Structure of the D-alanylgriseoluteic acid biosynthetic protein EhpF, an atypical member of the ANL superfamily of adenylating enzymes

Abstract

The structure of EhpF, a 41 kDa protein that functions in the biosynthetic pathway leading to the broad-spectrum antimicrobial compound D-alanylgriseoluteic acid (AGA), is reported. A cluster of approximately 16 genes, including ehpF, located on a 200 kbp plasmid native to certain strains of Pantoea agglomerans encodes the proteins that are required for the conversion of chorismic acid to AGA. Phenazine-1,6-dicarboxylate has been identified as an intermediate in AGA biosynthesis and deletion of ehpF results in accumulation of this compound in vivo. The crystallographic data presented here reveal that EhpF is an atypical member of the acyl-CoA synthase or ANL superfamily of adenylating enzymes. These enzymes typically catalyze two-step reactions involving adenylation of a carboxylate substrate followed by transfer of the substrate from AMP to coenzyme A or another phosphopantetheine. EhpF is distinguished by the absence of the C-terminal domain that is characteristic of enzymes from this family and is involved in phosphopantetheine binding and in the second half of the canonical two-step reaction that is typically observed. Based on the structure of EhpF and a bioinformatic analysis, it is proposed that EhpF and EhpG convert phenazine-1,6-dicarboxylate to 6-formylphenazine-1-carboxylate via an adenylyl intermediate.

Figure 1

Modified version of the biosynthetic pathway leading to AGA first proposed by Giddens et al. (2002 ▶).

Figure 2

Proposed function of EhpF based on the crystallographic observations and its homology to GriC from the grixazone pathway and an arylcarboxylate reductase from Nocardia. (a) Proposed conversion of PDC to 6-formyl-phenazine-1-carboxylate catalyzed by EhpF and EhpG. (b) The role of GriC and GriD in grixazone biosyntheis. (c) Vanillin production in Nocardia catalyzed by an arylcarboxylate reductase enzyme that features an N-terminal EhpF/GriC-like domain.

Figure 3

Structure of EhpF. (a) Ribbon diagram of the EhpF dimer. The subunits are colored by subdomain: residues 1–81 are shown in yellow, residues 82–267 are shown in magenta and residues 268–354 are shown in blue. (b) Ribbon diagram of the EhpF monomer colored in rainbow. The N- and C-termini are labeled. Bound PDC is shown in red.

Figure 4

Comparison of the structure of EhpF with that of DhbE, a representative member of the ANL superfamily of adenylating enzymes (May et al., 2002 ▶). Subdomains 2 and 3 of EhpF and subdomains 2 and 3 of domain I of DhbE (all shown in red) superimpose with an r.m.s.d. of ∼3.3 Å for 229 C^α atoms. Structures are shown in identical orientations after superpositioning. (a) Ribbon diagram of the EhpF monomer. Subdomain 1 is colored blue, while subdomains 2 and 3 are colored red. (b) Ribbon diagram of DhbE (PDB code 1md9; May et al., 2002 ▶). Subdomain 1 is shown in magenta, subdomains 2 and 3 are shown in red and the C-­terminal domain (which is lacked by EhpF) is shown in yellow.

Figure 5

Structure-based sequence alignment of the homologous regions of EhpF and DhbE. The alignment was constructed using the MATRAS algorithm (Kawabata, 2003 ▶).

Figure 6

EhpF may use alternative residues to perform key functions. While most of the nucleotide-binding residues found in ANL-superfamily enzymes are conserved in EhpF, two key residues are missing at first glance. Stereoview of the potential nucleotide-binding site of EhpF, illustrating that Arg330 and Lys172 are in reasonable positions to perform roles analogous to the highly conserved residues Arg428 and Lys519 of DhBE. Following automated structural alignment, Arg428, Lys519 (both shown in green) and AMP (shown in yellow) from DhbE (PDB code 1md9) are displayed together with a ribbon diagram of EhpF. Asp331, Arg330 and Lys172 of EhpF are shown in blue.

Figure 7

(a) Stereoview of the PDC-binding site. For clarity, Thr234 and Asn232 are shown but not labeled. (b) Stereo OMIT map and the final model of PDC. Positive difference density was calculated by omitting PDC from a round of refinement with REFMAC5. Density is contoured at 3σ.

Figure 8

Arg120 and Arg153 separate the PDC-binding site from the ATP-binding/catalytic site of EhpF. (a) Cartoon illustrating the location of the observed PDC-binding site relative to the location of the canonical ANL-superfamily active site. The structures of PDC-bound EhpF and DhbE in complex with AMP and 2,3-dihydroxybenzoate (PDB code 1md9) were superimposed to generate the figure. (b) Cartoon representation illustrating the location of Arg120 and Arg153. Pockets were identified and illustrated using the PyMOL plugin PocketPicker (Weisel et al., 2007 ▶). The orientation is the same as in (a). (c) Simulation of a gatekeeping function for Arg120 and Arg153. PocketPicker identifies a single large cavity running through the protein and connecting the PDC-binding and catalytic sites when Arg120 and Arg153 are truncated in silico to alanine residues.