Synthesis, inhibitory and activation properties of prenyldiphosphate mimics for aromatic prenylations with ubiA -prenyltransferase

4-Hydroxybenzoate oligoprenyl transferase from E. coli ( ubiA -prenyl transferase) is a crucial enzyme for ubiquinone biosynthesis. It catalyzes the formation of 3-oligoprenyl-4-hydroxy-benzoates like geranyl hydroxybenzoate (GHB, 23 ) from geranyl pyrophosphate (GPP, 22 ). Several analogues and mimics of geranyl pyrophosphate have been prepared for an examination of their ability to inhibit the enzyme. 7,11-Dimethyl-3-oxododeca-6,10-dienoic acid ( 2 ), 3-hydroxy-7,11-dimethyldodeca-6,10-dienoic acid ( 3 ), 2-hydroxy-4,8-dimethyl-3,7-nonadienylphosphonic acid ( 4 ), and tripotassium [[(4 E )-5,9-dimethyldeca-4,8-dienyl]phosphinato](difluoro)methyl-phosphonate ( 5 ) were synthesized from geraniol. ω -2, ω -1-Dihydroxylated farnesyl diphosphate 6 was prepared from trans,trans -farnesol. All compounds were tested for enzyme inhibition in a competitive assay with natural substrate. The effect of these compounds on ubiA -prenyl- transferase activity varied substantially, ranging from almost full inhibition to, surprisingly, enhanced enzymatic activity at low concentrations by some compounds. A special, EDTA-modifyable magnesium effect is discussed as potential reason.


Introduction
Prenyl transferases catalyze the electrophilic alkylation of electron-rich acceptor substrates by the hydrocarbon moiety of allylic isoprenoid diphosphates. 1Some of the more common acceptors are carbon-carbon double bonds (synthesis of isoprenoid chains), aromatic rings (e.g. for the synthesis of respiratory quinones, vitamins E), 2 amino groups (modification of tRNAs), 3 and sulfhydryl moieties (modification of proteins such as ras-farnesyl transferase, an anti-proliferation target). 4The products of prenyl transfer reactions are ultimately converted into over 30000 naturally occurring isoprenoid compounds.Isoprenoid biosynthesis is essential in all organisms.In Escherichia coli, for instance, the isoprenoids ubiquinone and bactoprenol are required for respiration and cell wall biosynthesis, respectively.The aromatic polyprenyl transferase encoded by ubiA is involved in the biosynthesis of prenylated quinones (Scheme 1). 5 In vivo, the enzyme transfers diphosphorylated acyclic oligoprenyl moieties (diphosphorylated terpene alcohols) to the meta-position of 4-hydroxybenzoic acid (PHB). 6ossible candidates for mimics of geranyl diphosphate should fulfill three requirements: they should show affinity to a prenyl diphosphate synthase or transferase; they should be stable under study conditions, especially against hydrolysis; and they should be easy to synthesize.With these requirements in mind, we approached two different classes of candidates for our approach.The first class consists of non-hydrolysable diphosphate analogues with phosphorus-based acidic moieties, the second one of mimics where all phosphates are replaced by non-phosphorus (acidic) moieties.

COOH
A natural example of the latter group is chaetomellic acid (1, Figure 1), which was isolated from the coelomycete Chaetomella acutiseta.It inhibits farnesyl-protein-transferase at IC50values in the nanomolar range. 7Chaetomellic acid is a competitor to farnesyl pyrophosphate (FPP), and does not contain any phosphate moiety.We were interested in the synthesis of the similar β-keto acid 2 and β-hydroxy acid 3 which might act as competitive inhibitors of ubiA-p-hydroxybenzoate oligoprenyl transferase (Figure 2).
As examples of the first class, several non-hydrolysable allylic diphosphate mimics are known from the literature.These include phosphonophosphates, 8 phosphonophosphinates, 9 and diphosphonates, 10 where the methylene group replaces the oxygen between phosphorus and carbon and the bridging oxygen between the two phosphorus atoms.All these phosphinates are inhibitors, but the bisphosphonates with an ester bond linked to prenyl are still enzyme cleavable and thus potential substrates.6b,11,12 Therefore, we excluded the latter and decided to try β-hydroxyphosphonic acid 4 as an inhibitor halfway between class I and class II.Such compounds were investigated by Kang et al. with respect to their inhibitory activity of protein farnesyltransferase. 13 A pure class II substrate is the stable diphosphate analog 5. Finally, 10,11-dihydroxy-3,7,11-trimethyldodeca-2,6-dienyl diphosphate (6), i.e. oxidized FPP, was included.Although it is a hydrolysable substrate in principle, we knew from earlier experiments that it is accepted poorly by ubiA-transferase and thus, it may act as an inhibitor.In this report, we present the synthesis of these compounds, along with a study of their impact on the magnesium depended activity of aromatic ubiA-oligoprenyl transferase of E. coli.

Syntheses
The synthesis of the β-keto acid 2 and β-hydroxy acid 3 is shown in Scheme  The synthesis of the geranyl-based phosphonate 4 (Scheme 3) started with the Doering oxidation of geraniol to geranial (9), which was reacted with dimethyl methylphosphonate anion to give racemic 10.This methyl phosphonic ester was cleaved with trimethylsilyl bromide in the presence of 2,4,6-collidine as acid scavenger to afford the free hydroxyphosphonic acid 4. 14,15 An attempt to oxidize β-hydroxy phosphonate 3 to the β-keto phosphonate by Doering oxidation failed.Scheme 3. Synthesis of geranyl-based β-hydroxy phosphonic acid 4.
The synthetic route to phosphonate 5 (Scheme 4), a hydrolytically stable analog of diphosphates, was identical to one known for a farnesyl derivative. 14The difluoromethylene link is a better oxygen mimic than methylene.Thus, geraniol was subjected to a standard C 2 -homologation to form ester 11, followed by ester reduction and conversion of the resulting alcohol into bromide 12.The Arbuzov reaction 16 of 12 with excess of triethyl phosphate delivered in high yield diester 13, which by base-induced hydrolysis was converted into the mono acid.Subsequent DMF-catalyzed reaction with oxalyl chloride in benzene at room temperature afforded ethyl phosphonochloridoate 14.Treatment of the intermediate mono phosphonate with excess N,N-dimethylaminotrimethylsilane in dichloromethane prior to acid chloride formation resulted in a faster reaction and a cleaner transformation into ethyl phosphonochloridoate 14. 14,17 Subsequent C-P coupling reaction was performed by dropwise addition of a THF solution of 14 to a solution of the anion 15 at -78 °C.The anion 15 was formed by the reaction of LDA with diethyl difluoromethylphosphonate. 18 The resulting diester 16 was cleaved with TMS-I in the presence of 2,4,6-collidine as acid scavenger 14 to produce the desired product 5.
The synthesis of (E,E)-10,11-dihydroxyfarnesyl diphosphate started from all-trans farnesol (Scheme 5), protected as acetate 17.For selective epoxidation of the 10,11-double bond we followed a method developed by van Tamelen et al. 19 Farnesyl acetate was treated with NBS in aqueous tert-butyl alcohol 20 or aqueous 1,2-dimethoxyethane 21 to give the desired bromohydrin 18 in 29% and 90% yield, respectively.Base induced the formation of epoxide 19, 21 which in turn, was ring-opened with aqueous perchloric acid to give triol 20. 22

UbiA-Transferase inhibition
The compounds described above were tested for their ability to inhibit ubiA-transferase using a membrane fraction of a cell disruption of the E. coli K12 strain pALMU3. 24Compounds were assayed in competition to 1 mM geranyl diphosphate (GPP) at concentrations ranging up to 1 mM (Scheme 6).These results, expressed as the geranyl hydroxylbenzoate (GHB) formation in the presence of the compound divided by the GHB formation in the absence of the compound (i.e.relative conversions) are shown in Figures 3 and 4. * In this assay, the membrane fraction contained 32 mg protein/mL, all others contained 0.67 mg protein/mL.For standard assay conditions see Experimental Section.
Only β-hydroxy acid 3 and β-keto acid 2 showed significant inhibitory activity regarding the conversion of geranyl diphosphate (Figure 3). 25 Their IC50 values are 0.36 and 0.58 mM, respectively.At 1 mM concentration, β-hydroxy carboxylate 3 showed almost complete inhibition of ubiA-transferase activity.β-Hydroxy phosphonate (from 4, remaining activity: 90% at 1 mM, data not shown) and difluoro phosphinophosphonate 5 did not show significant inhibitory activity.Although these molecules are structurally more related to the natural substrate than the carboxylic acids, the lower binding affinities have been attributed to differences in the pK a of the phosphate versus the phosphonate groups. 26Difluoromethylene analogues more closely match the pK a , but the fluorine atoms introduce steric interactions and other properties not found in the natural diphosphate substrates. 26Another reason for the low activity of 4 and 5 might be the structural difference compared to the natural substrates because in 4 the distance of the diphosphate mimicking moiety to the first double bond is shorter, in 5 it is longer by one atom than in natural oligoprenyl diphosphates.
(E,E)-10,11-Dihydroxyfarnesyl diphosphate 6 is only very poorly accepted as substrate, although sterically it must be acceptable to the enzyme as has been shown by us before in a substrate model. 12Surprisingly, under standard conditions as described above, its inhibitory effect is very weak too.
An interesting pattern was seen for inhibitors 3 and 5. Before exerting inhibitory activity, their addition initially leads to enhanced conversion, i.e. they act as activators.This effect is dependent on the concentration not only of the inhibitor but also of the enzyme concentration (Figure 4).β-Hydroxy acid 3 is a potent inhibitor with a biocatalyst membrane fraction containing 0.67 mg protein/mL.After an initial 10% increase in activity at 0.05 mM concentration (vide infra) inhibitory properties prevail and at 1 mM concentration of 3, enzyme activity is almost 0%.In the case of compound 5 the activation effect is even larger and up to 47% increase of activity at 0.1 mM concentration is observed at low enzyme concentration.The results for β-hydroxy carboxylate 3 (■, □) and for phosphinato(difluoro)methylphosphonate 5 (▲, ) are shown with squares and triangles, respectively.Empty data points □, : the membrane fraction contained 0.67 mg protein/mL; filled data points ■, ▲: the membrane fraction contained 2.68 mg protein/mL.For standard assay conditions see Experimental Section.
From our earlier experiments we know that even a small reduction in the concentration of the magnesium ions results in an increase of product formation (through enhanced enzyme stability).6b,27 Although elevated concentrations of magnesium ions ensure a high initial rate of the enzymatic reaction, they cause the enzyme denaturation much faster then in experiments where the MgCl 2 concentration is reduced.Thus, high salt concentration ultimately results in lower turnover.We suppose that the dual role played by MgCl 2 is particularly responsible for the increased enzymatic activity observed with small amounts of inhibitor.Because the tested inhibitors are diphosphate mimics, they are also able to form complexes with Mg 2+ -ions.Due to this complexation, especially with phosphinato(difluoro)methylphosphonate 5, an activity increase up to 47% was observed.In order to prove this hypothesis, an alternative Mg 2+scavenger, which should not have any inhibitory activity, was applied to the system.Indeed, experiments with ethylenediaminetetraacetic acid (EDTA) led to an enhancement of activity, similar in concentration dependence as observed before, but without any inhibition of the enzyme at higher concentrations (Figure 5).Of course, such an effect is more evident at low enzyme concentration, i.e. in systems where the Mg 2+ -concentration relative to that of the enzyme is high; thus [Mg] 2+ reduction results in a higher effect.Thus, the different activation/inhibition profiles of the compounds 2-6, displayed e.g. in Figure 3, result from an overlap of several contrasting effects: a competitive inhibition at the enzyme, the magnesium ion binding properties of inhibitor vs. GPP vs. enzyme, and a resultant irreversible influence on a (concentration dependent) enzymatic activity.
In summary, the different substances described show varied effects on ubiA-transferase.For all compounds, a very slightly to significantly increased enzymatic activity was observed at low concentration.This can be attributed to the Mg 2+ -complexating activity inherent to diphosphate mimics.At higher concentrations this positive effect is overruled by the inhibitory activity.The two best inhibitors were the β-oxidized carboxylic acids 2 and 3, the latter is clearly superior.β-Keto acids like 2 may act in their enol form and thus may also be considered as β-hydroxy Experimental Section General Procedures. 1 H and 13 C NMR spectra were recorded on a Varian Mercury 300 and a Varian Mercury 400 instrument using CDCl 3 as a solvent (unless noted otherwise) and (CH 3 ) 4 Si or CDCl 3 ( 13 C, δ 77.00) as internal standards. 31P and 19 F NMR spectra were recorded on a Varian Mercury 400. 31 P Chemical shifts are reported in ppm relative to an external standard of 85% H 3 PO 4 .Non-quantitative measurements were performed in the hydrogen decoupled modus; quantitative measurements were performed in the hydrogen-coupled modus.A known volume of an aqueous solution of diammonium phenylphosphonate (1.0 M) was added to the samples as a quantitative internal standard. 19F data were obtained using CFCl 3 as an external reference.Ion electrospray ionization (ESI) mass spectra were recorded on a Finnigan MAT TSQ 7000 or with a API 150Ex (Applied Biosystems) equipped with a turbo ionspray source.High-resolution ESI mass spectra were recorded on a Bruker 70e FT-ICR (Bruker Daltonics, USA) using nitrogen as drying gas at 150 °C.IR spectra were measured on a Bruker IFS 28 as thin oil films between KCl plates or as KBr pellets.HPLC was performed on a HP 1090 with integrated photo diode array detector (Multosphere 100-5 µm RP 18) or a Merck D-7000 (Lichrosphere 100-5µm RP 18).Samples were chromatographed with methanol/water mixtures containing 0.2% formic acid 80:20, flow rate: 1 mL/min.Tetrahydrofuran and diethyl ether were freshly distilled from sodium/benzophenone, dichloromethane was freshly distilled from calcium hydride.Other solvents and chemicals obtained from commercial sources were used without further purification.All moisture and air sensitive reactions were conducted under argon in vacuum-dried glassware.Flash column chromatography was carried out on silica [Merck: Kieselgel 60, particle size 0.040-0.063mm (230-240 Mesh ASTM), Art.-No.9385, Baker: silica for flash chromatography, particle size 0.030-0.060mm, Art.-No.7024-02] under a pressure of 1.4-1.6 bar.Thinlayer-chromatography was performed using silica plates from Merck (Kieselgel-60 F 254 on aluminum sheets with fluorescence indicator, Prod.-No.5554).
. To a solution of 16 (0.24 g, 0.54 mmol) in dry dichloromethane (3 mL) under argon were added 2,4,6-collidine (0.24 mL, 1.82 mmol) followed by iodotrimethylsilane (0.50 mL, 3.81 mmol) at 0 °C drop-wise over 5 min.The reaction was stirred at 0 °C for 3 h, the solvent was evaporated and the residue was evacuated under vacuum.Aqueous 2 M KOH (2 mL) was added, and the resulting solution was extracted with ether (3 x 5 mL).The aqueous phase was freeze-dried and the organic compound was extracted with methanol (3 x 5 mL).The solvent was evaporated to produce 5 (0.14 g, 78%) as a white, amorphous powder.(7).A suspension of NaH (0.75g, 60% suspension in mineral oil, 18.7 mmol) in THF (50 mL) was cooled to 0 °C, and a solution of methyl acetoacetate (2 g, 17 mmol) in THF (5 mL) was added dropwise.The mixture was stirred at 0 °C for 1 h.Then n-BuLi (11.7 mL, 1.6 M solution in THF, 18.7 mmol) was added dropwise at -5 °C.The mixture was stirred for 1 h while warming to room temperature and was subsequently cooled to 0 °C.Subsequently, a solution of geranyl bromide (1.84 g, 8.5 mmol) in THF (5 mL) was added dropwise.After stirring the mixture at room temperature for 3.5 h saturated aqueous NH 4 Cl solution (20 mL) was added.The aqueous layer was extracted with ether (3 x 50 mL) and the organic layers were combined.The organic phase was washed with water (40 mL) and brine (40 mL), dried (Na 2 SO 4 ) and concentrated.Flash chromatography (petroleum ether/ethyl acetate, 5:1) provided 7 (1.42 g, 66%) as a pale yellow oil. 1

Ethyl (4E)-5,9-dimethyldeca-4,8-dienoate (11).
To a solution of trans-geraniol (5.3 mL, 30 mmol) in ether (25 mL) at -20 °C was added a solution of phosphorus tribromide (1.4 mL, 15 mmol) in ether (15 mL) within 10 min, and the reaction mixture was stirred for 4 h.The reaction was quenched with water, extracted with petroleum ether, washed in turn with water, saturated aqueous NaHCO 3 , and brine.The organic layer was dried (MgSO 4 ) and evaporated at 30 °C to provide (E)-geranyl bromide (6.4 g, 99%) of as a labile yellow liquid.To a suspension of NaH (1.44 g of 60% suspension in mineral oil, 60 mmol) in dry THF (150 mL) under argon was slowly added diethyl malonate (9.14 mL, 60 mmol) at room temperature.The resulting solution was stirred for 30 min, and a solution of geranyl bromide (4.24 g, 19.6 mmol) in THF (10 mL) was added.After stirring for 6 h the reaction mixture was quenched with saturated aqueous ammonium chloride.After addition of ether (300 mL) the organic layer was separated and washed with brine (2 x 100 mL), dried (MgSO 4 ), evaporated, and the bulk of the diethyl malonate removed by Kugelrohr distillation in high vacuum to afford the crude product as residue (7.90 g, max.19.6 mmol).
To a stirred solution of the alcohol (3.1 g, 17.0 mmol) in dichloromethane (40 mL) was added triethylamine (4.8 mL, 35.2 mmol) followed by dropwise addition of methanesulfonyl chloride (1.9 mL, 19.3 mmol) at 0 °C within 15 min.After stirring at 0 °C for 1.5 h, the reaction mixture was diluted with dichloromethane, washed rapidly with hydrochloric acid (10%, 40 mL), saturated NaHCO 3 (40 mL), and brine (40 mL).The organic phase was dried (MgSO 4 ) and evaporated to produce the crude mesylate (4.4 g, 99 %) as a colorless oil.MS (EI): m/z (%) 260 (1) [M] + , 245 (4) [M-CH 3 ] + , 217 (49), 149 (39), 121 (53), 107 (44), 93 (99), 79 (100), 67 (99).The addition of anhydrous lithium bromide (4.6 g, 52.5 mmol) to a solution of the mesylate (4.4 g, 16.9 mmol) in dry THF (30 mL) at room temperature resulted in a mild exothermic reaction.The resulting suspension was stirred at room temperature for 23 h, and after dilution with ether the organic phase was washed with water (2 x 30 mL) and brine (30 mL), dried (MgSO 4 ) and evaporated to provide crude bromide 12 (4.0g, 93 %) as a pale yellow liquid.The crude product 12 without purification was immediately used for the next step.MS (EI): m/z (%) 244 (3 To a stirred solution of the monoacid (2.9 g, 10.8 mmol) in dichloromethane (10 mL) was added under argon N,N-dimethyltrimethylsilylamine (1.4 mL, 21.8 mmol).The mixture was stirred at room temperature for 2 h.The solvent was evaporated, the residue was dissolved in benzene (10 mL), and the solvent was evaporated under vacuum for 30 min.The residue was dissolved in dichloromethane (8 mL) containing 4 drops of DMF, and oxalyl chloride (1.5 mL, 18 mmol) was added under argon at 0 °C within 10 min.The solution was stirred at 0 °C for 75 min and then at room temperature for 45 min.After evaporation of the solvent the residue was dissolved in benzene, which in turn, was evaporated under high vacuum to give the unstable acid chloride 14 (3.15 g, 99%) as a labile orange oil that was immediately used without purification for the C-P coupling reaction.

Figure 4 .
Figure 4. GHB synthesis from geranyl diphosphate as a function of inhibitor and enzyme concentration.The results for β-hydroxy carboxylate 3 (■, □) and for phosphinato(difluoro)methylphosphonate 5 (▲, ) are shown with squares and triangles, respectively.Empty data points □, : the membrane fraction contained 0.67 mg protein/mL; filled data points ■, ▲: the membrane fraction contained 2.68 mg protein/mL.For standard assay conditions see Experimental Section.

Figure 5 .
Figure 5. GHB synthesis from geranyl diphosphate as a function of EDTA concentration (Mg 2+complex formation).□: Membrane fraction with 0.67 mg protein/mL; ■: membrane fraction with 2.68 mg protein/mL.For standard assay conditions see Experimental Section.