Discovery and Characterization of Antibody Probes of Module 2 of the 6-Deoxyerythronolide B Synthase

Abstract

Fragment antigen-binding domains of antibodies (F_abs) are powerful probes of structure-function relationships of assembly-line polyketide synthases (PKSs). We report the discovery and characterization of F_abs interrogating the structure and function of the ketosynthase-acyltransferase (KS-AT) core of Module 2 of the 6-deoxyerythronolide B synthase (DEBS). Two F_abs (AC2 and BB1) were identified to potently inhibit the catalytic activity of Module 2. Both AC2 & BB1 were found to modulate ACP-mediated reactions catalyzed by this module, albeit by distinct mechanisms. AC2 primarily affects the rate (k_cat), whereas BB1 increases the K_M of an ACP-mediated reaction. A third F_ab, AA5, binds to the KS-AT fragment of DEBS Module 2 without altering either parameter; it is phenotypically reminiscent of a previously characterized F_ab, 1B2, shown to principally recognize the N-terminal helical docking domain of DEBS Module 3. Crystal structures of AA5 and 1B2 bound to the KS-AT fragment of Module 2 were solved to 2.70 and 2.65 Å resolution, respectively, and revealed entirely distinct recognition features of the two antibodies. The new tools and insights reported here pave the way towards advancing our understanding of the structure-function relationships of DEBS Module 2, arguably the most well-studied module of an assembly-line PKS.

Graphical Abstract

Assembly-line polyketide synthases (PKSs) are multi-enzyme systems that synthesize structurally complex polyketide natural products, many of which have found therapeutic utility (1). Engineering assembly-line PKSs to produce novel bioactive agents has therefore been a longstanding goal, albeit a challenging one due to their sheer complexity and our limited understanding of their structure-function relationships. To advance this frontier, our laboratory and others have focused on achieving a fundamental understanding of the core mechanisms underlying the enzymology of assembly-line PKSs. Specifically, each PKS module of an assembly line harbors a ketosynthase (KS), an acyltransferase (AT), and an acyl carrier protein (ACP) domain that collaborate to elongate the growing polyketide chain via a decarboxylative C-C bond forming reaction with an unsubstituted or substituted malonyl extender unit (2). Additional enzymatic domains (e.g., ketoreductase (KR), dehydratase (DH), enoyl reductase (ER)) may also be present in some but not all modules of assembly-line PKSs (3–5). Owing to the universality of the chain elongation process, we have sought to understand the mechanisms by which the KS, AT, ACP, and their covalently bound acyl-chain species interact with each other in the context of the catalytic cycle of a PKS module (Figure S1).

Over the past two decades, X-ray crystallography and single-particle cryo-electron microscopy (cryoEM) have been especially powerful structural tools for structure-function analysis of assembly-line PKSs (6–17). More recent structural efforts have also relied on the use of fragment antigen-binding domains of antibodies (F_abs) as chaperones for crystallography and cryoEM (6, 12–15). For example, the F_ab 1B2 binds to the N-terminal helical docking domain of Module 3 of the 6-deoxyerythronolide B synthase (DEBS; Figure 1) and has proven invaluable for visualizing entire modules of two unrelated PKSs, the lasalocid PKS (13) and DEBS (14). We therefore sought to discover additional antibody probes to enhance our understanding of the KS-AT core of a representative assembly-line PKS module.

Figure 1.

F_abs specific to distinct regions of DEBS Modules 1, 2 and 3 have been identified to date via phage display (top). SDS-PAGE analysis (bottom right) of purified F_abs (AA5, AC2, and BB1) described in this report are shown under non-reducing (left lane) and reducing (right lane) conditions. For reference, a previously characterized F_ab (1B2) that binds specifically to the helical docking domain of Module 3 is also included.

Analogous to our previous phage display library screen of F_abs (3.7 × 10¹⁰) against DEBS modules 1 and 3 (6, 12, 15), three unique F_abs (AA5, AC2, and BB1) against DEBS Module 2 were identified (Figure 1). To stabilize Module 2 in its homodimeric state in the absence of the remainder of the DEBS 1 protein (Figure S1), this stand-alone module was fused to the N-terminal coiled-coil docking domain from Module 3 (hereafter referred to as D(3)-Module 2).

Binding analysis using size exclusion chromatography (SEC) revealed that all three F_abs recognized the homodimeric D(3)KS₂-AT₂ fragment of D(3)-Module 2 (representative data for AA5 is shown in Figure 2; data for other antibodies is shown in Figure S2). Antibody recognition was quantified via ELISA with apparent K_D values of AA5, AC2, and BB1 being 5.7 ± 0.06, 1.6 ± 0.02, and 8.2 ± 0.07 nM, respectively (Figure S3).

Figure 2.

1.5 molar equivalents F_ab AA5 were incubated with D(3)KS₂-AT₂ on ice for 1 h before injection onto a SEC column. Comparison of SEC UV Chromatographs of AA5 (pink), D(3)KS₂-AT₂ (blue), and AA5:D(3)KS₂-AT₂ (grey). SDS-PAGE analysis of the major SEC peak (grey) shows a near 1:1 stoichiometric mixture of AA5 and D(3)KS₂-AT₂.

The ability of all three F_abs to bind homodimeric D(3)KS₂-AT₂ tightly highlights the power of phage display libraries of naïve F_abs to yield high-affinity reagents. Control studies with a previously characterized F_ab, 1B2, which recognizes the D(3) docking domain of Module 3, validated our analysis of the newly isolated antibodies (Figure S4) (15).

To establish that AA5, AC2, and BB1 specifically recognized the KS-AT fragment of Module 2 and not its flanking D(3) domain, each antibody was counter-screened against D(5)KS₂-AT₂, a homodimeric construct harboring the distantly related N-terminal docking domain of Module 5 in lieu of D(3) (Figure S5). All three F_abs described in this report (AA5, AC2, and BB1) retained their affinity for the latter protein, while 1B2 was unable to bind D(5)KS₂-AT₂ (Figure S6). From this data we concluded that AA5, AC2, and BB1 were specific for the KS-AT core of Module 2.

To investigate whether AA5, AC2, and BB1 inhibited the catalytic activity of Module 2, turnover of a truncated derivative of DEBS harboring its loading didomain, Module 1, and Module 2 fused to the terminal thioesterase (TE) domain was measured spectroscopically in the presence of each F_ab (15). As shown in Figure 3A, AC2 and BB1 but not AA5 inhibited turnover of this bimodular PKS.

Figure 3.

(A) NADPH consumption by a truncated bimodular derivative of DEBS comprised of only its first two modules was monitored in the presence of 1.5 equivalents AA5 (pink), AC2 (green), BB1 (orange), or 1B2 (black). Control assays lacking M2-TE (grey) or any F_ab (blue) are also shown. (B) Effect of AA5, AC2, BB1, and 1B2 binding on DEBS Module 2-TE turnover at varied NDK-SNAC concentrations. (C) Effect of AA5 (5 μM), AC2 (5 μM) or BB1 (1 μM) on the activity of a dissociated form of Module 2 comprised of D(3)KS₂-AT₂, KR₂, and varying concentrations of ACP₂. Data points correspond to averages of technical triplicates.

As illustrated by the control 1B2 antibody, this assay is incapable of discriminating between inhibition of intermodular chain translocation between the ACP domain of Module 1 and the KS domain of Module 2 from inhibition of a core reaction (transacylation or elongation) catalyzed by KS₂-AT₂. To do so, the turnover of stand-alone Module 2-TE was assayed in the presence of a diffusible analog of its natural diketide substrate (hereafter referred to as NDK-SNAC, whose preparation is described in Figures S7–S13) (15, 18–20). AC2 and BB1 (but not AA5) inhibited Module 2-TE enzyme activity (Figure 3B). To further dissect this inhibition mechanism, Module 2 was dissociated into three fragments: D(3)KS₂-AT₂, KR₂, and ACP₂ (Figure S14). Prior efforts in our laboratory to reconstitute the catalytic activity of a PKS module from its dissected domains relied on ¹⁴C radiolabeling or LC-MS assays (21). Here we were able to quantify the activity of Module 2 by monitoring NADPH consumption in the presence of a stoichiometric excess of ACP₂ (Figure S15). Under conditions where neither KS acylation with its diketide substrate nor ACP acylation with a methylmalonyl extender unit nor β-ketoreduction of the elongated triketide product were rate-limiting, AC2 and BB1 inhibited turnover of dissociated Module 2 while AA5 did not (Figure 3C, Figure S16) (19, 22). Notably, the inhibitory mechanisms of AC2 and BB1 were different; whereas AC2 appeared to be a non-competitive inhibitor, the inhibitory effect of BB1 appeared predominantly competitive. The competitive nature of BB1 inhibition in this assay is distinct from its non-competitive inhibitory effect in the stand-alone M2-TE assay (Figure 3B). Taken together, our data suggests that TE-promoted substrate off-loading, not KS-promoted chain elongation, is rate limiting in the M2-TE turnover assay under saturating NDK-SNAC concentrations. This finding yet again highlights the utility of F_abs as mechanistic probes for PKSs. Further analysis of the distinct inhibitory mechanisms of AC2 and BB1 is warranted.

While crystal growth from complexes comprised of KS₂-AT₂ and either AC2 or BB1 has proven elusive thus far, we took advantage of AA5 and 1B2 as crystallography chaperones to obtain co-crystals that diffracted to 2.70 and 2.65 Å, respectively (Figure 4, Table S1, Figure S17). In both cases the homodimeric KS₂-AT₂ protein crystallized in an extended architecture analogous to all previously characterized high-resolution structures of this didomain core of assembly-line PKS modules (13–15, 23).

Figure 4.

Crystal structure(s) of (A) 1B2- (PDB ID 8EE0) and (B) AA5- (PDB ID 8EE1) bound D(3)KS₂-AT₂.

Similar to its previously reported co-crystal structure with D(3)KS₃-AT₃, 1B2 predominantly made contact with the N-terminal docking domain of D(3)KS₂-AT₂ (Figure S18) (15). While D(3) was well-resolved in this co-crystal structure, it was not resolved in the AA5 co-crystal structure. Meanwhile, AA5 principally contacted the hydrolase subdomain of the AT in an orientation that was directed away from its active site Ser residue. Additional contacts were also observed with the KS-AT linker. The remarkably high affinity of AA5 for KS₂-AT₂ appears to be dominated by several polar contacts (Figure S19). Overall, this structure validates the non-inhibitory nature of AA5 binding to Module 2. No major conformational changes were observed between the two snapshots of the KS₂-AT₂ fragment (Figure S20, RMSD 1.44 Å). Slight variation between the two KS₂-AT₂ models can be attributed to AT flexibility; this type of rotation of the AT domain has been previously described (23–24). Modeling studies were undertaken with the previously reported NMR structure of ACP₂ (PDB ID 2JU2) (25) to gain insight into the specificity-determining residues at the KS/ACP interface of DEBS Module 2. Recently the cryoEM structure of DEBS Module 1 with the ACP stalled in the elongation state (PDB ID 7M7F), was reported (14). To simulate the analogous binding pose for DEBS Module 2, the KS₂-AT₂ didomain from PDB 8EE0 (in complex with F_ab 1B2) was superposed onto the KS₁-AT₁ didomain and the solution NMR structure of ACP₂ was superposed onto ACP₁ (Figure S21). In agreement with prior experimental data, the model revealed favorable interactions between Loop 1 of ACP₂ and the KS₂-AT₂ (26), further supporting our reported structures.

This is the first reported structure of the KS-AT didomain of DEBS Module 2, arguably the most extensively studied module of an assembly-line PKS to date (11, 25–28). Our results highlight the utility of F_abs in studying PKSs. Three unique F_abs (AA5, AC2, and BB1) that recognize the KS-AT fragment of DEBS Module 2 were characterized. AA5 was highlighted as a crystallography chaperone, meanwhile AC2 and BB1 have unique inhibitory properties that are useful mechanistic probes of Module 2. This module is known to have an exceptionally broad substrate scope (27); further investigations into this property are warranted in a manner that is enabled by our reported structures and crystallographic chaperones.