Structural Elucidation of the Mechanism of Molecular Recognition in Chiral Crystalline Sponges

Abstract To gain insight into chiral recognition in porous materials we have prepared a family of fourth generation chiral metal–organic frameworks (MOFs) that have rigid frameworks and adaptable (flexible) pores. The previously reported parent material, [Co2(S‐mandelate)2(4,4′‐bipyridine)3](NO3)2, CMOM‐1S, is a modular MOF; five new variants in which counterions (BF4 −, CMOM‐2S) or mandelate ligands are substituted (2‐Cl, CMOM‐11R; 3‐Cl, CMOM‐21R; 4‐Cl, CMOM‐31R; 4‐CH3, CMOM‐41R) and the existing CF3SO3 − variant CMOM‐3S are studied herein. Fine‐tuning of pore size, shape, and chemistry afforded a series of distinct host–guest binding sites with variable chiral separation properties with respect to three structural isomers of phenylpropanol. Structural analysis of the resulting crystalline sponge phases revealed that host–guest interactions, guest–guest interactions, and pore adaptability collectively determine chiral discrimination.


Introduction
Theexistence of chirality in biology makes the production of pure enantiomers important to the manufacture of pharmaceuticals,a grochemicals,f lavorings,a nd fragrances. Thed evelopment of new techniques and materials for asymmetric synthesis and enantiomeric separation has therefore received great attention. Fore xample,w hereas some enantiomers of ad rug molecule have no activity,s ome can have different toxicology,b eing harmful or toxic. [1] Asymmetric synthesis is an elegant solution when available,b ut many times it is more economical to synthesize molecules as racemic mixtures and then separate the enantiomers.I nt his context, polysaccharide-and cyclodextrin-based chiral stationary phases (CSPs) have been extensively used for chiral chromatography with efficacyd etermined by the stereospecificity of surfaces and cavities. [2] However,t he chiral recognition mechanisms in these materials are not well understood, especially across ar ange of chiral compound types.
Metal-organic materials (MOMs) [3] are crystalline materials comprised of metal ions or clusters and organic ligands, making them modular and capable of exhibiting extra-large porosity.I np articular,c hiral MOMs (CMOMs) can be designed to combine homochirality and porosity such that, when the pore size shows agood match for the targeted guest, they can provide aw ell-defined stereospecific environment for the tight fit that is required for the discrimination and separation of enantiomers. [4] Previously,r acemic alcohols, ketones,amines,amides,acids,sulfoxides,and diols have been examined in terms of enantioselectivity by CMOM adsorbents,C SPs,a nd membranes. [4,5] Other types of porous solids investigated in this context include hydrogen-bonded frameworks, [6] covalent organic frameworks (COFs), [7] and porous organic cages. [8] Although more traditional porous materials such as zeolites have also been studied, the relative instability of the homochiral network [9] limits their potential. Thus far, few porous materials have been studied to specifically address the origin of enantioselectivity [5,6,10] and the nature of the interactions that promote enantioselective separation remains understudied. This is mainly because guest disorder, partial occupancyand high symmetry of the host can preclude accurate structural determination at the molecular level.
Herein, we address the mechanism of chiral recognition in af amily of CMOMs derived from the parent structure [Co 2 (man) 2 (bpy) 3 ](NO 3 ) 2 (man = S-mandelate,C MOM-1S, or R-mandelate,C MOM-1R,b py = 4,4'-bipyridine). [5b] X-ray crystallography can be apowerful tool in this context as it can provide in situ information about the supramolecular interactions that drive host-guest binding.T oo ur knowledge, CMOM-3S is the only CMOM that can serve as both ageneral-purpose crystalline sponge [11] and achiral stationary phase for enantioselective separation/identification of racemic mixtures. [12] We have therefore exploited the modular nature of CMOM-1S/R,f or which both the counterion and mandelate linker are amenable to substitution. Theresulting family of CMOMs detailed herein exhibits ah ard rigid framework and adaptable or soft pores. [13] These materials can therefore be classified as af ourth-generation MOFs. [14] The parent CMOM, CMOM-1S,can be readily fine-tuned through substitution of the counterion (BF 4 À ,C MOM-2S;C F 3 SO 3 À , CMOM-3S)o rl inker ligand (2-Cl, CMOM-11R;3 -Cl, CMOM-21R;4 -Cl, CMOM-31R;4 -CH 3 ,C MOM-41R). As detailed below,t his family of CMOMs can serve as chiral crystalline sponges (CCSs) to provide structural insight into the supramolecular interactions that occur between phenylpropanols and the pore surface of this family of CMOMs.

Results and Discussion
CMOM-1S/R is comprised of inexpensive,c ommercially available ligands and its modular nature enables the metal ions,linkers and anions to be substituted. Figure 1shows the crystal structure of this CMOM whereas Table 1l ists their structural components.I np rinciple,t his modularity should enable the generation of ad iverse platform of isostructural derivatives for the systematic study of the factors that influence chiral discrimination in CMOMs.

Chiral Resolution of Phenylpropanols
We selected three structural isomers of phenylpropanol (Scheme 1), namely 1-phenyl-1-propanol (1P1P), 1-phenyl-2-  propanol (1P2P) and 2-phenyl-1-propanol (2P1P) to investigate the chiral discrimination of our CMOM family towards their racemic mixtures.E nantiopure phenylpropanols are important intermediates used in the synthesis of pharmaceutical and parasiticide compounds. [16] Chiral resolution experiments were conducted using our previously reported procedure [5b] on activated CMOM crystals (see the Experimental Section in the Supporting Information for full details). The activated crystals were soaked in racemic mixtures of phenylpropanols for 5days before the crystalline solids were filtered and washed with cyclohexane to remove the excess of molecules adsorbed on the external surface.W et hen extracted the encapsulated phenylpropanols from the crystals using dichloromethane and we evaluated the enantioselective separation performance by analyzing the composition of the eluate through HPLC. Figure 2shows the results of the chiral separation of 1P1P, 1P2P,a nd 2P1P by CMOM-1S, 2S, 3S, 11R, 21R, 31R,a nd 41R.I nterestingly,t he outcomes differ in both enantiomeric excess (i.e., separation performance) and absolute values.For example,w hereas we observed a3 1% ee resolution of (R)-1P1P for 1S (66 % ee based on ad ifferent resolution method), [5b] we observed no separation and opposite enantioselectivity for 2S and 3S structures,with 0% and À42 % ee, respectively.A st he cationic framework for 1S, 2S,a nd 3S is the same in terms of structure and chirality,these differences in performance must be attributed to the EFA( i.e., NO 3 À , BF 4 À ,CF 3 SO 3 À ,respectively) and its impact upon pore shape and surface chemistry.C onversely,w hen we changed the functionality of the mandelate ligand, we observed alower ee (< 20 %) for 11R (2-Cl-man), 21R (3-Cl-man), 31R (4-Clman), and 41R (4-Me-man) vs.t hat observed for 1S. Furthermore, 21R,p referentially bonded to the opposite enantiomer preferred by 11R, 21R, 31R,a nd 41R.T hese results indicate that even subtle structural effects can strongly impact chiral discrimination in these CMOMs.
Looking at the three regioisomers of phenylpropanol (Scheme 1), we found that the difference in the position of the hydroxyl group strongly affects the enantioselectivity.I nt he case of 1S,( R)-1P1P was preferentially adsorbed, whereas it favored the adsorption of (S)-1P2P and (S)-2P1P.Asimilar phenomenon was observed for 3S but with the opposite binding selectivity (up to 42 % ee)f or these three phenylpropanols.T he highest degree of separation was achieved by 3S for 1P2P and 2P1P.O ur results should be placed in the following context:t he most widely used method of assigning the relative and absolute configurations of aseries of related compounds is through asymmetric synthesis or chiral separation. Ty pically in asymmetric synthesis,the structure of one of agroup of new compounds is determined by single-crystal Xray diffraction (SCXD), and, by analogy,t he same configuration is assigned to related compounds.T he reverse enantioselectivity observed herein suggests that the assignment of chirality by analogy is not reliable.

Chiral Recognition Mechanism Studies
Perhaps the most salient aspect of the results reported above is the variable enantioselectivity that occurs from subtle changes in the composition of the CMOMs.T ob etter understand these results we determined the nature of the host-guest binding sites from SCXRD studies of the guest loaded CMOM crystals.T he observation of 0% ee by 1P1Ploaded 2S stands out as being anomalous.I ntuitively,o ne would expect that all chiral porous materials will exhibit at least some degree of chiral discrimination unless size exclusion happens.I ndeed, whereas previous studies on homochiral hosts such as CMOMs,COFs and metal-organic cages have revealed examples of chiral materials with low or moderate enantioselectivity,t othe best of our knowledge, 0%ee has not yet been reported. [5,17] Thanks to the crystalline sponge nature of this class of CMOMs we are in position to elucidate the nature of the intermolecular interactions that resulted in 0% ee in 2S,achiral porous material. The structure of the 1P1P-loaded 2S reveals that the unit cell is doubled vs.as-synthesized 2S along b axis,and that there are six crystallographically independent 1P1P molecules in the structure.Asshown in Figure 3, the orientation between two enantiomeric pairs of 1P1P molecules is perpendicular, presumably to maximize the packing efficiency and guest interactions with the framework. In the first pair of 1P1P molecules,there are p-p interactions between them and two bpy ligands of the framework (Figure 3a). In addition, there are hydrogen bonding interactions between two 1P1P molecules and four surrounding BF 4 À counterions,w ith C p À H···F, C alkyl À H···F,a nd O À H···F interactions ranging from 2.605 to 3.651 .T he packing of the second pair of 1P1P molecules is similar to that of the first pair, but offsetting of two phenyl rings results in ad irectional C alkyl ÀH···p interaction between two 1P1P molecules (Figure 3b). In the third pair of 1P1P molecules,b oth phenyl rings of 1P1P exhibit close contacts with pyridyl moieties from the framework. Both hydroxyl groups interact with the same BF 4 À ion through hydrogen bonds.Hirshfeld surface analysis [18] of 1P1P molecules reveals that they are tightly encapsulated in the chiral channel (Figure 3d-f). Notably,each enantiomeric pair contains equal

Angewandte Chemie
Research Articles amounts of (S)-and (R)-1P1P,l eading to a0 %ee from crystallographic analysis,w hich is fully consistent with the experimentally measured discrimination results.
To better understand the variability of the absolute configuration of guests within the same scaffold, we determined the structures of 3S loaded with each of the three phenyl propanols.T he host-guest binding sites for 1P1P and 1P2P were discussed in our earlier paper. [12] Intermolecular pp,C À H···p,a nd hydrogen bonding interactions contribute to the effective enantioselective recognition of (S)-1P1P and (R)-1P2P.I nt he case of 2P1P,w ed id not observe p-p interactions between 2P1P and bpy ligands (Figure 4a). Instead, the phenyl ring of 2P1P forms hydrogen bonds and p-interactions through C p À H···X (X = O, Fand p). Hydrogen bonding and C alkyl À H···p interactions between the side chain of 2P1P and the framework dictates the orientation of 2P1P. One disordered 2P1P is located in ad ifferent binding site, wherein the terminal propyl alcohol interacts with triflate anions through C alkyl ÀH···X (X = O, F) and OÀH···O hydrogen bonds (Figure 4b). Thes trength and number of interactions within the chiral channel transform the binding sites for (R)-2P1P.T he Hirshfeld surface analysis of 2P1P reflects once more the close contacts detailed above.

The Role of the EFAs on the Mechanism of Molecular Recognition
We focused then on the analysis of the mechanism of enantiomeric separation working with three structures with different anions (1S, 2S,a nd 3S)a nd one racemic mixture, 1P1P.T he increasing size of anions used in the chiral channel of 1S, 2S,and 3S,that is,NO 3 À ,BF 4 À to CF 3 SO 3 À ,respectively,  results in decreasing pore volume from 1S to 3S;F igure 5 shows these differences.L ooking at the adsorption of 1P1P, whereas the cross-section of the largest pore cavity of the structures is similar (8 8 ), it is the pore shape and surface chemistry what defines the interactions with the 1P1P guest molecules.In2S,the guest molecules pack with higher density than those of 1S (Figure 3a nd Figure 5), while the 1P1P molecules in 3S were isolated between two junctions with the distance of 10.2 .T oa nalyze the supramolecular interactions of each guest molecule,w ep erformed the fingerprint plots to highlight specific close-contacts from host-single guest and guest-guest contributions;F igure 5c-i and the Supporting Information, Figures S9-S11 show the full interaction. Thei nteraction map was constructed by defining distances from the Hirshfeld surface to the nearest nucleus inside the surface (or internal, d i )and outside the surface (or external, d e )a st he first functions of distance explored for mapping on the surfaces.T he visual comparison between these three plots in terms of area in Figure 5c,g ,a nd k demonstrates the significant contribution of guest-guest interactions in 2S that arise from p-p and CÀH···p interactions.W enote that the shape of guest-guest interactions in 2S is spread widely over the range of de, di < 1.8 ,r ather than the narrow needle-shape found in 1S,i ndicating the substantially close contacts among the neighboring guests in 2S.Besides,the 1P1P molecules in the three structures shown in Figure 5a ll display strong interactions with the host structures,asevidenced by the highlighted area colored with red and cyan. From the above analysis,one can infer that the most strongly bound of two enantiomers within the host will hinder chiral discrimination ability.

The Impact of Guest Geometry on the Mechanism of Molecular Recognition
Following the analysis on the impact of the anion, on the enantiomeric separation, we compared the separation of the three molecules,1P1P,1P2P,and 2P1P,by3S.Looking across the three guest molecules loaded in the 3S structures,t heir positions within the chiral channels are consistent, with one binding site being common to all three PPs.F igure 6 illustrates this idea;t he phenyl rings of the PP molecules interact with the aromatic surface (orange) created by the phenyl and pyridyl rings of the framework, whereas the hydrophobic surface generated by the À CF 3 moieties repel the ÀOH moieties and attracts the alkyl CÀHg roups.H ere,t he hydroxy groups of the PPs are stabilized by the hydrophilic surface of the framework through hydrogen bonding interactions taking place in different positions,asindicated by the red arrows in Figure 6. While the position and inclination of the head of the PP phenyl rings remain broadly the same,the orientation of the tail varies as ar esult of weak interactions. Thes pecific chemical environment provided by the host appears to mimic enzyme binding sites in terms of the tight fit and ability to discriminate between enantiomers. [19] Unfortunately,o ur attempts to accomplish the X-ray analysis of the guest-included crystals of CMOMs 11R-41R failed due to the poor crystallinity of these systems.T his is not completely unexpected and it is very likely linked to the fact that these CMOMs show lower enantioselectivity.A ss uch, the lack of distinct binding sites has aclear impact by the derivatization of the mandelate linker ligands.

Conclusion
Modifying the pore chemistry of CMOMs as reported herein profoundly influences chiral discrimination properties. In each CMOM, adaptable pore size and shape resulted in tight binding sites that enable av ariety of host-guest and guest-guest interactions.T hat the CMOMs can serve as crystalline sponges enabled the use of X-ray crystallography to provide detailed analysis of short contacts at the molecular level and, in turn, provided insight into the molecular recognition phenomena that impact chiral separation. We have thereby demonstrated the feasibility of using aplatform of CMOMs as crystalline sponges for systematic study of chiral discrimination in porous materials by manipulation of chiral pores while retaining the same framework structure. Theresulting variability in enantioselectivity is quite dramatic considering the invariability of the cationic framework and means that this CMOM platform can be classified as afourthgeneration MOF with hard-soft features that enable chiral discrimination and functioning as ac rystalline sponge.T hat enantiomeric discrimination is driven by tight guest binding sites within the chiral cavity is likely to be ag enerally important feature of CMOMs that exhibit strong enantioselectivity.W hat is perhaps more important though is that the parent CMOM structure can be easily tuned to enable ad hoc enantiomeric separations.W ef oresee opportunities for the development of more sophisticated CMOMs with precisely controlled chiral environments that will open up ap athway for their use in stereospecific catalysis and for separation of enantiomers of biologically active compounds in the pharmaceutical industry. Figure 6. X-ray structural analysis study of the guest binding pockets of 1P1P (a), 1P2P (b), and 2P1P (c) in CMOM-3S.CMOM-3S discriminates between similar substrates.T he color of the Connolly surface represents the element that generates the corresponding part of the surface: Corange, Ored, Nblue, Fcyan, Hwhite. The carbon atoms of the substrates are colored magenta. The yellow,g reen, and red arrows indicate the aromatic, hydrophobic, and hydrophilic surface of the CMOM, respectively.