Exploring the compatibility of phosphopantetheinyl transferases with acyl carrier proteins spanning type II polyketide synthase sequence space

Abstract

Phosphopantetheinyl transferases (PPTases) play an essential role in primary and secondary metabolism. These enzymes facilitate the posttranslational activation of acyl carrier proteins (ACPs) central to the biosynthesis of fatty acids and polyketides. Modulation of ACP-PPTase interactions is a promising approach to both increase access to desired molecular outputs and disrupt mechanisms associated with disease progression. However, such an approach requires understanding the molecular principles that govern ACP-PPTase interactions across diverse synthases. Through a multiyear, course-based undergraduate research experience (CURE), 17 ACPs representing a range of putative type II polyketide synthases, from actinobacterial and nonactinobacterial phyla, were evaluated as substrates for three PPTases (AcpS, Sfp, and vulPPT). The observed PPTase compatibility, sequence-level analyses, and predictive structural modeling suggest that ACP selectivity is driven by amino acids surrounding the conserved, modified serine on the ACP. We propose that vulPPT and Sfp interactions with ACPs are driven primarily by hydrophobic contacts, whereas AcpS may favor ACPs that exhibit high net-negative charge density, as well as a broad electronegative surface distribution. Furthermore, we report a plausible, hitherto unreported hydrophobic interaction between vulPPT and a conserved ACP crease upstream of the invariant serine, which may facilitate docking. This work provides a catalog of compatible and incompatible ACP-PPTase partnerships, highlighting specific regions on the ACP and/or PPTase that show promise for future strategic engineering and inhibitor development efforts. One-Sentence Summary: Seventeen acyl carrier proteins from diverse type II polyketide synthases were evaluated for their compatibility with three phosphopantetheinyl transferases; results, along with sequence level-analyses and predictive structural modeling, reveal specific regions that can guide future strategic engineering efforts.

Keywords: acyl carrier protein; phosphopantetheinyl transferase; polyketide; synthase.

Graphical Abstract

Fig. 1.

The PPTase-catalyzed conversion of an inactive “apo” ACP to the active “holo” ACP. apo-ACPs are activated to their holo form via the PPTase-catalyzed installation of a CoA-derived 4´-phosphopantetheine (Ppant) arm onto the conserved serine at the N-terminus of ACP helix II.

Fig. 2.

The type II PKS ACPs addressed in this study belong to phylogenetically diverse BGCs. (a) The featured phylogenetic tree represents 6,322 type II PKS chain length factor (CLF) protein sequences identified in our previous global bioinformatic analysis of type II PKS BGCs. For methods associated with the construction of the tree, see McBride et al. (2023). The ring signifies the phylum classification for each terminal node. Cognate CLFs for the ACPs explored in this study are marked with a star. (b) Pairwise percent sequence identity (%ID) heatmap between all actinobacterial and nonactinobacterial ACPs selected for study. (c) Box plot showing the distribution and central tendency of ACP %IDs, both within and between the actinobacterial (act) and nonactinobacterial (non-act) groups. Boxes represent the interquartile range (IQR). Whiskers extend to data points up to 1.5x the IQR from the lower and upper quartiles. Median values are shown as black bands, means are shown as black crosses. (***) = p ≤ .001 (Welch’s ANOVA with the Games–Howell post-hoc test).

Fig. 3.

Sfp, AcpS, and vulPPT demonstrate broader substrate scope within the actinobacterial ACPs (top) than nonactinobacterial ACPs explored (bottom). The % holo was calculated based on the ratio of the area under the curve for the Abs₂₈₀ peak in the LC trace corresponding to holo versus the total area under the curve for the Abs₂₈₀ peaks in the LC trace corresponding to total ACP and multiplying by 100. The bar height represents the average % holo for the ACP–PPTase pair evaluated in triplicate, with error bars representing the standard deviation. All reactions were run at room temperature for 18 hr under the following conditions: ACP (80 µM), DTT (2.5 mM, from 1-M stock), MgCl₂ (10 mM, from 250-mM stock), CoA (0.8 mM, from 50-mM stock), PPTase (1 µM) in 50-mM sodium phosphate buffer, pH 7.6. Full names of each ACP and their corresponding sequence can be found in Supplementary Data (Tables S1 and S2) along with the LC-MS results for each phosphopantetheinylation reaction.

Fig. 4.

Predictive structural modeling of vulPPT with cognate and noncognate ACPs. (a) Predicted vulPPT–vulACP complex (predicted TM score = 0.89) following relaxation by molecular dynamics simulation. (b) vulACP Ile43 buried in a conserved vulPPT hydrophobic pocket. (c) Structural alignment of noncognate, nonactinobacterial ACPs carrying a (Ile/Leu)→Thr variation, and a possible hydrogen-bonded contact with vulPPT Trp₁₇₀. (d) Plausible hydrogen-bonded contacts between ferACP Gln₃₇/gloACP Gln₃₁ and Sfp Lys₃₁/vulPPT Arg₄₈. (e) Plausible hydrophobic interactions between vulPPT Phe₅₄ and a conserved vulACP hydrophobic crease.

Fig. 5.

Electrostatic rationale for AcpS–ACP interaction. (a) Reconstructed E. coli AcpP-AcpS complex (PDB: 5VCB). The electrostatic surface potential map of AcpS is shown. E. coli AcpP is depicted in lime green. (b) ACP net charge comparison between ACPs showing strong (>60% conversion) and poor (<60% conversion) compatibility with AcpS. Error bars represent 95% confidence interval. (***) = p ≤ .001 (Welch two-sample t-test). (c) Electrostatic surface potential maps of E. coli AcpP and nonactinobacterial ACPs vul/dac/bweACP. All ACPs are rendered in the same orientation. Color map is identical to panel (a). Mean electrostatic potentials (kcal/[mol∙e]) are shown.