Crystal Structure of the CorA Mg2+ Transporter
Vladimir V. Lunin1*, Elena Dobrovetsky2*, Galina Khutoreskaya2, Rongguang Zhang3, Andrzej
Joachimiak3, Declan A. Doyle4, Alexey Bochkarev2,5,6, Michael E. Maguire7, Aled M.
Edwards1,2,3,5,6, Christopher M. Koth2,8
1
Department of Medical Biophysics, University of Toronto, 112 College Street, Toronto, ON,
Canada M5G 1L6
2
Banting and Best Department of Medical Research, University of Toronto, 112 College Street,
Toronto, ON, Canada M5G 1L6
3
Structural Biology Center & Midwest Center for Structural Genomics, Biosciences Division,
Argonne National Laboratory, 9700 S. Cass Av. Argonne, IL 60439
4
Structural Genomics Consortium Botnar Research Centre, Oxford, Oxon OX3 7LD, United
Kingdom
5
Department of Medical Genetics and Microbiology, University of Toronto, 112 College Street,
Toronto, ON, Canada M5G 1L6
6
Structural Genomics Consortium, Banting Institute, 100 College Street, Toronto, ON, Canada
M5G 1L5
7
Department of Pharmacology, Case Western Reserve University, Cleveland, OH 44106-4965
*
These authors contributed equally to this work.
8
Present address: Vertex Pharmaceuticals Incorporated, 130 Waverly St., Cambridge, MA 02139 USA
Supplementary Information
Methods
Expression and purification of full-length and soluble domain of T. maritima CorA. The
full coding sequence of T. maritima CorA was amplified by PCR from genomic DNA using high
fidelity Pfu polymerase (Stratagene). The restriction site NdeI was added to the 5’ end of the
PCR product. The restriction site BamHI was added to the 3' end of the PCR product. The
digested PCR product was inserted directionally between the corresponding sites of a modified
pET15b (Novagen) T7 polymerase expression vector in which the thrombin cleavage site
(LVPR^GS) had been replaced with a TEV protease recognition site (ENLYFQ^G), generating
plasmid pET-TMCORA. The recombinant protein was expressed as a fusion to an N-terminal
six-histidine tag and a TEV protease site. Nine other prokaryotic CorA homologues were
similarly cloned, tested for expression and their purification attempted (data not shown). A
construct for expression of the soluble domain of CorA (residues 1-266) was generated by
inserting a stop codon (TGA) in plasmid pET-TMCORA, immediately after the sequence coding
for residue E266, using QuickChange mutagenesis (Stratagene).
E. coli cells containing the T. maritima CorA expression plasmids described above were
grown in 12 L Luria Broth at 37ºC to an OD600 of 0.6-0.8. IPTG was then added to a final
concentration of 1.0 mM. The cells were grown for an additional 12 hours at 16ºC, harvested by
centrifugation and frozen in liquid nitrogen. The frozen pellet was resuspended in 250 mL icecold Lysis Buffer (50 mM HEPES, pH 7.5, 500 mM NaCl, 5% glycerol, protease inhibitors
(Complete Protease Inhibitor Cocktail (Roche), according to the technical manual)) and lysed by
French press. All subsequent procedures were carried out at 4ºC unless otherwise specified. The
suspension was centrifuged for 30 minutes at 100 000 x g. The pellets were then solubilized in
250 ml Binding Buffer (50 mM HEPES, pH 7.5, 500 mM NaCl, 5% glycerol, 10mM imidazole,
1% n-dodecyl--D-maltopyranoside (DDM, Anatrace), protease inhibitors) and stirred gently for
12 hours. The sample was then centrifuged for 30 minutes at 100 000 x g, and the supernatant
loaded onto a 1 x 10 cm Ni-NTA gravity column equilibrated with Binding Buffer. The column
was washed with 20 column volumes of Wash Buffer (50 mM Hepes, pH 7.5, 500 mM NaCl,
5% glycerol, 35mM imidazole, 0.02% DDM). Bound protein was eluted with Elution Buffer (50
mM HEPES, pH 7.5, 500 mM NaCl, 5% glycerol, 200mM imidazole, 0.02% DDM) and
dialyzed overnight against Dialysis Buffer (50 mM HEPES, pH 7.5, 500 mM NaCl, 5%
glycerol). During dialysis, removal of the hexahistidine tag was facilitated by the addition of
histidine-tagged TEV protease (Invitrogen), according to the TEV protease technical manual
(Invitrogen). Digestions were monitored by SDS-polyacrylamide gel electrophoresis. The
resulting proteins contained three additional residues at the N terminus (Gly-Ser-His). TEV
protease and the histidine tag were separated from CorA by collecting the flow-through from a
second Ni-NTA column purification, as described earlier.
Expression of selenomethionine-labelled protein was essentially as described (Korolev et
al, 2002). The level of selenomethionine incorporation was regulated by the addition of
unlabelled methionine to the growth media and monitored by mass spectrometry (data not
shown).
Crystallization of T. maritima CorA. For crystallization, protein solutions were used
immediately after purification or stored at –78°C. Crystals were grown by the hanging drop
method at 22°C, at a concentration of 2-4 mg/ml. Initial ‘hits’ were obtained using the Nextal
MB Class I and II screens (Nextal). From these, initial conditions, the reservoir solutions were
optimized. In the case of full-length T.
aritime CorA, 2 µl protein was mixed with 2 µl
reservoir solution containing 20% (w/v) PEG 2000 (Fluka), 0.3M Mg(NO3)2 and 0.1M Tris pH
8.0. Needle-like crystals appeared after 3-5 days and matured to full size within 2-3 weeks. In
the case of the CorA soluble domain, 2 µl protein was mixed with 2 µl reservoir solution
containing 35% (w/v) PEG 3350 (Fluka), 0.2M MgCl2 and 0.1M Tris pH 8.5. Diamond-shaped
crystals formed after two days and matured to full size within 1-2 weeks.
Structure determination. Native crystals of full-length T. maritima CorA, which formed only
in the presence of magnesium, diffracted X-rays to 3.9 Å. Despite extensive efforts, heavy-metal
derivatives of full-length T. maritima CorA suitable for phase determination could not be
obtained. Selenomethionine-containing crystals were grown and a dataset collected, but the
crystals formed only from the selenomethionine-containing protein preparations that had no or
poor incorporation (<20%) of selenomethionine. The low occupancy did not enable us to solve
the selenium sub-structure from these data. Phase information for the crystals of the full-length
transporter was eventually obtained by first determining the 1.85 Å structure of the aminoterminal soluble domain of CorA (residues 1-244, Figure S1) and using this structure as a search
model to solve the structure of the full-length protein by molecular replacement.
The structure of a selenomethionine-labelled soluble domain (amino acids 1-266) was
solved using single wavelength anomalous dispersion (SAD). Diffraction data were collected
using a home source generator (Rigaku FR-E with RAXIS-IV++ detector) and at beamline X25
of the National Synchrotron Light Source at Brookhaven National Labs, to 1.85 Å. Data were
processed using HKL2000 (Otwinowski and Minor, 2000). Three selenium sites were found
using the interface Bake’N’Phase (Weeks et al, 2002), and the obtained phases were improved
by solvent flattening using RESOLVE (Terwilliger, 2000). The software packages ARP/WARP
(Perrakis et al, 1999) and REFMAC5 (Murshudov et al, 1997) were used to complete the model
and refine it to a final R/Rfree of 0.196/0.233 (see Table 1 of the paper). The final model
contains residues 13-117, 120-199, 207-244, 236 water molecules and two metal ions, assigned
as Mg2+ and Na+. The Mg2+ ion is coordinated by Asp89 O1 and five water molecules (Figure
5a of paper). Also four molecules of detergent (DDM) were modelled in regions of long,
continuous density within the detergent environment.
Data for the full-length protein were collected at beamlines 17ID and 19ID at the
Advanced Photon Source, Argonne National Labs. The full-length model was built first by
fitting the known structure of the N-terminal domain and fitting its last helix (residues 208-244)
into electron density. Five positions for the soluble domain, arranged as a pentameric ring, were
found using the program PHASER (Storoni et al, 2001). The soluble domain core (residues 13199) fit well in the calculated 2Fo-Fc electron density maps, while helix 207-244 was out of the
density. After fitting that helix and re-calculating 2Fo-Fc maps, five long helices, initially as 60residue polyalanine templates, were placed according to newly resolved features. When this
model was refined and helical templates fitted into the density, five additional 30-residue
polyalanine templates were added to the model, as new features developed. Data collected from
selenomethionine-labelled full-length protein crystals (with ~20% incorporation of
selenomethionine) were used to calculate an anomalous Fourier map with phases taken from this
current model. A peak search in this map revealed 3 peaks per soluble domain, corresponding to
methionine sidechains. Five additional peaks per molecule were found; two inside the pore as
well as three peaks outside the pore. These peaks were used to place methionine residues and
thread the amino acid sequence onto the model. The assigned sequence corresponds well with
two kinks in the stalk helix at positions Gly274 and Pro303. Five-fold averaging alternated with
solvent flattening using PHASES (Furey and Swaminathan, 2001) was used to improve the
quality of electron density maps. The final model was refined to 3.9 Å.
The final full-length model contains residues 9-315 and 323-349. The loop from residues
316-322 remains disordered or invisible in electron density maps. After refinement using
REFMAC5 and CNS (Brunger et al, 1998), the R-factor was 0.361, and the Rfree 0.406. All
statistics are summarized in Table 1, below.
DALI Analysis. The fold of the soluble, intracellular domain of each subunit is unique in
comparison to all other transporters and ion channels of known structure. The DALI (Holm and
Sander, 1993) search performed for the soluble domain structure (residues 9-200) found a closest
structural homologue with a Z-score of 3.0, and r.m.s.d of 3.5 Å over 88 Ca atoms in 11
fragments, and could not be counted as a structural homologue.
The DALI search performed for the full-length protein structure provided a list of
structures with Z-scores up to 10, r.m.s.d.’s down to 2.5 Å over 108 Ca atoms (i.e. PDB ID
1UUR). These structures contained three consecutive -helices and could be thus superimposed
with the two willow helices and beginning of the stalk helix in CorA. No other structural
features of T. maritima CorA were identified.
Table 1. X-ray refinement statistics
Soluble domain
Full-length
Resolution (Å)
1.85
3.9
Rwork/Rfree
0.196/0.233
0.361/0.406
Protein
1915
13805
Ligand/ion
134
n.d.
Water
236
n.d.
Protein
36.3
163.3
Ligand/ion
66.8
n.d.
Water
46.3
n.d.
0.017
0.007
1.73
1.16
Number of atoms
B-factors
R.m.s deviations
Bond lengths (Å)
Bond angles ()
Supplementary References
Brunger A.T., Adams P.D., Clore G.M., Delono W.L., Gros P., Grosse-Kunstleve R.W., Jiang
J.-S., Kuszewski J., Nilges N., Pannu N.S., Read R.J., Rice L.M., Simonson T., and Warren G.L
"Crystallography and NMR system (CNS): A new software for macromolecular structure
determination." Acta Cryst. D54, 905-921 (1998).
Furey, W. and Swaminathan, S. PHASES-95: A Program Package for the Processing and
Analysis of Diffraction Data from Macromolecules. Methods in Enzymology, Volume 276:
Macromolecular Crystallography, vol 277, Part B, chapter 31, eds. C. Carter & R. Sweet,
Academic Press, Orlando, Fl. (1997).
Holm, L. and Sander, C. Protein structure comparison by alignment of distance matrices. J. Mol.
Biol. 233, 123–138 (1993).
Kehres, DG, Lawyer, C.H. and Maguire, M.E. The CorA magnesium transporter gene family.
Microbial & Comparative Genomics 43:151-169(1998).
Knoop, V, M Groth-Malonek, M Gebert, K Eifler and K Weyand. Transport of magnesium and
other divalent cations: evolution of the 2-TM-GxN proteins in the MIT superfamily" Mol
Genet.Genomics:1-12 (2005).
Korolev, S., Ikeguchi, Y., Skarina, T., Beasley, S., Arrowsmith, C.H., Edwards, A.M.,
Joachimiak, A., Pegg, A.E. and Savchenko, A. The crystal structure of spermidine synthase with
a multisubstrate adduct inhibitor. Nature Struct. Biol. 9: 27 - 31 (2002).
Murshudov, G.N., Vagin, A.A., and Dodson, E.J. Refinement of Macromolecular Structures by
the Maximum-Likelihood Method. Acta Crystallogr. D53, 240-255 (1997).
Otwinowski, Z. and Minor, W. Processing of X-ray Diffraction Data Collected in Oscillation
Mode. Methods in Enzymology, Volume 276: Macromolecular Crystallography, part A, p.307326, 1997, C.W.Carter,Jr.&R.M.Sweet, Eds.,Academic Press (New York).
Perrakis A, Morris R and Lamzin VS. Automated protein model building combined with
iterative structure refinement. Nature Struct. Biol. 6:458-463, (1999).
Petrey, D. & Honig, B. GRASP2: visualization, surface properties, and electrostatics of
macromolecular structures and sequences. Methods Enzymol. 374, 492-509 (2003).
Rost, B., Fariselli, P. & Casadio, R. Topology prediction for helical transmembrane proteins at
86% accuracy. Protein Sci. 5, 1704-1718 (1996).
Rost, B., Yachdav, G. & Liu, J. The PredictProtein server. Nucleic Acids Res. 32, W321-W326
(2004).
Terwilliger, T. C. Maximum likelihood density modification. Acta Cryst. D56, 965-972 (2000).
Storoni, L.C., McCoy, A.J., and Read, R.J. Likelihood-enhanced fast rotation functions. Pushing
the boundaries of molecular replacement with maximum likelihood. Acta Cryst. D60, 432-438
(2004).
Weeks, C.M., Blessing, R.H., Miller, R., Mungee, R., Potter, S.A., Rappleye, J., Smith, G.D.,
Xu, H. and Furey, W. Towards automated protein structure determination: BnP, the SnBPHASES interface, Z. Kristallogr. 217, 686-693 (2002).