Asymmetric Biocatalytic Synthesis of 1‐Aryltetrahydro‐β‐carbolines Enabled by “Substrate Walking”

Abstract Stereoselective catalysts for the Pictet–Spengler reaction of tryptamines and aldehydes may allow a simple and fast approach to chiral 1‐substituted tetrahydro‐β‐carbolines. Although biocatalysts have previously been employed for the Pictet–Spengler reaction, not a single one accepts benzaldehyde and its substituted derivatives. To address this challenge, a combination of substrate walking and transfer of beneficial mutations between different wild‐type backbones was used to develop a strictosidine synthase from Rauvolfia serpentina (RsSTR) into a suitable enzyme for the asymmetric Pictet–Spengler condensation of tryptamine and benzaldehyde derivatives. The double variant RsSTR V176L/V208A accepted various ortho‐, meta‐ and para‐substituted benzaldehydes and produced the corresponding chiral 1‐aryl‐tetrahydro‐β‐carbolines with up to 99 % enantiomeric excess.

Abstract: Stereoselective catalysts for the Pictet-Spengler reactiono ft ryptamines and aldehydesm ay allow as imple and fast approach to chiral 1-substitutedt etrahydro-b-carbolines. Although biocatalystsh ave previously been employed for the Pictet-Spengler reaction,not as ingle one acceptsb enzaldehyde and its substituted derivatives. To addresst his challenge,acombination of substrate walking and transfer of beneficial mutationsb etween different wild-type backbones was used to develop as trictosidine synthase from Rauvolfia serpentina (RsSTR) into as uitable enzymef or the asymmetricP ictet-Spengler condensation of tryptaminea nd benzaldehyde derivatives. The double variant RsSTR V176L/V208A accepted various ortho-, metaand para-substituted benzaldehydes and produced the correspondingc hiral 1-aryl-tetrahydro-b-carbolines with up to 99 %enantiomeric excess.
An efficienta pproachf or the synthesis of 1-substituted tetrahydro-b-carbolines is the Pictet-Spengler reaction, in whicha 2-arylethylamine reacts with ac arbonyl compound, usually forming as ix-membered ring. [4][5][6] Various chemical protocols for the Pictet-Spengler reaction have been established, including many stereoselective approaches. [5,7] In nature this condensation reactioni sc atalysed by enzymes called Pictet-Spenglerases, [8] which are sub-categorised according to their substrates. For instance, norcoclaurine synthases( NCSs) [9] prefer dopamine as amine substrate,w hile strictosidine synthases (STRs) [10] acceptt ryptamine and its derivatives, leadingt ob-carbolines as products. While for norcoclaurine synthases ab road scope of carbonyl substratesh as been reported, [9] it has only recently been shown that STRs accepts mall aliphatic aldehydes besides the natural, highly functionalized aldehyde secologanin (4c, Figure1)and itsa nalogues. [10d] In contrastt ot he natural reaction of tryptamine (1)w ith secologanin( 4c), which gives the (S)-configured product, (S)-strictosidine, smalla ldehydes such as isovaleraldehyde led to the (R)-configured product. Computational methodss uggested the reason for this inverted absolute configuration to be an inverted binding mode:i nt he natural reactiont he indole ring of tryptamine is located at the back of the active-site pocket, while in the reaction with isovaleraldehyde it points to the outside of the active site. [11] Interestingly,a cceptance of benzaldehyde as as ubstrate has neither been reported for NCSs nor for STRs. In fact, only very recently av ariant of an orcoclaurine synthaseh as been shown to accept aldehydes branched in a-position, [9b] but benzaldehyde was not investigated in this work. An earlier study found no product formation from benzaldehyde and dopamineb y NCS from Thalictrum flavum. [9d] When testingh eterologously expressed wild-type strictosidine synthases originating from Catharanthus roseus (CrSTR), [10d, 12] Ophiorrhiza pumila,( OpSTR), [10d, 13] and Rauvolfia serpentina (RsSTR, PDB:f or example, 2V91) [10d, 14] with tryptamine (1)a nd benzaldehyde (2a,S cheme 1A), noneo fthem led to any trace of product formation.C onsequently,anew biocatalyst hadt ob ed eveloped for this type of substrate structure. Since STR hasb een shown to enablet wo binding modes of tryptamine and the aldehyde substrate (see above), [11] it is not obviousw hich region of the active site needs to be modified to improvea ctivity.M oreover,s ince benzaldehyde was not converted at all, as ubstrate walking [15] approach was followed. The idea is to adaptt he enzyme to a substrate possessing as tructure, figuratively speaking, between the target substrate and ac ompound known to be well accepted, for example, isovaleraldehyde( 4b). Adapting the catalystf or the structuralh omologue might also induce low activity for the target aldehyde 2a,a nd this activity can then be further improved. As as maller structural analogue of benzaldehyde, the a-substituted aldehyde2 -methylbutanal (4a, Figure 1) was selected, which wasa ccepted by the three STRs investigated, althought he conversions observed after 24 h were low (0.8 %f or OpSTR, using 10 mm of 1 and 50 mm of 4a,T able S1). For comparison, the isomeric aldehyde isovaleraldehyde( 4b)i st ransformed under the same conditions with 50-63 %conversion.
Since OpSTR provedt ob easynthetically useful catalyst in a previouss tudy, [10b] the initial mutationsw ere performed using this scaffold. In af ocusedl ibrary 13 residues inside and around the active-site pocket were addressed as wellasadditional residues identified by MD simulations (TableS7; see also Supporting Information "Selection of Sites for Mutagenesis"). [16,17] The 83 variantsw ere successfully expressed;h owever, only 19 variants werea ctive with aldehyde 4a.O ut of these active variants four displayed minimally higherc onversion than the wild type (Table 1, entries 2-5;T able S2). Next, double variants were prepared by pairwisec ombination of the beneficial substitutions in positions V147, I179 and L290. Additionally,t he V147I and I179V substitutionsw erec ombined with amino acid exchanges that had resulted in similara ctivity as the wild type (Y76W,F 197Y). This led to the identification of two OpSTR double variants (V147I/I179V,V 147I/L290I) displaying af ourfold increase of conversion for 4a compared to the wild type (Table 1, entries 6, 7; data for all double variants:T able S3). When the small library of doublev ariants wast ested with benzaldehyde (2a), formationo ft he desired product 3a was detected for the first time. Thereby OpSTR V147I/I179V provedt o be best (3 %conv. ;T able 1, entry 6).
Since the residues V147 and I179 seemed to represent hot spots in OpSTR,w eh ypothesized that exchanges at the corresponding positions might also be beneficial for the strictosidine synthase of Rauvolfia serpentina (RsSTR).S uch an approach [18] to identify hot spots in one enzyme and then transfer them to related enzymes may enable screening ab roader overall sequence space of possibly suitable candidates. The po-sitionsV 147 and I179 of OpSTR correspond to residues V176 and V208 in RsSTR. [19] When first investigating single variants for thesep ositions, V208A allowed already ar easonable conversion of benzaldehyde under standard assay conditions (13 %, Ta ble 1, entry 10). In the double variant V176L/V208A, the exchange of V176 to leucine even increased the conversion of 2a to 24 %( Ta ble 1, entry 11), while other amino acids in this positions uch as I, Fo rMwere less beneficial (TableS4).
To better understand the observed reactivity with respect to the non-naturala ldehydes of the V176L/V208A mutation we generated ab inding mode model (Figures 2a nd 3) based on X-ray crystal structure overlays and computational refinement (for computational details, see the Supporting Information). Our most likely model assumes the inverted binding mode that was found previously for 4b in OpSTR (PDB:6 s5q). [11] We think that the exchange V208A creates space in the back of the active-site pocket, making it better able to accommodate the phenyl moiety (Figure2). The substitution of Val176 by the larger leucine residue leads, in our model, to modifiedd ispersion interactions with the indole core, which might result in a more favourable positioning of the substrate for catalysis. Figure 1. Aldehyde 4a used for substrate walking as structural transformant between the well accepteds ubstrate 4b and the target substrate benzaldehyde (2a), which is not acceptedb ythe wild-typee nzymes. The structure of the natural substrate, secologanin (4c,Glc = b-d-glucosyl),i ss hownfor comparison. It is also worth to note that RsSTR V176L/V208A is about twice as active towards the previously reported substrate isovaleraldehyde( 4b)a st he wild-type enzyme, while its specific activity with the natural substrate secologanin (4c)i sr educed to approx. 58 %r elative to the wild type (Table S5).
Analysis of theo ptical purity of the obtained product 3a revealed an ee of 99 %w ith (R)-configuration. This absolute configuration is in line with the previously observed stereochemical outcome of RsSTR-catalysed Pictet-Spengler reactions of small aldehydes. [10b] Furthermore, the observed absolute configuration and the site of mutation that created space at the back of the active-site pocket (V208A)t oa ccommodate 2a also support as ubstrate binding mode in which the indole ring of tryptamine points out of the active site, as previously suggested by computational studies. [11] To elucidate the substrate scope of RsSTR V176L/V208A, substitutedb enzaldehyde derivatives were tested and revealed a tolerance of the variant for various meta-substituents,i nclud-ing halogens (F,C l, Br), methoxy and nitro groups (substrates 2c-2g). The best conversion was achieved with the m-bromo derivative( 2e,T able 2, entry 5). Presence of af luorine atom was also accepted in para-position (substrate 2b,T able 2, entry 2), while larger substituents like chloro, methoxy,o rn itro (2j-2l)w erenot tolerated there.
The reasont hat meta-substituents are well accepted while there is ac lear limitation in para-position is most likely steric hindrance, as am odel of (R)-1-phenyl-b-carboline (3a)i nt he active site of RsSTR V176L/V208A revealed free space in metaposition and restriction in para-position (Figure 3).
Since the ortho-position is closestt ot he carbaldehyde moiety, whichi st he site of reaction, it was expected that ortho-substitution would not be tolerated. However,i t turnedo ut that o-fluoro-( 2h)a nd even o-bromobenzaldehyde (2i)a re well-accepted substrates. Unexpectedly,t he conversion even improved significantly for the o-Br derivative (2i)c ompared to the unsubstituted 2a,r eaching up to 59 %( Table 2, entry 9).
The products 3b-i wereo btained in optical purities of 96-99 % ee,t he only exception being 3g,w hich was formedi n 90 % ee. This reduced optical purity can be ascribed to the substantialn on-enzymatic background reactivity of 1 with 2g (2.1 %c onversion in 24 h). The background reactivity of the other aldehydes is significantly lower ( 0.6 %).
Optimisation of the reaction conditions (see Supporting Information) allowed to improvet he ratio of tryptamine (1)t oa ldehyde 2 to 1:1.25 at at ryptaminec oncentration of 40 mm. Four selected biotransformations (2a,b,h,i)w ere carried out on preparative scale (5 mmol 1)t oc onfirm/establish the absolute configuration (Table S6). The products were isolated in 4-31 % yield and the optical purities obtained were between 96 and 98 % ee. Analysis of the optically enriched tetrahydro-b-carbolines by opticalr otation, circular dichroism (CD) spectroscopy and HPLC showed that all of them possess (R)-configuration (Table S15). Figure 2. Model of (R)-1-phenyl-b-carboline (3a)int he active siteo fwildtype RsSTR (white)a nd RsSTR V176L/V208A (black). Short non-bonded contacts between substrate and enzyme are showna sd ashed lines.  In summary,aP ictet-Spenglerase reactionw as developed for the stereoselective reactiono ft ryptamine with benzaldehyde derivatives. Because benzaldehyde was not accepted at all by the investigated wild-type enzymes,asubstrate-walking strategyw as applied, in which suitable hot spots identified in one STR backbone (OpSTR) were transferred to another (RsSTR). The RsSTR variant V176L/V208A turned out to accept a broad scope of benzaldehyde derivatives, particularly those substituted in meta-a nd ortho-position, allowing to obtain (R)configured products with up to 99 % ee. The suitable catalyst was created by testing ar ather smalll ibrary of variants (ca. 100) by combining rational design, single-site saturation and hot-spot transfer to other backbones. The concept and the catalyst developed open new approaches for the synthesis of important bioactive 1-aryltetrahydro-b-carbolinesi no ptically enriched form.