A Click Chemistry Approach towards Flavin-Cyclodextrin Conjugates—Bioinspired Sulfoxidation Catalysts

A click chemistry approach based on the reaction between alkynylflavins and mono(6-azido-6-deoxy)-β-cyclodextrin has proven to be a useful tool for the synthesis of flavin-cyclodextrin conjugates studied as monooxygenase mimics in enantioselective sulfoxidations.


Introduction
Within the last three decades, flavinium salts have been shown as useful tools for the organocatalytic activation of hydrogen peroxide and oxygen allowing various oxidations to proceed under mild conditions [1][2][3].In particular, extensive research on H 2 O 2 -sulfoxidations catalyzed by flavinium salts has resulted in several useful synthetic procedures for mild, chemoselective and stereoselective transformations of sulfides to sulfoxides [4][5][6][7][8][9].In these artificial systems, flavinium salts Fl mimic the function of flavin co-factors in monooxygenases [10] via the in situ formation of flavin-4a-hydroperoxide FlOOH, which subsequently oxidizes the substrate (Scheme 1).
Flavoenzymes usually give the corresponding oxygenated products chemoselectively and stereoselectively, which is provided by the stereoselective transfer of activated oxygen from flavin-4a-hydroperoxide to the substrate accommodated in an active site of the enzyme via non-covalent bonding interactions [11][12][13][14].The efficiency of an enzyme is usually sensitive to the modification in its active site as well as to a change in the substrate; both of which can lead to a significant loss of stereoselectivity [12,[14][15][16].Cyclodextrins have been extensively studied as biomimetic catalysts by utilizing their hydrophobic cavity for non-covalent binding of lipophilic substrates in neat aqueous media [17][18][19].Moreover, analogously to an enzyme active site, the cyclodextrin cavity offers a chiral environment for asymmetric transformations [20][21][22][23][24][25].
Souza and coworkers investigated artificial enzymes based on flavin and cyclodextrin moieties linked together with ether or ester functionality.The conjugates were tested in oxidation of mercaptans, NADH + models and in photooxidation of benzyl alcohols [26][27][28].Recently, we have designed sulfoxidation catalysts 1 and 2 (Figure 1) containing a redox-active flavinium "co-factor" and chiral substrate-binding site made using αor β-cyclodextrin; both parts were attached by amide bonds [29][30][31].The efficiency of these biomimetic catalysts was shown to depend strongly on the structure of the "co-factor", the relative position of both parts of the catalyst as well as the size of the cyclodextrin cavity.The best sulfoxidation catalysts were found to be alloxazinium-β-cyclodextrin conjugates 1a and 1b, which achieved an ee of up to 91% for tert-butyl methyl sulfide, 80% ee for aromatic sulfides and up to 64% ee for aliphatic sulfides with only 1 mol % catalyst in neat aqueous media.It was noteworthy that a small change in the catalyst structure such as turning the alloxazine relative to the cyclodextrin cavity (see catalyst 2) caused the stereoselectivity of the reaction to be completely lost [31].Flavoenzymes usually give the corresponding oxygenated products chemoselectively and stereoselectively, which is provided by the stereoselective transfer of activated oxygen from flavin-4ahydroperoxide to the substrate accommodated in an active site of the enzyme via non-covalent bonding interactions [11][12][13][14].The efficiency of an enzyme is usually sensitive to the modification in its active site as well as to a change in the substrate; both of which can lead to a significant loss of stereoselectivity [12,[14][15][16].Cyclodextrins have been extensively studied as biomimetic catalysts by utilizing their hydrophobic cavity for non-covalent binding of lipophilic substrates in neat aqueous media [17][18][19].Moreover, analogously to an enzyme active site, the cyclodextrin cavity offers a chiral environment for asymmetric transformations [20][21][22][23][24][25].
Souza and coworkers investigated artificial enzymes based on flavin and cyclodextrin moieties linked together with ether or ester functionality.The conjugates were tested in oxidation of mercaptans, NADH + models and in photooxidation of benzyl alcohols [26][27][28].Recently, we have designed sulfoxidation catalysts 1 and 2 (Figure 1) containing a redox-active flavinium "co-factor" and chiral substrate-binding site made using α-or β-cyclodextrin; both parts were attached by amide bonds [29][30][31].The efficiency of these biomimetic catalysts was shown to depend strongly on the structure of the "co-factor", the relative position of both parts of the catalyst as well as the size of the cyclodextrin cavity.The best sulfoxidation catalysts were found to be alloxazinium-β-cyclodextrin conjugates 1a and 1b, which achieved an ee of up to 91% for tert-butyl methyl sulfide, 80% ee for aromatic sulfides and up to 64% ee for aliphatic sulfides with only 1 mol % catalyst in neat aqueous media.It was noteworthy that a small change in the Molecules 2015, 20 3 Herein, we report a new synthetic approach towards flavin-cyclodextrin conjugates based on click chemistry, specifically the copper-catalyzed azide-alkyne [3+2] cycloaddition (CuAAC) [32][33][34][35].We used the same co-factor (alloxazine) and cavity (β-cyclodextrin) as in 1, to prove if the CuAAC methodology is compatible with the alloxazine unit since although click chemistry is well established among cyclodextrins [34,36], its use among flavins is still limited to only one isoalloxazine example [37,38].Herein, we report a new synthetic approach towards flavin-cyclodextrin conjugates based on click chemistry, specifically the copper-catalyzed azide-alkyne [3+2] cycloaddition (CuAAC) [32][33][34][35].We used the same co-factor (alloxazine) and cavity (β-cyclodextrin) as in 1, to prove if the CuAAC methodology is compatible with the alloxazine unit since although click chemistry is well established among cyclodextrins [34,36], its use among flavins is still limited to only one isoalloxazine example [37,38].

Synthesis of Conjugates 3 and 4
The strategy used towards the synthesis of new catalysts 3 and 4 corresponded to the synthetic pathway used for derivatives 1 and 2 (see Scheme 2).The synthetic route started with the preparation of flavin and cyclodextrin subunits functionalized with suitable groups for interconnecting both parts followed by the transformation of the neutral alloxazine into the corresponding alloxazinium moiety.Alkylation/quaternization of the N5 nitrogen is essential for the flavins to be active in the artificial oxygenations.Amide coupling of mono(6-amino-6-deoxy)-β-cyclodextrin with the corresponding flavin carboxylic acid used in the synthesis of 1 and 2 was replaced by a CuAAC between mono(6-azido-6-deoxy)-β-cyclodextrin 7 and an alloxazine possessing a terminal triple bond in its side chain.1-(Alkynyl)-3-methylalloxazine 5 and 1-methyl-3-propargylalloxazine 6 were prepared by alkylation of 3-methylalloxazine or 1-methylalloxazine, respectively, with a triple-bond containing alkylating agent.Mono(6-azido-6-deoxy)-β-cyclodextrin 7 was readily available using standard procedures starting from commercial β-cyclodextrin [39].
Molecules 2015, 20 4 the corresponding conjugates were formed in moderate yield with the exception of derivative 8b, which was not formed using method A despite great effort (Table 1).Lower yields of conjugates could be a result of redox activity of flavin moiety supporting undesired oxidation of Cu(I) during cycloaddition with less reactive azidocyclodextrine 7. Therefore we modified reaction conditions using copper sulfate, sodium ascorbate (to produce Cu(I) in-situ) and the tris[(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]amine (TBTA) ligand which is known to stabilize Cu(I) towards disproportionation and oxidation (method B) [40].Under these conditions, 8b was formed in good yield.Method B even gave the conjugates 8 and 9 in substantially higher yields, regardless of the position of the alkynyl group on the alloxazine subunit and the length of the spacer (Table 1).Importantly, the triazole spacer tolerated the subsequent reaction steps, which led to the transformation of the neutral alloxazine to alloxazinium moiety, i.e., the introduction of an ethyl group via reductive amination and oxidation, which proceeded in a quantitative yield in all cases.
Scheme 2. Synthesis of catalysts 3 and 4 using click chemistry approach.Method A: CuI, N,N-diisopropylethylamine, DMF; method B: TBTA, sodium ascorbate, CuSO4, DMF.The preliminary tests of the CuAAC reaction between 3-methyl-1-propargylalloxazine 5a and the model azides (benzylazide or p-tolylazide) gave the corresponding triazoles in excellent yield without decomposition of the alloxazine skeleton using conventional reaction conditions [32][33][34][35], i.e., copper iodide and Hünig's base in N,N-dimethylformamide (DMF) (see Supplementary Materials).Therefore, this method (A) was applied for the cycloaddition of alkynylalloxazines 5 and 6 with azide 7.In all cases, the corresponding conjugates were formed in moderate yield with the exception of derivative 8b, which was not formed using method A despite great effort (Table 1).Lower yields of conjugates could be a result of redox activity of flavin moiety supporting undesired oxidation of Cu(I) during cycloaddition with less reactive azidocyclodextrine 7. Therefore we modified reaction conditions using copper sulfate, sodium ascorbate (to produce Cu(I) in-situ) and the tris[(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]amine (TBTA) ligand which is known to stabilize Cu(I) towards disproportionation and oxidation (method B) [40].Under these conditions, 8b was formed in good yield.Method B even gave the conjugates 8 and 9 in substantially higher yields, regardless of the position of the alkynyl group on the alloxazine subunit and the length of the spacer (Table 1).Importantly, the triazole spacer tolerated the subsequent reaction steps, which led to the transformation of the neutral alloxazine to alloxazinium moiety, i.e., the introduction of an ethyl group via reductive amination and oxidation, which proceeded in a quantitative yield in all cases.

Catalytic Activity
To demonstrate the activity of the modified biomimetic systems in a catalytic oxidation, the prepared conjugates were subjected to preliminary screening in the enantioselective sulfoxidation of alkyl methyl sulfides and aryl methyl sulfides using hydrogen peroxide as a model reaction.The oxidation was performed in a buffered aqueous medium under the reaction conditions recently used in reactions catalyzed by analogous conjugates bearing an amide linker.Among the conjugates bearing a triazole spacer, the efficiency in the enantioselective sulfoxidations decreased with an increase in the spacer length and by "turning" the alloxazine moiety going from 3 to 4 (see Supplementary Materials).In Table 2, the reaction conversions and enantioselectivities of the sulfoxidations catalyzed by 3a, the most efficient conjugate with triazole linker, are compared with the catalytic activity of 1b, one of the most efficient conjugate from "amide" series with the same length of spacer.The reaction conversions of the non-catalyzed oxidation reactions (blanks) are given for information.The observed enantioselectivities obtained using 3a were lower than those found when using 1b but still comparable or even better than those obtained using other flavin-cyclodextrin conjugates or other chiral flavinium catalysts [7,9,[29][30][31].Interestingly, the enantioselectivities obtained using 3a were relatively high with difficult substrates such as benzyl, hexyl and butyl methyl sulfides, which are usually oxidized with low enantioselectivities.The lower enantioselectivities compared to 1b are probably a result of the lower rate of reaction found when using 3a, which is demonstrated by the lower reaction conversions obtained after 1 h; while complete conversion was observed after 60 min when using catalyst 1 [29][30][31], the reaction was not finished when using catalyst 3a.This fact causes that the non-stereoselective oxidation of the substrate with hydrogen peroxide (blank) competes significantly with the reaction catalyzed by 3a.This was in accordance with the observation that a higher catalyst loading increased the enantioselectivity of the reactions up to 50% ee in the case of substrates, which are susceptible to non-catalyzed oxidation (e.g., t-BuSCH 3 or BuSCH 3 ).The results show the replacement of amide by triazole spacer influences the catalytic activity of flavin-cyclodextrin conjugates.Most probably higher hydrogen bonding ability of amide function compared to triazole could positively influence the efficiency of amide type catalysts 1. Hydrogen bonds could (i) increase the intrinsic reactivity of hydroperoxide function in flavin-4a-hydroperoxide during oxygen transfer to a substrate (see Figure S6 in Supporting Materials for illustration) [41]; and (ii) affect relative arrangement of flavin and cyclodextrin subunits.Nevertheless, as shown recently [31], the catalytic activity of flavin-cyclodextrin conjugates is very sensitive on only minor changes in their structure and thus, further studies are necessary to fully understand the structure activity relationship.
Although flavin-cyclodextrin conjugates do not achieve enantioselectivities exhibited by enzymes (see Table 3), they seem to be promising biomimetic catalysts of sulfoxidations.They provide the oxidations within a relatively short time and in contrast to enzymes, they usually work at higher substrate concentrations.Biocatalytic systems with monooxygenases are more complicated as they require regeneration of NADPH.Cofactor regeneration is usually provided by glucose-6-phosphate dehydrogenase using glucose-6-phosphate as sacrificial reductant.On the other hand biocatalytic systems use oxygen as stoichiometric oxidant, which, advantageously, avoids any background (non-stereoselective) oxidation without participation of the catalyst since oxygen itself is not reactive enough to oxidize most sulfides under mild conditions.Undesired non-catalyzed oxidation unfortunately occurs in oxidations with hydrogen peroxide (see blank experiments in Table 2) thus decreasing overall stereoselectivity.Probably, use of oxygen in place of hydrogen peroxide in oxidations with flavin-cyclodextrin conjugates, analogously to enzymes (combined with a sacrificial reducing agent [2]), would be a way to eliminate some background non-selective oxidations.The other way to improve our catalytic systems could be to optimize the solvent conditions or to increase the reactivity of flavin subunit by introduction of a strong electron-withdrawing group to flavin benzene ring [6].

General Procedure
Substituted alloxazine was dissolved in dry N,N-dimethylformamide (DMF) and 2 equivalents of alkynylbromide and 5 equivalents of potassium carbonate were added.Reaction mixture was stirred under nitrogen atmosphere for 24 h.The reaction was followed by TLC (chloroform/methanol 100:3).Solid salts were filtered off and solvent was evaporated in vacuo.Residue was dissolved in dichloromethane and washed with water.Water layer was separated and extracted with dichloromethane.Collected organic layers were washed with water and dried with sodium sulfate.After evaporation of solvents, solid product was dried in vacuo.

Click Reactions
General Procedure A The mixture of alkynyl alloxazine 5 or 6 and 6-azido-6-deoxy-β-cyclodextrin 7 (1.2 equiv.) was stirred with copper(I) iodide (3 mol %) and N,N-diisopropylethylamine (2 equiv.) in dry N,N-dimethylformamide (DMF) at room temperature for 24 h.Conjugate was precipitated after addition of acetone and separated by centrifugation.Crude product was purified by flash chromatography on the column with reversed phase (gradient MeOH/H 2 O 1:9ÑMeOH/H 2 O 2:3).The obtained solid product 8 or 9 was dried in vacuo.

Catalytic Oxidations
Catalytic oxidations of alkyl or aryl methyl sulfides were performed in 1 mL thick-walled screw-capped vial (Supelco).The reaction mixtures were prepared by addition of sodium-phosphate buffer (pH = 7.5, 0.05 M, 300 µL), liquid substrate (3 ˆ10 ´5 mol), appropriate volume of catalyst solution (1 mol %-5 mol %) and finally aqueous hydrogen peroxide (2.3 mol equiv.).Stirring of the reaction mixture was provided by wrist action shaker (900 rpm) for 1 h at 25 ˝C.Then the reaction was quenched by addition of a solution of sodium dithionite in water (1.4 M, 170 µL).The mixture was extracted with CDCl 3 (3 ˆ0.5 mL).Collected organic layers were dried with molecular sieves (3Å) and used for 1 H-NMR determination of the conversion.Enantiomeric excess was determined by HPLC analysis on column with chiral stationary phase of the sample obtained by evaporation of CDCl 3 and dissolving a residue in heptane (0.5 mL).

Conclusions
We have demonstrated the CuAAC concept is useful for the preparation of flavin-cyclodextrin conjugates.The click chemistry represents alternative approach to the original procedure based on amide coupling and seems to be slightly more advantageous especially by reducing the number of reaction steps and improving the overall yield.Recently, Rotello found the isoalloxazine ring tolerates "click chemistry" when preparing flavin-polymer conjugates [37,38]; we have shown the CuAAC approach was tolerated by alloxazines.Therefore, it seems CuAAC can be used as a general tool for the construction of flavin (isoalloxazin or alloxazine) based supramolecular systems, which can be useful not only in organocatalysis, but also in photocatalysis [3,[26][27][28] and molecular recognition [49].The new conjugates show lower efficiency in the enantioselective sulfoxidation reaction when compared with catalyst 1, which possesses an amide linker.However, 3a is comparable with the other chiral flavin catalysts [1][2][3] and even overcomes efficiency of flavin conjugates with amide linker in butyl methyl sulfide oxidation achieving ee up to 50%.One should bear in mind that cyclodextrin-based catalysts usually achieve rather moderate enantioselectivities and the results found with conjugate 1 are rather exceptional.The extreme sensitivity to small structural changes observed among the flavin-cyclodextrin conjugates is a typical phenomenon that also occurs in the natural enzymes and its elucidation is currently being investigated in our laboratory.
a For procedures, see Experimental Section; b Product formation was not observed.
a For procedures, see Experimental Section; b Product formation was not observed.

Table 2 .
H 2 O 2 -sulfoxidations catalyzed by conjugate 3a and comparison with data for 1b and non-catalyzed reaction a .Conditions: substrate (0.1 mmol), H 2 O 2 (2.3 equiv.),phosphate buffer pH 7.5, 25 ˝C, catalyst loading 1 mol % (related to the substrate) if not stated otherwise; vigorous shaking for 1 h; b conversion determined by 1 H-NMR; c ee determined by HPLC on a chiral stationary phase (see Supplementary Material for details); d oxidations without catalyst; e data from ref. [30]; f quantitative conversion; g 5 mol % of the catalyst. a

Table 3 .
Comparison of the catalytic activity of flavin-cyclodextrin conjugates 1a and 3a with enzymes in sulfoxidations of selected sulfides.