Asymmetric Reductive Carbocyclization Using Engineered Ene Reductases

Abstract Ene reductases from the Old Yellow Enzyme (OYE) family reduce the C=C double bond in α,β‐unsaturated compounds bearing an electron‐withdrawing group, for example, a carbonyl group. This asymmetric reduction has been exploited for biocatalysis. Going beyond its canonical function, we show that members of this enzyme family can also catalyze the formation of C−C bonds. α,β‐Unsaturated aldehydes and ketones containing an additional electrophilic group undergo reductive cyclization. Mechanistically, the two‐electron‐reduced enzyme cofactor FMN delivers a hydride to generate an enolate intermediate, which reacts with the internal electrophile. Single‐site replacement of a crucial Tyr residue with a non‐protic Phe or Trp favored the cyclization over the natural reduction reaction. The new transformation enabled the enantioselective synthesis of chiral cyclopropanes in up to >99 % ee.

Over the last decades,the power of enzymatic catalysis has been recognized as an important tool for the stereoselective synthesis of active pharmaceutical ingredients,agrochemicals, and flavor compounds on al aboratory and industrial scale. However,biocatalysis has been largely applied for functionalgroup transformations that follow the natural reactivity of the applied enzymes,s uch as hydrolases,r eductases,o xidases,o r transaminases. [1] Theuse of enzymes for CÀCbond formation is less well established, although notable exceptions exist. Aldolases,h ydroxynitrile lyases,t hiamine diphosphate dependent enzymes,a nd terpene cyclases have been used for this type of reactions. [2] There is astrong desire to expand the biocatalytic toolbox with reactions that surpass the established metabolic pathways.T herefore,e ngineered and artificial metalloenzymes have been developed for biocatalytic CÀ Cb ond formations,i ncluding olefin cyclopropanation, [3] Suzuki coupling, [4] Diels-Alder reaction, [5] and others. [6] Herein, we report an ew type of enzymatic C À C-bond formation in which ac ombination of substrate design and protein engineering enabled asymmetric reductive cyclization using ene reductases.E ne reductases from the Old Yellow Enzyme (OYE) family are enzymes that reduce electrondeficient alkenes,a sp resent in a,b-unsaturated carbonyl compounds.The synthetic potential of this enzyme family for biocatalytic applications has been recognized in recent years. [7,8] According to the accepted mechanism, ah ydride is delivered from the reduced flavin mononucleotide (FMN) cofactor to the b-carbon to form an enolate,w hich is subsequently protonated with the assistance of Ty r-OH as aproton source. [9] We envisioned the use (or modification) of this enzyme family for reductive carbocyclizations by offering substrates that exhibit an additional internal electrophile and thus react intramolecularly with the generated enolate intermediate to produce cyclized products. [10,11] Toward this goal, carbonyl groups,alkylhalides,orepoxides could be used as the electrophile (Scheme 1).
Since the wild-type (WT) enzymes OPR3 (12-oxophytodienoic acid reductase 3) from tomato (Solanum lycopersicum)a nd YqjM from Bacillus subtilis have already shown their value in asymmetric reduction of alkenes in biocatalysis, [12] we used these enzymes to explore reductive carbocyclization reactions with substrates bearing various electronwithdrawing substituents and featuring w-halo alkyl groups with different chain lengths,w hich should give rise to different ring sizes (Scheme 2; Table 1). Interestingly,e ven the wild-type enzymes displayed measurable levels of reductive cyclization when 1a-Br was offered as as ubstrate (Table 1, entries 1a nd 2). However,w hen (E)-4-chlorobut-2-enal (1a-Cl)w as converted with wild-type OPR3 and YqjM, the natural reduction pathway was dominant over the reductive cyclization reaction (Table 1, entries 3a nd 4), thus indicating that with ac hlorine leaving group,t he g-carbon lacks sufficient electrophilicity.T he a,b-unsaturated ketone substrate 1b-Br was also transformed into ac yclic product though in lower amounts when compared to the more reactive aldehyde substrate 1a-Br (Table 1, entries 5a nd 6).
These initial experiments showed that alkene reduction and enolate trapping are competing pathways.F ollowing our proposal (Scheme 1), we reasoned that the desired enolate alkylation could be favored if protonation of the enolate could be prevented or suppressed by removing proton donors and residual water molecules from the active site.W e assumed that exchanging the proton donor (Tyr190 in OPR3) [9c] for an aprotic and apolar amino acid (Phe,T rp) should lead to beneficial effects because of an inability to directly reprotonate the intermediate as well as displacement of water from the active site.I ndeed, the OPR3 variants Y190F and Y190W showed a2 -fold improvement in the formation of cyclopropane carbaldehyde from 1a-Br com-pared to the wild-type enzyme (Table 1, entries 7and 8). For engineering YqjM, the critical Ty rr esidue at position 169 [13] was replaced with Phe or Trp. TheP he variant of YqjM converted (E)-4-bromobut-2-enal (1a-Br)s moothly in more than 99 %c onversion with ah igh preference for the C À C bond-forming reaction over the natural reduction reaction (Table 1, entry 9). In order to demonstrate the practical value of this transformation, a(E)-4-bromobut-2-enal (10 mm)was transformed with YqjM Y169F (0.05 mol %) on apreparative scale.T he very volatile cyclopropane carbaldehyde was further converted in asubsequent chemical reaction into the corresponding 2,4-dinitrophenylhydrazone in 79 %y ield of isolated product over two steps.
While the WT enzymes showed mainly alkene reduction of a,b-unsaturated ketones (substrate 1b-Br), YqjM Y169F turned out to be an excellent enzyme for the cyclization, producing acetylcyclopropane in 98 %c onversion (Table 1, entry 18).
All of the enzymes converted both substrates equally well in up to > 99 %conversion, with the OPR3 and YqjM variants Scheme 1. a) Reduction of an activated C=Cb ond by ene reductases. Activation of the double bond by hydrogen-bond formation is enabled, for example, by two His residues (shown as "AH"/"HA"). The FMN hydride (shown in blue) is transferred to the b-C. The resulting enolate is stabilizedb ytwo hydrogen bonds.R eprotonation at the a-carbon occurs through aconserved tyrosine residue. b) Proposed mechanism of reductive CÀCcoupling. In the absence of the Tyrresidue, the enolate attacks the internale lectrophilic carbon, thereby enabling the formation of cyclic compounds. Scheme 2. ReductiveC ÀCb ond formationsusing substrates 1a-Br-1b-Br.R eaction conditions:3 00 mLs tock solution (10 mm substrate, 1v ol% DMF, 1,2-DME as internal standard, and 15 mm NADH in 50 mm sodium phosphate-buffer at pH 7.5 and 150 mm NaCl) and enzyme (5 mm)i n300 mLs odium phosphate-buffer (50 mm,p H7.5, 150 mm NaCl) per tube, 180 min and 25 8 8Cat300 rpm. DMF = N,Ndimethylformamide, 1,2-DME = 1,2-dimetheoxyethane. showing ah igher preference for cyclopropanation activity ( Table 2). With the more electrophilic substrate 5-Br,t hese variants delivered almost exclusively the desired cyclopropanation product ( Table 2, entries 7-8, 10). Interestingly,W TY qjM and its Phe variant showed ar eversal in diastereoselectivity when switching from ac hlorine to the bromine substrate (Table 2, entries 4,5 vs.9,10). Very good enantioselectivity was observed for WT OPR3, which had as trong preference for the (R,S)-7 enantiomer (> 99 % ee;T able 2, entry 1). TheYqjM Y169F variant showed reversed as well as enhanced enantioselectivity compared to the wildtype enzyme,t hus indicating that subtle changes in the binding site might have large consequences for substrate recognition (Table 2, entry 5). We were able to rationalize the observed asymmetric induction through docking studies ( Figure S1 in the Supporting Information).
In conclusion, we have presented the first biocatalytic reductive carbocyclization using ene reductases.F or our cyclization approach, two strategies proved useful to suppress the natural reduction reaction. First, engineering of the enzymes by replacing the critical proton donor (Tyr190 in OPR3, Ty r169 in YqjM) with Phe or Tr pf avored the preference for the cyclization reaction. Second, substrate engineering could be used in as ynergistic way.
While an increase in the electrophilicity of the g-carbon favored the cyclization, this approach was often limited by the intrinsic instability of these substrates.D ecoration of the scaffold with substituents at the a-and b-position (as present in synthetically more attractive compounds) appeared to be am ore universal strategy,l eading to improved product selectivity by removing water molecules that could serve as potential alternative proton donors in the active site.W ew ere able to demonstrate that prochiral substrates can be cyclized to form chiral 1,2-disubstituted cyclopropanes with excellent enantioselectivity.C urrent efforts in our laboratory aim to expand the scope to larger ring systems by using protein and substrate engineering.   [b] ee [%] [b] (S,S)-7 ee [%] [b] (R,S)-7 [a] Conversions were determined by GC-FID analysis of the crude reaction mixture by using 1,2-DME as an internalstandard;n.d. = not detected. [b] Forassignment of de and ee products were analyzed as the corresponding alcohols after reduction of the samples using NaBH 4 .
Scheme 4. Biocatalytic conversion of substrate 8.R eaction conditions as for Scheme 2.  [b] (R,R)-10 ee [%] [b] (S,R)-10 [a] Conversions were determined by GC-FID analysis of the crude reaction mixture by using 1,2-DME as an internal standard. [b] Forassignmentofee,products were analyzed as the corresponding alcohols after reduction of the samples using NaBH 4 .

Conflict of interest
Theauthors declare no conflict of interest.