Elucidation of transient protein-protein interactions within carrier protein-dependent biosynthesis

doi:10.21203/rs.3.rs-72626/v1

Download PDF

Article

Elucidation of transient protein-protein interactions within carrier protein-dependent biosynthesis

https://doi.org/10.21203/rs.3.rs-72626/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 16 Mar, 2021

Read the published version in Communications Biology →

Version 1

posted

You are reading this latest preprint version

Fatty acid biosynthesis (FAB) is an essential and highly conserved metabolic pathway. In bacteria, this process is mediated by an elaborate network of protein•protein interactions (PPIs) involving a small, dynamic acyl carrier protein that interacts with dozens of other partner proteins (PPs). These PPIs have remained poorly characterized due to their dynamic and transient nature. Using a combination of solution-phase NMR spectroscopy and protein-protein docking simulations, we report a comprehensive residue-by-residue comparison of the PPIs formed during FAB in Escherichia coli. This work reveals the molecular basis of six discrete binding events responsible for E. coli FAB and offers insights into a method to characterize these events and those in related carrier protein-dependent pathways.

ONE SENTENCE SUMMARY: Through a combination of structural and computational analysis, a comparative evaluation of protein-protein interactions in de novo fatty acid biosynthesis in E. coli is performed.

Computational Biology

Chemical Biology

Spectroscopy

Solution-phase NMR Spectroscopy

Protein-protein Docking Simulations

Residue-by-residue Comparison

Carrier protein dependent synthases are responsible for the biosynthesis of a vast array of molecules, from primary metabolites to complex natural products ^1,2. These are generally organized as type I or type II enzymes, with the type I “megasynthases” containing multiple enzymatic domains and carrier proteins housed as large multi-domain proteins ³. In contrast, type II synthases exist as discrete proteins that must recognize and associate with one another in solution through an organized choreography of metabolic steps (Fig. 1a). In E. coli fatty acid biosynthesis (FAB) more than 25 partner proteins (PPs) are known to functionally bind to the acyl carrier protein (AcpP)(Fig. 1b) ^4–6, a small, four-helix bundle protein that shuttles intermediates between both fatty acid biosynthetic enzymes and regulatory proteins ⁷. AcpP must form specific protein•protein interactions (PPIs) with multiple partners, efficiently chaperoning intermediates through 30–35 discrete enzymatic steps to produce the membrane lipids that maintain homeostasis and facilitate cellular reproduction ^8,9. Simultaneously FAB generates fatty acid intermediates for cofactor biosynthesis and secondary metabolism¹⁰. This study further illustrates AcpP•PP recognition through unique PPIs with each of the FAB enzyme players (Fig. 1a). While presenting a combinatorial method to characterize these transient interactions, useful for both engineering and inhibitor design.

Throughout the iterative FAB cycle, substrates and intermediates are not only tethered to AcpP through a 4’-phosphopantetheine (PPant) thioester linkage ¹¹, but they are sequestered within the hydrophobic pocket of the AcpP helices, protecting intermediates from non-specific reactivity ¹². Reactions are controlled through this sequestration and presentation of the substrates appended to 4’-phosphopantetheine, a process termed chain flipping ^12,13. The rapid doubling times and relatively narrow distribution of fatty acid products require an efficient, high fidelity FAB ^14–16, suggesting that stochastic binding events of AcpP with its binding partners are unlikely. Activity studies and mechanism-based crosslinking experiments have demonstrated that acyl-AcpP binding and enzyme turnover are highly specific (Fig. 2c) ^17,18. A growing body of evidence suggests PPIs play a significant role in the mechanism of chain flipping and, therefore, the processivity of these pathways ^18–21. For example, recent studies have demonstrated that engineering enzyme specificity for a non-native AcpP-dependent enzyme can be accomplished by modifying the PPI residues for improved binding ²². Even single atom changes in the identity of AcpP-bound cargo have been demonstrated to impart significant perturbations to the structure of acyl-AcpP ²³.

Here we used ¹H-¹⁵N HSQC NMR titration studies to collect residue-by-residue information for six de novo FAB partner enzymes to characterize each intrinsic PPI with the E. coli AcpP. Experiments were performed to study the interfaces of AcpP with elongating ketosynthases FabB and FabF, reductases FabG and FabI, dehydratase FabA, and thioesterase TesA. These spectroscopic data combined with a combinatorial docking protocol benchmarked with crosslinked structures of AcpP in complex with FabA, FabZ, FabB, and FabF, provide atomic-resolution information on which residues of AcpP mediate each step in iterative de novo FAB. This combinatorial method was able to appreciate the unique challenges of modular synthases, with substrate identity effecting carrier protein structure and each enzyme forming unique interactions with the carrier. Due to the high sequence homology of AcpP with carrier proteins from other species ^25,21 and polyketide synthases ²⁵, this protocol is expected to extend for characterization of ACP•PP interactions for engineering and drug design across multiple systems.

NMR Titrations reveal dynamic AcpP interface

Previous work has established the utility of ¹H-¹⁵N HSQC-NMR titrations in the study of rapid and intricate PPIs ^26–28. In this study, uniformly labeled and perduterated ¹⁵N C8-AcpP (octanoyl-AcpP) (Fig. S6) was subjected to NMR titration using increasing concentrations of unlabeled PPs to detect the residues on AcpP that experience chemical shift migration (Fig. 2a, S1-3). Once saturated with partner enzyme, the extent of peak migration was quantified using the chemical shift perturbation (CSP) calculations (Fig. 2b, S1-3b)²⁹. The perturbed regions were then projected onto the 2D and 3D structures of the protein to identify regions affected by PP binding (Fig. 2c,d,4b). Furthermore, we utilized the TITAN NMR lineshape analysis program ³⁰ to analyze our spectra and obtain thermodynamic and kinetic parameters. Previous work has demonstrated the ability of chemical shift perturbations to identify critical protein-protein interactions, both in carrier protein-howemediated biosynthesis^31–33 and other systems ^34,35. Combining our titration data on FabF, FabI, FabG, and TesA with previous titrations on FabB ³¹ and FabA ²¹ allowed us to compare the binding interface on AcpP dictating PP recognition. With distinct residues involved in specific classes of enzyme binding, and internal residues that are likely important to the mechanistic process of chain flipping.

Docking to elucidate partner protein binding site

The ¹H-¹⁵N HSQC NMR titrations shown here can provide specific information of the AcpP residues involved in PP binding, however, the residues on the PP that mediate binding cannot be determined by this process. We have recently elucidated the x-ray crystal structures of several FAB enzymes crosslinked to AcpP, including FabA, FabZ, FabB, and FabF. These structures can indicate residues involved in PPIs on both proteins, however, each structure requires prior development of enzyme-specific crosslinking probes, which are not available in all cases. We sought to develop protein-protein docking protocols with Molsoft’s ICM software to predict structures of AcpP•PP complexes that have eluded experimental structural characterization ^36,37. Crystal structures of previously crosslinked AcpP-PPs were used to optimize this protocol. Briefly, it was identified that to accurately recreate complexes it was necessary to produce a water box in which the partner proteins were minimized. To maintain the applicability of the method to systems without crosslinked structures all docking was performed with an apo partner protein structure and the 2FAD crystal structure. Using expanded calculations to assist the general docking protocol (Fig. 3a), we were able to recapitulate crosslinked structure interfaces (Fig. 3b, Table S1) to sub 7 Å RMSD for the complex and sub 2 Å RMSD at the interface. Crosslinked structures were used in benchmarking as they give a learning set to examine which protocols and docking methods perform well. However, it must be noted that the comparison is imperfect, with the docked structures and NMR representing interactions in solution while the crosslinked structures are crystallized and covalently bound in a catalytic conformation. For example, the AcpP structure 2FAD and the AcpP crosslinked to FabA have a ~ 2 Å RMSD. The developed protocols were subsequently used to determine the binding interfaces of AcpP•FabI, AcpP•FabG, and AcpP•TesA (Fig. 3c, Fig. S4, Table S2, Fig. S4,S5) in conjunction with the NMR data. This methodology provides valuable context for matching the AcpP interactions to the PP structures, and the breadth of previously reported AcpP and PP activity and mutagenesis studies enable further validation of predicted AcpP•PP against past mutagenic experiments.

A combinatorial method to characterize modular synthase PPIs

To judge the ability of a combined method to accurately predict the structures of interacting enzymes, calculations were expanded to include CSP information. The interface residues identified through CSPs were given as focus residues for AcpP binding and known interacting residues specified below were given for partner proteins. Docked models of the enzymes FabA, FabB, and FabF were compared with and without known residues supplied. Data showed that all the enzymes tested had docking orientations which allowed for chain flipping of the substrate, demonstrating the ability of informed calculations to filter out nonproductive complexes. The average RMSD between the docked model and crosslinked crystal structures was: 9.29 Å in the informed calculation of FabF•AcpP binding and 20.08 Å (Table S7) without experimental knowledge. FabB models were 5.6 Å RMSD in the informed model and 10.5 Å in uninformed, and FabA models were 9.98 Å RMSD in the informed calculation and 9.89 Å in the uninformed. These data demonstrate that while docking alone was able to recreate the crosslinked structure, care must be taken to ensure the models are relevant. Leveraging experimental data ensured greater confidence in the models while yielding a suite of interface structures of relevant interacting orientations. To evaluate the ability of widely available online servers to recreate AcpP•Partner interfaces the AcpP•FabF complex was docked using the Cluspro, HADDOCK, and Rosie servers (Table S7). It should be noted that with informed residues and properly prepared structures online servers perform well. The structures used below are taken from the most stable docked orientation for simplicity, though the full suite of orientations remains relevant. However, these methods open the door to more studies of possible preliminary encounter complex states which may facilitate the fully bound form seen in crosslinking.

Ketosynthases: FabF and FabB

Elongating ketosynthases iteratively extend acyl-AcpP by two carbon units using malonyl-AcpP as a carbon source via a decarboxylative thia-Claisen condensation (Fig. 1a) ³⁸. C8-AcpP was titrated with increasing concentrations of the FabF ketosynthase (Fig. 2a) and compared to recently published data of C8-AcpP titrated with FabB. An octanoyl acylation state was selected to maintain consistency with prior work ^31,32 and was utilized for all titrations in this study. FabF displays perturbations throughout the AcpP, beginning at the end of helix I, with very little interaction on loop I until nearly the beginning of helix II. Helix II displays perturbations throughout until there is a small loss at the end of helix II, the visible residues are perturbed nearly throughout the AcpP until the end of helix IV where the perturbations drop off. The largest CSPs, Residues with CSPs greater than one standard deviation from the mean, in the FabF titration (Table S3) included I10 and L15 on helix 1 and F28 on loop 1 (Fig. 2b); D35, T39, V40 on helix 2; I54 on loop 2; and T64, V65, Q66, and A68 on helix 3. D35 and T39, the charged or polar residues which lie along the interface, appeared within interacting distance of N56’ and Q63' (residues identified from the models will be denoted by apostrophes throughout) ^31,32,39 (Fig. 4c,d). Surprisingly, a large number of these residues (I10, L15, F28, I54, T64, V65, and A68) are located within the acyl pocket or far from the interface yet show large perturbations. We hypothesized that these interior perturbations represent internal hydrophobic rearrangements that occur upon chain flipping during the binding event. Titan analysis calculated a K_d of 8.3 ± 9.8 µM with a k_off of 3512 ± 3341 s^− 1 and an approximately one to one stoichiometry(Table S4).

FabB performs the same ketosynthase reaction, but performs the first unsaturated elongation step, making a case study in specificity ⁴⁰. The NMR titrations were previously performed, but the data will be restated here for comparison. Overall FabB displays a slightly less broad set of CSPs than FabF, starting with little perturbation until the top of helix I. Perturbations continue once more at the end of loop I with more sparse interactions on helix II, without perturbation at residues 40 and 41 as well as a drop off in perturbation at the end of helix II. There are 2 CSPs on loop II and perturbations span most of helix III and loop III. Finally, helix IV sees sparse perturbations at the top of helix IV. In FabB the most perturbed residues were L15 of helix 1; D35, S36, L37, D38, and L42 on helix 2; E60 on helix 3; and T63 on loop 3. PPIs of AcpP with FabB contain a large number of predominantly acidic residues at the interface: D35, D38, and E60 (Fig. 4). Residues D35 interacts with K62’ and D38 forms a salt bridge with R65’ on FabB. E60 interacts with K150’ in the docked model. There is a small number of internal hydrophobic CSPs from L15 and L42 ³¹. Though FabF and FabB both perform the same fundamental chemical reaction, they appear to have distinct interfaces. FabF has a slightly larger interface (1023 Å²) when comparing docked models with FabB (962 Å²), perhaps consistent with its broader activity and the wider impact on CSPs compared to FabB⁴¹. This further agrees with data demonstrating a tighter binding for FabF than the previous calculated FabB Kd of 37.6 ± 6.6 µM. Though FabB and FabF share particularly similar structures and activity^37,39, their interactions with AcpP are unique.

Reductases: FabG and FabI

The condensation reaction performed by ketosynthases generates 3-oxoacyl-AcpP, which is subsequently reduced to 3R-hydroxyacyl-AcpP by FabG in a NADPH-dependent fashion (Fig. 1a)⁴². For NMR titrations, NAD⁺ was added along with FabG, as previous studies showed a difference in AcpP binding efficiency in the presence and absence of NAD^+ 43. The total perturbed residues span many residues across the AcpP. The perturbed residues begin at the end of helix I, with a few interactions across loop I. Helix II is perturbed to some degree across most of the AcpP, with only small regions seeing CSPs as low as background. Nearly all residues on loop II and through to helix IV are perturbed until the bottom of helix IV. The most perturbed most residues (Table S5, Fig. S7) on AcpP were: N25 and F28 of loop 1; D35, S36, L37, T39, V40, E47, and F50 on helix 2; and T64, Q66, and A68 on helix 3. Interface residues D35 and E47 interact with R19’ and R207’ of FabG respectively, while L37, T38, and V40 form hydrophobic interactions at the interface. N25, F28, F50, T52, T64, Q66, and A68 were all positioned away from the interacting face in the model. The identified region of interaction is in agreement with the binding region previously identified by mutagenesis and activity assays ^44,45. Binding calculations demonstrated a K_d of 20.5 ± 40.5 µM with a k_off of 5107 ± 2786 s^− 1 and a one to one stoichiometry(Table S4).

The final step of each elongation cycle in (saturated) FAB is catalyzed by the enoylreductase, FabI, which produces a saturated acyl-AcpP through NADH–dependent reduction of enoyl-AcpP (Fig. 1a)⁴⁶. FabI was also titrated with NAD⁺ present (Table S4, Fig. S8). Perturbations of AcpP by interaction with FabI are distributed throughout the ACP. With perturbations beginning on helix I and showing a few sparse interactions through helix I and onto loop I. However, significantly more interactions are seen on helix II, which shows interactions throughout only diminishing perturbation at the bottom of helix II. Finally, the loop II and helix III and IV see interactions fairly consistently until a drop in perturbations at the end of helix IV. The most perturbed residues of AcpP included I10 and L15 on helix 1; F28 in loop 1; D35, S36, L37, V43, M44, A45, and E47 in helix 2; A59 of helix 3; and Q66 and A68 of helix 4 are also highly perturbed. Similar to FabG, salt bridges likely form at residues D35 and E47, with E47 likely binding K43’ on FabI. And D35 interacting with R193’. Finally, the residues L37 and M44 on helix 2 form hydrophobic interactions with residues on the FabI interface, demonstrating a binding motif similar to FabG (Fig. 4). Uniquely, the perturbations and docked model of AcpP-FabI show not only the canonical AcpP helix II and III binding to the enzyme, but also additional interactions with helix 1. The identified binding region corresponds with previous mutational studies that first identified the AcpP–FabI interface ³⁹. Titan analysis calculated a K_d of 1.7 ± 1.2 µM with a k_off of 8500 ± 2700 s^− 1 and approximately one to one stoichiometry(Table S4).

The TesA E. coli thioesterase

Many organisms utilize a thioesterase to liberate fatty acids from the ACP. In E. coli, mature acyl-AcpPs are instead steered directly into other biosynthetic pathways via acyl transfer from AcpP, primarily for phospholipid biosynthesis. However, E. coli does possess the thioesterase TesA, which localizes in the bacterial periplasm ⁴⁷. Though TesA is not believed to be involved in the terminal step of E. coli FAB, it can hydrolyze acyl-AcpP in vitro and has been used as a tool for FAB engineering to increase free fatty acid titer when overexpressed within E. coli ⁴⁸. Perturbations predominantly occurred in unique locations compared to FAB enzymes (Table S6, Fig. S10), the perturbations are relatively minor throughout with small perturbations in helix and loop I. There are a larger number of perturbations on helix II, with more than half of the residues being perturbed over the background. Loop II and helix III show a diminished level of perturbation relative to helix II, this trend continues with few perturbations identified on loop III and helix IV. Overall TesA appears to effect less residues than the other proteins tested. The largest observed residues include loop 1 at S27 and D31; helix 2 at T42, M44, and A45; loop 2 at G52; and loop 3 at T63. Residue D31 appears to interact with R77’ of TesA upon binding. Additionally, D35 appears to interact with the TesA backbone or sidechain at S43’. The internal AcpP residues A45 and L42, located within the central hydrophobic core, are both perturbed upon TesA binding. S27 lies in the loop following helix I and near the interface of AcpP and the enzyme, likely experiencing or stabilizing loop motions upon salt bridge formation by D31. M44 appears somewhat distal from the interface near the acyl cargo, although in the case of FabI is part of the interface. T63 appears in the docked model to be positioned to interact with the hydrophobic surface region of TesA (Fig. 4). Titan analysis calculated a K_d of 12.5 ± 7 µM with a k_off of 9716 ± 820 s^− 1 (Table S4), though these data demonstrate more error due to the small number and small migration of peaks. The small number of interactions demonstrates that the TesA interface is not optimized for AcpP interactions, further suggesting that it could be engineered to provide a classical interface and increase the interactions and turnover.

Elucidation of dynamic AcpP•PP interface throughout E. coli FAS

Combining these NMR titrations and docked structures provides a powerful data set of functional PPIs in E. coli FAB (Fig. 1a). When compared against each other, these CSPs demonstrate two important concepts to shape our understanding of acyl carrier protein dependent synthases. Firstly, AcpP•PP interactions are commonly understood as electrostatic, with the acidic AcpP surface binding to a “positive patch” at the surface of the partner enzyme. However, the majority of the largest CSPs found in these studies correspond to hydrophobic residues (Fig. 3a,b) spanning the interface, acyl pocket, and back of the AcpP. But the data still suggests that electrostatic interface interactions are critical to the protein-protein binding event. Secondly, each enzyme enumerated above binds with AcpP transiently; the weak nature of these interactions is necessary for the “fast exchange” NMR chemical shifts and agrees with both our new data and previously known AcpP binding affinities ^29,44. Where AcpP and partner proteins are binding and dissociating fast enough to resolve as a single migrating peak on the spectra, rather than two distinct peaks. These findings demonstrate that recognition between AcpP and its PPs are dynamic processes, driven both by the electrostatic interface and conformational dynamism of the AcpP.

Across the six elongating enzymes tested, half of the residues with perturbations one standard deviation above the mean were at the interface, while the other half of perturbed residues lied in the pocket of AcpP. This is most likely a result of the substrate chain flipping into the PP. Approximately one-third of the largest perturbations, just 10 of 29, are unique to a single partner. More perturbations are shared by three or more of the six enzymes examined than are unique. Each partner, excluding TesA, displays perturbations at the “top” of the acyl pocket, at the start of helix 2 and the helix 3 to the beginning of helix 4. These interactions are likely those responsible for positioning S36 for substrate delivery. TesA is the only enzyme studied which is known to not be an AcpP FAB partner in vivo but has been demonstrated to have a low level of activity in vitro. Correspondingly, AcpP does not appear to form the interactions with TesA that are essential for efficient interactions. For other enzymes, it is not unreasonable that AcpP•PP interfaces would predominantly be shared sets of AcpP residues, with a few residues forming unique interactions that contribute to selectivity, given the small size of AcpP and the positively charged binding surfaces of PPs.

Our experiments found that recreating the interface of ACP•PP was feasible and in agreement with experimental structures of crosslinked complexes. While our study utilized a system that has a breadth of known information, the ubiquity of the active sites of modular synthases is such that even when partner residues are not known the correct binding site can be inferred. Furthermore, we believe the ability of NMR to appreciate the subtle differences in binding residues on the carrier protein is important to understand selectivity and substrate selection for inhibition or engineering. The combinatorial method synergizes the sensitivity and substrate accuracy of NMR experiments and the ability of informed docking to generate accurate models. This can be used to guide future engineering efforts, leveraging in silico screening for efficiency and economy. Furthermore, given the ability to recreate these interfaces, inhibitor screening should also be possible. The small size, simple electrostatic surfaces, and breadth of background knowledge of partner structures makes carrier protein dependent pathways ideal systems for computationally guided engineering and inhibition.

Taken together, these CSPs reveal a striking distinction between enzyme classes (Fig. 4a,b). Though AcpP contains only 77 amino acids, the residues involved in each binding are distinct, illuminating how one small protein can interact with dozens of partners. Each PP binding, excluding TesA, induces perturbations at the “top” of the acyl pocket, at the start of helix two and helix three to the beginning of helix four. Although AcpP•PP interactions are typically understood as predominantly electrostatic in nature, half of the largest CSPs correspond to hydrophobic residues (Fig. 4c,d). This represents an evolution in the understanding of type II FAS AcpP•partner recognition, demonstrating that unique residues are used for PPIs with different PPs. We propose a model wherein specific surface interactions are critical for creating allosteric movements within the central channel, triggering the chain flipping event. This may explain the stringent control of reactivity, yet broad range of substrates, necessary for FAB function. The disparate binding motifs found across this iterative pathway provides a compelling model for how a simple < 10 kDa protein performs unique interactions for each of six enzymes, while still displaying similarities within classes. This basic model can be extended across the known AcpP interactome, currently at 27 proteins, each of which may demonstrate similarly unique PPIs. Although this constitutes a broad sampling of each enzyme, further study can provide detail into each protein’s dynamics and allosteric control. This study provides a foundation with which to expand upon our understanding of PPI driven specificity, PPI redesign, and inhibitor development.

Crosby, J. & Crump, M. P. The structural role of the carrier protein – active controller or passive carrier. Nat. Prod. Rep. 29, 1111–1137 (2012).
Lai, J. R., Koglin, A. & Walsh, C. T. Carrier Protein Structure and Recognition in Polyketide and Nonribosomal Peptide Biosynthesis. Biochemistry 45, 14869–14879 (2006).
Schweizer, E. & Hofmann, J. Microbial Type I Fatty Acid Synthases (FAS): Major Players in a Network of Cellular FAS Systems. Microbiol. Mol. Biol. Rev. 68, 501–517 (2004).
Gully, D. & Bouveret, E. A protein network for phospholipid synthesis uncovered by a variant of the tandem affinity purification method in Escherichia coli. PROTEOMICS 6, 282–293 (2006).
De Lay, N. R. & Cronan, J. E. In vivo functional analyses of the type II acyl carrier proteins of fatty acid biosynthesis. J. Biol. Chem. 282, 20319–20328 (2007).
Gully, D., Moinier, D., Loiseau, L. & Bouveret, E. New partners of acyl carrier protein detected in Escherichia coli by tandem affinity purification. FEBS Lett. 548, 90–96 (2003).
Roujeinikova, A. et al. Structural Studies of Fatty Acyl-(Acyl Carrier Protein) Thioesters Reveal a Hydrophobic Binding Cavity that Can Expand to Fit Longer Substrates. J. Mol. Biol. 365, 135–145 (2007).
Tan, Z., Black, W., Yoon, J. M., Shanks, J. V. & Jarboe, L. R. Improving Escherichia coli membrane integrity and fatty acid production by expression tuning of FadL and OmpF. Microb. Cell Factories 16, (2017).
Shokri, A. & Larsson, G. Characterisation of the Escherichia coli membrane structure and function during fedbatch cultivation. Microb. Cell Factories 3, 9 (2004).
Morris, T. W., Reed, K. E. & Cronan, J. E. Lipoic acid metabolism in Escherichia coli: the lplA and lipB genes define redundant pathways for ligation of lipoyl groups to apoprotein. J. Bacteriol. 177, 1–10 (1995).
Majerus, P. W., Alberts, A. W. & Vagelos, P. R. Acyl Carrier Protein, IV. The Identification of 4′-phosphopantetheine as the Prosthetic Group of the Acyl Carrier Protein. Proc. Natl. Acad. Sci. U. S. A. 53, 410–417 (1965).
Cronan, J. E. The chain-flipping mechanism of ACP (acyl carrier protein)-dependent enzymes appears universal. Biochem. J. 460, 157–163 (2014).
Colizzi, F., Masetti, M., Recanatini, M. & Cavalli, A. Atomic-Level Characterization of the Chain-Flipping Mechanism in Fatty-Acids Biosynthesis. J. Phys. Chem. Lett. 7, 2899–2904 (2016).
Cronan, J. J. & Rock, C. O. Biosynthesis of Membrane Lipids. EcoSal Plus 3, (2008).
Liu, T., Vora, H. & Khosla, C. Quantitative analysis and engineering of fatty acid biosynthesis in E. coli. Metab. Eng. 12, 378–386 (2010).
Worthington, A. S., Rivera, H., Torpey, J. W., Alexander, M. D. & Burkart, M. D. Mechanism-Based Protein Cross-Linking Probes To Investigate Carrier Protein-Mediated Biosynthesis. ACS Chem. Biol. 1, 687–691 (2006).
Finzel, K. et al. Probing the Substrate Specificity and Protein-Protein Interactions of the E. coli Fatty Acid Dehydratase, FabA. Chem. Biol. 22, 1453–1460 (2015).
Miyanaga, A. et al. Structural Basis of Protein–Protein Interactions between a trans-Acting Acyltransferase and Acyl Carrier Protein in Polyketide Disorazole Biosynthesis. J. Am. Chem. Soc. 140, 7970–7978 (2018).
Miyanaga, A., Iwasawa, S., Shinohara, Y., Kudo, F. & Eguchi, T. Structure-based analysis of the molecular interactions between acyltransferase and acyl carrier protein in vicenistatin biosynthesis. Proc. Natl. Acad. Sci. U. S. A. 113, 1802–1807 (2016).
Weissman, K. J., Hong, H., Popovic, B. & Meersman, F. Evidence for a Protein-Protein Interaction Motif on an Acyl Carrier Protein Domain from a Modular Polyketide Synthase. Chem. Biol. 13, 625–636 (2006).
Beld, J., Cang, H. & Burkart, M. D. Visualizing the chain-flipping mechanism in fatty acid biosynthesis. Angew. Chem. Int. Ed Engl. 53, 14456–14461 (2014).
Sarria, S., Bartholow, T. G., Verga, A., Burkart, M. D. & Peralta-Yahya, P. Matching Protein Interfaces for Improved Medium-Chain Fatty Acid Production. ACS Synth. Biol. 7, 1179–1187 (2018).
Sztain, T. et al. Modifying the Thioester Linkage Affects the Structure of the Acyl Carrier Protein. Angew. Chem. Int. Ed. 0,.
Farmer, R., Thomas, C. M. & Winn, P. J. Structure, function and dynamics in acyl carrier proteins. PLOS ONE 14, e0219435 (2019).
Lohman, J. R. et al. The crystal structure of BlmI as a model for nonribosomal peptide synthetase peptidyl carrier proteins. Proteins 82, 1210–1218 (2014).
Rajagopal, P., Waygood, E. B., Reizer, J., Saier, M. H. & Klevit, R. E. Demonstration of protein-protein interaction specificity by NMR chemical shift mapping. Protein Sci. Publ. Protein Soc. 6, 2624–2627 (1997).
Zuiderweg, E. R. P. Mapping Protein−Protein Interactions in Solution by NMR Spectroscopy. Biochemistry 41, 1–7 (2002).
Pellecchia, M. et al. Structural insights into substrate binding by the molecular chaperone DnaK. Nat. Struct. Biol. 7, 298 (2000).
Williamson, M. P. Using chemical shift perturbation to characterise ligand binding. Prog. Nucl. Magn. Reson. Spectrosc. 73, 1–16 (2013).
Two-Dimensional NMR Lineshape Analysis | bioRxiv. https://www.biorxiv.org/content/10.1101/039735v1.
Milligan, J. C. et al. Molecular basis for interactions between an acyl carrier protein and a ketosynthase. Nat. Chem. Biol. 15, 669 (2019).
Nguyen, C. et al. Trapping the dynamic acyl carrier protein in fatty acid biosynthesis. Nature 505, 427–431 (2014).
Mj, J. et al. Manipulating Protein-Protein Interactions in Nonribosomal Peptide Synthetase Type II Peptidyl Carrier Proteins. Biochemistry https://pubmed.ncbi.nlm.nih.gov/28920687/ (2017) doi:10.1021/acs.biochem.7b00884.
Joseph, P., Poluri, K. M., Sepuru, K. M. & Rajarathnam, K. Characterizing Protein–Glycosaminoglycan Interactions Using Solution NMR Spectroscopy. Methods Mol. Biol. Clifton NJ 1229, 325–33 (2015).
Ahuja, S. et al. A Model of the Membrane-bound Cytochrome b5-Cytochrome P450 Complex from NMR and Mutagenesis Data. J. Biol. Chem. 288, (2013).
Abagyan, R. & Totrov, M. Biased probability Monte Carlo conformational searches and electrostatic calculations for peptides and proteins. J. Mol. Biol. 235, 983–1002 (1994).
Abagyan, R., Totrov, M. & Kuznetsov, D. ICM—A new method for protein modeling and design: Applications to docking and structure prediction from the distorted native conformation. J. Comput. Chem. 15, 488–506 (1994).
Joshi, V. C. & Wakil, S. J. Studies on the mechanism of fatty acid synthesis: XXVI. Purification and properties of malonyl-coenzyme A—Acyl carrier protein transacylase of Escherichia coli. Arch. Biochem. Biophys. 143, 493–505 (1971).
Rafi, S. et al. Structure of Acyl Carrier Protein Bound to FabI, the FASII Enoyl Reductase from Escherichia Coli. J. Biol. Chem. 281, 39285–39293 (2006).
Garwin, J. L., Klages, A. L. & Cronan, J. E. Structural, enzymatic, and genetic studies of beta-ketoacyl-acyl carrier protein synthases I and II of Escherichia coli. J. Biol. Chem. 255, 11949–11956 (1980).
Yu, X., Liu, T., Zhu, F. & Khosla, C. In vitro reconstitution and steady-state analysis of the fatty acid synthase from Escherichia coli. Proc. Natl. Acad. Sci. 108, 18643–18648 (2011).
Toomey, R. E. & Wakil, S. J. Studies on the mechanism of fatty acid synthesis XV. Preparation and general properties of β-ketoacyl acyl carrier protein reductase from Escherichia coli. Biochim. Biophys. Acta BBA - Lipids Lipid Metab. 116, 189–197 (1966).
Charov, K. & Burkart, M. D. A Single Tool to Monitor Multiple Protein–Protein Interactions of the Escherichia coli Acyl Carrier Protein. ACS Infect. Dis. 5, 1518–1523 (2019).
Zhang, Y.-M., Wu, B., Zheng, J. & Rock, C. O. Key Residues Responsible for Acyl Carrier Protein and β-Ketoacyl-Acyl Carrier Protein Reductase (FabG) Interaction. J. Biol. Chem. 278, 52935–52943 (2003).
Ac, P., Ym, Z., Co, R. & Sw, W. Cofactor-induced Conformational Rearrangements Establish a Catalytically Competent Active Site and a Proton Relay Conduit in FabG. Structure (London, England : 1993) https://pubmed.ncbi.nlm.nih.gov/15016358/ (2004) doi:10.1016/j.str.2004.02.008.
Heath, R. J. & Rock, C. O. Enoyl-Acyl Carrier Protein Reductase (fabI) Plays a Determinant Role in Completing Cycles of Fatty Acid Elongation in Escherichia coli. J. Biol. Chem. 270, 26538–26542 (1995).
Cho, H. & Cronan, J. E. ‘Protease I’ of Escherichia coli functions as a thioesterase in vivo. J. Bacteriol. 176, 1793–1795 (1994).
Shin, K. S., Kim, S. & Lee, S. K. Improvement of free fatty acid production using a mutant acyl-CoA thioesterase I with high specific activity in Escherichia coli. Biotechnol. Biofuels 9, 208 (2016).

Acknowledgements: We thank S. Opella, X. Huang, and the UCSD biomolecular NMR facility for support and guidance.

Funding: This work was funded by NSF IOS-1516156 and NIH R01GM095970.

Author Contributions: T.B. performed NMR experiments and developed docking protocol. R.A. provided support and guidance and A.P. provided assistance for docking. T.S. aided preparation of matplotlib and pymol scripts. DJL provided NMR data and interpretation. M.A.Y. assisted protein preparation and analysis. M.D.B. supervised the study and analysis.

Data availability statement: The docked models that support the findings of this study are available from the corresponding author upon reasonable request.

There is NO Competing Interest.

Supplemental942020.pdf
Supplemental Materials: Supplementary Methods, Figures S1 to S11, Tables S1-S6

Download PDF

Journal Publication

published 16 Mar, 2021

Read the published version in Communications Biology →

Version 1

posted

You are reading this latest preprint version

Elucidation of transient protein-protein interactions within carrier protein-dependent biosynthesis

Status:

Journal Publication

Version 1

Abstract

Figures

Main Text

Results

NMR Titrations reveal dynamic AcpP interface

Docking to elucidate partner protein binding site

A combinatorial method to characterize modular synthase PPIs

Ketosynthases: FabF and FabB

Reductases: FabG and FabI

Discussion

References and Notes

Declarations

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1