Structure and Function of a Dual Reductase–Dehydratase Enzyme System Involved in p-Terphenyl Biosynthesis

Abstract

We report the identification of the ter gene cluster responsible for the formation of the p-terphenyl derivatives terfestatins B and C and echoside B from the Appalachian Streptomyces strain RM-5–8. We characterize the function of TerB/C, catalysts that work together as a dual enzyme system in the biosynthesis of natural terphenyls. TerB acts as a reductase and TerC as a dehydratase to enable the conversion of polyporic acid to a terphenyl triol intermediate. X-ray crystallography of the apo and substrate-bound forms for both enzymes provides additional mechanistic insights. Validation of the TerC structural model via mutagenesis highlights a critical role of arginine 143 and aspartate 173 in catalysis. Cumulatively, this work highlights a set of enzymes acting in harmony to control and direct reactive intermediates and advances fundamental understanding of the previously unresolved early steps in terphenyl biosynthesis.

Graphical Abstract

INTRODUCTION

Para-substituted terphenyls (p-terphenyls) are microbial natural products that consist of a chain of three benzene rings connected together at the para position. p-Terphenyls possess diverse biological activities including phosphodiesterase inhibition, neuroprotection, DNA topoisomerase inhibition, and antioxidant and immunosuppressive functions.^1–3 Examples of bacterial and fungal p-terphenyls include terfestatins, echosides, atromentin, terphenyllins, and terferol (Figures 1, S1)^1,4,5 Terphenyl metabolites are also structurally related to the microbial bis-(indolyl)quinones terraquinone, cochliodinols, and asterriquinones (Figures 1, S1). Similar to the p-terphenyls, indolylquinones possess a wide array of biological activities including antiretroviral, antidiabetic, antitumor activities and cytotoxic properties.^6–8

Figure 1.

Examples of microbial p-terphenyls (A) and bis(indolyl)benzoquinone compounds (B). Structures in blue are metabolites isolated from Appalachian Streptomyces RM-5–8.⁵

The biosynthesis of p-terphenyls and bis-(indolyl)quinones begins with the condensation of two precursors derived from the aromatic amino acids, l-Phe/l-Tyr and l-Trp, respectively (Figure 2). This central condensation reaction in p-terphenyls is catalyzed by a tridomain nonribosomal peptide synthetase to yield polyporic acid (2, PPA).⁹ PPA is proposed to subsequently undergo reduction and dehydration to give terphenyl triol 3, but the specific enzyme contributions and/or sequence of events have not been defined.^2,9 As part of an effort to explore the microbial diversity and corresponding metabolic potential of actinomycetes associated with thermal vents emanating from underground coal mine fires in Appalachia,^5,10–18 we previously reported the discovery of two new p-terphenyl glycosides (terfestatins B and C from Streptomyces RM-5–8), each decorated with the rare unsaturated sugar 4-deoxy-α-l-threo-hex-4-enopyranuronate (Figure 1, iv).^5,19 Herein, we report the elucidation of the putative terfestatin/echoside biosynthetic gene cluster (BGC) from Streptomyces RM-5–8. In addition, we highlight the collaborative role of two enzymes encoded by this gene cluster (TerB and TerC) in the conversion of PPA (Figure 2, 2) to terphenyl triol 3. TerB acts as a reductase to convert 2 to a tetraol intermediate, while TerC is a dehydratase that converts the tetraol to 3 (Figure 2). Cumulatively, this work advances our understanding of an early biosynthetic transformation central to the biosynthesis of terphenyls and corresponding new genomic, biochemical, and structural tools.

Figure 2.

Biosynthesis of terfestatins. (A) Putative ter BGC from Streptomyces RM-5–8 highlighting terB and terC (red) and (B) proposed biosynthetic pathway for terfestatins.

RESULTS AND DISCUSSION

Genomics and the Putative Terfestatin BGC/Pathway.

The 10.5 Mb Streptomyces RM-5–8 genome was previously sequenced.²⁰ Genome analysis using the anti-SMASH algorithm with a cluster finder^21,22 revealed multiple natural product BGCs including those encoding for type I and type II-derived polyketides, nonribosomal peptides, hybrid polyketide-nonribosomal peptides, and terpenes. One Streptomyces RM-5–8 BGC [the putative ter cluster, 39.4 kb, 33 open reading frames (ORFs); GenBank accession number MN877931, Figure 2A, Table S1] displayed the signature for terfestatin/echoside biosynthetic genes and additional genes anticipated for the corresponding Streptomyces RM-5–8 terfestatin/echoside metabolites.^2,3,9,23,24 Polymerase chain reactions followed by sequencing confirmed any ambiguous regions (Supporting Information Materials and Methods). The core biosynthetic genes of the ter cluster displayed 82% homology to the previously reported Streptomyces sp. LZ35 ech cluster, responsible for the biosynthesis of echoside (Table S1).^2,3 Analysis of the ter cluster revealed the inclusion of genes encoding for the following putative functions: a tridomain nonribosomal peptide synthetase (NRPS, terA), an oxidoreductase (terB), a nuclear transport factor 2 (NTF2) family protein (terC), (Figures S2 and S3) a nitroreductase transferase (terD), two methyltransferases (terF/L), and three glycosyltransferases (terO/P/Q). Based on these comparative analyses, we propose the biosynthetic pathway in Figure 2B for the biosynthesis of terfestatins.^2,3,9,24 Within this context, TerD is proposed to act as an aminotransferase to convert phenylalanine to phenylpyruvic acid (1). TerA is a 948 amino acid A-T-TE tridomain NRPS with a signature Phe adenylation domain lacking a C domain.²⁵ TerA has a similar architecture as the other terphenyl condensing enzymes and is responsible for the dimerization of two phenylpyruvic acid (1) molecules to produce PPA (2).^2,3 Conserved domain analyses of TerB and TerC (Figure S3; 314 and 166 amino acids, respectively) identified TerB as a member of the short-chain dehydrogenase–reductase superfamily^26,27 and TerC as a member of the NTF2 superfamily (Figure S4; Table S1). The NTF2 superfamily includes enzymes with diverse functions such as dehydrogenases, dehydrases, and cyclases.^28–31 Methyltransferases TerF and TerL and glycosyltransferases TerO, TerP, and TerQ are expected to subsequently catalyze the final maturation of Streptomyces RM-5–8 terfestatin/echoside metabolites. Although the putative reductase/dehydratase TerB/TerC (and corresponding homologs encoded by related BGCs) pair is proposed to participate in the conversion of 2 to 3, this remained speculative.

TerB and TerC In Vitro Studies.

For these studies, the synthesis of 2 followed previously reported methods,³² while 3 was synthesized using a new approach (Supporting Information Material and Methods). The terB and terC genes were cloned into a pET28a expression vector and used to transform Escherichia coli (BL21) followed by sequence verification. The corresponding proteins N-His₆-TerB and N-His₆-TerC (herein referred to as TerB and TerC, respectively) were overproduced in E. coli (BL21) and purified to homogeneity via Ni²⁺-affinity chromatography (Figure S5). In vitro enzymatic reactions containing 0.4 mM 2 and 20 mM NADH in the presence of TerB afforded ≤10% conversion to a new product consistent with 3 based on HPLC-DAD-ESI/MS (Figure 3A). Although substitution of TerB with TerC under the same reaction conditions resulted in no product formation, a dramatic increase in the formation of 3 (94% conversion) was observed when equimolar concentrations (6.6 μM) of both TerB and TerC were present (Figures 3A, S6, and S7). As confirmation, the product of the TerB/C-catalyzed reaction co-eluted with a synthetic 3 standard (Figure 3A). No product was detected in the absence of NADH in the presence of one or both enzymes. No TerB/C complex formation was observed via size exclusion chromatography in the absence or presence of cosubstrates. Thus, while the two enzymes have a cooperative effect on the production of 3, any TerB/C interactions must be weak and/or transiently similar to other biosynthetic enzyme systems that catalyze specific steps together.^33–35 Cosubstrate specificity in this multi-enzyme reaction was also assessed (Figure 3B). At equimolar concentrations (0.4 mM of both 2 and the reducing cosubstrate), NADH and NADPH led to similar levels of 3. In contrast, at higher reducing cosubstrate concentrations (0.4 mM 2 and 2 or 20 mM reducing cosubstrate), NADH led to two- to fourfold increased production of 3 as compared to NADPH. It is also important to note that product 3 was unstable under the reaction conditions (Figures S7 and S8). After ~75 min of the TerB/C-catalyzed reaction, decomposition of 3 was detectable and this product instability was independent of TerB/C and/or reducing cosubstrate concentrations (Figure S8). Finally, reaction reversibility was evaluated through incubation of pure 3 (0.4 mM) in the presence of the oxidized cosubstrate [2 mM NAD or nicotinamide adenine dinucleotide phosphate (NADP)] and catalysts (TerB and TerC) which led to minimal turnover (Figure S9).

Figure 3.

Conversion of PPA (2) to 3 under different conditions. (A) HPLC chromatograms of reactions performed in Tris 50 mM pH 8.0 containing 0.4 mM 2 and different combinations of 10 mM NADH and/or 6.6 μM of each of TerB and/or TerC as following: (i) NADH, (ii) NADH/TerB, (iii) NADH/TerC, (iv) TerB/TerC, (v) NADH/TerB/TerC, and (vi) NADH/TerB/TerC spiked with 3; conversion percentage of 2 to 3 in the presence of increasing concentrations of (B) 0.4, 2, and 20 mM NADH and NADPH in the presence of wild-type (wt) TerB and wt TerC; and (C) 2 and 20 mM NADH and NADPH in the presence of wt TerB and wt TerC, TerC_R143A, TerC_H159A, or TerC_D173A. The reactions were performed in Tris 50 mM pH 8.0 containing 0.4 mM PPA, 6.6 μM wt TerB, and 6.6 μM wt or mutant TerC. All samples were incubated at 30 °C for 60 min.

TerB and TerC X-ray Crystal Structures.

To gain additional mechanistic insights, the X-ray crystal structures of apo-TerB (PDB ID 6D2V), ternary structure-TerB (PDB ID 6ND7), apo-TerC (PDB ID 6D34), and binary-TerC (PDB ID 6WF4) were determined. The apo and ternary (PPA (2)/NADP) TerB structures were solved to 1.9 and 1.35 Å, respectively. The apo structure revealed bound NADP carried with the protein during purification and crystallization (Figures 4 and S10 and Table S2). NADP in the binding pocket is also most likely oxidized over the course of expression and purification and would not supply an active hydride nucleophile once co-crystallized with 2 leading to stability of the aromatic substrates in the binding cleft. NADP seems to be in the same orientation in both apo and ternary structures. The 2′-phosphate group of NADP ribose seems to form electrostatic interactions with the sidechains of Thr12 and Arg34 and the main chain nitrogen of Arg34 and Phe35 (Figure S11). These interactions may slow NADP release after catalysis, consistent with the higher turnover with NADH versus NADPH (Figure 3B), suggesting that multiple NAD(P)H are required for the conversion of 2 to 3.³⁶ Other TerB-nicotinamide interactions include Gln14, Val15, Val79, Arg81, Arg81, and Lys149. The quinone carbon atoms 2′/5′ and 3′/6′ of PPA are 2.6–2.7 Å away from the 4-hydride of NAD(P)H pyridine. The quinone’s oxygen atoms also make electrostatic interactions with Arg8, Thr116, Tyr145, and Arg268. The 2 in the ternary complex does not seem to be turned over efficiently due to the presence of the oxidized form of the insufficient amount of NADP.

Figure 4.

X-ray crystal structures of substrate-bound TerB and TerC. (A) Overall structure of TerB (cyan) with 2 and NADPH shown as sticks in their binding sites. Residues that interact with the ligands shown as sticks. (B) 2fo–fc Density for the 2 and NADPH ligands shown at 2RMSD. Both ligands are very well resolved and the phosphate group differentiating NADPH from NADH is clearly present. (C) Active site and surrounding residues of interest. Compound 2 is recruited by W141, Y181, and Y188 with R268 latching above 2. (D) TerC jelly-roll fold and residue sidechains exposed to the binding site at the center of the protein (magenta) are shown around the substrate species of 2 (gray) and the interesting water molecule (red sphere). (E) TerC active site showing distances between the conserved Asp and Arg residues, water molecule, His nitrogen atom, and 2. (F) 2fo-fc Density of the active site at 1RMSD.

The apo and binary (PPA, 2) TerC structures were solved to 2.1 and 1.97 Å, respectively (Figure 4). TerC reflects an NTF-2 like jellyroll fold common in monomeric enzymes of its size.³⁷ The active site is lined with a combination of aromatic and four methionine residues, capable of stabilizing the binding of compounds structurally similar to 2.³⁸ Densities of the two terminal aromatic rings of 2 are clearly present and co-planar with each other. However, the density of the central ring of 2 is not planar, with density both above and below where the quinone ring would be, indicating the presence of enzymatic activity in TerC. TerC-observed molecular interactions with 2 included Arg143, His159, and Asp173 (Figure 4), potentially implicating these residues in catalysis. Both TerB and TerC lack interfaces that fit sterically and do not have compatible charges that would enable the formation of a physical complex^39,40 consistent with our in vitro assays.

TerC Mutagenic Studies.

To elucidate the role of TerC Arg143, His159, and Asp173 (Figure 4), three targeted variants (TerC R143A, H159A, and D173A) were created (Table S3). The corresponding N-His₆-TerC mutants were overproduced in E. coli and purified to homogeneity. Circular dichroism confirmed proper folding of the purified TerC mutants compared to the wild-type (Figure S12). In vitro reactions in the presence of 6.6 μM TerB, 0.4 mM 2, 2 or 20 mM NAD(P)H, and 6.6 μM of each of the TerC variant were compared to reactions containing wild-type TerC (Figure 3C). Based on this comparative assessment, mutagenesis of TerC H159 had no impact on catalysis under optimized conditions (6.6 μM TerB, 0.4 mM 2, and 20 mM NADH). In contrast, mutagenesis of TerC R143 led to ~50% reduction in turnover and mutagenesis of TerC D173 led to nearly complete inactivation under the same reaction conditions. These data implicate Asp173 as a primary contributor to TerC catalysis with a significant contribution from Arg143.

To test the ability of the TerC mutants in the Ter B/C system to catalyze the conversion of 3 to 2, 3 (0.4 mM) was incubated in the presence of 2 mM NAD(P) in the presence of TerB/C (Figure S13). Similar to the forward reaction, TerC_D173A had the most detrimental effect on activity followed by TerC_R143A with minimal effect with TerC_H159A.

CONCLUSIONS

Functional annotation of biosynthetic genes leads to the discovery of new enzymatic-catalyzed chemistry, new bio-catalysts, and bioactive compounds.^41–45 In this study, we characterized a poorly understood essential step in the biosynthesis of p-terphenyls and revealed how an unprecedented dual reductase-dehydratase system enables the formation of 3 from 2 prior to downstream methylation and/or glycosylation tailoring reactions. These biosynthetic reactions are central to terphenyl and bis-(indolyl)quinone formation, which are microbial compounds with diverse functions.

It is clear from the in vitro reactions (Figure 3A) and the apo- and substrate-bound TerB/C crystal structures (Figure 4) that the two enzymes possess distinct catalytic functions and that they do not interact structurally. The TerB/C dual system is an example of two enzymes that function together but do not form an observable complex. Similar systems have been shown before to catalyze an individual biosynthetic step in the biosynthesis of natural products.^46–49 For example, the LazDE and LazBF pair systems work as a cyclodehydratase and dehydratase, respectively, in the biosynthesis of lactazole A.³⁵ MtmGIV/C work together to install the mithramycin trisaccharide chain,⁴⁷ while SqnG1/G2 glycosyltransferases were found to be essential for the efficient production of saquayamycins.⁴⁶ The two enzymes SipS4/S15 catalyze the transfer of d-xylosamine in the biosynthesis of sipanmycin,⁴⁹ while both TylM2/M3³³ are required for the transfer of mycaminose in tylocin. The previous examples, along with our TerB/C reported herein, could represent the tip of the iceberg and suggest that others systems exist by cooperating to catalyze a specific function in natural product biosynthesis.

Based on our functional and structural analyses, we propose that TerB catalyzes the transfer of two hydrides from NAD(P) H to 2 forming [1,1′:4′,1″-terphenyl]-2′,3′,5′,6′-tetraol (Figures 5A, 2a). Consistent with this, the quinone ring of 2 is observed in the TerB electron density (Figure 4). Compound 2a exists in resonance with 3′,5′,6′-trihydroxy-[1,1′:4′,1″-terphenyl]-2′-one (Figures 5A, 2b). TerB catalyzes a second round of NAD(P)H-mediated reduction of 2b to form 2′,3′-dihydro-[1,1′:4′,1″-terphenyl]-2′,3′,5′,6′-tetraol (Figures 5A, 2c). Thus, TerB catalyzes the transfer of four hydride equivalents from NAD(P)H to 2 which explains the need of the reaction for excess NAD(P)H (Figure 5A). The preference of NADH over NADPH at concentrations ≥2 mM (Figure 3B) can be explained by the tight binding of the phosphate moiety to the TerB active site making it difficult for the oxidized nicotinamide to be replaced by fresh NADPH. A limited amount of 3 formed in the presence of TerB alone could be explained by spontaneous dehydration of 2c (Figure 3A). On the other hand, we show that TerC acts as a dehydratase that leads to a ninefold increase in the formation of 3 when present. TerC alone has no effect on 2 or 3, suggesting that its native substrate is another compound. TerC is a homolog to another NTF-2 member, limonene-1,2-epoxide hydrolase (LEH).⁵⁰ LEH catalyzes the hydrolysis of an aromatic epoxide, styrene oxide, to form a dihydroxy substrate using an Arg-Asp-H₂O triad. Indeed, our structure analysis suggests key roles for TerC Arg143 and Asp173 (Figure 4) that were confirmed by mutagenesis (Figure 3C). Thus, we propose a general acid-base mechanism that involves Arg143 and Asp173 side chain participation (Figure 5B). In this putative mechanism, Asp173 abstracts a proton from the hydroxy group at C-3′ of 2c forming an oxyanion at C-3′ which subsequently attacks 2d electrophilic C-2′. The hydroxyl group at 2d C-2′ is then protonated by Arg143 and eliminated as water generating a C-2′–3′ epoxide intermediate (2e). The epoxide is rearranged to restore aromaticity and subsequent formation of 3. This proposal is also consistent with the TerC-2 co-crystal electron density observed in the binding pocket. Compound 3 is unstable and readily oxidizes to give the corresponding quinone unless it undergoes downstream methylation or glycosylation. The role of TerC is another example on how nature found a way to facilitate difficult chemistry.

Figure 5.

Proposed mechanism for roles of TerB (A) and TerC (B) in the conversion of 2 to 3. B denotes Arg143 or Asp173 side chain participation. Based on the determined TerC structure, some steps may be H₂O-mediated.

In short, we characterized the functions and protein structures of TerB and TerC and their key role in the biosynthesis of an important class of biologically active natural products. It seems that the presence of TerC in gene clusters determines whether the dimerization of 1 and other α-keto acids will lead to the formation of hydroquinone or the triol intermediate. This could set the stage as a key biomarker for the annotation of cryptic BGCs. Our work sets the stage to advance the study of tailoring enzymes that work together for the formation of unstable intermediates of natural products. The findings of this work can also be extended to other dimer natural products arising from keto acids such as the bis(indolyl)quinone molecules for the combinatorial biosynthesis of aromatic triol derivatives.

METHODS

Strains and Materials.

The genetic DNA of Streptomyces sp. RM-5–8 was previously isolated and sequenced.²⁰ All primers were purchased from Integrated DNA Technologies, and E. coli 5α and BL21(DE3) competent cells were purchased from New England Biolabs. All DNA sequencing was conducted with the primers T7 promoter and T7 terminator (Table S3). Crystallization screens Index HT, PEGRx HT, Crystal Screen HT, and SaltRx HT were obtained from Hampton Research. Rigaku Wizard Classic screens were purchased from Molecular Dimensions. N-(2-hydroxyethyl)-piperazine-N′-ethanesulfonic acid (HEPES) buffer was purchased from Gold Biotechnology in a sodium salt form. All other reagents and chemicals were purchased from Sigma-Aldrich or Fisher Scientific and were used without further purification unless otherwise stated. PD-10 columns and Ni-NTA superflow columns were purchased from GE Healthcare. All solvents used were of ACS grade and purchased from Fisher Scientific or Pharmco-AAPER.

DNA Extraction, Genome Sequencing, and Analyses.

Cell growth, DNA extraction, and whole genome sequencing were reported previously.²⁰ Homology searches of the generated contigs were carried out using BLASTX and Position-Specific Iterated BLAST (PSI-BLAST). Gene alignments and analyses were performed using Geneious Pro 11.1.5.⁵¹ The ter BGC and each ORF were manually checked using the Basic Local Alignment Search Tool (BLAST). Low-quality regions were sequence verified by the polymerase chain reaction and Sanger sequencing. Putative gene functions were identified via BLAST. Homology searches of the generated contigs were carried out using BLASTX and PSI-BLAST. The sequence of the putative ter BGC has been deposited under GenBank accession number MN877931.

Gene Synthesis and Protein Overproduction and Purification.

The terB and terC genes were obtained from synthetic nucleic acids (GenScript). Both genes were digested using NdeI and HindIII restriction enzymes (New England Biolabs). The corresponding fragments were cloned into the E. coli expression vector pET28a (Novagen) and confirmed via sequencing. The validated expression plasmids (pSEterB and pSEterC) were subsequently transformed into E. coli BL21 (DE3) competent cells (New England Biolabs). Production strains for terC mutants were constructed in a similar fashion. All studies employed the corresponding N-terminal-His₆ fusion proteins.

For protein production, 1L of LB broth (Becton, Dickinson and Company) supplemented with kanamycin (70 μg mL⁻¹) was inoculated with 0.1% (v/v) of overnight pSETerB-E. coli and pSETerC-E. coli in the BL21 (DE3) seed culture and grown at 37 °C with shaking (250 rpm). Cultures were induced at OD₆₀₀ of ~0.6–0.8 with isopropyl-β-d-thiogalactopyranoside (0.5 mM final concentration) and allowed to grow for an additional 16 h at 21 °C. Cells were harvested by centrifugation and stored in lysis buffer (10 mM imidazole, 50 mM sodium monobasic phosphate, and 300 mM NaCl, pH 8.0) at −80 °C until used. All subsequent steps were carried out on ice. Cells were allowed to thaw and were subsequently lysed by sonication (Fisherbrand model 705 Sonic Dismembrator with a microtip, 700 W, 2 × 20 s pulses, 15 s between pulses). Insoluble debris was removed by centrifugation at 17 000g for 1 h. The supernatant was collected and filtered using 0.22 μm filters, and the desired N-His₆-TerB and N-His₆-TerC fusion proteins were purified via HiTrap nickel-nitrilotriacetic acid (Ni-NTA) affinity chromatography using standard protocols and AKTA Purifier 10 (GE Healthcare) or the Bio-Rad NGS chromatography system. Buffer exchange of each sample was performed using a PD-10 column (GE Healthcare) eluted with 50 mM Tris and 100 mM NaCl, pH 8.0, to yield 11 and 13 mg L⁻¹ of TerB and TerC, respectively. Fractions were collected and concentrated using Amicon Ultra Centrifuge columns 10 000 MWCO (EMD Millipore) and stored in 50 mM Tris and 100 mM NaCl, pH 8.0, at −80 °C. Protein concentrations were determined by Bradford assay (Bio-Rad) using bovine serum albumin as a standard. Purity and presence of proteins were confirmed by SDS-PAGE gel electrophoresis. The N-His₆-TerB and N-His₆-TerC fusion proteins (referred to as TerB and TerC) were used for all studies. Production of TerC mutants for the overall study followed the same protocol.

Analytical Assays.

Standard in vitro assays were conducted in 1.5 mL tubes in a volume of 100 μL of Tris 50 mM (pH 8.0) containing a final concentration of 0.4 mM 2 and 10 mM NAD(P)H. After preincubation of the reaction mixture at 30 °C for 10 min, the reactions were initiated with the addition of 6.6 μM of each enzyme and allowed to proceed for 60 min (unless otherwise noted) at 30 °C. Reversible reactions were performed in 100 μL of Tris 50 mM (pH 8.0) containing a final concentration of 0.4 mM 3, 10 mM NAD(P) and 6.6 μM of each enzyme. Reactions were quenched by the addition of 100 μL of MeOH and mixing followed by centrifugation (22 000g, 15 min, 4 °C) to remove precipitated proteins. The supernatants were analyzed by high-performance liquid chromatography (HPLC) using method A (see General methods) to calculate the conversion rate based on the area of the substrate and the prenylated product peaks.

Crystallization, Diffraction Data Collection, and Structure Determination of TerB and TerC.

Both enzymes were purified as above and concentrations were adjusted to 10 mg mL⁻¹ in 50 mM HEPES and 50 mM NaCl, pH 7.8. All crystal screening was carried out in 96-well plates made with a Mosquito crystallization robot and Hampton HR series or Rigaku Wizard Classic crystallization screens. Rod crystals of TerC apo were found in Rigaku Wizard Classic 3&4 well H10: 30% v/v 2-propanol, 100 mM Tris/HCl pH 8.5, 30% w/v PEG 3350. These screening hits were optimized to a final condition of 25% isopropyl alcohol, 50 mM sodium chloride, and 100 mM HEPES buffer pH 7.5, 30% PEG 3350. Rhombohedra TerB apo crystals were found in HR2–134 well H11 and were not reproducible (HR2–134 H11 well condition: 0.1 M potassium thiocyanate, 30% w/v PEG MME 2,000). Protein co-crystallization with PPA was carried out by mixing 2:1 protein solution with 3 mM PPA in an isopropanol vehicle and screening via Mosquito as before. Rod crystals of TerC and PPA that were similar to apo formed readily but only exhibited 2D diffraction when screened. Cube crystals were formed in HR2–086 well F9, with condition: 20% v/v 2-propanol, 0.1 M MES pH 6.0, 20% w/v PEG MME 2,000. TerB, and PPA crystals grew readily in multiple conditions as octahedrons up to 300 μm in length. The final optimized condition of TerB PPA crystals was 25% PEG 3350, 200 mM NaCl, 100 mM HEPES pH 7.5. TerC crystals did not require further cryoprotection, and TerB crystals were cryoprotected via 20% glycerol, in some cases spiked with 2.0 M potassium iodide for SAD phasing. Diffraction data for TerC apo, TerC PPA, and TerB apo were collected at sector 21 LS-CAT at Argonne National Laboratory. TerB PPA data were collected at sector 23 GM/CA-CAT at Argonne National Laboratory. On beam-line data processing was performed using iMosflm.⁵² Final data processing was performed in Dials or XDS for all structures.^53,54 The TerC apo structure was solved in Phenix using 3ebt as the search model for molecular replacement.⁵⁵ TerC with PPA was solved in Phenix using the TerC apo structure as the search model for molecular replacement. TerB PPA was solved using heavy atom SAD data from iodide-soaked crystals collected at λ = 1.319 Å and a native dataset. Anomalous data were handled in SHELXE and used to make the anomalous map.⁵⁶ TerB apo was found to be the same unit cell as TerB PPA, so phases from the TerB PPA structure were used to build the model with arp_warp after rigid body refinement.⁵⁷ In all cases, Coot and Phenix.refine were used for refinement and hands-on model building.⁵⁸ Final structural models were analyzed and figures were produced in PyMol (Schrodinger).

Mutagenesis of TerC.

Point mutations were generated by PCR and the Q5 Site Directed Mutagenesis kit (New England Biolabs) using the appropriate primer pair (Table S3, engineered codons are underlined) and pSEterC as the template. PCR for each reaction included 25 ng template DNA, 1× of Q5 Hot Start High Fidelity master mix (New England Biolabs), and 0.5 μM of each primer. The PCR program included an initial hold at 98 °C for 30 s followed by 25 cycles of 98 °C for 10 s, annealing temperature ranging from 56 to 68 °C for 20 s, 72 °C for 2 min 20 s, and 1 cycle of 72 °C for 3 min. All amplicons were confirmed and quantified using gel electrophoresis and a ND-1000 Spectrophotometer (Thermo). This was followed by kinase, ligase & DpnI (KLD) treatment according to the manufacturer’s protocol. The mutated plasmids were transformed to E. coli DH5α chemically competent cells (New England Biolabs) and plated on LB agar supplemented with kanamycin 70 μg mL⁻¹overnight at 37 °C. Single colonies were used to inoculate 5 mL cultures of LB broth supplemented with kanamycin and incubated for 16 h at 37 °C and 225 rpm. Plasmid DNA was isolated from each culture using the QIAprep Spin Miniprep kit (Qiagen) according to the manufacturer’s instructions. The isolated plasmid DNA for each desired mutant was confirmed by sequencing and subsequently transformed into chemically competent E. coli BL21 (DE3) (New England Biolabs) according to the manufacturer’s instructions, plated on LB agar supplemented with kanamycin, and incubated at 37 °C. Isolated single colonies were used to inoculate LB medium broth supplemented with kanamycin and incubated overnight at 37 °C and 225 rpm. A glycerol stock [with a final concentration of 15% glycerol (v/v)] of each culture was prepared and frozen at −80 °C. Protein overproduction and purification followed the same protocol as that for wild-type proteins (described in the previous section). To confirm proper folding of the purified proteins, the CD spectra of the wild-type TerC and three mutants in their substrate-free state were measured using a Jasco J1500 spectropolarimeter at 25 °C using quartz cells (Alpha Nanotech) with a pathlength of 0.1 cm. The enzymes were dissolved at a concentration of approximately 200 μg mL⁻¹ in 10 mM sodium phosphate buffer and 100 mM sodium fluoride. Low noise CD spectra were measured by averaging 5 scans, and the final spectra were corrected by subtracting the corresponding baseline with the buffer..