Light-Driven Purification of Progesterone from Steroid Mixtures Using a Photoresponsive Metal–Organic Capsule

Chemical separations are expensive, consuming 10–15% of humanity’s global energy budget. Many current separation methods employ thermal energy for distillation, often through the combustion of carbon-containing fuels, or extractions and crystallizations from organic solvents, which must then be discarded or redistilled, with a substantial energetic cost. The direct use of renewable energy sources, such as light, could enable the development of novel separations processes, as is required for the transition away from fossil fuel use. Metal–organic capsules, which can selectively bind molecules from mixtures, can provide the foundation for these novel separations processes. Here we report a tetrahedral metal–organic capsule bearing light-responsive diazo moieties around its metal-ion vertices. This capsule can be used to selectively separate progesterone from a mixture of steroids in a process driven by visible light energy. Our process combines biphasic extraction and selective binding of progesterone with the light-driven release of this molecule in purified form. Ultimately, our process might be adapted to the purifications of the many other fine chemical products that are bound selectively by capsules.


Materials and Methods
Unless stated otherwise, all the reagents were purchased from commercial sources and used without further purification.Ligand A was prepared following literature procedure with slightly modified reaction condition. 1Compound B was synthesized according to reported procedures. 2 Self-assembly reactions were performed in CD3CN.
Slice-selective NMR experiments were performed on a Bruker AVIII HD Smart Probe spectrometer. 3We thank Dr Peter Gierth and the NMR facility at the University of Cambridge for setting up the slice-selective NMR experiments.

Photo Irradiation
The irradiations at 350 nm and 500 nm were performed in-situ placing the NMR tubes inside a Rayonet photochemical chamber reactor (40 cm deep, 25 cm diameter, 16 × 14 W light sources, operating temperature 32 °C).

Binding of Other Steroids by trans-1
Figure S26: Host-guest studies using other steroids as prospective guests (mestranol, cholesterol and 7-dehydrocholesterol).
The steroids mestranol, cholesterol and 7-dehydrocholesterol were investigated as potential guests for trans-1 (c = 0.97 mM) because the above-mentioned steroids also show good solubility in cyclopentane, as does progesterone.Initially a solution of trans-1 was prepared in a NMR tube as described above.Five equivalents of each steroid was then added, as a solid. 1 H NMR spectra were recorded 30 minutes after the addition of each steroid.NMR spectra indicated that the other steroids were not encapsulated by trans-1.

Determination of Partition Coefficients
Using slice-selective 1 H NMR: The values of the partition coefficients were determined by sliceselective 1 H NMR. 1 mg of each steroid was initially dissolved in 250 l of CD3CN in an NMR tube, then 250 l of cyclopentane was added.The sample was then shaken for 30 minutes to enable the steroids to distribute between the two solvents.The NMR tubes were then allowed to stand for 10 minutes to ensure that the two clear layers were well separated.At equilibrium, the distributions of the steroids were calculated using NMR integration using Phloroglucinol as an internal standard.Using bulk extraction experiment: The partition coefficient of each steroid was measured from bulk extraction experiments.100 mg of each steroid was added into a 1:1 mixture of acetonitrile and cyclopentane (10 ml).The sample was then shaken for 30 minutes for the steroid to distribute between the two solvents.The layers were then separated and evaporated to obtain the solid steroid.Then the partition coefficient of each steroid was calculated using following equation: The partition coefficients of progesterone, mestranol, cholesterol and 7-dehydrocholesterol were determined to be 9.20 ( 0.07), 9.34 ( 0.34), 0.34 ( 0.01) and 0.65 ( 0.02), respectively.
The partition coefficients presented in Fig. 2 were calculated using this extraction experiment.
Table S1.Partition coefficients for progesterone, mestranol, cholesterol and 7-dehydrocholesterol between acetonitrile and cyclopentane with no cage, in the presence of trans-1, and in the presence of disassembled 1.

Extraction and Washing
Initially, trans-1 (c = 0.97 mM) was prepared in an NMR tube in 500 l CD3CN as described above.Five equivalents of each steroid as solids were added.Progesterone was the only steroid encapsulated by trans-1, in slow exchange on the 1 H NMR chemical shift time scale, while the other steroids remained unbound.500 l of cyclopentane was then added to the mixture.The sample was then shaken for 30 minutes for the steroids to distribute between the two solvents.The NMR tubes were allowed to settle for 10 minutes to ensure that the two clear layers were well separated.Cyclopentane containing unbound steroids was then decanted from the NMR tube.For cholesterol and 7-dehydrocholesterol, only one extraction was sufficient, while for progesterone and mestranol, five repeated extractions were required to remove the excess steroids.After removal of cyclopentane, the CD3CN was evaporated using a flow of N2.Finally, fresh CD3CN was used to record the NMR spectrum.

Washing with Cyclopentane to Remove Excess Progesterone
Figure S44: 1 H NMR (CD3CN, 500 MHz, 25 °C) spectra of (i) trans-1 (c = 0.97 mM) in the presence of excess progesterone (c = 4.85 mM).Removal of excess progesterone with (ii) 500 l and (iii) 2.5 ml cyclopentane washings.The peaks of the unbound progesterone are highlighted.

Purification of Progesterone from a Mixture of Steroids Using trans-1, Cyclopentane and Light
Initially trans-1 (c = 5.09 mM) in 500 l of CD3CN was prepared in an NMR tube, as confirmed by 1 H NMR. Mestranol (c = 25.4 mM), progesterone (c = 25.4 mM), cholesterol (c = 25.4 mM) and 7-dehydrocholesterol (c = 25.4 mM) were then added. 1 H NMR indicated that one equivalent of progesterone with respect to trans-1 was encapsulated by the cage, while the other steroids remained unbound.Then, 500 l of cyclopentane was added to the mixture.The sample was then shaken for 30 minutes for the steroids to distribute between the two solvents.The NMR tube was left for 10 minutes to ensure that the two clear layers were well separated.Slice selective 1 H NMR then measured the distribution of steroids in both layers.Cyclopentane containing the unbound steroids was then decanted from the NMR tube.Five extractions, each time with 500 l of cyclopentane, were required to remove all of the unbound steroids.The cyclopentane washes were combined and evaporated.The CD3CN remaining the in NMR tube was evaporated under a flow of N2.Fresh CD3CN was used to dissolve the red solid product inside the NMR tube. 1 H NMR indicated the residual solid to be pure progesterone⊂trans-1.To this sample, 500 l of 3,14 4 5 2 P cyclopentane was added, followed by the irradiation at 350 nm. 1 H NMR after irradiation indicated the release of progesterone from the cage.Slice selective 1 H NMR showed the distribution of progesterone in both layers.Five extractions were again required to maximise the yield of pure progesterone.The CD3CN in NMR tube was evaporated using a flow of N2.Fresh CD3CN was used to dissolve the solid product inside the NMR tube.Finally, the sample was irradiated at 500 nm to reassemble the cage. 1 H NMR indicated the residual solid to be pure trans-1.Six purification cycles were performed to recover 68% ( 4%) of the pure progesterone from the mixture.

Purification of Progesterone from a Stoichiometric Mixture of Steroids Using trans-1, Cyclopentane and Light
Initially trans-1 (c = 6.36 mM) in 500 l of CD3CN was prepared in a NMR tube, as confirmed by 1 H NMR. Mestranol (c = 6.36 mM), progesterone (c = 6.36 mM), cholesterol (c = 6.36 mM) and 7-dehydrocholesterol (c = 6.36 mM) were then added. 1 H NMR indicated that progesterone was encapsulated by the cage while mestranol, cholesterol and 7-dehydrocholesterol remained unbound.Then, 500 l of cyclopentane was added to the mixture.The sample was then shaken for 30 minutes for the steroids to distribute between the two solvents.The NMR tube was allowed to stand for about 10 minutes to ensure that the two clear layers were well separated.Cyclopentane containing mestranol, cholesterol and 7-dehydrocholesterol was then decanted from the NMR tube.Fresh CD3CN was added to the NMR tube to make the volume up to 500 l. 1 H NMR indicated the residual solid to be pure progesterone⊂trans-1. 500 l of cyclopentane was then added before irradiation at 350 nm.An 1 H NMR spectrum was recorded after irradiation indicating the release of progesterone from the cage.Slice selective 1 H NMR was performed, showing the distribution of progesterone between both layers.Pure progesterone was then recovered by extraction.

Purification of Progesterone from a Mixture Containing Six Steroids Using trans-1, Cyclopentane Extraction, and Light
A mixture of 11-α-hydroxyprogesterone (6.36 mM), testosterone (6.36 mM), progesterone (6.36 mM), mestranol (6.36 mM), cholesterol (6.36 mM) and 7-dehydrocholesterol (6.36 mM) in 500 l of CD3CN was prepared in a NMR tube, as confirmed by 1 H NMR.Then, 500 l of cyclopentane was added to the mixture.The sample was then shaken for 30 minutes for the steroids to partition between the two solvents.The NMR tube was allowed to stand for 10 minutes to ensure that the two clear layers were well separated.Cyclopentane containing progesterone, mestranol, cholesterol and 7-dehydrocholesterol was then decanted from the NMR tube.Then the cyclopentane was evaporated and then a solution of trans-1 (6.36 mM) in 500 l of CD3CN was added. 1 H NMR indicated that progesterone was encapsulated by cage 1 while mestranol, cholesterol and 7-dehydrocholesterol remained unbound.Then, 500 l of cyclopentane was added to the mixture.The sample was then shaken for 30 minutes for the steroids to partition between the two solvents.The NMR tube was allowed to stand for 10 minutes to ensure that the two clear layers were well separated.The cyclopentane layer containing mestranol, cholesterol and 7dehydrocholesterol was then decanted from the NMR tube. 1 H NMR indicated the residual solid to be pure progesterone⊂trans-1. 500 l of cyclopentane was then added before irradiation at 350 nm.The released progesterone was then recovered by extraction.Finally, the sample was irradiated at 500 nm to reassemble the cage. 1 H NMR indicated the residual solid to be pure trans-1.(iii) after addition of cage trans-1 followed by extraction of unbound mestranol, cholesterol and 7-dehydrocholesterol with cyclopentane; (iv) after irradiation at 350 nm to purify progesterone, followed by 500 nm irradiation to reassemble the cage.

Volume Calculations
In order to determine the available void space within the structure of cage trans-1, Molovol 4 calculations based on the optimized structures obtained in this study were performed.The cavity volume was calculated to be 715 Å 3 using the parameters tabulated below.

S41
The molecular volumes of steroids were calculated using MoloVol from their MM3-optimized structures.

MM3 Models
Structures were calculated using Scigress, 5 running MM2 followed by MM3 calculations until convergence occurred.

Figure S62.
MM3-optimized molecular model of trans-1 and progesterone⊂trans-1.The radii, measured as half of the distance between farthest-spaced hydrogen atoms, were calculated to be 18.9 Å and 24.8 Å for trans-1 and progesterone⊂trans-1 respectively, consistent with the increase in hydrodynamic radius observed by DOSY (Figures S7 and S20).

Figure S16 :
Figure S16: Encapsulation of progesterone as a guest.

Figure
Figure S17: 1 H NMR (CD3CN, 500 MHz, 0 °C) spectrum of progesterone⊂trans-1. Peaks corresponding to unbound progesterone are labelled with an asterisk.The peaks corresponding to encapsulated guests are indicated by filled circles.The methyl peaks of the encapsulated testosterone guest are assigned as indicated.

5 .
Binding of Progesterone by trans-1 in the Presence of Other SteroidsInitially trans-1 (c = 0.97 mM) was prepared in a NMR tube as described above.Mestranol (c = 0.97 mM or c = 4.85 mM), progesterone (c = 0.97 mM or c = 4.85 mM), cholesterol (c = 0.97 mM or c = 4.85 mM) and 7-dehydrocholesterol (c = 0.97 mM or c = 4.85 mM) were then added individually.It was found that progesterone was the only steroid encapsulated by trans-1, showing a slow exchange on the 1 H NMR chemical shift time scale while other steroids remain unbound.

Figure S37 :
Figure S37: Comparison of 1 H NMR spectra (CD3CN, 500 MHz, 25 °C) of (i) progesterone⊂trans-1 and (ii) trans-1 in the presence of the mixture of steroids.The peaks of the progesterone encapsulated within trans-1 are highlighted.

Figure S52 :Figure S53 :
Figure S52: Purification cycles to extract pure progesterone from a mixture containing excess steroids.

Figure S56 :
Figure S56: Plots showing purification of progesterone from the mixture.Multiple cycles needed to be performed to recover the maximum amount of progesterone.

Figure S57 :Figure S58 :
Figure S57: Extraction of progesterone from a mixture of steroids.

Figure S60 :
Figure S60: Extraction of progesterone from a mixture of structurally similar steroids.

FigureFigure
Figure S64. 1 H NMR (CD3CN, 400 MHz, 0 °C) spectrum of testosterone⊂trans-1. Peaks corresponding to unbound testosterone are labelled with asterisks.The peaks corresponding to encapsulated guests are indicated by filled circles.The methyl peaks of the encapsulated testosterone guest are assigned as indicated.d / ppm

Figure S66 .Figure
Figure S66.trans-1 was observed to bind testosterone in preference to progesterone, as noted below.

Figure S68 .
Figure S68.Comparison of 1 H NMR spectra (CD3CN, 400 MHz, 0 °C) of (i) progesterone⊂trans-1, (ii) testosterone⊂trans-1 and (iii) trans-1 in the presence of progesterone and testosterone.The peaks corresponding to free testosterone and progesterone are labeled.The expanded spectra between 0 and -3 ppm at left indicated selective binding of testosterone with trans-1 in the presence of progesterone.

S46Figure
Figure S69. 1 H NMR (CD3CN, 400 MHz, 0 °C) titration of testosterone⊂trans-1 a) with b) 0.5, c) 1.5, d) 2.5, e) 3.5, and f) 4.5 equiv. of progesterone.The peaks corresponding to free testosterone and progesterone are labeled.The 1 H NMR spectra did not change with the progressive addition of progesterone, indicating a higher binding affinity for testosterone.

Figure
Figure S70. 1 H NMR (CD3CN, 400 MHz, 0 °C) titration of progesterone⊂trans-1 a) with b) 1.5, c) 3.0 and d) 4.5 equiv. of testosterone.The peaks corresponding to free testosterone and progesterone are labeled. 1H NMR indicates displacement of progesterone by testosterone, indicating a higher binding affinity of testosterone than for progesterone.

S47Figure S71 .
Figure S71.Partition of progesterone between acetonitrile and cyclopentane a) with no cage and b) in the presence of trans-1 and testosterone.The partition coefficient of progesterone was unaltered by the presence of the cage, and determined to be 9.20 ( 0.07).

Figure S72 .
Figure S72.Extraction of progesterone from a mixture of progesterone and testosterone.

Figure S73 .
Figure S73.Purification of progesterone from testosterone. 1 H NMR spectra (400 MHz, CD3CN, 0 °C) of (i) cage trans-1 after the addition of 4.5 equiv. of progesterone and testosterone; (ii) after the extraction of progesterone with cyclopentane.The peaks corresponding to free testosterone and progesterone are labeled.