Structure of sugar-bound LacY

Global Fold.

An X-ray crystal structure of LacY G46W/G262W cocrystallized in the presence of 5 mM β-d-galactopyranosyl-1-thio-β-d-galactopyranoside (TDG) was determined to 3.5-Å resolution (Fig. 1), and, importantly, crystals were not obtained in the absence of an added galactoside. Data collection and refinement statistics are summarized in SI Appendix, Table S1. The protein crystallized in a different crystal form and space group than other LacY structures, suggesting an overall tertiary structure different from the previous inward-facing conformations (11–14). Crystals of the G46W/G262W mutant are in space group C222₁ with cell dimensions a = 101.6 Å, b = 121.9 Å, and c = 264.0 Å, and the structure was determined by molecular replacement using the partial structure of LacY (13) as the search molecule. The protein occupies 29% of the cell volume.

Two independent molecules in the asymmetric unit that are adjacent to one another, as if in a bilayer, but have opposite-facing orientations (Fig. 1 and SI Appendix, Fig. S1). The Trp replacements on the periplasmic side are illustrated (SI Appendix, Fig. S2). Both molecules are in an almost occluded, outward-facing conformation with a single molecule of TDG in the central sugar-binding site. However, the two structures (termed “A” and “B”) differ slightly in the substrate orientation and in the arrangement of substrate-binding side chains, suggesting that they sample nearby states around the same part of the transport cycle. A space-filling view of molecule A from the periplasmic side (Fig. 2A) reveals bound TDG through an opening that is too narrow to allow entrance or exit (∼3 Å at the narrowest point) (45). In contrast, the cytoplasmic side of the molecule is tightly sealed with residues Glu64 (helix II), Thr348 (helix XI), Tyr350 (helix XI), Phe334 (helix X), Phe140 (helix V), Gly141 (helix V), and Pro280 (helix VIII) providing a zipper-like motif (Fig. 2B).

The overall rmsd between 380 Cαs of the mutant with bound TDG and WT LacY without any substrate bound [Protein Data Bank (PDB) ID code 2V8N] is 4.3 Å, consistent with very different states. Difference–distance analysis identifies two rigid bodies in these structures, one comprising most of the N-terminal helix bundle (168 residues) and the other comprising most of the C-terminal helix bundle (188 residues); the Cαs can be separately superimposed with rmsds of 1.2 Å. Thus, the mutant structure is related to WT apo LacY by an approximately 30° rotation of one rigid domain against the other about an axis perpendicular to the membrane plane. (Fig. 1 and SI Appendix, Fig. S3). Helix VI moves ∼12 Å closer to helix VII, and helix IV moves ∼12 Å closer to helix IX on the cytoplasmic side. The relative displacements of all helices are listed in SI Appendix, Tables S2 and S3.

Substrate-Binding Site.

The conformation of the high-affinity TDG molecule is clearly defined in the almost occluded central cavity (Figs. 1 and 3) and allows assignment of likely hydrogen-bond interactions with the protein (SI Appendix, Table S4), although the errors in interatomic distances are high at 3.5-Å resolution. Electron density for the sugar is seen in an omit map after refinement without TDG (SI Appendix, Fig. S4). The sugar is bound in a manner that is highly consistent with previous mutational, functional, and spectroscopic analyses (reviewed in refs. 1, 2, and 46).

One of the galactopyranosyl rings of TDG lies against Trp151 (helix V), confirming biochemical (36) and spectroscopic (37) findings that indicate hydrophobic stacking between the bottom of the galactopyranosyl ring and the indole ring of the aromatic Trp side chain (Fig. 4). Although Trp151 can be replaced with Tyr or Phe, progressively decreased affinity is observed, and total replacement of the aromatic side chain at position 151 completely abolishes binding and transport. In addition, a red shift in the phosphorescence spectrum of single-Trp151 LacY observed upon sugar binding is explained by their apposition in the structure.

Glu269 (helix VIII) is the acceptor of hydrogen bonds from the C4-OH group of the galactopyranosyl ring, making it a likely candidate for a primary role in specificity. The structure explains the finding that even conservative replacement with an Asp abolishes binding and inactivates transport (4, 39, 47).

The η1 NH₂ group of Arg144 (helix V) donates a hydrogen bond to O5 in the ring and also is within hydrogen-bond distance of C6-OH. The η2 NH₂ group of Arg144 donates hydrogen bonds to C2′-OH of TDG and to Glu126 Oε2 (Fig. 4). Conservative replacement of Arg144 with Lys or neutral replacements virtually destroys binding and transport (4, 32).

Glu126 (helix IV) acts as hydrogen-bond acceptor from the C2′-OH of TDG, explaining why conservative replacement with Asp causes markedly diminished binding affinity and removal of the carboxyl group abolishes binding and transport (4, 33).

His322 (helix X) acts as a hydrogen-bond donor/acceptor between the εNH of the imidazole ring and the C3-OH of TDG and is stabilized by a hydrogen-bond donor/acceptor with the δNH of the imidazole and the OH of Tyr236. All replacements for His322 tested exhibit a marked decrease in binding affinity and little or no transport activity (4, 48, 49).

Finally, Asn272 (helix VIII) likely donates a hydrogen bond to the C4-OH of TDG, confirming the recent biochemical finding (42) indicating that Asn272 also plays an essential role in sugar binding. Gln is the only replacement that is well tolerated by LacY.

Additional Important Residues.

In addition to the residues that play a direct role in galactoside binding, Cys148 (helix V), well known with respect to substrate protection against alkylation (reviewed in ref. 46), is close to bound TDG but is not sufficiently close to interact directly. Similarly, replacement of Ala122 (helix IV) with bulky side chains or alkylation of A122C with bulky thiol reagents causes LacY to become specific for the monosaccharide galactose, and disaccharide binding and transport are blocked (50). However, Ala122 does not make direct contact with TDG.

Asp240 (helix VII) and Lys-319 (helix X) are weakly salt bridged (51, 52), and mutation of either residue does not markedly alter affinity (4).

Glu325 (helix X) and Arg302 (helix IX) (Figs. 4 and 6) are directly involved in coupled H⁺ translocation. Neutral replacement of either residue yields mutants that are defective in all transport reactions that involve lactose/H⁺ symport but catalyze equilibrium exchange and/or counterflow as well as or better than WT (1, 2). Neither residue is sufficiently close to the bound sugar to make direct contact.

Materials.

TDG was obtained from Carbosynth Limited. Buffers were from Sigma-Aldrich. Talon Superflow Resin was purchased from BD Clontech. Dodecyl-β-d-maltopyranoside (DDM) and n-nonyl-β-d-glucoside (NG) were obtained from Affymetrix. All other materials were of reagent grade and were obtained from commercial sources.

Growth, Expression, and Purification.

Plasmid pT7-5 encoding LacY G46W/G262W/His6 was expressed in E. coli C41. The detailed procedure is described in SI Appendix. Membranes were prepared and solubilized in DDM, and LacY was purified on a Co(II) column and eluted in 20 mM Hepes/0.2% NG/200 mM imidazole (pH 6.5). Details are given in SI Appendix.

Crystallization, Data Collection, and Structure Determination.

TDG was added (5 mM final concentration) to a protein solution [10 mg/mL in 20 mM Hepes (pH 6.5)/0.2% NG] before crystallization trials. Crystallization screens were set up using the hanging-drop vapor-diffusion method on a Mosquito Crystal Robot (TTP Labtech) in a 96-well plate with a drop ratio of 100 nL well solution:100 nL protein. Crystals appeared in 24 h and grew to 500 × 150 × 150 µm in 1 wk. Crystals were harvested after 7 d and screened for diffraction. Crystals were reproducible and appeared in wide range of PEG concentrations. They also appeared at different pHs and with different buffers. The best-diffracting crystals grew in 1.0 M NaCl/0.05 M Tris (pH 8.0)/24% PEG 600. No cryoprotectant was added before data collection.

Diffraction data were collected at the Lawrence Berkeley National Laboratory Advanced Light Source Beamline 8.3.1, at −170 °C and at a wavelength of 1.115 Å. The highest-resolution crystals diffracted to 3.5 Å. Data were processed with either XDS (54) or HKL2000 (55).

The structure was solved by molecular replacement in PHENIX (56) using LacY 2CFQ as the search model. Coot was used for density fitting (57), and refinement was done with PHENIX (56) and Refmac from the CCP4 suite (58). Hydrogens were included in riding positions. In the final cycles of refinement, translation, libration, screw-motion parameters (59) for a single group (the entire structure) were refined, and Ramachandran angles were restrained. SI Appendix, Table S1 shows data and refinement statistics. The MolProbity server (60) was used for structural validation leading to improvement. Difference–distance analysis was done using Rapido (61), and rigid body motions were analyzed using the Dyndom server (62).