Molecular Basis for the Substrate Promiscuity of Isopentenyl Phosphate Kinase from Candidatus methanomethylophilus alvus

Abstract

Isopentenyl phosphate kinases (IPKs) catalyze the ATP-dependent phosphorylation of isopentenyl monophosphate (IP) to isopentenyl diphosphate (IPP) in the alternate mevalonate pathways of the archaea and plant cytoplasm. In recent years, IPKs have also been employed in artificial biosynthetic pathways called “(iso) prenol pathways” that utilize promiscuous kinases to sequentially phosphorylate (iso) prenol and generate the isoprenoid precursors IPP and dimethylallyl diphosphate (DMAPP). Furthermore, IPKs have garnered attention for their impressive substrate promiscuity toward non-natural alkyl-monophosphates (alkyl-Ps), which has prompted their utilization as biocatalysts for the generation of novel isoprenoids. However, none of the IPK crystal structures currently available contain non-natural substrates, leaving the roles of active-site residues in substrate promiscuity ambiguous. To address this, we present herein the high-resolution crystal structures of an IPK from Candidatus methanomethylophilus alvus (CMA) in the apo form and bound to natural and non-natural substrates. Additionally, we describe active-site engineering studies leading to enzyme variants with broadened substrate scope, as well as structure determination of two such variants (Ile74Ala and Ile146Ala) bound to non-natural alkyl-Ps. Collectively, our crystallographic studies compare six structures of CMA variants in different ligand-bound forms and highlight contrasting structural dynamics of the two substrate-binding sites. Furthermore, the structural and mutational studies confirm a novel role of the highly conserved DVTGG motif in catalysis, both in CMA and in IPKs at large. As such, the current study provides a molecular basis for the substrate-binding modes and catalytic performance of CMA toward the goal of developing IPKs into useful biocatalysts.

Graphical Abstract

INTRODUCTION

Isoprenoids are a large class of natural products that display a diverse range of biological activities (antimicrobial, antiviral, antiparasitic, and anticancer) along with various utilities as biofuels, commercial feedstocks, and food and cosmetic additives.^1–3 Their variety in structure and function, however, often obscures the fact that all natural isoprenoids are constructed from only two precursors (isopentenyl diphosphate, IPP and dimethylallyl diphosphate, DMAPP), which are utilized by various prenyltransferases to either decorate existing scaffolds or form longer-carbon-chain prenyl diphosphates.⁴ The elongated chains can then be cyclized either before or after transfer onto a variety of preexisting scaffolds, most of which are amenable to additional modification by downstream enzymes.^5–7

To access these versatile precursors, cells utilize one of two highly conserved biosynthetic pathways: the methylerythritol phosphate (MEP) pathway or the mevalonate (MVA) pathway.⁴ The MVA route, utilized mostly by eukarya and archaea, begins with the condensation of three units of acetyl-CoA in two steps, followed by reductive elimination of the CoA to form MVA. The primary alcohol of MVA is then phosphorylated by mevalonate-5-kinase (M5K), and the resulting MVA-5-phosphate undergoes a combination of diphosphorylation and decarboxylation to form IPP. Based on the specific order of the final two steps, an organism’s MVA pathway is further classified into “classical” and “alternate” routes (Figure 1). In most eukaryotes, the diphosphorylation of MVA-5-phosphate happens before the decarboxylation as per the original proposal and is thus called the “classical” MVA pathway (solid box in Figure 1).^8–16 Contrarily, in the archaea and plant cytoplasm, the decarboxylation of MVA-5-phosphate to isopentenyl monophosphate (IP) occurs first, followed by phosphorylation of IP to IPP (“alternate” MVA pathway, dashed box in Figure 1).^8–10,17 The latter bifurcation is a recent development in our understanding of isoprenoid metabolism, and its identification was catalyzed by the discovery of the isopentenyl phosphate kinase (IPK, Figure 1) from Methanocaldococcus jannaschii (MJ), which performs the final phosphorylation step of the alternate MVA pathway.⁸ At the time, homologs of the final two enzymes in the classical pathway had yet to be discovered in archaeal genomes, so the identification of IPK activity in MJ led the authors to hypothesize that the organism (and archaea at large) performs these steps in reverse.⁸ Subsequent studies confirm the existence of the missing decarboxylase enzyme (MVA-5-phosphate decarboxylase) in archaea, and a handful even propose offshoots of the alternate pathway that still culminate in the IPK-catalyzed reaction.^{11–13,15,16}

Figure 1.

Mevalonate pathways found in nature. The bifurcation of the general scheme occurs in the final two steps, where the diphosphorylation occurs either before (classical, solid box) or after the decarboxylation (alternate, dashed box). Abbreviations: CoA = coenzyme A, AAT = acetoacetyl-CoA thiolase, HMGS = 3-hydroxy-3-methyl-glutaryl-CoA synthase, HMGR = 3-hydroxy-3-methyl-glutaryl-CoA reductase, M5K = mevalonate 5-kinase, PHMK = phosphomevalonate kinase, PMD = phosphomevalonate decarboxylase, DPMD = diphosphomevalonate decarboxylase, IPK = isopentenyl phosphate kinase, and IDI = isopentenyl diphosphate isomerase.

Given the centrality of IPP and DMAPP to isoprenoid synthesis, it was unsurprising that other studies have sought to optimize these natural biosynthetic routes for increased isoprenoid production.^18–24 However, a combination of factors stemming from the inherent complexity of the chemistry and its regulation within cells have made recreating the natural pathways in bioreactors a challenging endeavor. To overcome these barriers, several groups have opted to simplify the process by supplying the readily available isoprene alcohols (isopentenyl alcohol, IOH and dimethylallyl alcohol, DMAOH) for phosphorylation by promiscuous kinases.^25–30

The resulting chemoenzymatic platforms, known as “(iso)prenol pathways,” have all converged to a single scheme where IOH and DMAOH are first phosphorylated by an alcohol kinase (acid phosphatase PhoN from Shigella flexneri,²⁶ hydroxyethylthiazole kinase ThiM from Escherichia coli,^28,29 and choline kinase from Saccharomyces cerevisiae^25,27) to the corresponding IP and dimethylallyl monophosphate (DMAP), respectively. These compounds then act as substrates in situ for an IPK that phosphorylates them to IPP and DMAPP. The simplicity of these systems, combined with their encouraging product titers, have indicated that they are poised to revolutionize industrial isoprenoid synthesis with proper optimization. Importantly, the commonality of IPKs among the various (iso)prenol pathways clearly highlights the importance of the enzyme class as biocatalysts for isoprenoid synthesis.

IPKs themselves are Mg²⁺-dependent enzymes known to be highly efficient catalysts with their natural substrates (k_cat/K_M ~ 10⁶ M⁻¹ s⁻¹).^8,12,31–35 A handful of studies also document the enzymes’ substrate specificity toward non-natural alkyl-monophosphates (alkyl-Ps),^31,35,36 which is further augmented by engineering studies producing IPK variants capable of diphosphorylating the long-chain isoprene units geranyl and farnesyl monophosphate (GP and FP, respectively).^33,37 These studies confirm that the generalized enzyme class is highly promiscuous toward monophosphorylated substrates, including those bearing chemoselective functionalities (alkynes, azides, and terminal alkenes).^35,36 Furthermore, subsequent coupling reactions with aromatic prenyltransferases (PTs)³⁵ and terpene cyclases³³ demonstrate that this promiscuity can be utilized in platforms for the biocatalytic synthesis of natural and non-natural isoprenoids, which could have major implications for drug discovery and other medicinally relevant syntheses.

Currently, X-ray-crystallographic structures of three different IPK homologs [MJ, Thermoplasma acidophilum (THA), and Methanothermobacter thermautotrophicus (MTH)] are available in the Protein Data Bank (PDB).^33,38 Based on these structures, IPKs are classified under the amino acid kinase (AAK) superfamily alongside N-acetyl-l-glutamate kinase (NAGK), aspartokinase (AK), glutamate kinase (GK), carbamate kinase (CBK), uridine monophosphate kinase (UMPK), and fosfomycin resistance kinase (FK), and they exist as homodimers with subunits composed of two roughly symmetrical domains.^40–45 Examination of ternary complexes with both substrates (IP and ATP) and products (IPP and ADP) reveal the active site to lie between the C- and N-terminal domains responsible for binding the phosphoryl donor (ATP) and acceptor (IP), respectively.^33,38 Furthermore, the structural studies published thus far present a useful depiction of the catalytic roles of various active site residues. However, current knowledge on the molecular recognition of non-natural substrates is limited by the rather low number of IPK structures available and the nonexistence of their complexes with non-natural substrates, and mutational studies of the alkyl-monophosphate binding pocket conducted thus far are focused only on improving efficiency with GP and FP.^33,37 Therefore, it is important to study the structure–activity relationships between IPKs and their non-natural substrates to truly capitalize on the enzymes’ utility for non-natural precursor synthesis.

Within this context, we present herein structural analyses of an IPK homolog from the archaeon Candidatus methanomethylophilus alvus (CMA). Specifically, we examine three different crystal structures of the wild-type (WT) CMA: the apo form, a ternary complex bound to the natural substrate pair (IP and ATP), and a ternary complex bound to the non-natural substrate benzyl monophosphate (BP)³⁵ and ADP. We further identify a set of three residues in the IP-binding pocket responsible for interacting with the carbon chains of the alkyl-Ps and perform mutational studies to understand their effects on CMA’s substrate profile. Using a library of synthetic alkyl-monophosphate (alkyl-P) analogues, we identify two mutations (Ile74Ala and Ile146Ala) that result in increased promiscuity of the CMA variants, particularly toward aliphatic alkyl-P analogues with carbon chains 3–4 atoms longer than IP or DMAP. Kinetic analysis of the Ile74Ala and Ile146Ala enzyme variants further reveal opposite preferences for alkyl-P analogues bearing either rigid or flexible carbon tails (respectively). Subsequent crystallographic studies produce three additional ternary complexes of the CMA mutants each bound to a unique non-natural alkyl-P substrate, and comparison of the active sites between the mutant structures implies that their opposite substrate preferences arise from a differential distribution of the extra space in the binding pocket caused by the Ile-to-Ala mutation. Additionally, collective analysis of six CMA structures highlights a conformational change in the ATP-binding site that may provide new insight into the enzyme’s catalytic mechanism. Thus, the current study advances our understanding of the structural dynamics of the IPKs’ ATP-binding site while also acting as a first step toward optimizing them as biocatalysts for the synthesis of non-natural isoprenoids.

RESULTS AND DISCUSSION

CMA Crystallization and Structure Solution.

Previous studies involving five novel IPK homologs demonstrated a generalized promiscuity toward a variety of non-natural alkyl-P analogues.³⁵ Therefore, with the goal of understanding the molecular determinants of the IPKs’ substrate profiles, we pursued structural studies of CMA, which displays promiscuous activity with several non-natural alkyl-P analogues.³⁵ These efforts have yielded three crystal structures of WT CMA: the apo form, a ternary complex bound to the natural substrates/products IP/IPP and ATP/ADP, and a ternary complex bound to the non-natural phosphoryl acceptor BP and ADP. Structures were solved by molecular replacement using a model generated from either the MTH structure bound to ADP (3LL9, 35% sequence identity with CMA)³⁸ or the WT CMA apo structure. Refinement was performed to R_free values of 23–27% against data sets with a resolution of 2.2–2.5 Å. Crystallographic statistics are further summarized in Table 1.

Table 1.

Data Collection and Refinement Statistics for Crystal Structures of WT CMA

PDB ID	7LNV	7LNU	7LNT
ligands	Apo	A: IP/IPP, ATP/ADP	BP, ADP
		B: IP, ADP
wavelength	1.54	1.54	1.54
resolution rangea	37.3–2.2 (2.3–2.2)	45.4–2.5 (2.6–2.5)	28.1–2.4 (2.43–2.35)
space group	C121	P212121	P212121
unit cell (a, b, c; Å)	194.39, 45.63, 113.87	40.89, 74.50, 171.69	40.70, 74.66, 170.69
unit cell (α, β, γ; °)	90, 121.2, 90	90, 90, 90	90, 90, 90
total reflections	170,868	70,194	101,941
unique reflectionsa	43,957 (4382)	18,440 (1623)	22,224 (2058)
multiplicitya	3.9 (2.5)	4.2 (2.5)	4.6 (3.0)
completeness (%)a	99.8 (99.0)	97.1 (87.4)	98.8 (93.6)
mean I/σIa	9.91 (1.01)	11.96 (1.90)	16.90 (1.00)
Wilson B factor	29.91	30.42	28.17
R merge a	0.12 (0.75)	0.087 (0.55)	0.097 (0.34)
CC1/2a	0.99 (0.55)	0.99 (0.65)	1.01 (0.87)
reflections used in refinementa	43,940 (4379)	18,357 (1623)	22,222 (2058)
reflections used for Rfreea	2002 (200)	937 (82)	1072 (109)
R work a	0.225 (0.311)	0.229 (0.248)	0.187 (0.224)
R free a	0.280 (0.358)	0.289 (0.296)	0.233 (0.292)
total number of non-hydrogen atoms	5867	4091	4116
non-hydrogen atoms in macromolecules	5610	3808	3817
non-hydrogen atoms in ligands	32	119	90
number of solvent molecules	225	164	209
protein residues	737	502	499
RMSD (bonds)	0.005	0.002	0.003
RMSD (angles)	0.72	0.66	0.83
Ramachandran favored (%)	95.56	96.96	97.37
Ramachandran allowed (%)	4.16	2.63	2.23
Ramachandran outliers (%)	0.28	0.41	0.40

^aValues in parentheses denote the highest resolution shell.

The asymmetric unit of the apo structure contains three copies of the CMA monomer, as well as three malonate anions and two molecules of glycerol. Each monomeric subunit in the apo structure is bound to a malonate at the same position as the terminal phosphate moieties of one of the substrates (α-phosphate of alkyl-P and γ-phosphate of ATP) or products (β-phosphate of alkyl-PP and β-phosphate of ADP) in other ternary complexes. The two glycerol molecules are observed in subunit B: one near the C-terminal loop, and the other in the cleft between the N- and C-terminal domains.

Unlike the apo form, both CMA ternary complexes contain two monomers each in their asymmetric units. The first is cocrystallized with the natural substrates IP and ATP (CMA•IP•ATP), but the two monomers contain differing combinations of reactants and/or products bound in the active site. Interestingly, electron density for both pairs of reagents (IP/ATP and IPP/ADP) can be modeled in the active site of monomer A with approximately equal occupancy (IP/IPP: 0.61/0.41 and ATP/ADP: 0.40/0.34). Moreover, monomer B contains a mixed pair of reactant and product (IP/ADP). The second ternary complex is cocrystallized with the non-natural substrate BP and ATP, but both monomers contain BP and ADP bound in the active site (CMA•BP•ADP). The presence of reactant–product mixed pairs in the solved structures (IP/ADP and BP/ADP) likely results from ATP hydrolysis during crystallization. The CMA•BP•ADP complex also contains two molecules of glycerol from the cryoprotectant solution that bind in the dimer interface.

Overall Fold and Quaternary Structure.

The asymmetric unit of each CMA structure contains two monomers arranged in a homodimeric structure (Figure 2A), with the apo form containing an additional monomer whose dimer interface is parallel with a crystallographic twofold rotation axis. A similar, noncrystallographic axis is observed roughly perpendicular to the central β-sheets of the dimer subunits, with the interface parallel to the monomers’ αC helices (Figure 2A). Each CMA monomer exhibits an open αβα sandwich fold common to the AAKs (Figure 2B),^33,38–44 resulting in an N-terminal domain (residues 1–162) responsible for binding the alkyl-P substrate (IP/IPP or BP) and a C-terminal domain (residues 163–259) binding the nucleotide.

Figure 2.

Overall structure of CMA•IP•ATP. The dimer interface is visualized in (A) with Chain A shown in green and Chain B in red. Monomer B is highlighted in (B) with the secondary structure elements colored red (α helices), yellow (β strands), and green (3₁₀ helices). This color scheme is mirrored in the topology map depicted in (C). The ligands in (A) and (B) are shown in cyan (IP/ADP) and yellow (IPP/ATP).

The core β-sheet is composed of eight strands (β3, β6, β2, β1, β9, β13, β14, and β12) surrounded on either side by bundles of three or four α-helices (αC, αA, αH on one side and αD, αE, αG, αF on the other) (Figure 2B,C). Additional secondary structural elements include three β-hairpins (β4-β5, β7-β8, and β10-β11), three 3₁₀-helices (η1, η2, and η3), and an additional helix (αB) capping the alkyl-binding site in the N-terminal domain.

The dimer interface is formed exclusively through interactions of the N-terminal domains and includes four salt bridges (Arg73-Glu87 between αC helices and Glu118-Arg140 between αD helices and β7-β8 hairpins), six hydrogen bonds (H-bonds; side chains of Cys104 on η3 and backbone carbonyls of Ser98 on β3 to side chains of Ser103 on η3; and side chains of Arg122 on αD to backbone carbonyls of Asp139 on β7), and various hydrophobic contacts involving the monomers’ αC, αD, and η3 helices and β3 strands (Figure 2A). Thus, contribution from all three types of interactions affords the formation of constitutive dimers. Analysis using the PISA tool from the PDBe^45,46 confirms that the dimer interface buries an average of 1585 Ų of solvent accessible area per monomer (~3170 Ų total). Furthermore, each structure is assigned a complexation significance score (CSS): 0.932 for the apo form, 0.637 for CMA•IP•ATP, and 0.657 for CMA•BP•ADP. These values indicate that, in the absence of bound substrates/products, the interface is essential to assembly formation by significantly lowering the free energy of solvation, the contribution of which lessens upon binding the substrates (see Supporting Information Table S3).

Active Site.

As with other IPK structures described thus far,^33,38 the active site of CMA spans the phosphoryl acceptor and donor binding pockets in the N- and C-terminal domains, respectively. The N-terminal domain (residues 1–162) is responsible for binding the alkyl-P substrate (IP or BP) and contains the β7-β8 hairpin, two α-helices (αB and αC), and two 3₁₀-helices (η1 and η2) that create the hydrophobic and H-bonding interactions required to anchor this amphipathic substrate in the binding site (Figure 2B). The secondary structural elements of the pocket largely mirror those observed in previous IPK structures with one major exception in the η2 helix (observed as a β hairpin in all previous structures).^33,38 Additionally, the part of the secondary structure that contains the η2 helix and the αB-η2 and η2-αC loops folds exclusively into the “closed” conformation observed with THA and MTH³⁸ as opposed to the “open” conformation found in MJ.³³

The hydrophobic environment required to sequester the organic portion of the alkyl-P/PP reagents is created by the side chains of Ala45 (β2-αB loop), Ala53 (αB helix), Ile58 (η2 helix), Ala71, Ile74, Met75, Thr78 (αC helix), Val136 (β7 strand), Phe144, and Ile146 (β8 strand) (Figure 3). The side chain orientations of these residues remain largely static between the four monomers of CMA•IP•ATP and CMA•BP•ADP, with only slight perturbations caused by the rotameric differences in residues such as Met75 and Ile146. As such, the binding of IP/IPP and BP in the two ternary complexes (Figure 3) remains largely the same, with the planes of the isopentenyl and phenyl moieties aligning nearly parallel to one another. However, the protrusion of BP’s phenyl ring further into the pocket likely accounts for the higher K_M value previously observed for the WT enzyme’s reaction with BP (0.09 mM) compared to IP (0.036 mM).³⁵

Figure 3.

Active sites of CMA•IP•ATP and CMA•BP•ADP. The interactions of IP with ATP/ADP across monomers A and B of CMA•IP•ATP are depicted in (A). The carbon atoms of IP are colored yellow/dark blue, the carbon atoms of ATP/ADP are colored blue/green, phosphates are colored green/orange and sky blue/red, water molecules are modeled as light orange/cyan spheres, active site residues are colored gray/lime, and important H-bonds are shown with black/magenta dashed lines (for monomer A/B, respectively). The interactions of IPP and ADP in monomer A of CMA•IP•ATP are depicted in (B). The carbon atoms of IPP and ATP are colored yellow and green, respectively; active site residues are colored gray; and water molecules are modeled as light orange spheres. Important H-bonds are shown with black dashed lines. The active site of CMA•BP•ADP monomer B overlaid with IP is depicted in (C). The carbon atoms of IP, ADP, and BP are colored yellow, green, and magenta (respectively); active-site residues are colored gray; and water molecules are modeled as light orange spheres. Important H-bonds are shown with black dashed lines.

The residues interacting with the phosphate moieties of the alkyl-P are mostly conserved across the IPKs and several exhibit mobility in their side chain orientation across different substrate/product-bound forms (Figure 3). For example, the catalytic residue His50 (αB helix)^33,38 undergoes a significant conformational change in response to substrate binding. In the apo structure, the imidazole side chain is situated away from the alkyl-binding pocket (Figure S1A,B), but in response to substrate/product binding in CMA•IP•ATP and CMA•BP•ADP, it moves closer to the alkyl-binding pocket to create a H-bond between N_ε of the imidazole ring and the terminal phosphates of IP/IPP and BP (Figures 3 and S1A,B).

Interestingly, this His50 imidazole ring rotates almost ~180° in monomer B of CMA•IP•ATP compared to monomer A (Figure 3A) and shifts by ~10° in both monomers of CMA•BP•ADP while maintaining the H-bond with the terminal phosphate of IP or BP, respectively (Figures 3A,C and S1A). The mobility of this residue is further highlighted by the differences in the H-bond partner observed across the four monomers of CMA•IP•ATP and CMA•BP•ADP. In monomers A and B of CMA•IP•ATP (bound to IP/IPP and IP, respectively; Figure 3A,B) and monomer A of CMA•BP•ADP (bound to BP), N_ε interacts with a nonbridging O atom of P_α in IP/BP, while in monomer B of CMA•BP•ADP (bound to BP), it makes an additional interaction with the C1-P_α bridging O atom (Figure 3C). In addition, His50 interacts with a nonbridging O atom of IPP’s P_β in monomer A of CMA•IP•ATP (Figure 3B). Similar interactions are conserved across substrate-bound IPK structures, as well as other AAKs (UMPK, FomA),^33,38,42,44 and the observation of both IP- and IPP-specific interactions of His50 in monomer A of CMA•IP•ATP may imply that this particular structure captured a snapshot of the postphosphorylation phase of the reaction, wherein the products (IPP and ADP) were just about to leave the active site.

The alkyl-substrate/product (IP/IPP and BP) is further anchored in the active site through H-bonding interactions of the terminal phosphates with the backbone amides of Ala45, Gly46 (β2-αB loop), and Gly149 (β8-αE loop), as well as water-mediated interactions with the side chain of Lys5 (β1 strand) and the side chain and backbone amide of Asp150 (β8-αE loop) (Figure 3). Among these, the indirect interaction of Asp150 through a ubiquitous water molecule (WAT1 in Figure 3) is present in most monomers (apo and ternary) and is proposed as a catalytically relevant interaction among the AAK family.^33,38,42,44

The nucleotide (ATP/ADP) is accommodated in the C-terminal domain (residues 163–259) through interactions that are conserved across the AAK family.^33,38,42,44 Specifically, the adenosine moiety binds at the distal end of the pocket surrounded by two α-helices (αF and αG), the β10-β11 hairpin, and the αF-αG loop (Figure 2B). Among these, only the β10-β11 hairpin (residues 176–186) lingering over the adenine can be fully visualized in the nucleotide-bound monomers of the ternary complexes. In monomer A of both CMA•IP•ATP and CMA•BP•ADP, the αF helix, the αF-αG loop, and the N-terminal end of the αG helix (residues 192–215) cannot be modeled based on uninterpretable electron density. While packing contacts do differ between the monomers of these structures, these differences do not appear to contribute to the disorder. The high dynamic motion of this region indicated by the missing density is consistent with similar observations in other IPK structures and its proposed role in nucleotide sequestration and product release.^33,38,42,44

The adenine ring of ATP/ADP is stabilized through H-bonding interactions between the N-atoms of the adenine ring and the backbone carbonyls (Tyr175 and Ala177) and amides (Tyr175) of the β10-β11 hairpin (Figure 3). However, the stacking interaction previously observed between the adenine ring and a hydrophobic side chain on the αG helix (Val212 in CMA and Ile202 in THA) is not observed in any monomers of the WT complexes. In fact, Val212 is located ~8 Å away from the adenine ring in both, and the proposed hydrophobic stacking is instead replaced by a π-cation interaction with the side chain of Lys215 (αG helix, Figure 3). The new interaction is facilitated by a nucleotide-binding-induced rotation of the central axes of the αF and αG helices (~10 and ~20°, respectively) compared to the apo form. This is quite interesting given Lys215’s proposed role in stabilizing the phosphate moieties of the nucleotide as a member of the “Lys triangle” (alongside Lys5 and Lys14) observed in other IPK structures.^33,38

At the proximal end of the nucleotide-binding pocket, the ribose is stabilized through multiple H-bonds to its hydroxyl groups. Specifically, the side chain of Asp178 (β10-β11 hairpin) serves as both a direct H-bonding partner with the ribose C2′-OH and as an intermediary between this hydroxyl and the backbone amide of Lys180 (β10-β11 hairpin; Figure 3). The side chain of Lys180 also H-bonds directly to the C3′-OH alongside the side chain of Asp170 (β9-β10 loop; analogous to Asp181 in THA).³⁸ Further into the cleft, the phosphate moieties of ATP/ADP form an extensive network of H-bonds and electrostatic interactions to stabilize their negative charges (Figure 3). Both Gly8 and Ser9 of the β1-η1 loop (analogous to the β1-αA loop in other IPK structures)^33,38 provide H-bonds through their backbone amides largely to the P_β-O atoms of ATP/ADP, but Gly8 also contacts a P_γ-O of ATP in monomer A of CMA•IP•ATP. Additionally, the side chains of Ser9 and Ser169 (β9 strand) form H-bonds with the P_β-O atoms (both bridging and nonbridging) (Figure 3).

The other two invariant members of the Lys triangle (Lys5 on the β1 strand and Lys14 on the η1-αA loop) are involved in phosphate stabilization and display mobility in their side chain orientations across the CMA structures. Lys5 interacts primarily through its side chain amine with the nonbridging P_β-O atoms of ATP and ADP, although it also seems to exert considerable influence over the terminal phosphates of both substrates (IP/ATP) through its H-bond with WAT1 (Figure 3). In monomer A of CMA•IP•ATP (bound to IP/IPP and ATP/ADP), WAT1 is observed bridging the free O atoms of P_γ in ATP and P_α in IP, implying Lys5 may play a role in positioning the substrates for catalysis by also interacting with WAT1. In contrast, Lys14 in monomer A of both ternary complexes makes H-bonds with the phosphate moieties of ATP (nonbridging P_γ-O) or ADP (nonbridging P_α-O or P_β-O), but in monomer B of both complexes (Figure 3A,C), the side chain sways away from the active site cleft altogether as observed for the apo structure (see Supporting Information Figure S1). This is partly due to uninterpretable electron density of Lys14’s side chain in these monomers, indicating the importance of the mobility of this residue in stabilizing the pentacoordinate transition state as noted for other IPKs.^33,38

The monomers housing substrate/product mixed pairs (IP/ADP or BP/ADP) also contain two water molecules that appear to mimic ATP’s P_γ and one of its nonbridging O atoms (Figure 3C). Together, these waters mediate an intricate web of indirect H-bonds between the terminal phosphates of IP/BP and ADP and the surrounding residues. Finally, while IPK reactions are dependent on the presence of Mg²⁺ and both ligand-bound structures were crystallized with this ion, it cannot be unambiguously identified in either complex.

Mutational Studies.

Based on our analysis of the alkyl-binding pockets in CMA•IP•ATP and CMA•BP•ADP, we selected bulky residues in contact with the alkyl groups of IP and BP (Ile74, Val136, and Ile146) for Ala-substitution studies, the goal of which was to understand the effects of creating more space in the active site on the promiscuity of CMA toward alkyl-P analogues. Prior studies of IPKs found that mutating analogous residues to Ala resulted in new activities with the longer-chain prenyl analogues GP and FP.^33,37 Furthermore, sequence alignment of known IPK sequences (Supporting Information Figure S2) suggested the Ala-substitutions of analogous Val, Ile, and Thr residues were likely to create similar substrate profiles in other IPK homologs, such that our results would likely be generalizable to the class of IPKs as a whole. Therefore, we carried out Ile74Ala, Val136Ala, and Ile146Ala mutations in CMA and purified the recombinant proteins as described previously for the WT enzyme.³⁵ All three enzyme variants were subsequently evaluated for phosphorylating activity using a focused library of alkyl-Ps (Figure 4A) as a primary screen, several of which were previously shown to act as substrates for WT CMA.³⁵

Figure 4.

(A) Library of alkyl-P analogues utilized for the current study. (B) Results of screening CMA variants against the library of alkyl-P analogues using the PK-LDH assay. Conversions were calculated by measuring the absorbance at 340 nm of each enzymatic reaction just before and 1 h after the addition of the CMA variant at 37 °C and comparing it to a positive control (n = 2). Appropriate controls were conducted to account for any ATPase activity. Each reaction consisted of 2 U PK, 2 U LDH, 0.6 mM NADH, 1.5 mM PEP, 2 mM ATP, and 4 μg of IPK incubated in a buffered solution (25 mM Tris pH 7.8 and 5 mM MgCl₂). All positive reactions were verified using high-resolution mass spectrometry (HRMS). Conversion data for WT CMA were obtained previously.³⁵

The library of alkyl-P analogues was divided into five categories based on structural similarity. Nonallylic alkyl-Ps (1 and 3–10, with 1 being the natural substrate IP) were distinguished by their lack of a C2-C3 double bond and varied in the linear chain length from 3 to 6 atoms beyond the phosphate moiety. These analogues also tested the placement of the double bond at the C3-C4 position (1 and 6), the addition of chemoselective alkynes (3 and 7) and azide (8) groups, and complete saturation of the carbon chain (4, 5, 9, and 10). For the second subset of alkyl-Ps, the allylic DMAP (2) served as a template upon which various alterations were made (11–27). These included the removal of DMAP’s methyl branches (11, 12, and 18), the addition of various saturated hydrocarbons onto DMAP’s C4 and/or C5 (14, 15, 19–21, 23, and 26), the cyclization of C4 and C5 using different-sized carbon linkers (16 and 17), and the introduction of chemoselective functionalities such as alkynes (24 and 25), azides (18 and 27), and a terminal alkene (22). GP (28) served in a role similar to DMAP as a template for the third set of analogues (28–33) in which different groups were added to C8. These included a methyl group (29), a hydroxy group (30), a propargyl ether (31), a benzylic ether (32), and a third prenyl group (33, FP). The fourth subset of alkyl-Ps (34–39) included conjugated systems such as dienes (34–36) and cinnamyl analogues (37–39). The former differed in the inclusion/placement of a seventh carbon as a methyl group, while the latter added either a methyl group (38) or an ethene (39) to the cinnamyl-P template (37). The last subset included benzylic alkyl-Ps (40–46, with 40 being BP) that differed in the choice of the para-substituent. These included a fluorine atom (41), a chlorine atom (42), a methyl group (43), a nitro group (44), a methoxy group (45), and a methylenedioxy group that formed a bicyclic structure by also attaching at the meta-position (46). Additionally, several members of the library were included based on the utility of their diphosphate products in PT-catalyzed alkylation, which themselves are implicated in various biocatalytic applications.^47–51

Each of the three Ala-substituted variants of CMA (Ile74Ala, Val136Ala, and Ile146Ala) was screened against the alkyl-P library using the pyruvate kinase-lactate dehydrogenase (PK-LDH) assay.³⁵ The resulting screening data were plotted alongside what were previously obtained for WT CMA (Figure 4B and Supporting Information Table S4)³⁵ and revealed that 36 out of 46 total substrates were utilized by one or more variants. Specifically, Ile74Ala, Val136Ala, and Ile146Ala phosphorylated 35, 25, and 32 analogues, respectively, compared to 22 analogues accepted by the WT CMA. Among the three variants, Ile74Ala and Ile146Ala displayed substantially broadened substrate scope, while that of Val136Ala was similar to the profile of WT CMA. In addition, Val136Ala produced the lowest activity and promiscuity among the three variants for almost any given analogue (see Supporting Information Table S4). Therefore, we have omitted this variant from further discussion of substrate promiscuity and enzyme kinetics.

Based on the previous data, WT CMA generally favored aliphatic analogues bearing 4–6 atoms in their unbranched alkyl chains (1–10 and 12–19), as well as minimally substituted benzylic analogues (40 and 41). It also exhibited an intolerance toward both long-chain (≥seven nonbranching atoms) and conformationally constrained alkyl groups such as dienes.³⁵ In contrast, both Ile74Ala and Ile146Ala displayed expanded substrate tolerance with moderate (20–50%) to good (>50%) activity with analogues bearing up to eight atoms in their unbranched chains (20–22 and 24–27). In addition, Ile146Ala was able to phosphorylate 34, while a majority of the conjugated analogues (34, 37, and 38) served as substrates for Ile74Ala.

Furthermore, both Ile74Ala and Ile146Ala readily accepted the benzylic analogues with bulkier substituents (43–45) as substrates with moderate to good activities. Ile74Ala was even able to accept the bicyclic 46 with decent conversion (~25%), which was somewhat surprising given the even larger bulk of the methylenedioxy group in 46 compared to the substituents of 43–45. These results indicated that increasing the available space of the binding pocket allowed bulkier alkyl-P analogues to be accepted as substrates for Ile74Ala and Ile146Ala.

Further comparison of the substrate profiles between the Ile74Ala and Ile146Ala variants demonstrated that they both generally accepted the same analogues with slightly different preferences. Specifically, the Ile74Ala variant displayed higher conversions with 3, 20, 25, 37, 38, 43, 45, and 46, while the Ile146Ala mutant preferred 4, 8, 11, 15–19, 21, 24, 34, and 44 (Figure 4B and Supporting Information Table S4). Structural comparison between the alkyl-P analogues implied that both conformational flexibility and three-dimensionality in the alkyl chain played a prominent role in determining the CMA mutants’ substrate profiles. For example, Ile74Ala preferentially phosphorylated rigid, linear alkyl-Ps (37, 38, 43, 45, and 46), while Ile146Ala displayed generally higher conversions with longer, more flexible analogues (8, 18, 19, 21, and 24), as well as the cyclized but nonlinear 15 and 16. Based on this analysis, the variants’ preferences for their substrates seemed to arise from the introduction of additional space at different positions in the pocket relative to the isopentenyl group.

Collectively, these novel activities are relevant to the biocatalytic potential of IPKs for multiple reasons. For example, the two variants’ new or increased ability to phosphorylate analogues bearing chemoselective functionalities (alkynes on 24 and 25, azides on 18 and 27, terminal alkenes on 22, and dienes on 34) suggested that they bear additional utility in non-natural isoprenoid synthesis compared to the WT. Furthermore, longer-chain allylic and substituted benzylic diphosphates are known substrates of PT-catalyzed alkylation, which themselves have downstream biocatalytic applications.^47–51 Alongside their demonstrated utility in drug diversification,^49–51 the proposed engineering of PTs to accept chemoselective functionalities allows for the coupling of drug molecules to solid supports (for target-fetching studies) and fluoro- or chromophores (for bioimaging). As such, the efficient enzymatic production of the corresponding diphosphates is of great interest.

Kinetics.

To assess the ability of Ile74Ala and Ile146Ala to phosphorylate non-natural alkyl-P analogues, kinetic studies were performed with selected analogues using the PK-LDH assay under standardized conditions (see Materials and Methods).³⁵ The kinetic parameters of the WT,³⁵ Ile74Ala, and Ile146Ala variants of CMA with different alkyl-P substrates have been compiled in Table 2. We have also included the parameters for the Val136Ala mutant with the natural substrates 1 and 2 in the Supporting Information (see Table S6).

Table 2.

Pseudo-First-Order Kinetic Parameters for CMA Variants with Mutations in the IP-Binding Pocket

	kcat (s−1)			KM (mM)			kcat/KM (mM−1 s−1)
analogue	WTa	Ile74Ala	Ile146Ala	WTa	Ile74Ala	Ile146Ala	WTa	Ile74Ala	Ile146Ala
1	19 ± 1	42 ± 2	68 ± 2	0.018 ± 0.005	0.22 ± 0.02	0.16 ± 0.02	1000	190	430
2	24 ± 2	5.6 ± 0.3	33 ± 3	0.07 ± 0.01	0.40 ± 0.07	0.5 ± 0.1	360	14	60
11	8.0 ± 0.7	2.7 ± 0.3	5.6 ± 0.2	0.13 ± 0.02	1.2 ± 0.3	0.52 ± 0.06	63	2.2	11
13	16.5 ± 0.9	7.8 ± 0.5	16.6 ± 0.7	0.06 ± 0.01	0.6 ± 0.1	0.22 ± 0.03	280	13	80
14	2.7 ± 0.1	1.24 ± 0.08	0.94 ± 0.06	0.46 ± 0.07	0.43 ± 0.08	0.39 ± 0.07	5.9	2.9	2.4
15	0.51 ± 0.02	0.41 ± 0.03	0.289 ± 0.008	0.39 ± 0.05	1.2 ± 0.2	0.23 ± 0.03	1.3	0.34	1.3
16	1.22 ± 0.08	0.62 ± 0.06	3.4 ± 0.2	0.33 ± 0.06	1.1 ± 0.2	0.36 ± 0.06	3.8	0.6	10
17	0.277 ± 0.006	0.115 ± 0.005	1.31 ± 0.08	0.56 ± 0.04	0.45 ± 0.06	0.8 ± 0.2	0.5	0.26	1.6
19	0.41 ± 0.04	4 ± 1	0.92 ± 0.03	0.9 ± 0.2	3 ± 1	0.34 ± 0.05	0.5	1.2	2.7
21	ND	0.21 ± 0.01	1.05 ± 0.03	ND	0.41 ± 0.07	0.59 ± 0.06	ND	0.50	1.8
22	ND	0.65 ± 0.10	2.20 ± 0.05	ND	1.6 ± 0.4	0.55 ± 0.05	ND	0.4	4.0
24	0.17 ± 0.05	1.4 ± 0.3	1.47 ± 0.08	12 ± 4	7.0 ± 1.8	1.9 ± 0.2	0.013	0.20	0.8
25	ND	0.69 ± 0.05	1.4 ± 0.1	ND	0.51 ± 0.10	7 ± 1	ND	1.4	0.20
26	ND	0.25 ± 0.01	0.022 ± 0.003	ND	0.35 ± 0.06	0.4 ± 0.1	ND	0.7	0.06
27	0.15 ± 0.03	1.4 ± 0.2	1.79 ± 0.08	5.0 ± 1.3	0.8 ± 0.3	0.89 ± 0.10	0.03	1.6	2.0
34	ND	1.72 ± 0.09	1.16 ± 0.05	ND	0.83 ± 0.10	1.9 ± 0.2	ND	2.1	0.60
37	ND	0.31 ± 0.06	ND	ND	1.1 ± 0.5	ND	ND	0.3	ND
38	ND	6 ± 1	ND	ND	1.7 ± 0.6	ND	ND	3	ND
40	2.0 ± 0.1	0.41 ± 0.02	0.98 ± 0.03	0.09 ± 0.01	0.11 ± 0.03	0.42 ± 0.04	21	4	2.3
41	0.93 ± 0.02	2.07 ± 0.10	0.87 ± 0.02	0.11 ± 0.01	0.47 ± 0.06	0.22 ± 0.02	8.4	4.4	3.9
43	ND	1.05 ± 0.07	0.65 ± 0.01	ND	0.08 ± 0.02	0.58 ± 0.04	ND	13	1.1
44	ND	0.46 ± 0.03	0.74 ± 0.03	ND	1.2 ± 0.2	0.94 ± 0.10	ND	0.39	0.79

^aReported previously by Kumar et al.³⁵

Comparison of kinetic parameters for the natural substrate 1 between WT³⁵ and the Ile74Ala and Ile146Ala mutants indicated that, while the k_cat values increased 2–3 fold for the variants, the catalytic efficiency (k_cat/K_M) decreased 2–5 fold, largely due to an order of magnitude increase in K_M values. The structural isomer of the natural substrate (DMAP, 2) displayed a similar decrease in efficiency between the WT and Ile146Ala (6-fold) but a more dramatic decrease between the WT and Ile74Ala (26-fold). Similar efficiency trends were observed for demethylated and chlorine-substituted DMAP analogues (12 and 13, respectively), and these changes were the result of both increased K_M and decreased k_cat values for both variants. Thus, in general, the WT CMA displayed better or comparable catalytic efficiencies for the DMAP analogues with ≤four atoms in their unbranched alkyl chains compared to Ile146Ala and Ile74Ala, largely due to the increased K_M values. These results suggested the importance of hydrophobic interactions from the side chains of Ile74 and Ile146 in stabilizing the binding of the natural substrate 1 and its structurally similar analogues. Furthermore, the loss of hydrophobic contacts in mutants Ile146Ala and Ile74Ala could also explain the 2–3-fold increase in k_cat for 1. It is likely that the movement of the linear isoprene unit within the mutated pocket resulted in a concomitant movement of the phosphate group, which could potentially orient one of its free O atoms closer to the γ-phosphate of ATP.

As the length of the alkyl-P’s carbon chains increased beyond that of the natural substrate, the trends in efficiency also began to shift. For example, Ile74Ala and Ile146Ala displayed relatively similar values of K_M and k_cat/K_M with the singly and doubly methylated DMAP analogues (14 and 15, respectively) compared to the WT enzyme. Given the decreased hydrophobic packing introduced by the Ile-to-Ala mutations, the similar kinetic parameters between the three variants suggested that each pocket had either slightly too little (WT) or slightly too much (Ile74Ala and Ile146Ala) space to optimally accommodate the alkyl chains of 14 and 15. The same argument appeared to hold with the combination of WT or Ile74Ala with the cyclized DMAP analogues (cyclopentyl 16 and cyclohexyl 17), both of which displayed higher efficiencies with Ile146Ala when compared with either the WT (3-fold lower) or Ile74Ala variants (6–17-fold lower).

Interestingly, while most alkyl-P analogues containing ≥six atoms in their linear alkyl chain (20, 22, and 24–27) did not serve as substrates for WT CMA,³⁵ both Ile74Ala and Ile146Ala were able to phosphorylate 19, 21, 22, and 24–27 with good efficiencies (≥1 mM⁻¹ s⁻¹).

Among these longer-chain analogues (except 25 and 26), Ile146Ala displayed generally higher catalytic efficiencies compared to Ile74Ala, derived mostly from lower K_M values. Notably, a single-atom increase in the chain length of 26 compared to 21 caused a substantial decrease in k_cat for Ile146Ala (47-fold) while maintaining a similar K_M, whereas the K_M and k_cat values of Ile74Ala with these two analogues remained similar. Furthermore, the introduction of rigid alkyne moieties in 24 and 25 caused 3- and 17-fold increases in K_M, respectively, for Ile146Ala compared to their saturated counterparts (21 and 26), while the K_M of Ile74Ala increased dramatically only for 24 (17-fold). These results implied that Ile74Ala prefers longer, more conformationally hindered alkyl chains compared to Ile146Ala, which better tolerated shorter and more flexible alkyl-P analogues.

Similarly, among the dienes, the catalytic efficiency of Ile146Ala with 34 was threefold lower than that of Ile74Ala mostly due to a higher K_M value. However, the catalytic efficiency of Ile74Ala with the phenylated analogue 38 was an order of magnitude higher than that of 37 due to a higher k_cat value. This was interesting considering 38 bore an additional C3-methyl group compared to 37. Thus, it appeared that the additional C3-methyl group on 38 may have played a role in orienting the monophosphate group for phosphorylation.

Among the benzylic analogues (40, 41, 43, and 44), the catalytic efficiency of Ile74Ala and Ile146Ala decreased 2–9-fold for minimally substituted analogues (40 and 41) compared to the WT, which was due to a combination of higher K_M and lower k_cat values. Conversely, while activity could not be detected for WT CMA with analogues bearing bulky substituents on the benzene ring (43 or 44),³⁵ the mutated variants displayed good catalytic efficiency (0.4–13 mM⁻¹ s⁻¹) with these substrates. Thus, while Ile-to-Ala mutations likely decreased the hydrophobic packing against the phenyl rings, the concomitant increase in space allowed Ile74Ala and Ile146Ala to better accommodate longer substituents, matching the overall trends observed with the aliphatic alkyl-P analogues.

CMA Mutant Crystallization and Structure Solution.

To better understand the structure–activity relationships between the mutants and their non-natural alkyl-P substrates, we crystalized the two most promiscuous variants of CMA, Ile74Ala and Ile146Ala, bound to various non-natural alkyl-P analogues that were not substrates for the WT enzyme.³⁵ Collectively, our efforts yielded three structures, each with a unique combination of the CMA variant and the alkyl-P analogue: Ile74Ala bound to 38 and ADP (I74A•38•ADP), Ile146Ala bound to the diphosphate product of 26 (26-P) and ADP (I146A•26-P•ADP), and Ile146Ala bound to 27 and ADP (I146A•27•ADP). Crystallographic statistics for each of these mutant structures are summarized in Table 3.

Table 3.

Data Collection and Refinement Statistics for Crystal Structures of CMA Mutants

PDB ID	7N9D	7LNX	7LNW
CMA variant	Ile74Ala	Ile146Ala	Ile146Ala
ligands	38, ADP	A: 26-P, ADP	27, ADP
		B: 26, ADP
wavelength	1.54	0.979	1.54
resolution rangea	27.0–2.1 (2.2–2.1)	39.4–2.3 (2.4–2.3)	28.7–2.3 (2.4–2.3)
space group	P622	P212121	P1211
unit cell (a, b, c; Å)	172.29, 172.29, 80.96	40.56, 73.28, 166.93	49.25, 93.00, 52.14
unit cell (α, β, γ; °)	90, 90, 120	90, 90, 90	90, 92.017, 90
total reflections	447,878	141,521	86,142
unique reflectionsa	41,415 (3953)	22,730 (2111)	21,021 (2067)
multiplicitya	10.8 (5.8)	6.2 (4.2)	4.1 (3.1)
completenessa (%)	99.3 (96.6)	98.9 (92.7)	99.5 (97.7)
mean I/σIa	12.3 (1.0)	10.1 (1.7)	8.86 (1.67)
Wilson B factor	29.46	35.75	27.23
R merge a	0.097 (0.74)	0.12 (0.78)	0.14 (0.49)
CC1/2a	0.992(0.662)	0.998(0.710)	0.996 (0.754)
reflections used in refinementa	41,414 (3953)	22,714 (2111)	20,994 (2067)
reflections used for Rfreea	2000 (191)	1998 (185)	2004 (192)
R work a	0.205 (0.291)	0.245 (0.315)	0.202 (0.232)
R free a	0.239 (0.348)	0.305 (0.403)	0.255 (0.288)
total number of non-hydrogen atoms	4252	3808	3990
non-hydrogen atoms in macromolecules	3857	3663	3825
non-hydrogen atoms in ligands	92	90	82
number of solvent molecules	303	55	83
protein residues	508	482	501
RMSD (bonds)	0.005	0.002	0.003
RMSD (angles)	0.91	0.61	0.69
Ramachandran favored (%)	97.42	95.35	97.36
Ramachandran allowed (%)	2.18	4.23	2.23
Ramachandran outliers (%)	0.40	0.42	0.41

^aValues in parentheses denote the highest resolution shell.

Similar to the ternary complexes of WT CMA, the asymmetric unit of each mutant structure contains two monomers arranged in a dimeric structure, and almost all of these monomers contain a mixed pair of substrate/product bound in the active site. However, I146A•26-P•ADP was unique among the six current structures in that monomer A contained a pair of products (26-P/ADP). In addition, glycerol molecules are found near the N-termini of each monomer B in I74A•38•ADP and I146A•26-P•ADP (one per structure). Analyses of the mutant structures’ dimer interfaces using the PISA tool^45,46 reveal similar values for the average area buried by dimerization (~1567 Ų per monomer and ~3134 Ų total).

Comparison of Alkyl-P-Binding Sites between CMA Variants.

Unsurprisingly, the core architecture of the alkyl-binding pocket remains intact between the WT and mutant structures of CMA, and the single Ile-to-Ala mutation does not seem to affect the orientation of the side chains of the other residues in contact with the alkyl groups (Figure 5). However, the structure of I74A•38•ADP did capture novel contributions to the alkyl-binding site that have not been observed in previous IPK structures. Specifically, the side chains of Val208 and Thr209 from the DVTGG motif in the αF-αG loop (proposed to be important for nucleotide phosphate recognition;³⁸ colored green in Figure 5) contribute additional hydrophobicity to the alkyl-binding pocket occupied by 38, and the side chain hydroxyl of Thr209 forms a direct H-bonding interaction with a free oxygen of 38’s phosphate moiety. Additionally, the backbone amide of Val208 in monomer A lies ~3.1 Å away from N_η of the catalytic His50 (red dashed line, Figure 5), implying a potential role of Val208 in orienting the imidazole ring for catalysis. Though these interactions only appear in I74A•38•ADP, it is unlikely the mutation of Ile74 is the specific cause given the residues’ locations at opposite ends of the alkyl-binding pocket. Therefore, this novel conformation of I74A•38•ADP indicates that the DVTGG motif may be important not only for phosphate recognition of the nucleotide but also for interacting with the phosphoryl acceptor substrate during IPK catalysis (discussed later).

Figure 5.

Alkyl-binding sites of I74A•38•ADP and I146A•26-P•ADP. The interactions of 38 and 26-P are depicted in (A) and (B), respectively. The carbon atoms of 38 and 26-P are depicted in yellow and magenta (respectively), water molecules are modeled as yellow or orange spheres, and important polar contacts are shown with black dashed lines.

Additionally, the orientation of the alkyl-Ps was found to differ between the WT and mutant structures. For example, in I74A•38•ADP, the plane of 38’s isoprene unit tilts ~30° out of line with that of IP with a concomitant movement of 38’s phosphate moiety further into the pocket (~1 Å, Figure 6A). This is not surprising given the location of Ile74 at the back of the alkyl-binding pocket (Figure 3). However, this movement of the isoprene phosphate in I74A•38•ADP compared to that in CMA•IP•ATP increased the distance between the electro-philic P_α of 38 and the nucleophilic P_γ-O atom of ATP and may explain the sevenfold decrease in k_cat for 38 when compared to IP (1) for the Ile74Ala mutant (Table 2).

Figure 6.

Comparison of the orientations of 38 (A), 26 (B), 26-P (C), and 27 (D) with their natural comparators IP and IPP in the binding pockets of CMA variants. The carbons of 38, 26/26-P, and 27 are colored cyan, purple, and tan (respectively), while their phosphate groups are shown in orange and red. The carbons of IP and IPP are colored yellow, while their phosphate groups are shown in green and sky blue.

Similar orientational differences between the bound alkyl-P analogues are observed when comparing the ternary complexes of the WT and Ile146Ala. Throughout the four monomers captured in I146A•26-P•ADP and I146A•27•ADP, the plane of the isoprene moiety in each substrate/product (26/26-P/27) is rotated ~180° compared to their counterparts in the WT complexes such that C2 of each carbon chain points toward the top of the alkyl-binding pocket instead of the bottom as seen for IP (Figure 6B–D). Furthermore, the terminal propyl/methylazide groups attached to the isoprene units of 26/27 (respectively) cannot be superimposed between any of the monomers, which appeared to be due in large part to ambiguous electron density for the last three atoms in each chain (see Supporting Information Figure S3). While the rotation of the isoprene moieties of 26/26-P/27 is not completely certain, these ambiguities are informative about the higher flexibility of 26 and 27 compared to the other alkyl-Ps and the ability of Ile146Ala to accommodate it.

Based on these structural observations and the associated kinetic trends, we are able to draw reasonable conclusions about why the Ile-to-Ala mutations caused differential alteration of the CMA substrate profile. Perhaps the most obvious point highlighted by our data is the profound influence of the mutated residues’ locations in the pocket on the distribution of the additional space. In the WT ternary complexes, the sec-butyl side chains of Ile74 and Ile146 are both directly in contact with the alkyl group of IP/BP such that the only major difference between them is their location relative to the plane of the alkyl group (Figure 3C). As such, both Ile-to-Ala mutations seem to introduce roughly equivalent amounts of space to the binding pocket but differ in where this space is added. Since Ile74 is located nearly parallel to the plane of IP’s isopentenyl group, trimming back its sec-butyl side chain creates additional space at the distal end of the binding pocket. In contrast, Ile146 pushes into the pocket along one of the faces of IP’s isopentenyl group, and the corresponding mutation to Ala causes the additional space to be diffused along the side and up toward the top of the pocket. Thus, the two mutations have opposite effects on the generalized length and width of the pocket: Ile74Ala creates a longer but similarly narrow pocket to the WT enzyme, while Ile146Ala causes the pocket to widen while maintaining a similar length.

The differential distribution of space between the two variant pockets also appears to explain the differences in their substrate and kinetic profiles. Ile74Ala generally displays higher turnovers and more favorable kinetic parameters with alkyl-Ps bearing rigid carbon tails, such as 38 with which it was crystallized. Alongside the structural data of I74A•38•ADP, it becomes clear that the enzyme’s generally lower K_M values for rigid alkyl-Ps arise from its more restricted binding pocket compared to Ile146Ala. The increased movement of the carbon chains in more flexible alkyl-Ps likely cause a greater number of steric clashes with the walls of the narrow Ile74Ala pocket, which would decrease binding affinity for these substrates.

In contrast, the similar lengths of the WT and Ile146Ala pockets likely preclude several of the rigid alkyl-Ps (which also tended to be longer than the natural substrate) from binding. Alongside this generalized inability to utilize rigid alkyl-Ps as substrates, the Ile146Ala mutant also displays an opposing preference for alkyl-Ps bearing flexible carbon chains (such as 26 and 27 with which it was crystallized). Once again, the structural data for this variant (I146A•26-P•ADP and I146A•27•ADP) provides an explanation for the trend. The rotation of the isoprene plane does not appear to be catalytically relevant considering the phosphate moieties were still largely superimposable (Figure 6), but the ambiguous electron densities for the remainders of the aliphatic tails points to a multitude of conformations for 26 and 27 that are catalytically active. This would explain the Ile146Ala variant’s greater promiscuity toward and higher catalytic efficiencies with flexible alkyl-Ps compared to Ile74Ala. Such mutational effects on promiscuity need to be considered when CMA is further engineered for non-natural isoprenoid synthesis.

Comparison of Nucleotide-Binding Sites between CMA Variants.

Secondary structural elements and interactions with the nucleoside portion of ADP (involving Asp170, Tyr175, and Asp178) and the diphosphate moiety (involving Lys5, Gly8, Ser9, Lys14, and Ser169) are largely maintained between the WT and mutant structures (Figure 7A), though monomer A of I146A•27•ADP does display an additional α helix (residues 202–206) in the middle of its αF-αG loop (Figure 7B). Furthermore, both subunits of I74A•38•ADP display a novel conformation for the C-terminal nucleotide-binding domain of CMA, wherein the αF and αG helices are rotated 20–30° toward the active-site cleft compared to the same helices in the WT and Ile146Ala ternary complexes (Figure 7B). As seen with the WT structures, the crystal packing contacts differ between the monomers and the mutant complexes, but those present in I74A•38•ADP do not appear to stabilize the αF-αG loop into the novel orientation.

Figure 7.

(A) Nucleotide-binding site of monomer A in I74A•38•ADP. The carbons of ADP are colored green, water molecules are modeled as yellow spheres, and important polar contacts are shown with black dashed lines. (B) Conformational changes of the nucleotide-binding domain of the WT apo structure (red) and the I146A•27•ADP (green) and I74A•38•ADP (yellow) complexes.

With the rotation of the αF and αG helices, the secondary structural elements of the C-terminal domains align well with their counterparts in the product-bound structure of THA (PDB ID: 3LL5),³⁸ restoring two key interactions with the nucleotide observed in this structure that had yet to be observed with CMA. Namely, Val212 is close enough to the adenine ring to engage in hydrophobic stacking, and the side chain of Lys215 interacts with the diphosphate moiety of ADP (Figure 7A). From the perspective of IPKs, however, the true novelty of this conformation is in the observation of intelligible electron density for the intervening αF-αG loop, which is found lying directly over the active-site cleft with the DVTGG motif common to the AAKs covering the phosphate moieties.³⁸ Although none of the previous IPK structures visualize this closed conformation,^33,38 the fosfomycin-bound structure of FomA from Streptomyces wedmorensis (FK, PDB ID: 3D41) displays a similar conformation for this part of the nucleotide-binding pocket.⁴² With this closed conformation, the side chains of Asp207 and Thr209 and the amide of Gly211 from the DVTGG sequence make water-mediated interactions with the phosphates of the nucleotide (Figure 7A).

In addition, monomer A of the I74A•38•ADP complex provides insight into a potential binding site of Mg²⁺, which has yet to be cocrystallized with an IPK.

Shown as a green sphere in Figure 8, this potential Mg²⁺ ion is octahedrally coordinated by four water molecules, a nonbridging P_α oxygen of ADP, and a nonbridging P_β oxygen of ADP. Superposition of this active site with FomA (PDB ID: 3D41),⁴² which does contain a bound Mg²⁺, reveals that assignment of the corresponding electron density to this ion is reasonable by homology. However, the combination of a slight distortion in the octahedral coordination geometry and the lengths of the interactions (2.2–2.6 Å) do not allow us to unambiguously assign Mg²⁺ to this position, so it is modeled as a water molecule in the final structure (green sphere in Figure 8). Still, the similar coordination shell and position in the active site points to the possibility of a Mg²⁺ being replaced by water during cryoprotection or data collection and potentially explains the closed conformation observed in this complex (I74A•38•ADP). Collectively, the analysis of six structures of CMA variants presented herein provide for the first time both open and closed conformations of the nucleotide-binding domain relevant to IPK catalysis.

Figure 8.

Proposed Mg²⁺-binding site in I74A•38•ADP in monomer A. The water modeled in place of Mg²⁺ is colored dark green, while the remaining waters are shown as yellow stars. The carbons of 38 and ADP are colored green and purple (respectively), active-site residues are colored teal, and important polar contacts are shown with black dashed lines.

Investigating the Catalytic Roles of Val208 and Thr209 from the DVTGG Motif.

As discussed above, the I74A•38•ADP complex displayed a conformation of the αF-αG loop in which the DVTGG motif lies directly over the active site (Figure 7). Furthermore, residues Val208 and Thr209 within this group were observed to be directly interacting with the bound acceptor substrate (Figure 5A). The entirely hydrophobic side chain of Val208 appeared to help cap the IP-binding pocket, while the amphipathic side chain of Thr209 seemed to perform double duty in contacting both the hydrocarbon and phosphate portions of the substrate. The DVTGG motif has been known for its interactions with the nucleotide,³⁸ so specific contacts with the phosphoryl acceptor could imply additional catalytic roles of the residues in question. However, because the novel loop conformation was only observed in the I74A•38•ADP complex, it was unclear whether the interactions are maintained during WT IPK catalysis.

To understand the role of Val208 and Thr209 in catalysis, we carried out Val208Ala, Thr209Ala, and Thr209Ser mutations into WT and Ile74Ala CMA and created six novel variants (Val208Ala, Thr209Ala, Thr209Ser, Ile74Ala/Val208Ala, Ile74Ala/Thr209Ala, and Ile74Ala/Thr209Ser). The single Ala substitution of Val208 in WT CMA was meant to test its proposed role in capping the IP-binding site by trimming back the isopropyl side chain. Meanwhile, the Ser and Ala mutations at Thr209 were intended to stifle one of its two observed interactions: the role of the OH group in stabilizing the phosphate moiety of the acceptor substrate (maintained in Thr209Ser) and contributing additional hydrophobicity to the IP-binding site (maintained in Thr209Ala). The six new variants were screened against the library of alkyl-P analogues, and when plotted against the original profiles of CMA WT³⁵ and Ile74Ala (Figure 9 and Supporting Information Table S7), the new DVTGG variants displayed generally lower activity. Profiles of variants containing the Thr209Ser mutation were the most akin to their original versions (WT or Ile74Ala), showing similar or moderately reduced conversions (<15%) in most cases, whereas the Val208Ala mutation appeared to cause comparatively more disruption to the enzymes’ promiscuity. When applied to the WT enzyme, Val208Ala maintained activity with most of the same alkyl-P analogues compared to WT CMA, though these conversions were generally reduced by >20%.

Figure 9.

Results of screening DVTGG variants of CMA WT (A) and I74A (B) against the library of alkyl-P analogues using the PK-LDH assay. Conversions were calculated by measuring the absorbance at 340 nm of each enzymatic reaction just before and 1 h after the addition of the CMA variant at 37 °C and comparing it to a positive control (n = 2). Appropriate controls were conducted to account for any ATPase activity. Each reaction consisted of 2 U PK, 2 U LDH, 0.6 mM NADH, 1.5 mM PEP, 2 mM ATP, and 4 μg of IPK incubated in a buffered solution (25 mM Tris pH 7.8 and 5 mM MgCl₂). All positive reactions were verified using HRMS. Conversion data for WT CMA were obtained previously.³⁵

Reductions were even more pronounced for the double mutant Ile74Ala/ Val208Ala, which completely lost activity with several alkyl-Ps compared to Ile74Ala (20–22, 24–26, 34, 37, 38, and 40–46). The most dramatic changes, however, were observed with the Thr209Ala mutation. Both the single (Thr209Ala) and double (Ile74Ala/Thr209Ala) mutants bearing this mutation lost nearly all activity such that the former phosphorylated four alkyl-Ps (1, 2, 10, and 13) while the latter only utilized the natural substrate 1 (colored red in Figure 9). Overall, the substrate profiles of the six variants implied that Val208 and Thr209 were indeed relevant to CMA-catalyzed reactions in both the WT and Ile74Ala variants.

To understand the specific effects of these mutations on catalysis, we carried out kinetic studies of the six mutants with the natural substrate IP (1). The resulting parameters are summarized in Table 4. In the case of both Val208Ala mutants, specificity for 1 decreased (14–29-fold) through a combination of increased K_M (3–4-fold) and decreased k_cat (3–9-fold). Similar changes were observed for the Thr209Ser mutants, though the decreases in specificity were only 5–10-fold for this case. Taken together, these changes suggested the hydrophobicity of Val208 and Thr209’s side chains do play a role in catalysis with respect to the alkyl-P.

Table 4.

Pseudo-First-Order Kinetic Parameters for IP (1) with CMA Variants Bearing Mutations in the DVTGG Motif

	kcat (s−1)	KM (mM)	kcat/KM (mM−1 s−1)
Val208Ala	5.4 ± 0.1	0.075 ± 0.010	72
Thr209Ala	0.038 ± 0.002	0.05 ± 0.02	0.8
Thr209Ser	7.2 ± 0.3	0.037 ± 0.008	200
Ile74Ala	4.9 ± 0.2	0.75 ± 0.08	6.5
Val208Ala
Ile74Ala	0.035 ± 0.002	0.5 ± 0.1	0.08
Thr209Ala
Ile74Ala	8.5 ± 0.6	0.44 ± 0.09	19
Thr209Ser

As implied by their substrate profiles, kinetic data for variants containing the Thr209Ala mutation indicated that specificity for IP dropped significantly compared to the WT or Ile74Ala enzymes (1200–2400-fold), with the main contribution being decreased k_cat values (500–1200-fold). Though increases in K_M were also observed (2–4-fold), it was clear from the differences in k_cat that the side chain OH group of Thr209 was essential for achieving high reaction rates, especially when compared to the Thr209Ser mutants (activity largely retained compared to the WT and Ile74Ala). Similar changes were observed when ATP was the limiting substrate (Supporting Information Table S8), though the near full order of magnitude increase in K_M suggested this hydroxyl still contributed to ATP binding through the network of hydrogen bonds. As such, the combined mutational studies of Val208 and Thr209 suggested that the DVTGG motif is directly involved in phosphotransfer. Furthermore, high conservation of the motif across IPK sequences (Supporting Information Figure S2) implied that the current findings are applicable to the wider enzyme class.

In terms of specific residue interactions, the loop conformation observed in the I74A•38•ADP complex (Figure 5A) implied that the hydrophobic side chain of Val208 helps cap the IP-binding site and orient the alkyl-P for catalysis. The amphipathic side chain of Thr209 contributes additional hydrophobicity to this capping while also helping to stabilize the acceptor substrate’s phosphate moiety through hydrogen bonding.

Thus, while this snapshot was originally captured in a singular mutant structure, the current results and the dynamic nature of the loop suggested that the conserved DVTGG motif adopts this conformation during the catalytic cycles of CMA and other WT IPKs.

CONCLUSIONS

In summary, the current study has provided an in-depth analysis on the promiscuity of the IPK from CMA with respect to its three-dimensional structure, as well as engineering studies aimed at altering this promiscuity and elucidating a novel role of the DVTGG motif in catalysis. Our investigation began by solving three crystal structures of the WT enzyme in different ligand-bound forms (apo, ternary complex with natural substrates, and ternary complex with non-natural alkyl-P and ADP), which allowed us to identify the molecular determinants of its alkyl-substrate specificity. Based on this analysis, three bulky residues from the alkyl-binding pocket (Ile74, Val136, and Ile146) were mutated to Ala to introduce additional space within the pocket and thereby increase the length of alkyl chains that could theoretically fit. Screening and kinetic studies using a library of alkyl-P analogues revealed enhanced promiscuity and differential substrate preferences for the Ile74Ala and Ile146Ala variants toward longer-chain alkyl-Ps, which led us to pursue their substrate-bound crystal structures. Our efforts yielded an additional three structures differing in their bound alkyl-P substrate: a ternary complex of Ile74Ala bound to a non-natural substrate bearing a rigid phenylated alkyl chain (38) and two ternary complexes of Ile146Ala bound to different non-natural substrates bearing somewhat flexible alkyl chains (26, 27). In-depth structural analysis revealed similarities in the oligomeric state and tertiary folding between the CMA enzymes and previously characterized IPKs, yet several secondary structural elements were found to be unique to CMA. Comparison of the six current structures also highlighted the high rigidity of the alkyl-P-binding pocket even in the midst of mutation. As anticipated, the reduction of bulky side chains in contact with the alkyl-P introduced additional space into the binding pocket, providing the mutated enzymes with enhanced promiscuity. Closer investigation of the Ile74Ala and Ile16Ala structures further demonstrated the critical importance of the mutated residue’s location relative to the plane of the natural substrate in determining the associated variant’s substrate profile. Additionally, our characterization of a novel loop conformation in the nucleotide-binding C-terminal domain highlighted the potential relevance of the DVTGG motif in IPK substrate binding and catalysis. Subsequent mutational studies of the Val and Thr residues (observed directly interacting with the alkyl-P substrate) confirmed these hypotheses for the IPK from CMA, and the high conservation of the DVTGG motif suggested our conclusions could likely be generalized to IPKs at large. Overall, the current study has provided a firm foundation for both future engineering studies of the IPKs as well as their biocatalytic utility for the synthesis of non-natural isoprenoids.

MATERIALS AND METHODS

General Reagents and Materials.

Unless otherwise stated, all chemicals and reagents were purchased from Sigma-Aldrich, Acros, Alfa-Aesar, or TCI and were reagent grade or better. Enzymes for molecular biology were purchased from New England Biolabs.

Expression and Purification of CMA Variants.

Preparation of the CMA WT construct has been described previously,³⁵ and similar protocols were used for the CMA mutants. Briefly, Escherichia coli Rosetta2 cells were transformed with pET28a vectors containing a codon-optimized synthetic cmaIPK gene (GenScript Biotech, Piscataway, NJ, USA) inserted between the NdeI and EcoRI sites. Expression began with inoculation of 5 mL of cultures of Luria-Bertani (LB) broth containing 50 μg mL⁻¹ kanamycin (KAN), which were subsequently incubated overnight at 37 °C and 200 rpm. The starter cultures were then used to inoculate 1 L of cultures of LB broth (50 μg mL⁻¹ KAN) grown at 37 °C and 200 rpm. When the OD₆₀₀ reached 0.6–0.8 (~4 h), the cultures were induced with 0.5 mM IPTG (final concentration) and incubated at 20 °C and 200 rpm for 16–20 h. Cells were then separated from the media through centrifugation, resuspended in lysis buffer (50 mM NaH₂PO₄ pH 8.0, 300 mM NaCl, 10 mM imidazole), and frozen at –80 °C.

Purification began with three cycles of freezing at –80 °C and thawing at room temperature. Cultures were lysed through a combination of lysozyme treatment (30 min on ice) and sonication (40 min total, in 10 min cycles of 10 s pulses and 25 s breaks at 30% max amplitude). Centrifugation was used to separate the soluble fraction from the insoluble cell debris, and the supernatant was subsequently purified using Ni-NTA chromatography [5 mL HisTrap Fast Flow column (Cytiva, Marlborough, MA, USA); a stepwise gradient of 10–500 mM imidazole in 50 mM NaH₂PO₄ pH 8.0 and 300 mM NaCl over 11 column volumes]. Fractions containing the purified protein were pooled and concentrated in an Amicon centrifugal concentrator, and repeated cycles of dilution and reconcentration in storage buffer [25 mM Tris pH 8.0, 50 mM KCl, and 10–20% (v/v) glycerol] removed a majority of the imidazole (final concentration < 1 mM). The purity of the concentrated protein was checked by SDS-PAGE, and pure proteins were drop-frozen in liquid N₂ and stored at −80 °C.

Crystallization.

WT CMA crystals were grown using a hanging-drop vapor diffusion method at a constant 16 °C. Each 2 μL drop comprised a 1:1 mixture of enzyme (12.3 mg mL⁻¹) and crystallization buffer. For the apo condition, the crystallization buffer consisted of 0.2 M sodium malonate, pH 7, and 20% (w/v) PEG 3350. In this case, crystals formed flowerlike plate clusters, which reached optimal size in 2 days. All ligand-bound WT CMA crystals were grown in 0.1 M ammonium acetate, 0.1 M bis-tris, pH 5.5, and 15% (w/v) PEG 10,000 crystallization buffer. The enzyme mixture for these ligand-bound crystals was incubated at room temperature for 3 h and contained final concentrations of 2.0 mM alkyl-P [IP/BP], 4.0 mM ATP, and 2.0 mM MgCl₂.

Crystals of CMA Ile74Ala grew in a 0.6 μL sitting drop subjected to vapor diffusion at ambient temperature. The drop was composed of a 1:1 mixture of buffered enzyme–ligand mixtures (12.5 mg mL⁻¹ CMA Ile74Ala, 2 mM 38, 4 mM ATP, 25 mM Tris, pH 8.0, 50 mM KCl, and 5 mM MgCl₂) and crystallization buffer (3.5 M sodium formate, pH 7.0), and the resulting rectangular crystals grew to optimum size over the course of 6 months (initial small crystals grew within 36 days). The CMA Ile146Ala crystals were grown in a similar manner to those of Ile74Ala except for the individual enzyme–ligand mixtures (12.5 mg mL⁻¹ CMA Ile146Ala, 2 mM 26/27, 4 mM ATP, 25 mM Tris, pH 8.0, 50 mM KCl, and 5 mM MgCl₂) and the crystallization buffer [0.1 M bis-tris, pH 5.5, 0.1 M NaCl, and 28% (w/v) PEG 3350]. The Ile146Ala crystals grew to optimal size within 2 days, and crystals from the mixture of Ile146Ala and 26 were cryoprotected using crystallization buffer supplemented with 15% (v/v) glycerol. These crystals were then stored in liquid N₂ for 2 weeks before data collection at the Stanford Synchrotron Radiation Lightsource (SSRL).

Data Collection and Structure Solution.

Before in-house data collection, crystals of CMA WT were cryoprotected by sequentially immersing them in their respective crystallization buffers containing increasing concentrations of glycerol [10, 15, and 20% (v/v)]. Similarly, the Ile74Ala/38 and Ile146Ala/27 crystals were cryoprotected with 5.0 M sodium formate and 0.1 M bis-tris, pH 5.5, 0.1 M NaCl, 28% (w/v) PEG 3350, and 15% (v/v) glycerol [respectively] before mounting. Diffraction data for these five structures were collected using a Rigaku MicroMax 007HF X-ray generator utilizing CuKα radiation with a Dectris Pilatus P200K detector, and the data sets were processed, merged, and scaled using the HKL-3000 suite.⁵² The diffraction data for the ternary complex of Ile146Ala and 26 were collected at SSRL Beamline 9–2 (λ = 0.97946 Å). This data set was processed using XDS⁵³ and then scaled and merged using AIMLESS in the CCP4 suite.⁵⁴

The apo CMA structure was solved by molecular replacement using PHASER⁵⁵ and the previously determined ADP-bound structure of MTH (PDB ID: 3LL9).³⁸ Once refined, the apo CMA structure was used to determine the structures of all five ternary complexes by molecular replacement. Model building of the enzymes was done using COOT,⁵⁶ and ligands that were not found in the library database were built and imported into COOT using a combination of JLigand⁵⁷ and eLBOW in the PHENIX suite. Refinement of the models was done using phenix.refine in the PHENIX suite.⁵⁸

Synthesis of Alkyl Monophosphate (Alkyl-P) Analogues.

The synthetic methods of all alkyl monophosphates utilized in this study (isolated as the ammonium salts) have been described previously in detail.³⁵

Site-Directed Mutagenesis.

CMA point mutants were generated through PCR extension of recombinant pET28a-cmaIPK plasmids with mutagenic primers (see Supporting Information Table S1). After mutagenesis, PCR products were treated with DpnI to digest methylated template DNA, and the product of the digestion was then transformed, expressed, and purified as described above.

High-Throughput Screening.

Purified CMA point mutants were screened against the library of alkyl-Ps using the PK-LDH assay as described previously.³⁵ Briefly, reactions were conducted in a high-throughput manner using 96-well plates. Each well contained 150 μL of reaction mixture composed of 4 μg of IPK, 2 U PK, 2 U LDH, 0.6 mM NADH, 1.5 mM PEP, 2 mM ATP, and 1 mM alkyl-P in buffer (25 mM Tris pH 7.8 and 5 mM MgCl₂). Before the IPK was added (10 μL), an absorbance reading at 340 nm and 37 °C was obtained to establish each well’s initial A₃₄₀ before any NADH was consumed. After the addition of IPK, the A₃₄₀ was monitored every 30 s for 1 h at 37 °C, and the difference between the final A₃₄₀ at 1 h and the initial A₃₄₀ before the addition of IPK was used to calculate the percentage conversion of NADH. All positive reactions were subsequently confirmed using HRMS as described in the Supporting Information. Additionally, two control reactions were conducted: one without any alkyl-P to establish the baseline and one using ADP instead of ATP to identify full conversion.

Kinetic Assays.

Pairs of enzymes and substrates were kinetically characterized using the PK-LDH assay as described previously.³⁵ For studies of the alkyl-P analogues as substrates, each well contained 150 μL of buffered reaction mixture (25 mM Tris pH 7.8 and 5 mM MgCl₂) composed of 2 U PK, 2 U LDH, 0.6 mM NADH, 1 mM PEP, 2 mM ATP, 0–5 mM alkyl-P, and a concentration of IPK suitable to observe ≤20% conversion of NADH in 30 min. For studies of ATP, the concentration of alkyl-P (1) was held constant at 2 mM and ATP was varied from 0 to 2 mM. The addition of IPK initiated the reaction, and the A₃₄₀ was measured every 30 s for 30 min at 37 °C. Initial rates were determined from the slope of the line of best fit for the time period in each reaction giving ~10% conversion after excluding the first 3 min (considered an equilibration period). Slopes were corrected for the degradation of NADH using the slope of a control reaction containing no alkyl-P. The kinetic constants k_cat and K_M and their associated errors were determined by inputting the initial rate data for each substrate concentration into GraphPad Prism and conducting a nonlinear regression. Values for the specificity constant k_cat/K_M were obtained by performing the calculation in Microsoft Excel and propagating the errors.

ABBREVIATIONS

IPP

isopentenyl diphosphate

DMAPP

dimethylallyl diphosphate

MEP

methylerythritol phosphate

MVA

mevalonate

IP

isopentenyl monophosphate

IPK

isopentenyl phosphate kinase

MJ

Methanocaldococcus jannaschii

IOH

isopentenyl alcohol

DMAOH

dimethylallyl alcohol

DMAP

dimethylallyl monophosphate

GP

geranyl monophosphate

FP

farnesyl monophosphate

PT

aromatic prenyltransferase

THA

Thermoplasma acidophilum

MTH

Methanothermobacter thermautotrophicus

AAK

amino acid kinase

NAGK

N-acetyl-l-glutamate kinase

AK

aspartokinase

GK

glutamate kinase

CBK

carbamate kinase

UMPK

uridine monophosphate kinase

FK

fosfomycin resistance kinase

CMA

Candidatus methanomethylophilus alvus

WT

wild-type

BP

benzyl monophosphate

alkyl-P

alkyl monophosphate

PK-LDH

pyruvate kinase-lactate dehydrogenase

HRMS

high-resolution mass spectrometry