, Fearnley IM, Walker JE, Hirst J (2002) Definition of the nuclear encoded protein composition of bovine heart mitochondrial complex I. Identification of two new subunits. J Biol Chem 277(52):50311–50317. 27. van Rahden VA, et al. (2015) Mutations in NDUFB11, encoding a complex I component of the mitochondrial respiratory chain, cause microphthalmia with linear skin defects syndrome. Am J Hum Genet 96(4):640–650. 28. Panelli D, et al. (2013) The mechanism of alternative splicing of the X-linked NDUFB11 gene of the respiratory chain complex I, impact of rotenone treatment in neuroblastoma cells. Biochim Biophys Acta 1829(2):211–218. 29. Chen R, Fearnley IM, Peak-Chew SY, Walker JE (2004) The phosphorylation of subunits of complex I from bovine heart mitochondria. J Biol Chem 279(25):26036–26045. 30. Lim SC, et al. (2013) Mutations in LYRM4, encoding iron-sulfur cluster biogenesis factor ISD11, cause deficiency of multiple respiratory chain complexes. Hum Mol Genet 22(22):4460–4473. 31. Ghezzi D, et al. (2009) SDHAF1, encoding a LYR complex-II specific assembly factor, is mutated in SDH-defective infantile leukoencephalopathy. Nat Genet 41(6):654–656. 32. Maio N, et al. (2014) Cochaperone binding to LYR motifs confers specificity of iron sulfur cluster delivery. Cell Metab 19(3):445–457. 33. Runswick MJ, Fearnley IM, Skehel JM, Walker JE (1991) Presence of an acyl carrier protein in NADH:ubiquinone oxidoreductase from bovine heart mitochondria. FEBS Lett 286(1-2):121–124. 34. Riemer J, Bulleid N, Herrmann JM (2009) Disulfide formation in the ER and mitochondria: Two solutions to a common process. Science 324(5932):1284–1287. 35. Cavallaro G (2010) Genome-wide analysis of eukaryotic twin CX9C proteins. Mol Biosyst 6(12):2459–2470. 36. Carroll J, et al. (2005) The post-translational modifications of the nuclear encoded subunits of complex I from bovine heart mitochondria. Mol Cell Proteomics 4(5): 693–699. 37. Darshi M, Trinh KN, Murphy AN, Taylor SS (2012) Targeting and import mechanism of coiled-coil helix coiled-coil helix domain-containing protein 3 (ChChd3) into the mitochondrial intermembrane space. J Biol Chem 287(47):39480–39491. 38. Aras S, et al. (2015) MNRR1 (formerly CHCHD2) is a bi-organellar regulator of mitochondrial metabolism. Mitochondrion 20:43–51. 39. Sharpley MS, Shannon RJ, Draghi F, Hirst J (2006) Interactions between phospholipids and NADH:ubiquinone oxidoreductase (complex I) from bovine mitochondria. Biochemistry 45(1):241–248. 40. Urbani A, Warne T (2005) A colorimetric determination for glycosidic and bile saltbased detergents: Applications in membrane protein research. Anal Biochem 336(1): 117–124. 41. Kabsch W (2010) XDS. Acta Crystallogr D Biol Crystallogr 66(Pt 2):125–132. 42. Battye TGG, Kontogiannis L, Johnson O, Powell HR, Leslie AGW (2011) iMOSFLM: a new graphical interface for diffraction-image processing with MOSFLM. Acta Crystallogr D Biol Crystallogr 67(Pt 4):271–281. 43. McCoy AJ, et al. (2007) Phaser crystallographic software. J Appl Cryst 40(Pt 4):658–674. 44. Adams PD, et al. (2010) PHENIX: A comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr D Biol Crystallogr 66(Pt 2):213–221. 45. Emsley P, Lohkamp B, Scott WG, Cowtan K (2010) Features and development of Coot. Acta Crystallogr D Biol Crystallogr 66(Pt 4):486–501. 46. Krogh A, Larsson B, von Heijne G, Sonnhammer ELL (2001) Predicting transmembrane protein topology with a hidden Markov model: Application to complete genomes. J Mol Biol 305(3):567–580. 47. Tusnády GE, Simon I (1998) Principles governing amino acid composition of integral membrane proteins: Application to topology prediction. J Mol Biol 283(2):489–506. 48. Bernsel A, Viklund H, Hennerdal A, Elofsson A (2009) TOPCONS: Consensus prediction of membrane protein topology. Nucleic Acids Res 37(Web Server issue):W465-8. 49. Jones DT (1999) Protein secondary structure prediction based on position-specific scoring matrices. J Mol Biol 292(2):195–202.

Zhu et al.

Structure of subcomplex Iβ of mammalian respiratory complex I leads to new supernumerary subunit assignments.

Mitochondrial complex I (proton-pumping NADH:ubiquinone oxidoreductase) is an essential respiratory enzyme. Mammalian complex I contains 45 subunits: ...
NAN Sizes 1 Downloads 10 Views