UV–vis spectra of various forms of the IDI-2-bound FMN. Reaction conditions are given in the Experimental Procedures. (A) An 80 µM solution of E•FMN_ox (dashed line) was photoreduced under anaerobic conditons to give E•FMN_red (solid black line). Addition of 2 mM IPP led to the intermediate spectrum (dotted line). If 2 mM IPP was pre-incubated with E•FMN_ox prior to photoreduction, the neutral semiquinone is formed (solid gray line). (B) Spectra of dithionite (10 mM) reduced E•FMN_red (80 µM, dashed line) and the flavin intermediates formed upon addition of 2 mM IPP (solid black line), (R)-[2-²H]-IPP (solid gray line), or DMAPP (dotted line).

UV–vis absorbance spectra of the oxidized (-•-•-) and reduced (----) forms of enzyme-bound 5-deazaFMN (A) and 1-deazaFMN (B) bound to IDI-2 at pH 7.0 and 25 °C. E•5-deazaFMN_ox was photoreduced, while E•1-deazaFMN_ox required chemical reduction with sodium dithionite (see Experimental Procedures for details). When 1 mM IPP was added to the reduced enzymes, a unique flavin species (—) formed. The spectra of these species are consistent with the neutral reduced coenzymes.

Stopped-flow spectrophotometry of flaivn intermediate formation. (A) Pre-steady-state changes in A₄₃₅ recorded over 400 ms for the reaction of 68 µM E•FMN_red with 2 mM IPP (black line), 2 mM DMAPP (light gray line), and 2 mM (R)-[2-²H]-IPP (dark gray line). When enzyme was mixed against buffer, there was no change in A₄₃₅ (dotted line). The time courses at 435 nm were fit to a single-exponential equation () to determine the amplitude of the absorbance change (A), the observed rate (k_obs), and the initial A₄₃₅ (C). For IPP: A = 0.11 ± 0.001 AU (absorbance units); k_obs = 52 ± 0.9 s⁻¹; and C = 0.20 ± 0.001 AU. For DMAPP: A = 0.053 ± 0.0007 AU; k_obs = 14.8 ± 0.4 s⁻¹; and C = 0.223 ± 0.0008 AU. For (R)-[2-²H]-IPP: A = 0.12 ± 0.001; k_obs = 40.9 ± 0.5 s⁻¹; and C = 0.21 ± 0.001 AU. (B) Pre-steady state accumulation of the flavin intermediate in the reaction of 68 µM E•FMN_red (reduced with 10 mM dithionite) with 2 mM IPP. Spectra from bottom to top at 435 nm were obtained 0.001, 0.013, 0.025, 0.037, 0.075, and 0.399 s after mixing.

Single-turnover kinetics of flavin intermediate formation and decay over 5 s. IDI-2 was mixed in a molar excess over either IPP (black) or (R)-[2-²H]-IPP (gray), and the formation and decay of the flavin intermediate was followed at 435 nm. The time courses were fit to double-exponential equations to extract the amplitudes and the observed rates for the fast (increasing A₄₃₅) and slow (decreasing A₄₃₅) phases. Eight injections were performed with each substrate, and the average rates and amplitudes were determined. For IPP: A_f = 0.14 ± 0.01 AU, k_f = 34 ± 2.4 s⁻¹, A_s = −0.028 ± 0.009 AU, and k_s = 2.1 ± 0.4 s⁻¹. For (R)-[2-²H]-IPP: A_f = 0.14 ± 0.007 AU, k_f = 23 ± 0.9 s⁻¹, A_s = −0.043 ± 0.001 AU, and k_s = 0.9 ± 0.05 s⁻¹.

X-band EPR spectra recorded for reaction mixtures containing 80 µM IDI-2, 200 µM FMN, and 2 mM IPP. The FMN coenzyme was either photoreduced (A, light gray) or chemically reduced by 10 mM dithionite (A, black) or 10 mM NADPH (A, dark gray) prior to the addition of 2 mM IPP. After 2 min, the reactions were flash frozen and stored in liquid nitrogen until the spectra were recorded. The concentration of FMN_sem was determined to be 0.39, 0.86, and 0.72 µM for dithionite reduced, NADPH reduced, and photoreduced samples, respectively. These values correspond to 0.5, 1, and 0.9% of the total enzyme concentration. (B) In contrast to these reaction mixtures, the photoreduced E•FMN_ox•IPP mixture (gray) yielded 29.6 µM neutral semiquinone, corresponding to 37% of the total enzyme concentration. The spectrum of the NADPH reduced reaction mixture (black) is shown for comparison. Experimental conditions: temperature, 100 K; modulation amplitude, 5 G; microwave power, 1 mW; and spectrometer frequency, 9.601 GHz. The spectra are an average of five 3 min scans.

Pre-steady-state formation of DMAPP in the rapid-mix chemical quench experiment. The reaction contained 19.7 µM IDI-2, 100 µM FMN, 100 mM dithionite, and 520 µM IPP in 100 mM HEPES, 5 mM MgCl₂, and 1 mM DTT (pH 7.0 and 37 °C). Both enzyme and IPP solutions contained all of the other reaction components. Individual injections were quenched at varying times (0.003, 0.004, 0.005, 0.006, 0.007, 0.008, 0.009, 0.010, 0.030, 0.050, 0.100, 0.500, 0.750, and 1.000 s) by 25% HCl/MeOH after mixing IDI-2 and [1-¹⁴C]-IPP (specific activity = 1.92 µCi/µmol). The workup and extraction procedures to detect DMAPP are described in the Experimental Procedures. The concentration of DMAPP at each time point was calculated and normalized by the total enzyme concentration. The data fit a line with a slope of 1.4 ± 0.01 s⁻¹ and a y-intercept of 0.04 ± 0.01. The inset shows a closer view of the early time points.

Substrate deuterium KIE on V_max. (A) ¹H NMR assay showing the conversion of IPP (C4′ methyl singlet at δ = 1.72 ppm) to DMAPP ((E)-methyl singlet at δ = 1.70 ppm and (Z)-methyl singlet at δ = 1.66 ppm). (B) Plot showing the formation of DMAPP from IPP (white circles) and (R)-[2-²H]-IPP (black circles) as a function of time. The integrated (Z)-methyl peak areas of DMAPP were normalized by the 2 mM acetate signal (δ = 1.84 ppm) to calculate the concentration of DMAPP at each time point. The initial velocities were determined from these plots to be 0.170 ± 0.001 µM s⁻¹ for IPP and 0.093 ± 0.002 µM s⁻¹ for (R)-[2-²H]-IPP, yielding ^DV_max = 1.8 ± 0.04.

Solvent kinetic isotope effects on V_max. The initial velocity (v_o) of DMAPP formation at 269 µM IPP was determined in triplicate under anaerobic conditions at pL 7.0, 7.5, 8.0, and 8.5. The velocities in H₂O (white circles) were consistently higher than the velocities in D₂O (black circles). The corrected ^D₂OV_max was calculated to be 1.4 ± 0.08 at pL 8.0 (see Experimental Procedures). The ^D₂OV_max at pL 8.0 was taken to be the most accurate value because this range appears to be the pL-independent region of v_o for both solvents.

Isoprenoids comprise a large and ubiquitous class of metabolites that participate in numerous physiological processes (–). All isoprenoids are derived initially from the condensation of two isoprene units, isopentenyl pyrophosphate (IPP, 1) and dimethylallyl pyrophosphate (DMAPP, 2). Two biosynthetic pathways for the production of IPP and DMAPP are known, the mevalonate (MVA) pathway found in animals, fungi, and archaebacteria, and the non-mevalonate or methyl erythritol phosphate (MEP) pathway found in most eubacteria, green algae, and the chloroplasts of higher plants (, , ). In the MVA pathway, DMAPP must be generated from IPP by an isopentenyl diphophate/dimethylallyl diphosphate isomerase (IDI, ) (). In the MEP pathway, both IPP and DMAPP are coproduced from 4-hydroxy-3-methyl-2-butenyl diphosphate in the same enzymatic reaction (catalyzed by IspH). However, most of the organisms that utilize the MEP pathway still contain IDI, presumably to regulate the cytosolic pools of these two essential isoprenoid building blocks (–).

Two types of structurally unrelated IDIs, which likely operate by distinct chemical mechanisms, have been identified in nature. The type I IDI (or IDI-1), which has been extensively studied, uses divalent metal ions (Mg²⁺ and Zn²⁺) to elicit an acid/base mediated 1,3-antarafacial proton addition/elimination reaction through a 3° carbocation intermediate (–). The type II IDI (IDI-2), which was discovered more recently, requires flavin mononucleotide (FMN), a reduced nicotinamide adenine dinucleotide cofactor (either NADH or NADPH), and divalent metal ions (Mg²⁺) for catalysis (). Since its initial isolation from Streptomyces sp. CL190, IDI-2 enzymes have been found in many eubacteria and archaebacteria as part of either the MVA or the MEP biosynthetic pathways (–). Subsequent biochemical and structural characterization of IDI-2 (, –) has revealed that the IDI-2-FMN_ox complex is reduced with stoichiometric amounts of NAD(P)H via stereospecific transfer of the pro-S hydride and that the resultant IDI-2-FMN_red is capable of performing multiple turnovers (). Recent redox titrations of IDI-2 from Staphylococcus aureus and Thermus thermophilus in the presence and absence of IPP (1) have shown that the apoenzyme thermodynamically stabilizes the neutral flavin semiquinone in the presence of IPP (, ). It was also found that the apoenzyme reconstituted with 5-deazaFMN is inactive, while 1-deazaFMN supports catalysis (, ). These data suggest that the flavin coenzyme is not simply required to maintain the active site structure, which has been proposed (); rather, it plays an active role in the chemical transformation of IPP (1) to DMAPP (2). Cumulatively, these results are consistent with a mechanism involving a cryptic redox cycle (), where a single electron is transferred from FMN_red to IPP (1) to generate a semiquinone and a substrate radical (5 and 6, respectively, ) (, ). Substrate deprotonation and single electron transfer back to FMN_sem completes the isomerization and regenerates FMN_red for another round of catalysis. However, a recent study by Johnston et al. using radical clock mechanistic probes has called this single electron transfer mechanism into question ().

As an alternative to the electron transfer mechanism, IDI-2 may catalyze the isomerization reaction solely by acid/base chemistry (mechanisms A–D, ) in a manner similar to the IDI-1 catalyzed reaction (). All of these mechanisms are based on the following assumptions: (i) Substrate binding leads to the rapid accumulation of the neutral reduced FMN (4) in both the forward and the reverse directions (as observed in the single- and multiple-turnover stopped flow studies), (ii) the rate limiting step occurs before DMAPP formation in the forward direction (as indicated by the rapid-quench studies), and (iii) the rate limiting step involves acid/base chemistry (as suggested by the KIE studies). For simplicity, the substrate binding steps and the conversion of FMNH⁻ to FMNH₂ are not shown in . In one possible acid/base mechanism (), the FMN coenzyme may play a role in the stabilization of a 3° carbocation intermediate (7) through cation–π interactions, similar to the function proposed for a conserved tryptophan residue (Trp161 in E. coli) in the IDI-1 enzyme. This mechanism, however, has a few loopholes. For example, it is not clear as to why 5-deazaFMN does not support catalysis, unless this cation–π interaction depends strongly on the N5 atom or a hydrogen bonding network maintained by the N5 atom. Also, it is not clear as to why the neutral form of the reduced coenzyme (4) accumulates upon the addition of substrate, as the anionic form should be better suited to stabilize a carbocation intermediate through electrostatic interactions.