Figure 1.
Af-GlnK3 and its physiological role in ammonium uptake.
A) Top view of the trimer Af-GlnK3, highlighting the ligand binding sites between the monomers and the protruding T-loops that are required for blocking ammonium transport. B) As discussed previously [1], [33], ammonium is actively taken up by Amt proteins and used to aminate glutamate in the ATP-dependent reaction of glutamine synthetase (GS). This reaction is coupled to glutamate∶oxoglutarate amidotransferase (GOGAT) that forms two molecules of glutamate from glutamine and 2-oxoglutarate.
Table 1.
Data collection and refinement statistics.
Figure 2.
Structural differences between (A) the ATP:Mg2+:2-OG complex and (B) the ADP complex of Af-GlnK3.
In (A) the key ligand 2-oxolutarate requires the presence of ATP for binding and is located at the base of the T-loop (blue), with its ã-carboxy group forming a hydrogen bond to the conserved K58. Residue Q39 is the only protein ligand to the Mg2+ ion (grey sphere), and it is this residue that in the ADP complex (B) attains the exact position of 2-OG in (A), forming an analogous hydrogen bond to K58. The resulting tilt and shift of the base of the T-loop leads to a stable â-hairpin structure in (A), compared to a less well-ordered loop in (B) that moves inward by 20°towards the trimer. In both structures, the respective other T-loop conformation is indicated.
Figure 3.
Binding mode of the ligands ATP, Mg2+ and 2-oxoglutarate to Af-GlnK3.
The stereo image shows a view into the ligand-binding cleft located at the interface of two monomers, one of which (dark green) provides the T-loop (blue) and B-loop regions to the binding site, the other monomer (light green) the C-loop. The Mg2+ ion (grey sphere) shows octahedral coordination by all three phosphate groups of ATP, by the á-carboxy and á-keto functions of 2-oxoglutarate and by the ã-amido oxygen atom of residue Q39.
Figure 4.
Structural consequences of ligand binding to GlnK proteins.
(A) With the ligand 2-OG (shown in CPK representation) placed in a wedge-like manner at the base, the T-loops of Af-GlnK3 are pried apart in a locked conformation. In the trimer, residues R47 of the monomers are 46.2 Å apart, a distance too large to be able to insert into the substrate channels of the cognate ammonium transporter. (B) Structure of the E. coli ortholog GlnK as seen in complex with the ammonium transporter AmtB (PDB-ID 2NS1) [27]. The T-loops are ordered and are positioned to fit the substrate channels of the transporter trimer, at a distance of 31.3 Å between residues R47.
Figure 5.
Binding of ATP and ADP to Af-GlnK3.
Contrary to observations made with the homologous Af-GlnK2 the titrations of (A) ATP and (B) ADP in the presence of 25 mM Mg2+ do not show signs of cooperative behavior. The integrated heat data (bottom panel) was fit with a model implying identical sites. The resulting enthalpies and dissociation constants underline that ATP is a much stronger ligand than ADP (Table 2).
Figure 6.
2-OG binding to Af-GlnK3 and temperature dependence.
A) A titration at 30°C shows an initial endothermic event indicating an entropy-driven process that is followed by a strongly exothermic, enthalpy-driven event. In the analysis of population microstates (bottom panel) this translates to an initial accumulation of the singly occupied species (▴) due to negative cooperativity for the second site (▪), but strong positive cooperativity for binding the third ligand (•). Only singly or fully occupied binding sites will be present in relevant amounts. B) At 70°C the initial binding event becomes exothermic, leading to a very different overall shape of the experimental curve (top panel). However, analysis of the population microstates shows the same qualitative behavior as in (A).
Table 2.
Thermodynamic parameters derived from calorimetric titrations of ATP and ADP nucleotides to Af-GlnK3.
Table 3.
Thermodynamic parameters derived from calorimetric titrations of 2-oxoglutarate to Af-GlnK3 pre-incubated with MgATP.
Figure 7.
ITC analysis of 2-OG binding to the Af-GlnK3 variants F86I and F86P at 30°C.
A) With respect to the wild type, the F86I variant shows reduced anticooperativity. The population microstate analysis (bottom panel) reveals that initially the singly-occupied species (▴) is populated, but that the negative cooperativity then is weaker so that the state with two bound ligands (▪) does accumulate before it yields to the fully occupied state (•) around a molar ratio of protein vs. ligand of 3. B) This effect is further enhanced in the F86P variant, where cooperativity is hardly seen in the microstate analysis and the binding sites are occupied sequentially. However, unlike in Af-GlnK2 that natively has P86, 2-OG is still bound.