Fig. 1. The glutaminase domains are present but disordered in the GlmS and GlmS–GlcN6P structures. a, Molecular packing of the GlmS crystal. The asymmetric unit contains one monomer, constituted by one synthase domain and one glutaminase domain that is not observed in the electron density. The interface of the functional dimer is formed by two synthase domains (colored in blue and cyan) belonging to neighboring asymmetric units. The N-terminal residues of the synthase domains are indicated for one dimer. A large volume in which no model could be traced can accommodate the glutaminase domain. b, SDS-PAGE analysis (10 % polyacrylamide gel) of dissolved crystals of GlmS. Lanes 1 and 4, molecular mass markers; lane 2, purified GlmS before crystallization (2.6 μg); lane 3, dissolved crystals of GlmS (2.15 μg).

Fig. 2. TLS refinement of the GlmS and GlmS–GlcN6P structures with two subdomains of ^CGlmS being treated as different TLS groups. (a) The C_α trace of ^CGlmS of the GlmS structure is colored according to the isotropic B-factors ranging from dark blue (B = 25 Å²) to red (B = 110 Å²). (b) Comparison of the displacement parameters of ^CGlmS before and after the anisotropic refinement. Up: GlmS crystal. Down: GlmS–GlcN6P crystal. B_TLS, the contribution from the TLS motion, B_res, the residual B-factors after applying the TLS model, and their sum B are shown. The isotropic B-factors, B_iso, are also indicated. (c) Mean displacements given as the TLS tensor eigenvalues, when each subdomain of the synthase domain is treated as a rigid group in the TLS refinements. Group 1 includes residues 243–449 and group 2, residues 450–608. Left: GlmS structure, right: GlmS–GlcN6P structure. (a) Eigenvalues of the translation tensor. (b) Eigenvalues of the libration tensor. (c) Eigenvalues of the screw tensor. The eigenvalues are shown as a cumulative stack bar. The S tensor was made symmetric by referring it to a coordinate system whose origin is at the center of reaction for the rigid group.

Fig. 3. In the absence of sugar, the C-tail is disordered and the synthase site is accessible to solvent. (a) 2F_o-F_cal electron density map contoured at 1 σ shows no density for residues 602–608 of the C-tail in the GlmS structure, which indicates its disorder. Two synthase domains (stick models in yellow and orange) belonging to neighboring asymmetric units form an extensive interface. In the GlmS–GlcN6P structure (synthase domains in pink and magenta coils), the C-tail covers the GlcN6P product (sphere model) in the synthase site. (b) Synthase site of the GlmS–GlcN6P structure showing the GlcN6P ligand and residues distant less than 4 Å. A 2F_o–F_cal electron density map contoured at 1 σ is superimposed on the model. (c) General view of the sugar-binding site. The synthase active site is formed at the interface of two synthase domains belonging to different monomers (in dark blue and light blue). When both the synthase and glutaminase domains are ordered as in the GlmS–Fru6P structure⁴ (glutaminase domain in cyan), the C-tail (in red) is sandwiched between one synthase domain and one glutaminase domain of the same monomer. The sugar-binding site is made by the backbones of the His-loop (in green) and C-tail. The Trp74 indole group from the glutaminase domain also participates in shielding the synthase site from solvent. The opening of the synthase site may result from a movement of the C-tail, the His-loop and/or the glutaminase domain.

Fig. 4. Conformational changes linked to the ordering of the glutaminase domains. The synthase domains are indicated in yellow and orange, pink and magenta, light and dark blue, and light and dark green for the GlmS, GlmS–GlcN6P, Glms–Fru6P and GlmS–Glc6P–DON structures, respectively. The glutaminase domains of the GlmS–Fru6P and GlmS–Glc6P–DON structures are indicated in cyan and green, respectively. (a) The ordering of the glutaminase domains is accompanied by a conformational change of Lys503^# and a translation shift of helix CF. In the GlmS–Fru6P structure, the H bonds that Arg539^# makes with the carboxyl group of Asp29 and the carbonyl group of Ala75 ensure the contact between the glutaminase and synthase domains of different monomers. In the GlmS–GlcN6P and GlmS structures, Arg539^# hydrogen binds to the carbonyl group of Ile510^# (not shown) and to the side-chain carbonyl group of Asn600. The salt bridge between the ε-amino group of Lys503^# and the oxygen carboxylate of Glu535^# is conserved in all structures through the translation of helix CF. In addition, Lys503^# hydrogen bonds to the main-chain carbonyl group of Asn600 in the GlmS–GlcN6P and GlmS structures but to the terminal carboxyl group of Glu608 in the GlmS–Fru6P structure. (b) Closer view of the synthase site showing the interactions between the C-tail and the synthase domain of the same monomer, the sugars and the glutaminase domains. Whereas the C-tail is disordered in the GlmS structure, in the GlmS–Fru6P structure, it is linked to the synthase domain of the same monomer via H bonds between the carbonyl group of Val605 and the NH group of Ala353, between the ε-amino group of Lys603 and both the carbonyl group of Gly398 and the carboxylate group of Glu396, and between the side-chain carboxylate group of Glu608 and the hydroxyl group of Tyr332. Most of these interactions are observed in the GlmS–GlcN6P structure. In the GlmS–Fru6P structure, the glutaminase domain and the C-loop are linked by four H bonds, between the hydroxyl group of Tyr28 and the carbonyl group of Leu601, the carbonyl group of Tyr25 and NH group of Lys603, the guanidinium group of Arg26 and both the carbonyl group of Lys603 and the hydroxyl group of Thr606. In the GlmS, GlmS–GlcN6P and GlmS–Fru6P structures, containing respectively no sugar, a cyclic sugar or a linear sugar, the location of His504^# is different. The imidazole N^δ of His504^# hydrogen binds to the hydroxyl 01 atom of GlcN6P and to the hydroxyl 05 atom of Fru6P. The H-bonds between the sugar and the catalytic residues Lys485 and Glu488 are direct for the GlmS–Fru6P structure but mediated by water molecules for the GlmS–GlcN6P structure. There are two direct H bonds between the C-tail and GlcN6P (through the carbonyl oxygen atom of Ala602 and both hydroxyl O1 and the amino group of GlcN6P), while all the contacts between the C-tail and Fru6P are indirect. (c) A view of a molecular surface section of the sugar-binding pocket in the GlmS–GlcN6P structure. Open-chain Fru6P (blue stick) lies deeper in the sugar-binding pocket than cyclic GlcN6P (pink stick). When sugar is bound at the synthase site, the C-tail (in coil) covers the synthase site. (d) Lys503^# adopts very different conformations in the complexes with linear or cyclic sugars. In the structures of GlmS–Fru6P, GlmS–Glc6P–DON and ^CGlmS-2-amino-2-deoxy-glucitol-6P structures (synthase domains in dark and light gray), the ε-amino group of Lys503^# makes an ionic interaction with the carboxylate group of the terminal amino acid (residue 608 of the C-tail), thereby strengthening the interactions between the C-tail and the synthase domain. Moreover, the conformation of the His-loop main chain allows a deeper burying of the sugar inside the synthase site, which is completely inaccessible to solvent. In the structures of GlmS and GlmS–GlcN6P, the C-tail is in a relaxed conformation and Lys503^# participates in shielding the synthase site from solvent in the absence of an ordered glutaminase domain.

Fig. 5. Step-by-step formation of the ammonia channel. The accessible surface in the different structures calculated with PyMol (http://pymol.sourceforge.net/) and a probe of 1.3 Å radius is represented following the same color scheme as Fig. 4 and is superposed to the accessible surface of the channel in the GlmS–Glc6P–DON structure, represented as a gray mesh surface. The positions of Glc6P and DON as observed in the GlmS–Glc6P–DON structure are indicated in stick representation to locate the synthase and glutaminase sites in all structures. (a) The GlmS structure. The glutaminase domain and the C-tail are not ordered and the synthase site is open to solvent. The channel is not formed because only the His-loop forms one of its rims. (b) The GlmS–GlcN6P structure. In the presence of cyclic sugars, the C-tail is in a relaxed conformation and Lys503^# participates in shielding the synthase site from solvent in the absence of ordered glutaminase domains. (c) The GlmS–Fru6P structure. Because of the ordering of the glutaminase domains, the channel is almost formed but not continuous because the indole group of Trp74 is inserted between the two observed adjacent cavities. (d) The Glms–Glc6P–DON structure. The rotation of the Trp74 indole group opens the channel, which is fully functional.

Data collection and refinement statistics

Links between the sugar conformation, the presence of the glutaminase domains and the postions of Lys503^#, the C-tail and helix CF

SB, The number of salt-bridges. HB, the number of hydrogen bonds.