STRUCTURE NOTE X-ray Crystal Structure of MTH938 From Methanobacterium thermoautotrophicum at 2.2 Å Resolution Reveals a Novel Tertiary Protein Fold Kalyan Das, 1,2 Rong Xiao, 1,3 Elisabet Wahlberg, 1 Fred Hsu, 4 Cheryl H. Arrowsmith, 4 Gaetano T. Montelione 1,3 * and Edward Arnold 1,2 * 1 Center for Advanced Biotechnology and Medicine, Piscataway, New Jersey 2 Department of Chemistry, Rutgers University, Piscataway, New Jersey 3 Department of Molecular Biology and Biochemistry, Rutgers University, Piscataway, New Jersey 4 Ontario Cancer Institute and Department of Medical Biophysics, University of Toronto, Toronto, Ontario, Canada Introduction. We have determined the crystal structure of MTH938 (Fig. 1), a hypothetical protein encoded by the Methanobacterium thermoautotrophicum (Mthe) genome (DNA bases 843,263– 862,747), 1 at 2.2 Å resolution by Se-Met multiwavelength anomalous diffraction (MAD) techniques. Se-Met labeled MTH938 crystallized with the symmetry of space group P4 1 2 1 2 with one dimer per asymmetric unit. The dimensions of each monomer of 111 amino acid residues are about 26  30  32 Å 3 . A Dali search 2 with this MTH938 structure found no significant structural similarity (highest Z-score of 2.7) with any existing protein. The crystal structure of MTH938 reveals a new tertiary fold consisting of three -sheets and three -helices (Fig. 1). There is a disulfide bond between residues Cys 5 and Cys 87 in each monomer. As Mthe is an anaerobic archaea and the cystine pair is not conserved in the amino acid sequence alignment (Fig. 1), the potential structural and functional significance of the disulfide bond is uncertain. It is interesting that the only eukaryotic homolog in this sequence cluster is an unnamed human protein, suggesting possible lateral gene transfer into the human genome. Two larger sheets, one from each monomer, associate as a ten-strand mixed -sheet [Fig. 2(a)] that forms the base of a cleft [Fig. 2(b)]. Molecular modeling and electrostatic potential calculations 3 suggest that this cleft could poten- tially bind double-stranded nucleic acid with interacting elements from A and the tip of 5 of either subunit of the MTH938 dimer. The dimer interface surface area of 262 Å 2 , however, corresponds to only about 5.5% of the surface area of a monomer. Dynamic light scattering and gel filtration chromatography also indicate that MTH938 is monomeric in solution. Further biochemical and struc- tural investigations on this protein are in progress. Methods. DNA from M. thermoautotrophicum, bases 843,263 to 862,747, section 74 of 148, was cloned into expression vector pET15b and transformed into Esche- richia coli BL21-DE3 cells. The selenomethionine deriva- tive of MTH938 was prepared following a published proto- col. 4 Purified Se-Met labeled MTH938 containing a 10 amino acid N-terminal linker with a hexa-His tag was concentrated to about 10 mg/ml in 20 mM Tris-HCl, pH 8.0, with 100 mM NaCl and 5 mM -mercaptoethanol. Crystals grown in hanging drops containing 20% PEG 3350, 0.2 M ammonium chloride, and 0.1 M sodium cacodylate at pH 6.2 were used for X-ray diffraction data collection. Diffraction intensity data (Table I) were col- lected using the Advanced Photon Source (APS) Beamline 14BM-D, Argonne National Laboratory, from a single frozen crystal (100 K) at three wavelengths. The wave- lengths selected were the peak (1) and inflection (2) of the Se K-edge, and at a higher energy remote wavelength (3). The data were processed and scaled to 2.2 Å resolu- tion using Denzo and Scalepack, 5 respectively. The sum- mary of X-ray data statistics is listed in Table I. Four Se sites, corresponding to two molecules per asymmetric unit, were located using direct methods as implemented in SnB 2.1 6 and MAD phases were calculated to 2.7 Å resolution based on the anomalous signal from the Se sites using SOLVE version 1.18 7 with a figure of merit (FOM) of 0.69. The MAD phases were further improved and extended to 2.2 Å resolution using RESOLVE version 1.04 7 and ARP V5.1. 8 The model was built manually into electron density maps calculated using phases obtained from these proce- dures. Cycles of model building, using O version 6.1 9 followed by least squares refinement using CNS 10 with bulk solvent correction, yielded the final structure that Grant sponsor: NIH; Grant number: P50 GM62413; Grant sponsor: New Jersey Commission on Science and Technology; Grant sponsor: Ontario Research and Development Challenge Fund. *Correspondence to: Prof. Gaetano Montelione, CABM—Rutgers University, 679 Hoes Lane, Piscataway, NJ 08854. E-mail: guy@cabm.rutgers.edu; or Prof. Edward Arnold, CABM—Rutgers University, 679 Hoes Lane, Piscataway, NJ 08854. E-mail: arnold@cabm.rutgers.edu. Received 4 June 2001; Accepted 6 July 2001 PROTEINS: Structure, Function, and Genetics 45:486 – 488 (2001) © 2001 WILEY-LISS, INC.