Crystal structure of the YgfY from Escherichia coli, a protein that may be involved in transcriptional regulation

Introduction.

The ygfY gene from Escherichia coli encodes an 88 amino acid protein of unknown function, and is a member of a large sequence family present in both, prokaryotes and eukaryotes. A Psi-Blast search of the nonredundant and environmental nonredundant databases revealed 249 family members.1 Of these sequences, 152 are from environmental samples, mostly from a Sargasso Sea sample filtered to include only microbial cells.2 The first cycles of Psi-Blast iteration yielded 78 homologous bacterial proteins, including 42 from the Sargasso Sea. Plant and fungal proteins emerged in the second cycle, and insect and mammalian sequences emerged in the third cycle. While the bacterial proteins have approximately the same size as YgfY (except for incomplete sequences from environmental samples), the eukaryotic proteins are at least double in size. None of the sequence homologues has known biochemical function. The cellular role is also unknown, except for a remote relationship (E-score = 10⁻⁴) to a protein from Saccharomyces cerevisiae, YOL071W/EM15 (GI:6324501), which is required for sporulation and for transcriptional induction of the early meiotic-specific transcription factor IME1.3 This 162-amino-acid residue yeast protein exhibits 25% identity to YgfY over a stretch of 60 amino acids.

The crystal structure of YgfY was determined at 1.2 Å resolution as a part of our structural genomics project (http://s2f.umbi.umd.edu), revealing a five-helix fold similar to that of the homologous protein NMA1147 from Neisseria meningitidis (30% identity) determined recently by NMR methods.4 Yet, there are inconsistencies in the details of these two structures that reflect the different levels of accuracy of the two methods of structure determination. We propose that the functional region is different from the one proposed based on the NMR structure, and highlight a particularly important structural difference associated with the proposed activity center.

Materials and Methods.

YgfY from E. coli was amplified using PfuTurbo DNA polymerase (Stratagene), genomic DNA, and 5′- and 3′-end primers. In addition to the native gene sequence, a sequence consistent with a thrombin cleavage site was introduced, and a NdeI restriction site was designed to convert the 6xHis-tagged construct into a construct coding for native protein. The PCR product was introduced into the pET100/D-TOPO expression vector by TOPO directional cloning procedure (Invitrogen). For wild-type protein production, the E. coli strain BL21 Star (DE3) was transformed with the recombinant plasmid. An expression screen showed that the His-tagged protein was soluble, whereas the native protein was insoluble. Wild-type protein was produced by growing the cells at 37°C in Super Broth medium supplemented by ampicillin (100 μg/mL). Once the cell culture reached A₆₀₀ = 0.6, expression was induced by the addition of 1 mM IPTG, and after 3 h the cells were harvested by centrifugation. To prepare selenomethionine (SeMet) containing protein, the E. coli B834 (DE3) strain was transformed with the recombinant vector. The cells were grown at 30°C in minimal medium supplemented with ampicillin (50 μg/mL), SeMet, and 19 amino acids other than methionine. When the cell culture reached A₆₀₀ = 0.5, 1 mM IPTG was added, and after 3 h the cells were harvested.

Cells were suspended in 20 mM Tris HCl (pH 8.0), 0.5 M NaCl, and 5 mM imidazole, and lysed by passage through a French press. The soluble fraction was loaded on Ni-NTA metal affinity column (Qiagen). Protein was eluted with 20 mM Tris HCl (pH 8.0), 0.5 M NaCl, and 250 mM imidazole. To remove the N-terminal sequence containing the 6xHis tag, human thrombin (Haemtech) was added at 1:2000 molar ratio and incubated overnight at 4°C in Tris HCl (pH 8.0), and 500 mM NaCl. Thrombin was removed by passing the protein mixture through benzamidine column (Amersham Biosciences). The cleaved and uncleaved protein and the N-terminal peptide were separated on a second Ni-NTA column. The protein was dialyzed against a buffer of 50 mM NaCl and 20 mM Tris HCl (pH 7.5), and further purified with a size-exclusion chromatography column. Protein expression, thrombin cleavage, and sample purity were assessed by SDS-PAGE, and the molecular weight was determined by mass spectrometry. The oligomeric state of the protein was measured by analytical size-exclusion chromatography using a G75 Superdex column (Amersham Pharmacia), and by dynamic light scattering using a Dyna Pro instrument (Protein Solutions).

Crystals were obtained by the vapor diffusion method in hanging drops at room temperature. Both SeMet-containing and wild-type protein crystals were formed in 1.1 M sodium malonate solution buffered with 0.1 M Tris HCl (pH 8.0). Crystals belonging to space group P2₁2₁2 with two molecules in the asymmetric unit were obtained for both the SeMet and wild-type proteins (Table I), and were used for structure determination. For the wild-type protein, crystals were also obtained in space group I222 with one molecule in the asymmetric unit.

For diffraction data collection, crystals were flash-cooled at 100 K in the crystallization solution. Multiple wavelength diffraction (MAD) data of selenium- and gold-containing crystals (Table I) were collected on the IMCA-CAT 17-ID beamline at the Advanced Photon Source (Argonne National Laboratory, Argonne, Illinois). The beamline was equipped with an ADSC Quantum 210 Charge-Coupled Device (CCD) detector. The home facility was used to characterize crystals and to acquire the 2.0 Å diffraction data of the wild-type protein I222 crystal. X-rays were supplied by a Siemens rotating anode which was equipped with MAR345 image plate. Ultra high resolution data at 1.2 Å for a wild-type protein crystal were collected at IMCA-CAT. Data processing was performed using the computer programs HKL5 and CrystalClear (Rigaku MSC).

The structure of YgfY was determined by combining selenium and gold MAD data from the P2₁2₁2 crystal form. The Se sites were identified using the computer program SHELXD,6, 7 but the phase quality was inadequate for map interpretation. Thus, a wild-type YgfY crystal was soaked with 1 mM KAu(CN)₂ for three days, and a two-wavelength diffraction data set was collected (Table I). Two gold sites were identified in a difference Fourier map calculated with phases derived from the Se data. Phase combination was carried out using the MLPHARE program8 as implemented in CCP4.9 The phases were improved by solvent flattening and noncrystallographic symmetry averaging using RESOLVE.10 The resulting electron density map at 2.0 Å resolution was traced using the graphics program O.11

The wild-type structure was refined at 1.2 Å resolution, initially with the CNS program,12 and then with the program REFMAC.13 Unisotropic temperature factors were introduced towards the end of the refinement, and hydrogen atoms were included in the last cycle (the addition of hydrogen atoms reduced R_free by 1%). For the I222 crystal form, the structure was determined at 2.0 Å resolution by Molecular Replacement, using BEAST,14 and then refined using CNS.

Structure analysis was carried out using the computer programs: Procheck15 for analysis of geometry, and Molscript16 and Raster3D17, 18 for depiction of structure.

Results and Discussion.

Cleavage of the His-tag left three amino acids residues N-terminal to the authentic polypeptide (Gly-Ser-His). Of these, only the main chain of the histidine residue is seen in the electron density map. In the following discussion, we use a numbering scheme based on the authentic sequence with the first residue being Met1. The structure is well ordered, and all residues of the authentic native sequence are included in the model. For the 1.2-Å resolution structure, three side chains in one molecule and four side chains in the second molecule of the asymmetric unit were modeled with alternate conformations.

The structure of YgfY was determined at ultra high resolution (1.2 Å) in the P2₁2₁2 crystal form, and at moderately high resolution (2.0 Å) in the I222 crystal form (Table II). The packing of molecules in the two crystal forms is very similar and, since both were obtained under the same crystallization condition, it is unclear whether the change of space group was due to slight variation in the growth conditions or due to perturbations during flash-cooling. The molecules pack into dimers in both crystal forms, which differ only by a rotation of two degrees between the two molecules. The structure is essentially the same within the limit of the resolution. Exchange between P2₁2₁2 and I222 space groups was reported previously.19, 20

In contrast to the dimeric packing in the crystals, dynamic light scattering measurements and analytical size-exclusion chromatography indicated that the protein exists in solution primarily in a monomeric form. Close inspection of the dimer interface revealed that despite the suggestive packing, the surfaces do not match well and a layer of solvent molecules fills the gap. Moreover, the amino acid residues at the interface are not conserved in the YgfY sequence family. Thus, we propose that the biological unit is a monomer.

YgfY folds into a compact, five-α-helical bundle [Fig. 1(A)], similar to the fold of the sequence family member, NMA1147 from Neisseria meningitides, determined by NMR.4 The closest SCOP fold classification to YgfY fold is of the cyclin-like domains, characterized by five α-helices with one α-helix (helix 3) surrounded by the other four,21, 22 helices 1–3 forming an antiparallel up-down cluster, and helices 4–5 running orthogonally to one another. In addition to the cyclin family, this super family also includes the transcription factor IIB (TFIIB) core domain and retinoblastoma tumor suppressor domains. These are all protein binding domains. Interestingly, helix 3 of YgfY and NMA1147 is exposed to solvent because the ensuing loop and helix 4 are positioned above the C-terminus of helix 3. As discussed below, we propose that the exposed surface of helix 3 is involved in the active center of the molecule.

The RMS deviation between super-positioned YgfY and NMA1147 α-carbon atom pairs is 2.2 Å. Although the two structures share the same overall fold, there are many differences in details. These include inefficient packing in the NMR structures that results in holes or pockets that are not observed in the X-ray structure, different side chain interactions including those between core residues, and a key conserved residue that forms a strikingly repulsive electrostatic arrangement in NMA1147 (Asp54 in NMA1147, Asp51 in YgfY), whereas in YgfY it forms a perfect ion pair with a conserved arginine (Arg18 in NMA1147, Arg15 in YgfY). We believe some of these discrepancies are due to errors in the NMR structure determination that may have arisen because of the relatively low number of inter-residue constraints (10 per amino acid residue).

A pair of residues, conserved in the sequence family, forms an ion pair, Arg15 and Asp51 [Fig. 1(B)]. The only exception is a sequence from Arabidopsis thaliana, where Asp51 is replaced by an asparagine. The same residues are conserved in the sequence of NMA1147 (Arg18 and Asp54). However, in the NMR structure, the Asp54 side-chain is buried and does not form electrostatic interactions to compensate for the charge burial [Fig. 1(C)]. The Arg15-Asp51 salt bridge is located at the center of a surface spanning the contact region of helices 1 and 3, the loop between helices 1 and 2, and the loop between helices 3 and 4. In addition to the residue involved in the salt bridge, several other residues tend to be conserved or conservatively replaced in the sequence family, including Met1 (an initial methionine is always conserved but in YgfY it is well defined and involved is specific interactions), Arg8, Trp11, Gly16, and Met17. At the edge of this region, Glu19, Asp21 are conserved and Phe55 is on the surface and conservatively replaced. In addition to forming a salt bridge, the precise positioning of Arg15 and Asp51 is assured by the interaction of Asp21 carboxyl group with the backbone amide of Arg15, by the interaction of the guanidinium group with the backbone carbonyl groups of residues 47 and 49, and by the carboxyl group of Asp51 interaction with the backbone amide of the invariant Gly16. This exquisite arrangement maintains the structural integrity of the loops that connect helices 1 and 2, and helices 2 and 4. The residue conservation and structure pattern support the notion that this region is likely to be important for function.

A region juxtaposed to the above surface, on the other side of Glu19 and Phe55, comprises a shallow depression that is equally balanced with polar and hydrophobic residues. The residues associated with the depression are not conserved in the sequence family. In contrast, a deep pocket lined with hydrophobic residues is present in the structure of NMA1147. When the X-ray and NMR structures are compared, it emerges that the large pocket in the NMR structure is a consequence of inefficient packing that leads to the exposure of core residues and to a deep solvent-accessible pocket. Whereas Liu and colleagues implicated this region in function,4 the X-ray structure does not support this conclusion.

Although the YgfY sequence family contains three conserved residues with functional groups that potentially could be involved in enzymatic reaction, their disposition does not resemble an arrangement typical of enzyme active sites. Moreover, the relatively flat nature of the associated surface suggests that it is more likely to be involved in protein–protein interaction. The structural analogy to the cyclin-like domains involved in protein–protein interactions supports this hypothesis, because often (although not always) a common fold indicates common general function. Yet, the interaction regions of the cyclins and TFIIB do not map to our proposed site of action in the YgfY family. Nevertheless, because the yeast protein homologue was shown to be involved in transcriptional induction of the early meiotic-specific transcription factor IME1,3 it is tempting to speculate that TFIIB is the true functional analogue of YgfY, and YgfY family members are also associated with complexes that regulate transcription.

Acknowledgements

We thank John Moult and Eugene Melamud for the use and help with their bioinformatics web site (http://s2f.carb.umbi.umd.edu) and the Structural Genomics team at CARB for stimulating discussions. We thank the staff of IMCA-CAT at the Advanced Photon Source for their help during data collection. The IMCA-CAT facility is supported by the companies of the Industrial Macromolecular Crystallographic Association, through a contract with IIT. Use of the Advanced Photon Source was supported by the U.S. Department of Energy, Basic Energy Sciences, Office of Science, under contract W-31-109-Eng-38. The Keck foundation provided generous support for the purchase of X-ray equipment at CARB. The PDB coordinates entry codes of ygfY are 1X6I and 1X6J.