Structurally divergent dynamic combinatorial chemistry on racemic mixtures

Structurally Divergent Reactions on Racemic Mixtures are atypical processes in Nature. The few examples reported in the literature take place in organic solvents and are driven by the reagents’ interaction with bulky chiral catalysts. Herein, we describe a dynamic combinatorial approach to generate structural divergence from racemic building blocks. The divergence is due to a stereospecific electron-donor – electron-acceptor interaction of diastereomeric macrocycles, leading to structurally distinct pseudorotaxanes. The equilibrated dynamic combinatorial library contains, amongst various macrocycles, two different types of [2]catenanes that are non-isomeric. The formation of these [2]catenanes is due to a spontaneous stereo and structurally divergent assembly of the building blocks.

In a round-bottom flask, 5 (3.00 g, 5.60 mmoles,1 equiv) was dissolved in DMF (200 mL) under N2 atmosphere followed by addition of K2CO3 (3.88 g, 28.68 mmoles, 5 equiv). After stirring half an hour at r.t., MeI (1.4 mL, 22.48 mmoles, 4 equiv) was added and further stirred overnight. Water was added, and the compound was extracted with CH2Cl2. The organic phase was dried over MgSO4 anhydrous and the solvent removed in vacuo to yield an oily liquid, which was dried under high pressure. At this stage the compound was impure but used as it was; no yield was determined at this stage. 1 7), 29.0, 28.9, 23.9(3), 23.9(0). There are more peaks in the 13 C NMR spectrum than expected because of the impurity. TOF MS ASAP+: m/z calcd for C30H42O6S2 [M+H] + 563.2501, found 563.2504.

Synthesis of 7:
In a round-bottom flask, 6 (3.95 g) was dissolved in acetone (5 mL), followed by addition of MeOH (500 mL) and 2 M NaOH (250 mL); the reaction was stirred overnight at reflux. Upon cooling, concentrated HCl was added until the pH became acidic and a yellow solid precipitated. The precipitate was filtered and washed with CH2Cl2 and acetone. The precipitate was collected and dried to yield 7 as a yellow solid (1.05 g, 3.12 mmoles, 58% over 2 steps). 1

Synthesis of 8:
Compound 7 (1.05 g, 3.10 mmoles, 1 equiv) and N-hydroxysuccinimide (1.44 g, 12.50 mmoles, 4 equiv) were dissolved in DMF (110 mL) and cooled to 0 °C. EDC×HCl (2.41 g, 12.50 mmoles, 4 equiv) was added and the reaction was stirred for 15 minutes in the melting ice bath. The reaction was further stirred overnight at r.t. The solvent was removed under reduced pressure and a small amount of acetone was added. The suspension formed was added dropwise to a 1 M HCl solution vigorously stirred. The precipitate obtained was collected by vacuum filtration and dried under reduce pressure to obtain a brownish precipitate (1.40 g, 2.60 mmoles, 85%). 1

Synthesis of 9:
In a flame-dried round-bottom flask, S-trityl-L-cysteine (612.2 mg, 1.70 mmoles, 2.2 equiv) was added to a solution of 8 (401.8 mg, 0.76 mmoles, 1 equiv) in DMF (40 mL) under N2. Et3N (1 mL) was added and the reaction was stirred overnight at r.t. The solvent was removed under pressure and a small volume of acetone was added to the mixture, which was precipitated dropwise into a 1 M HCl solution vigorously stirred. The yellow precipitate was collected by filtration and dried (727.8 mg, 0.76 mmoles, quant.). 1

Synthesis of R,R-2:
In a round-bottom flask, 9 (500.4 mg, 0.49 mmoles) was dissolved in CH2Cl2 (10 mL) followed by addition of TFA (10 mL) and SiEt3H (1 mL). The reaction mixture was stirred 30 minutes at r.t. and the volatiles were subsequently removed. Et2O was added, removed under pressure and re-dissolved in Et2O to ensure the precipitation of the desired product. To this, n-hexane was added and the solid was filtered and further washed with n-hexane to yield R,R-2 as an yellow precipitate that was dried (228.2 mg, 0.42 mmoles, 86%). 1  NDI-Trt-R-Cys and NDI-Trt-S-Cys were synthesised following the previously published procedures: NDA (1 equiv) and S-trityl-L-cysteine (for NDI-Trt-R-Cys) (2.1 equiv) (or Strityl-D-cysteine in the case of NDI-Trt-S-Cys) were dissolved in DMF (for <300 mg of NDA use 5 mL, otherwise use 10 mL) in a 20 mL microwave tube (maximum amount of NDA is 1 g). Dry triethylamine (1 mL) was added and the mixture was sonicated until a clear solution was formed. The solution was heated in a microwave reactor for 5 min at 120 o C. When the reaction was finished, the solvent was removed under reduced pressure. The residue was redissolved in acetone (1-2 mL) and the solution was added dropwise to a vigourously stirred 1 M HCl aqueous solution. The suspension was filltered and the precipitate formed was dried under high pressure. The 1 H NMR spectra matched the ones reported in literature. NDI-Trt-S-Cys (201.0 mg, 0.21 mmoles, 1 equiv) was dissolved in a mixture of CH2Cl2 (3 mL) and TFA (3 mL), followed by the addition of Et3SiH (0.3 mL). The reaction mixture was stirred for half an hour at r.t. and the volatiles were removed under reduced pressure. Et2O was added the suspension followed by the addition of n-hexane. The precipitate formed was filtered and dried (74.9 mg, 0.16 mmoles, 75%). 1  R,R-1 was synthesised using the same procedure as S,S-1.

96%
S,S-10 and R,R-10 were synthesised using the published protocols and the same procedure as for NDI-Trt-R-Cys and NDI-Trt-S-Cys. Deprotection procedure: The protected species was dissolved in a mixture of CH2Cl2 (3 mL) and TFA (3 mL), followed by the addition of Et3SiH (0.3 mL). The reaction mixture was stirred for 3 hours at r.t. and the volatiles were removed under reduced pressure. Et2O was added to the residue and the solid formed was filtered and dried.

Supplementary Note 1 -Methods for the integration of the HPLC peaks:
Solutions of identical concentrations of the building blocks (R,R-1 and R,R-2) were prepared by dissolving each compound in a water:acetonitrile mixture. The UV-vis spectra of these solutions were recorded and the absorbance spectra of the two building blocks were superimposed. We chose the wavelength for which the absorbance was equal for both building blocks (in this case it was 389 nm).
First method -for well-separated peaks:

S21
The chromatogram was loaded in XCalibur v.2.0.7 and, using the Integration function, each known (based on MS analysis) peak was integrated. The results (the values of the integrals) were exported to Excel. The value obtained for each species was divided by the number of chromophores forming the species. (e.g., a catenane has 4 chromophores, so the value was divided by 4; for a dimer, we divided by 2). In this way, each adjusted integral value is directly comparable between different types of species with different numbers of chromophores. After this, the sum of all species in the chromatogram was calculated, and the value representing each individual species was divided by this sum and multiplied by 100 to get the percentage (yield) for each species.

Second method -for overlapping peaks:
QtiPlot v.8.9 was used for this method. The chromatogram was replotted in Qtiplot (the original data was exported as a text .csv file then imported in QtiPlot). Then, parts of chromatogram with overlapped peaks were analysed using Lorentzian or Gaussian deconvolution (this is an algorithm that reconstructs the entire peak which can therefore be integrated). For each of these peaks, a value representing the integrated area was generated. For the well-separated peaks in such chromatograms, the integration option in QtiPlot was used. After integrating all peaks, the data was exported to Excel and the steps from the first method were repeated: the value obtained for each species was divided by the number of chromophores forming that species. As before each adjusted value is directly comparable between different types of species with different number of chromophores. After this, the sum of all species in the chromatogram was calculated, and the value representing each individual species was divided by this sum and multiplied by 100 to get the percentage (yield) for each species.
For triplicate measurements, the areas were integrated three times. After the integration was done three times for each peak, the average for the values was calculated   The VT-NMR experiment is useful in confirming that the structure is a catenane. In the case of an isomeric macrocycle, the 1 H NMR peaks would get broader as the temperature increases due to high thermal energy. The sharpness of the peaks and the relative constant shift as the temperature increases is supportive of an interlocked geometry for this molecule.   The unlabelled peaks did not ionise and could not be identified.
From these experiments, we can conclude that Kd2>Kd3 (from Supplementary Figure  25). 5 10   The unlabelled peaks did not ionise and could not be identified. The large Cotton effect observed for the species with the retention time between 5 and 8 min is related to the overall low optical activity of the library members. The VT-NMR experiment is useful in confirming that the structure is a catenane. In the case of an isomeric macrocycle, the 1 H NMR peaks would get broader as the temperature increases due to high thermal energy. The sharpness of the peaks and the relative constant shift as the temperature increases is supportive of an interlocked geometry for this molecule. Cat II RSRR decomposes during the variable temperature UV / CD experiment. We suspect that the decomposition is caused by a combination of two factors: light and temperature (photo-and thermal degradation) as in the VT-NMR experiment there is no decomposition seen. D stands for donor (BDT) and A for acceptor (NDI). For clarity, spline curves were added; these do not represent mathematical models for kinetic data.

S52
To a solution of Y (2.5 mM), R,R-1 was added (as solid) to make the total DCL concentration 5 mM. After the pH was adjusted to 8 -8.5, an HPLC chromatogram was recorded every 55 min. The experiment shows that the DCL is under thermodynamic control with the formation of Cat I RRRR as the major species. This library resembles with the one formed from R,R-1 and R,R-2 in 1:1 ratio. showing the formation of the Cat I RRRR (black). The description of each species can be seen in legend (D stands for donor (BDT) and A stands for acceptor (NDI)). For clarity, spline curves were added; these do not represent mathematical models for kinetic data.    these do not represent mathematical models for kinetic data.

Cat I RRRR
Based on pure statistical distribution, the ratio between Linear S,S-1 dimer : Linear Z : Linear Y should be 1:2:1. This is not the case here, therefore linear Y is more thermodynamically stable than the other two linear species. The same is true for the corresponding cyclic dimers.

S63
To a solution of Y (2.5 mM), S,S-1 was added (as solid) in order to make a 5 mM total DCL concentration. After the pH was adjusted to 8-8.5, an HPLC chromatogram was recorded every 55 min. The experiment shows that the DCL is under thermodynamic control with the formation of Cat II RSRR as major species. This library resembles with the one formed from S,S-1 and R,R-2 in 1:1 ratio.  showing the formation of the Cat II RSRR (black). The description of each species can be seen in legend (D stands for donor (BDT) and A stands for acceptor (NDI)). For clarity, spline curves were added; these do not represent mathematical models for kinetic data.

Cat I isomers RSRR and RSSR
in the presence of 1 M NaNO3, d) R,R-2:R,R-1:S,S-1 (5:2:1 molar ratio, 5 mM total concentration) library in the presence of 1 M NaNO3. Absorbances recorded at 389 nm. ii) and iii) The values in the tables represent the percentage of each species identified in the chromatograms above. The integrations were done in triplicate and the RMSD is reported in parenthesis along with the value. Each row of the table corresponds to the chromatogram bearing the same identification in the figure above. The unlabelled peaks did not ionise and could not be identified.
It is important to mention that the DCL in Fig. S81 (d) is not representative for an SDRRM as ratio R,R-1:S,S-1 is 2:1. This library was setup in an effort to maximise the proportion of Cat I RRRR and Cat II RSRR in the DCL. The experiment was done to prove the similar behaviour of the NDI enantiomers R,R-1 and S,S-1.

UV-Vis titration of NDI-R,R/S,S-serine into Y
Supplementary Figure 85. Reverse-phase HPLC analysis of R,R-2 (5 mM total concentration) library. Absorbance was recorded at 389 nm.

Testing reversibility with DTT
As can be seen below, both libraries were reformed during the experiments, proving their reversible character. Cat II RSRR has reformed in a lower yield, because the concentration was reduced compared to Cat I RRRR.

Computational studies
Geometry optimisation was performed using

DCLSim
DCLs containing the building blocks with and without the NDI-serine competitors were simulated using DCLsim. 5 The results are presented below. For each simulated DCL, we present the building block (and competitor) concentration along with the Gibbs free energy of formation of each library member from its constituent building blocks (DG) as well the simulated and observed DCL distribution in percentages. In the DCLs containing competitors, the Ka column refers to the association constant for the competitor to the corresponding library member. We indicated that the competitor does not interact with a library member by adding a "0" in the Ka column. The equilibria responsible for the DCL distribution are presented under each table.