Structural Basis for Heavy Metal Detoxification by an Atm1-Type ABC Exporter

See allHide authors and affiliations

Science  07 Mar 2014:
Vol. 343, Issue 6175, pp. 1133-1136
DOI: 10.1126/science.1246489


Although substantial progress has been achieved in the structural analysis of exporters from the superfamily of adenosine triphosphate (ATP)–binding cassette (ABC) transporters, much less is known about how they selectively recognize substrates and how substrate binding is coupled to ATP hydrolysis. We have addressed these questions through crystallographic analysis of the Atm1/ABCB7/HMT1/ABCB6 ortholog from Novosphingobium aromaticivorans DSM 12444, NaAtm1, at 2.4 angstrom resolution. Consistent with a physiological role in cellular detoxification processes, functional studies showed that glutathione derivatives can serve as substrates for NaAtm1 and that its overexpression in Escherichia coli confers protection against silver and mercury toxicity. The glutathione binding site highlights the articulated design of ABC exporters, with ligands and nucleotides spanning structurally conserved elements to create adaptable interfaces accommodating conformational rearrangements during the transport cycle.

Crossing the Membrane

Adenosine triphosphate (ATP)–binding cassette (ABC) transporters couple ATP hydrolysis to the translocation of a wide variety of substrates across cell membranes. Srinivasan et al. (p. 1137) describe the structure of a yeast mitochondrial transporter involved in Fe-S protein biogenesis. The structure reveals bound glutathione, which suggests that glutathione is part of the translocated substrate. J. Y. Lee et al. (p. 1133) describe the structure of a bacterial ABC transporter that confers protection against silver and mercury. This protein also binds glutathione derivatives. The structure provides insight into how ligand interactions are coupled to ATP hydrolysis.

The Atm1/ABCB7/HMT1/ABCB6 family of adenosine triphosphate (ATP)–binding cassette (ABC) exporters has been implicated in transition metal homeostasis and detoxification processes (14). In eukaryotes, the Atm1/ABCB7 ortholog is present in the inner membrane of mitochondria and is required for the formation of cytosolic iron-sulfur cluster–containing proteins (5); mutations of this protein result in mitochondrial accumulation of iron (6) and are responsible for X-linked sideroblastic anemia (7). ABCB6 was first identified as a porphyrin transporter present in the outer membrane of mitochondria (8), but it may also play a role in arsenic resistance (9) and is present in the Golgi (10), lysosomal (11), and plasma membranes (12). Other homologs have been implicated in heavy metal detoxification, including cadmium resistance by heavy metal tolerance factor–1 (HMT1) in Caenorhabditis elegans and Drosophila melanogaster (1315). The available evidence (16) suggests that for some of these transporters, the substrates may be related to the thiol-containing glutathione (GSH), a tripeptide of sequence γ-Glu-Cys-Gly. GSH is an abundant cellular antioxidant that mediates resistance to electrophiles, heavy metals, and xenobiotics (1719).

As a starting point for addressing the molecular mechanism of this family of ABC transporters and their physiological role(s) in metal homeostasis, we have solved the structure of the Atm1-family ABC exporter from Novosphingobium aromaticivorans (NaAtm1) at 2.4 Å resolution (20). NaAtm1 shares ~45% sequence identity with Saccharomyces cerevisiae Atm1, human ABCB7, and HMT-1 (fig. S1), and hence it should provide a useful model for these eukaryotic transporters [see Srinivasan et al. (21)]. Reflecting the basic architecture of the ABC exporter family that was first elucidated for the Sav1866 transporter (22) and subsequently found in other ABC exporters (2327), NaAtm1 forms a homodimer, with each subunit containing six transmembrane (TM) helices fused to the nucleotide-binding domain (NBD) (Fig. 1). NaAtm1 adopts an inward-facing conformation that most closely resembles (fig. S2) the structures of a Thermotoga maritima ABC transporter (24) and the human ABCB10 transporter (26).

Fig. 1 Structural representations of NaAtm1.

(A) Ribbon diagram illustrating the dimeric structure of NaAtm1, viewed normal to the molecular two-fold axis with the membrane-spanning domains and NBDs oriented toward the top and bottom, respectively. The approximate position of the membrane is designated by the gray bilayer, with the periplasm- and cytoplasm-facing surfaces toward the top and bottom, respectively. (B) Binding sites for GSSG illustrated in the same orientation as (A), with the ligand depicted as space-filling models; primary and secondary binding sites are represented by green and yellow carbons, respectively. (C) Representation of one NaAtm1 subunit emphasizing the secondary structure arrangement, based on the Sav1866 nomenclature (22). ICL, intracellular loop.

Because functional studies for NaAtm1 have not been reported, we implemented three approaches to identify potential substrates for transport: (i) the ability of ligands to stimulate adenosine triphosphatase (ATPase) activity (16, 28, 29); (ii) in vitro transport assays measuring uptake into E. coli membrane vesicles; and (iii) in vivo susceptibility assays to assess the ability of NaAtm1 to restore tolerance to heavy metals in sensitive E. coli strains (30, 31). As observed for S. cerevisiae Atm1p (16), the ATPase activity of NaAtm1 was stimulated by GSH and related compounds, with the highest activity (defined by the ratio of the catalytic rate constant kcat to the Michaelis constant Km) observed for metallated, aromatic hydrocarbon–conjugated, and oxidized GSH derivatives (Table 1 and figs. S3 and S4). For the most active substrates, the ATPase activity was stimulated by a factor of 3 to 7 over the basal rate. Effective stimulation of ATPase activity by ligands appears to require both free α-carboxyl and α-amino groups, because glycine and the γ-peptides GSH, γ-Glu-Cys, and ophthalmic acid (glutathione with the thiol group replaced with methyl) exhibited stronger ATPase activities relative to Cys-Gly and monocarboxylic acids (Table 1); these properties contrast with yeast Atm1, in which the GSH thiol group is critical for efficient ATPase activity (16). However, the presence of α-amino and carboxyl groups in the transported ligand, although apparently necessary, is not sufficient for ATPase activity, as evidenced by the low activity of many amino acids (fig. S5). Further demonstration of the substrate specificity of NaAtm1 was provided by direct transport assays, where we observed ATP-driven, NaAtm1-dependent uptake of oxidized glutathione (GSSG) into E. coli membrane vesicles (Fig. 2A). The ability of NaAtm1 to transport GSSG distinguishes it from the E. coli CydDC ABC transporter that is selective for reduced glutathione (32).

Table 1 Kinetic constants for the ATPase activity of selected substrates derived from a Michaelis-Menten type analysis.

The measured ATPase activity is fit to the equation v/[E]T = kcat[S]/(Km + [S]) + kbasal, where v is the measured ATPase activity, [E]T is the total concentration of NaAtm1, [S] is the substrate concentration, and kbasal is the basal rate of ATPase activity, which is measured to be 8.8 ± 0.8 min–1 in 1 mM ATP.

View this table:
Fig. 2 In vitro and in vivo assays of NaAtm1 function.

(A) Amount of GSSG transported per milligram of protein accumulated inside vesicles prepared from E. coli membranes overexpressing wild-type NaAtm1 or the E523Q (Glu523 → Gln) Walker B motif ATPase-defective mutant. (B) Optical density of cells after 12 hours of growth in the presence of the indicated AgNO3 concentrations. E. coli strain GG44 (Cu+/Ag+ sensitive) was transformed with either an empty pET15 plasmid or a pET15 plasmid encoding NaAtm1. (C) Optical density of cells after 12 hours of growth in the presence of the indicated HgCl2 concentrations. E. coli strain GG48 (Zn2+/Cd2+/Hg2+ sensitive) was transformed as in (B). Error bars represent SD of three independent measurements.

The greatest stimulation of ATP hydrolysis by NaAtm1 was observed for the Ag and Hg complexes of GSH (Table 1). In view of the toxicity of these metals at low intracellular concentrations, we tested the in vivo relevance of the ATPase results by means of metal susceptibility assays (30, 31). Consistent with a role for NaAtm1 in exporting Ag-GSH and/or Hg-GSH, expression of NaAtm1 in metal-sensitive E. coli strains conferred protection against otherwise toxic concentrations of Ag+ and Hg2+ (Fig. 2, B and C). Taken together, these results suggest that NaAtm1 mediates resistance to heavy metal toxicity, likely through export of metallated GSH complexes, although other functional properties probably remain to be identified.

As illustrated with GSSG, the major binding site is located ~5 Å into the transmembrane region, dominated by interactions from helices TM5 and TM6 and accessible from the cytoplasmic side (Fig. 3, A and B, and fig. S6). This location is ~15 Å closer to the cytoplasmic surface than the ligand-binding sites in the mouse P-glycoprotein structure (23, 33), but the majority of those interactions also involve TM5 and TM6 as well as the dimeric equivalent TM11–12. Key interactions are formed between the free amino and carboxyl groups of GSSG and the protein. The α-carboxyl group of the γ-Glu forms hydrogen bonds with the amide NHs of Gly319 and Met320 in TM6 and with the side chain of Gln272 (TM5), while the α-amino group of Glu interacts with the peptide carbonyl of Asp316 (TM6) and the side chains of Asn269 and Gln272 (TM5). At the other end of the glutathione tripeptide, the α-carboxylate of the Gly forms a hydrogen bond to the side chain of Tyr156 (TM3) on the opposing subunit from the one interacting with the γ-Glu residue of GSH (fig. S7). The residues on TM3 and TM6 interacting with GSSG are positioned near helical irregularities, including a stretch of 310 helix between residues 314 and 317 following a kink at the conserved Pro314 in TM6, and a π helix between residues 158 and 162 in TM3; these may represent regions that undergo conformational changes during the transport cycle (34). The nonpolar contacts between the protein and ligand are primarily formed by the side chains of the broadly conserved residues Leu265 and Leu268 packing against the Gly end of GSSG, and by the more variable side chains of Met317 and Met320 interacting near the disulfide of GSSG (fig. S8).

Fig. 3 Binding sites of glutathione derivatives to NaAtm1.

(A) The primary binding site of GSSG, with the substrate and interacting side chains depicted as bonds and the polypeptide backbone in ribbons. The substrate carbons are colored light gray; the protein side-chain carbons are green. Hydrogen bonds between the protein and ligands are represented by yellow dashes. (B) The secondary binding site for GSSG. (C) Interactions between the γ-Glu of GSH and surrounding residues of NaAtm1. The carbon atoms of GSH are shaded light gray; green carbons denote side chains directly interacting with GSH, and blue carbons indicate side chains interacting with (green) residues interacting with GSH. (D) ATPase activity (kcat) of selected mutations in the ligand-binding site; rate constants in the absence of substrate and in the presence of 10 mM GSH are shown. Error bars represent SD of three independent measurements. Amino acid abbreviations: A, Ala; F, Phe; G, Gly; L, Leu; M, Met; N, Asn; Q, Gln; R, Arg; T, Thr; Y, Tyr.

A second, lower-occupancy, binding site for GSSG is positioned ~5 Å from the primary site toward the cytoplasmic surface. The relative positions of the two GSSG sites are suggestive of a potential binding mode for a [2Fe:2S](GS)4 complex (35), although the actual physiological relevance of this site is not known. The binding site involves conserved arginine residues 206 and 323, along with the more variable Arg210 (Figs. 1B and 3B). Residue Thr324 is the counterpart of residue Glu433 in human ABCB7 that is mutated to Lys in X-linked sideroblastic anemia (7). Two additional pairs of Arg residues are positioned in the translocation pathway toward the periplasmic side [the conserved Arg91 and the more variable Arg313 (fig. S9)] that perhaps represent a transient binding site for ligands in the outward-facing conformation.

The functional importance of these interactions was assessed by evaluating the consequences of site-directed mutagenesis and substrate variation on the ATPase activity. Residues Asn269, Gln272, and Gly319 that form hydrogen bonds to the α-amino and carboxylate groups of the γ-Glu are highly conserved, and substitution of these residues alters the ATPase activity (Fig. 3, C and D). Although mutations of these residues (Asn269 → Ala, Gln272 → Ala, Gly319 → Leu) result in varying degrees of ATPase activity, they all effectively minimize GSH stimulation. The Asn269 → Ala substitution both enhances the basal ATPase activity and eliminates GSH sensitivity, whereas incorporation of a bulky residue at position 319 (Gly → Leu) or Gln272 → Ala mutation substantially reduces both the GSH sensitivity and the basal ATPase activity. The minimal ATPase activity of the Gln272 → Ala variant, together with the ability of the free amino acids Gly and Met to stimulate ATPase activity as effectively as GSH, suggests that the binding of ligands with α-amino and carboxyl groups to the Gln272 site in TM5 can trigger the conformational changes associated with ATP hydrolysis (Fig. 3C and fig. S10). As a potential coupling mechanism, Gln272 is adjacent to the side chain of Tyr195 in TM4 that forms a hydrogen bond with the backbone carbonyl of Leu315 in TM6, thereby linking TM4 and TM5 to TM6 and the following NBD. Substitution of these residues yields interesting phenotypes (Fig. 3D); the Tyr195 → Phe variant exhibits substrate-insensitive ATPase hyperactivity, which suggests that the native Tyr195 contact to TM6 may stabilize the inward-facing conformation. Conversely, because the Gln272 → Ala variant is essentially ATPase-inactive, interactions of this residue with the binding of an amino acid motif may be required to unlock the protein from the cytoplasm-facing conformation (Fig. 3C). The double mutant Tyr195 → Phe/Gln272 → Ala, however, has a high basal ATPase activity without GSH stimulation, indicating that when neither residue can potentially interact with TM6, the transporter can facilely undergo the conformational changes associated with ATPase activity.

The interconversion between inward- and outward-facing conformations of ABC exporters coupled to ATP hydrolysis is generally described within the framework of the alternating access transport mechanism (36). However, an analysis of MsbA structures solved in multiple conformational states emphasized that this interconversion cannot be described in terms of rigid body movements of subunits, but rather rearrangements that occur at the interfaces between certain TM helices (27). More specifically, the alternating conformations of ABC exporters may be distinguished by the separation of either TM1 and TM2 (outward) or TM4 and TM5 (inward) from the remaining four TM helices within the same subunit (22, 27). Examination of the available structures of ABC exporters identifies four elements that are generally conserved structurally: TM1–2, TM4–5, and TM3&6 (Fig. 4A), together with the NBD that is the hallmark of this family. These pairs of TM helices reflect the approximate symmetry in the transmembrane domain (TMD) noted in the Sav1866 structure (22), where TM1–3 and TM4–6 are related by a two-fold axis between TM3 and TM6 perpendicular to the molecular two-fold axis. The conformational changes associated with transport involve rearrangements in the relative orientations of these conserved elements (Fig. 4B), mediated by their interactions with the ligand and ultimately coupled to the NBDs through intracellular loop 1 (ICL1) between TM1 and TM2, intracellular loop 2 (ICL2) between TM4 and TM5, and the covalent connection from TM6 to the NBDs (Fig. 1). ABC exporters can be considered to have an articulated construction; as this work demonstrates, transported ligands span these structurally conserved elements in the TMD to create adaptable interfaces accommodating conformational rearrangements during the transport cycle. Mutations in residues implicated in mediating such contacts have been identified in cystic fibrosis transmembrane conductance regulator (CFTR) variants, including residues Arg347 and Asp993 (37, 38) in TM6 and TM9, respectively. These residues spatially correspond to positions 316 and 157 in TM6 and TM3 on different subunits of NaAtm1 [based on the TM3&6 structural alignment of NaAtm1 with the Sav1866-based homology model of CFTR (39)], near the GSH binding site. This suggests that the interactions observed here are of broader relevance for influencing the conformational equilibria essential for proper functioning of ABC exporters.

Fig. 4 Structural conservation and the articulated construction of ABC exporters.

(A) Structural conservation of the individual TM1–2, TM3&6, and TM4–5 elements in the TMDs of ABC exporter structures. The NaAtm1 structure was used for the reference, with the TM1–2 and TM4–5 superpositions horizontally displaced from TM3&6 by 20 Å to the left and right, respectively. The GSSG ligand in the primary binding site is depicted by a space-filling model with the carbons colored green. The underlying symmetry in the TMD, first noted in the Sav1866 structure (22), relates TM1–3 and TM4–6 by a two-fold rotation axis passing between TM3 and TM6 and normal to the plane of the page. (B) Variations in the relative orientations of conserved TMD elements observed in the structures of ABC exporters. The TMDs of different exporters were aligned with NaAtm1 using only TM3&6 for the superposition; for clarity, only TM3&6 from NaAtm1, depicted with black ribbons, is shown. Although the individual TM1–2 and TM4–5 elements are structurally conserved as shown in (A), substantial changes in their relative positions are evident between these structures (especially TM4–5), reflecting tertiary structure changes associated with the transition between inward- and outward-facing conformations. The binding site for the GSSG ligand spans between these elements, serving to couple ligand binding to changes in TMD and transporter conformation. The Cα traces are colored purple [Sav1866; PDB 2HYD (22)], blue [TM287/288; PDB 3QF4 (24)], black (NaAtm1), green [ABCB10, PDB 4AYT (26)], yellow [mouse Pgp, PDB 3G5U (23)], orange [C. elegans Pgp, PDB 4F4C (25)], and red [MsbA, PDB 3B5W (27)], ordered from outward- to inward-facing conformations.

Supplementary Materials

Materials and Methods

Figures S1 to S10

Tables S1 and S2

References (4054)

References and Notes

  1. See supplementary materials on Science Online.
  2. Acknowledgments: Supported by NIH grant GM45162. We gratefully acknowledge the Gordon and Betty Moore Foundation, the Beckman Institute, and the Sanofi-Aventis Bioengineering Research Program at Caltech for their generous support of the Molecular Observatory at Caltech, and the staff at Beamline 12-2, Stanford Synchrotron Radiation Lightsource (SSRL), for their assistance with data collection. SSRL is operated for the U.S. Department of Energy and supported by its Office of Basic Energy Research and by National Institute of General Medical Sciences grant P41GM103393 and National Center for Research Resources grant P41RR001209. We thank G. Meloni, J. Kaiser, J. Hoy, R. Lill, S. Molik, I. Booth, S. Conway, and A. Laganowsky for stimulating discussions and providing reagents. Coordinates and structure factors have been deposited in the Protein Data Bank of the Research Collaboratory for Structural Bioinformatics, with IDs 4MRN, 4MRP, 4MRR, 4MRS, and 4MRV for NaAtm1 in apo form and in complexes with GSH, selenomethionine, GSSG, and GS-Hg, respectively.

Stay Connected to Science

Navigate This Article