Supplementary Figure 1 Legend
Ribbon diagram of the CorA soluble domain. α-helices (red) and β-sheets (yellow) are numbered according to the manuscript text. Intervening loops are coloured green. A single Mg2+ ion bound to Asp89 is indicated (purple sphere) with its associated hydration shell (small red spheres). This was modelled as magnesium for several reasons. First, the putative ion was hexacoordinated to oxygen atoms (Asp89 and five water molecules), at bond lengths of 1.95-2.0 Å and bond angles of about 90° (Figure 5a in paper). Second, the crystal was grown in the presence of 200 mM magnesium chloride, the only divalent cation present. Third, crystals only formed in the presence of magnesium. Full-length crystals also formed only in the presence of reservoir solutions containing at least 200 mM Mg2+.
Supplementary Figure 2 Legend
Close-up view of the CorA periplasmic surface. The highly conserved CorA “YGMNF” signature sequence, located at the periplasmic surface, is highlighted on two adjacent subunits (shown as yellow and green). The structure has been rotated ~90º from the viewing position in Figure 1a of the paper, towards the viewer. Residues are depicted as stick models.
Supplementary Figure 3 Legend
Secondary structure predictions of CorA homologues. The protein secondary structures of the T. maritima CorA (GI15643327), S. Typhimurium CorA (GI60391925), S. Typhimurium ZntB (GI11878225), A. thaliana AtMgt1/Mrs2-1 (GI18412911), A. thaliana AtMgt10/Mrs2-10 (GI25360983), and Homo sapiens Mrs2 (GI10190702) were predicted using the implementation of PROF (Rost et al, 1996) on the PredictProtein server ( Rost et al, 2004). The prediction for the T. maritima CorA was virtually identical to that determined by crystallography. For all proteins, the sequences were initially aligned to the C-terminus of the T. maritima and S. Typhimurium CorAs. Then, -helical domains (pink) and -sheets (blue) were placed on the sequences as predicted by PROF. For convenience in presentation of the 3 eukaryotic sequences, additional N-terminal sequence that had no predicted structure using PROF was omitted, as indicated by the open end of the bar and from the numbering under each sequence. The human Mrs2 protein has additional sequence at the C-terminus past the concentration of basic amino acids following TM2 that was also omitted.
Supplementary Figure 4 Legend
There are over 250 CorA homologs and around 200 Mrs2 homologs listed in the current NCBI and Swiss-Prot databases. Homology between proteins of this superfamily is strongest at the C-terminus and becomes progressively weaker moving C- to N-terminal. The C-terminal regions of several CorA homologs from the “MPEL” subfamily of the Eubacteria and Archaea (Kehres, Lawyer and Maguire (1998) and Knoop et al. (2005)) and selected Mrs2 mitochondrial eukaryotic homologs were aligned. All sequences shown end at the protein’s C-terminus except those followed by dashes. These latter sequences have variable length C-terminal extensions. The accession numbers and identity of the sequences shown are Mus: NP 001013407 Mus musculus Mrs2: Hs: CAI17108 Homo sapiens, Mrs2-like from chromosome 6; At1: AAN73211 Arabidopsis thaliana Mrs2-1; At10: AAN73219 Arabidopsis thaliana Mrs2-10 (AtMGT1); Sc: NP 014979 Saccharomyces cerevisiae Mrs2p; Tm: NP 228371 Thermotoga maritima CorA; Ph: NP 142150.1 Pyrococcus horikoshii CorA; Af: NP 069620.1 Archaeoglobus fulgidus DSM 4304 CorA; Mm: NP 634664.1 Methanosarcina mazei GOE1 CorA; Hp: NP 208136.1 Helicobacter pylori 26695 CorA; Wg: NP 871276.1 Wigglesworthia glossinidia (endosymbiont of Glossina brevipalpis) CorA; St: P0A2R8 Salmonella enterica serovar Typhimurium CorA; Mt: NP 335721.1 Mycobacterium tuberculosis CDC1551 CorA; Bb: CAE32308.1 Bordetella bronchiseptica CorA; Cc: NP 421975.1 Caulobacter crescentus CB15 CorA; Bc: NP 832870.1 Bacillus cereus ATCC 14579 CorA; Sc: CAC36370 Streptomyces coelicolor A3(2) CorA. Based on the crystal structure of CorA presented herein, the C-terminal portion of CorA/Mrs2 proteins is composed of an amphipathic -helix, followed by TM1, followed by a predicted “Mg2+-binding loop”, followed by TM2, followed by a “Lysine Ring” (see figure). Within these domains, several additional features can be seen. 1) Immediately preceding TM1 is an amphipathic -helix in which charged or polar residues tend to align every 1-2 turns of the helix. These correspond to the polar rings that line the interior face of the “funnel” of the soluble domain (see text, denoted by A in figure). 2) Within TM1, there is a highly conserved T residue towards the N-terminal end (B) and a (P/L)(P/L)T sequence in the middle of TM1 (C). 3) At the C-terminal end of TM1, the signature (Y/F)GMNF sequence is evident (D). 4) In the Mg2+ binding loop, several negatively charged residues are apparent in all sequences which includes an X1xMPELX2 sequence in the prokaryotic CorA homologs (E), where “X” is a charged residue. Notice however, that X1 and X2 are always of opposite charge (arrows). 5) TM2 characteristically begins with an aromatic residue and contains, near its center, a conserved LxxM sequence (F). 6) A lysine-rich sequence immediately follows TM2 in all homologs (G). Further, in those homologs with C-terminal sequence extensions, the extension tends to have additional lysine and arginine residues (not shown). In the prokaryotic CorA homologs, the lysine-rich sequence tends to be followed by “WL”.
Supplementary Figure 5 Legend
View of the electrostatic potential of the CorA surface. Regions of positive charge are mapped as blue on the surface, whereas negatively charged regions are coloured red. The structure has been rotated ~45º clockwise and ~20º towards the viewer from that depicted in Figure 1a. Electrostatic potential was determined using the program GRASP (Petrey & Honig, 2003).