Reactivity and Selectivity of Iminium Organocatalysis Improved by a Protein Host

Abstract There has been growing interest in performing organocatalysis within a supramolecular system as a means of controlling reaction reactivity and stereoselectivity. Here, a protein is used as a host for iminium catalysis. A pyrrolidine moiety is covalently linked to biotin and introduced to the protein host streptavidin for organocatalytic activity. Whereas in traditional systems stereoselectivity is largely controlled by the substituents added to the organocatalyst, enantiomeric enrichment by the reported supramolecular system is completely controlled by the host. Also, the yield of the model reaction increases over 10‐fold when streptavidin is included. A 1.1 Å crystal structure of the protein–catalyst complex and molecular simulations of a key intermediate reveal the chiral scaffold surrounding the organocatalytic reaction site. This work illustrates that proteins can be an excellent supramolecular host for driving stereoselective secondary amine organocatalysis.

Designing host-guest systems for organocatalysis has increasingly interested chemists. [1] Amolecular host provides as pecific environment that dictates both selectivity and reactivity of the organocatalysis,thereby providing additional means for reaction control. Nature provides such examples, including proteins.T hese proteins present ahydrophobic and inherently chiral scaffold favorable for organocatalysis. [2] Similar concepts have been tested previously.S mall, metalfree reagents including flavin, [3] thiazolium ylides, [4] pyridoxamine, [5] or selenocysteine [6] have been added to ap rotein covalently for oxidation, as well as CÀCa nd CÀNb ond formation reactions.R ecently,anovel artificial enzyme was created using p-stacking organocatalysts whose selectivity can be improved by modifying the protein scaffold. [1a,7] In contrast, iminium formation by as econdary amine has become am ajor mode of substrate activation in organocatalysis, [8] but ap rotein-based host for this type of catalysis has not been achieved.
Streptavidin (Sav)possesses astrong, noncovalent affinity to d-biotin. [9] Since Sav mostly binds to the ureido moiety of biotin, the valeric acid sidechain can be chemically modified for other applications,s uch as the production of artificial enzymes carrying either nonmetal [1a, 7] or metal cofactors. [10] Herein, we have adapted this technology and prepared Savbased hybrid catalysts carrying biotinylated secondary amine functionalities (Scheme 1). This work demonstrates that Sav, as aw ater-soluble protein, can provide am icroenvironment favorable for stereoselective secondary amine organocatalysis.
Several biotinylated organocatalysts were synthesized based on known catalytically active secondary amine motifs ( Figure 1), including 4-imidazolidinone (1)(2)(3)(4), proline (5 and 6), and pyrrolidine (7 and 8)derivatives.T he compounds 1-5 were prepared by copper-catalyzed Huisgen 1,3-dipolar cycloadditions of alkynylated biotin and azido-functionalized    precursors (5-47 %o verall yield). [11] Thec atalysts 6-8 were synthesized by direct coupling of biotin to either the Bocprotected amino-proline or amino-pyrrolidine starting materials,f ollowed by deprotection (48-63 %o verall yield). The catalysts 1-3 and 5-8 were obtained as single stereoisomers, but an inseparable 1:1m ixture of cis-a nd trans-isomers was formed for 4. Then ewly prepared biotinylated catalysts were used in the Michael addition of nitromethane to cinnamaldehyde (Table 1), at ype of 1,4-addition that is often used in pharmaceutical synthesis. [12] Our previous analysis of this reaction indicates that water can hamper the reaction progress when used as as olvent, but enhances turnover rate when used as an additive. [13] Hence,this sets the basis on how ap rotein scaffold can affect the efficiency of this organocatalytic reaction. Fori nitial screening,e ach biotinylated catalyst (20 mol %) was incubated with cinnamaldehyde and five equivalents of nitromethane at 25 8 8C(see the Supporting Information). No activity was observed in either deionized water or in acidic buffer (pH 2.0-6.0). In contrast, at pH 7.0 and 8.0 the proline and pyrrolidine catalysts (5)(6)(7)(8)w ere able to mediate the Michael addition as detected by GC-MS, whereas such observations were absent for 1-4.A tp H9.0, however, significant background reaction was observed, rendering these reaction conditions unfavorable for developing stereoselective organocatalysis.Inanacidic (< pH 7.0) or an on-buffered aqueous environment, secondary amines are likely to be protonated (pK a % 8) and deprotonation of nitromethane (pK a % 10) is unfavorable,a nd these factors likely stall the progress of the organocatalysis.
To see if the performance of organocatalysis can be improved by anchoring to aprotein surface,the catalysts 5-8 were introduced to streptavidin (Sav). Fort he proline derivatives 5 and 6,t he reaction yields at pH 7.0 were not affected by the addition of Sav and remain at about 5-6 % ( Table 1; see the Supporting Information). In contrast, the reactivities of the pyrrolidines 7 and 8 were significantly improved when Sav was included, showing about af ive-and twofold increase,respectively,inthe reaction yields at pH 7.0 (k cat /k uncat = 10 and DDG°= 5.71 kJ mol À1 for Sav-7). No reaction enhancement was observed in control reactions with either l-proline,p yrrolidine,o rSav-biotin. This data highlights the synergistic effect observed by the introduction of 7 and 8 into the protein scaffold ( Figure 2). While these two diastereomeric catalysts differ by one stereoconfiguration at the 3'-position, the yield of the reaction catalyzed by Sav-7 is slightly higher. Indeed, the reactivity of Sav-7 is comparable to that of aw ater-compatible derivative of the Jørgensen-Hayashi catalyst in that asimilar reaction yield per mol %of catalyst was obtained. [12c] Whereas many traditional organocatalysts such as the MacMillan and Jørgensen-Hiyashi catalysts contain either bulky or hydrogen-bonding groups adjacent to the reacting nitrogen atom for stereoselectivity control, these substituents are absent in 7 and 8.H owever,t he transformation taking place on the surface of the inherently chiral Sav is anticipated to improve the stereoselectivity.T his hypothesis is examined by chiral-phase LC analyses of the catalytically enhanced reactions.I nt he absence of Sav,n oe nantioselectivity was observed. In the Sav-7 reaction, the majority of the product contains an S configuration at C3, giving an enantiomeric   ratio of 80:20. Conversely,t he R stereoisomer was preferentially formed in the Sav-8 reaction (Table 1, entries 12 and  13). These results clearly indicate that the binding of the catalysts to Sav causes noticeable improvement in enantioselectivity. Acrystal structure at aresolution of 1.1 was obtained to pinpoint the interactions between the secondary amine catalyst and protein scaffold. Thea mide bond in the valeric moiety undergoes hydrogen bonding with Ser88, which is frequently seen when biotin is functionalized by amide-bond formation. [10f,h,j] In previous Sav-based artificial metalloenzymes,t he location of the metal catalyst is often unclear because either the ligand is inherently flexible or the metal catalyst dissociated during crystallography. [10f,h,j] In contrast, the location of the secondary amine catalyst is clearly revealed in this work. Thepyrrolidinyl moiety forms ahydrogen bond with Ser112, which has been shown to be important in artificial metalloenzyme design. [10f,h,j] Other residues surrounding the organocatalytic reaction are also revealed, including Leu124 and Lys121. Notably,L ys121 was also found to play critical role in the development of the Sav-based anion-p enzyme dictating the activity of the catalytically important tertiary amine. [1a] However,this residue likely plays ad ifferent role here,a si ti sd istant from the catalytic amine atom (8.1 and ca. 7.3 based on the X-ray and MD simulations,r espectively). Since Lys121 is within proximity to C3 and the phenyl moiety of the intermediate (4)(5), it likely dictates how the nucleophile approaches the iminium intermediate for reaction.
Molecular dynamics (MD) simulations were performed to investigate the origin of reaction stereoselectivity (see the Supporting Information for details). Our previous computational analysis indicates that the iminium and deprotonated nitromethane intermediates are formed in one step and thus they are extremely transient before forming as tereogenic center at C3. [13] Hence,t he reaction stereoselectivity is most likely dictated by the hemiaminal tetrahedral intermediate which forms prior to the iminium/deprotonated nitromethane pair. C1 of the hemiaminal tetrahedral intermediate can be either R or S,which exerts astrong effect on how C3 is being exposed for nucleophilic reaction (see Schemes S1and S2 in the Supporting Information). Interestingly,arepresentative snapshot of the (1S)intermediate from the MD simulation of Sav-7 overlays well with the crystal structure complex (Figure 3). According to the population analysis obtained from the MD simulations,t his (1S)i ntermediate dictates the stereoselectivity of the reaction. Upon dehydration, this intermediate will be converted into the iminium intermediate which exposes the C3 Si face for nucleophilic addition, whereas the opposite face is shielded by the protein (Figure 4). Consequently,f ormation of the stereoisomer (S)-10 a is favored. Similarly,f or the iminium derived from the (1R)intermediate,the C3 Si face of the iminium intermediate is exposed for reaction. In the case where 8 is used, formation of the (1R)i ntermediate is favored and the product stereoisomer is reversed (see Figure S7). However,t his intermediate is noticeably more flexible,a nd analysis of the MD trajectory of this complex indicates that while the majority of the conformers (ca. 90 %) yield the (3R)a dduct, there is apopulation of conformers that yield the opposite stereoisomer (see geometrical analysis and animation movies deposited in the Supporting Information).
Attempts to further enhance the reaction yield were made by prolonging the reaction time or raising the pH, but no improvement was found. Instead, there was as ubstantial amount of precipitation observed after 6-42 hours,and it can be caused either by imine formation between free amines of the protein and the aldehyde moieties or by aggregation resulting from surface binding of the hydrophobic reactants.  To suppress the precipitation, different organic solvents were included as additives (Table 1). They can be separated into three categories:n onpolar (C 6 D 6 ,C DCl 3 ), polar aprotic (EtOAc,T HF,D MSO,M eCN), and polar protic (MeOH) solvents. [10i,14] In all cases precipitate formation was suppressed but the reaction yield was improved only when either THF,E tOAc, or MeOH were included. However,r eaction yields fluctuated severely when THF or EtOAc was used, and cinnamic acid was observed as aside product. Also,inTHF up to 50 %of1,2-addition product was observed. Thepresence of MeOH enhances the reaction yield up to an average value of 80 %w ithin 18 hours per mol %c atalyst used. This data transforms into ar ate enhancement (k cat /k uncat )o f8and a DDG°= 5.12 kJ mol À1 with 10 %a dditional side products observed in the uncatalyzed background reaction. Such efficiency has not been seen in previously reported aqueous organocatalytic system. [12c, 15] Furthermore,w hile the previously reported Sav-based p-stacking organocatalytic system operates best at low pH, [1a] in the Sav-7 system such conditions were found to be detrimental for catalysis (ca. 5% at pH 5.0 and no product observed at pH 3.0).
Since MeOH proved to be the most productive cosolvent, the ratio of this solvent to buffer was screened for optimal reaction yield and enantioselectivity.I ncreasing the amount of MeOH steadily enhances both the yield and enantioselectivity of the model reaction (see Figure S8). Ar atio of 1:1o fb uffer/MeOH at pH 7.0 proved to be the optimal reaction conditions.Inthe control experiment where the Sav protein scaffold is omitted, no enantioselectivity was observed. Together,t hese data indicate that the stereoselectivity originates from the binding between Sav and 7,w hich remains intact even in the presence of asignificant amount of MeOH.
Thea pplicability of the Sav-7 system was explored by testing cinnamaldehyde derivatives and the respective ketones as alternative substrates (Table 2). Almost all of the employed aldehydes are tolerated by Sav-7,giving acceptable to good yields (> 35 %) and enantioselectivities (entries 1-7). Thel ow yield of 4'-nitrocinnamaldehyde (entry 5) is most likely caused by its low solubility in the buffer/MeOH mixture.I nc ontrast, all of the ketone counterparts give poor yields,a nd only the ones shown in Table 2a fford adetectable yield.
In conclusion, ah ybrid organocatalytic system based on the streptavidin-biotin technology was developed in this work. This system facilitates secondary amine catalyzed reactions.S urprisingly,b ys imply exploiting the scaffold of the wild-type streptavidin, as parsely substituted cyclic secondary amine can be used to catalyze reactions with high enantioselectivity.T his approach bypasses the need of striking ab alance between reactivity and stereoselectivity as frequently seen in traditional organocatalyst design. [8] Molecular simulations and protein crystallography have been particularly insightful, as they reveal how the protein dictates the orientation of the substrate and reaction stereoselectivity. This work lays the basis for protein engineering in which ad esignated scaffold can be modified for optimal organocatalysis. [10d,f,g,i-k,16] Furthermore,s ince this work enables organocatalysis in an isolated environment of as upramolec-ular complex, its compatibility with other species including reagents,o ther catalysts,a nd intermediates are greatly enhanced. [1a,c] Hence,t his protein-based organocatalytic system will facilitate the development of tandem reaction sequences and hybrid catalysts as tools in chemical biology. [17]