Figure 1.
Scheme of the net reaction catalyzed by alanine transaminase (glutamic acid-pyruvic acid transaminase, GPT).
In the first half-reaction (1) L-alanine is converted to pyruvate with the concomitant conversion of the Lys-PLP Schiff-base linked cofactor to free Lys and PMP (in AlaA, the catalytic lysine residue is Lys240). In the second and last half-reaction (2), a molecule of 2-oxoglutarate is converted to L-glutamate and PMP is recycled back to the enzyme’s resting state cofactor (Lys240-PLP). In the net reaction scheme, a lonely electron pair is shown beside the reactive amine groups of L-alanine and L-glutamate. In the case of valine-pyruvate transaminase (AvtA), the incoming oxo acid is 3-methyl-2-oxobutanoate yielding L-valine as the corresponding α-amino acid instead of L-glutamate.
Table 1.
Bacterial strains and plasmids.
Table 2.
Crystallographic data processing and refinement statistics of AlaA.
Figure 2.
Overall and active site structure of E. coli AlaA in complex with acetate.
(A) Ribbon representation of the overall structure of AlaA. One of the chains is shown in green while the other is shown in domain colors: the central, large domain in cyan, the small domain in violet, and the N-terminal arm of both chains is shown in deep blue. The PLP cofactor is shown in spheres and CPK colors. (B) Detail of the interaction between the N-terminal arm (H1-plug-H2) motif of one AlaA subunit (green) with helix H4 from the other subunit (cyan), shown with the experimental σA-weighted electron density map (2mFo–DFc) contoured at 1.0 σ level. Relevant helices and residues that participate in the interaction are labeled. (C) Inset of the active site. Active site residues are shown in sticks and color-coded as in (B); the carbon atoms of Lys240-PLP are in grey and in yellow in the acetate anion. Polar interactions within 3.5 Å of the PLP cofactor and acetate are represented by dotted lines. The experimental σA-weighted electron density map (2mFo–DFc) contoured at 1.0 σ.
Figure 3.
Comparison of AlaA with structurally homologous enzymes.
(A) Phylogenetic tree based on structure-based multiple sequence alignments of AlaA obtained from PDBeFold [58]. Functionally related enzymes are shaded in like colors; alanine transaminases in gold, tyrosine aminotransferases (TyrAT) in cyan, aspartate aminotransferases (AspAT) in pink, kynurenine aminotransferases (KAT) in green, aspartate decarboxylases (CobD, AspDC, in orange), histidinol phosphate aminotransferases (HspAT) in grey; other transaminases of unknown function or with unique substrate preferences are not shaded. (B) Cartoon representation of alanine transaminases of known structure, highlighting the overall fold structure, catalytic residues, cofactor status and N-terminal motifs of AlaA (PLP, acetate), PfAlaAT (PMP, PDB 1xi9), HvAlaAT (DCS, PDB 3tcm) and human ALT2 (PLP, PDB 3ihj). In AlaA and HvAlaAT the N-terminal H1-plug-H2 motifs are fully structured, whereas in PfAlaAT and ALT2 different segments of the N-terminal arm are disordered. The most representative PfAlaAT monomeric structure (present in three out of four copies in the crystal asymmetric unit) lacks interpretable electron density for the eight-residue segment (from Ala14 to Leu20, delimited by orange spheres) spanning the plug. In ALT2, the N-terminal 65-amino-acid residues fold into a long β-hairpin structure that swaps domain and extends toward the opposite subunit (the start and end of the swapped β-hairpin are marked with asterisks), partially covering the active site cavity and ending in a ten-residue unstructured segment (spanning Ile95 to Gln104 until the anchor Pro105 residue). This disordered region (delimited by yellow spheres) is also located over the substrate-binding pocket and therefore may have functional and structural roles akin to those of the plug motif described in AlaA and PfAlaAT.
Figure 4.
Active site architecture of alanine aminotransferases.
Catalytically important amino acids and the cofactor of alanine transaminases: AlaA (A), PfAlaAT (PDB 1xi9) (B), human ALT2 (PDB 3ihj) (C) and HvAlaAT (PDB 3tcm) (D) are represented in sticks and color coded as in Figure 3. Dashed lines represent polar interactions. (A) Carbon atoms of the acetate anion are shown in green. (B) PfAlaAT in the PMP form with a disordered N-terminal segment. The pyridoxamine ring of PMP is packed between Ile207 and Tyr127 (not shown for clarity). (C) In ALT2 all active-site residues pinpointed in AlaA (A) and PfAlaAT (B) are structurally conserved, except for the residue equivalent to Gly41 (AlaA)/Gly38 (PfAlaAT). Gly96 (ALT2), which falls in a disordered loop (Ile95 to Gln104), is a likely candidate to assume the substrate-binding role of Gly41 (AlaA) because of its proximity. (D) The suicide inhibitor complex of D-cycloserine (DCS) in the active site of HvAlaAT assumes an identical configuration to that of AlaA, likewise accompanied by a fully ordered N-terminal motif that is closer in structure to the bacterial/archeal enzymes than to the human ALT2 despite the 44% sequence identity (by comparison, HvAlaAT and AlaA are only 28% sequence identical).
Figure 5.
Dual substrate recognition in alanine aminotransferases: Catalytic pocket for dicarboxylic acid substrates.
Schematic representation of the active site of AlaA in complex with acetate (A) and of T. termophilus α-aminoadipate transaminase LysN (PDB 2zyj) (B, orange) and A. thaliana LL-aminopimelate aminotransferase (PDB 3ei5) (B, blue) crystallized in complex with the glutamate external aldimine of PLP (PGU). AlaA residues Tyr15, Arg18 and Tyr129 (shadowed) are equivalent to residues known to stabilize the γ-carboxylate moiety of PGU, including Arg23 in α-aminoadipate transaminase and Tyr37 and Tyr152 in LL-aminopimelate aminotransferase.
Figure 6.
Homology models of the AlaC and AvtA alanine aminotransferases from E. coli.
(A) Overall architecture of the homology models of E. coli AlaC (left) and AvtA (right) shown in ribbon representation and chain colors; PLP is shown in spheres and CPK colors. (B) Sequence alignment of AlaA, AlaC and AvtA. Conserved positions are shaded in blue; consensus active site residues are listed in blue underneath the alignment; secondary structural elements are shown (helices in red and strands in green). Despite the low sequence identity between AlaC and AvtA when aligned with AlaA (21% and 23%, respectively), the residues of the active site are strictly conserved with exception of Val104 and Ser105, which interact with the phosphate group mainly through backbone atoms. (C–D) Modeled active site configuration of AlaC (C) and AvtA (D). While the global fold was modeled on the basis of the closest bacterial homologous structures, the position and orientation of the PLP cofactor, which is crucial for catalysis, was modeled after the structure of the AlaA active site.
Figure 7.
Effect of cultivation parameters on generation times of E. coli K-12 (wild-type), ΔalaA, ΔalaC and ΔavtA strains.
Histogram summarizing the generation times of E. coli K-12 (wild-type strain, WT) and L-alanine aminotransferase single-gene knock-out (KO) strains (ΔalaA, ΔalaC and ΔavtA) in different media under aerobic conditions (with shaking) (A) or under oxygen-limiting conditions (static cultures) (B). Error bars are standard errors from the mean (SEM) derived from N independent assays (N = 3–6). Statistical significance of differences between the generation times of WT and KO strains (under the null hypothesis of no difference) is assessed using a one-tail t-Student test for paired data and is depicted atop the bars according to P-value (P>0.05, no label; P<0.05, *; P<0.01, **; P<0.001, ***).
Figure 8.
Growth competition between ΔalaA, ΔalaC and ΔavtA deletion mutants and the wild-type strain.
The ratio of mutant to total (mutant plus WT) bacterial cells is plotted against time (in d) for two independent experiments. In experiment 1 (open symbols) the co-cultures were dialy back diluted to 1∶10,000 (13.29 generations/day) and experiment 2 (filled symbols) to 1∶100,000 (16.61 generations/day). Data points were used to best fit exponential lines to determine slopes, which were divided by the number of generations to calculate the average growth rate differences of the mutant strains. The starting ratios are indicated by the y-intercepts.