Embedding and Positioning of Two FeII 4L4 Cages in Supramolecular Tripeptide Gels for Selective Chemical Segregation

Abstract An unreported d,l‐tripeptide self‐assembled into gels that embedded FeII 4L4 metal–organic cages to form materials that were characterized by TEM, EDX, Raman spectroscopy, rheometry, UV/Vis and NMR spectroscopy, and circular dichroism. The cage type and concentration modulated gel viscoelasticity, and thus the diffusion rate of molecular guests through the nanostructured matrix, as gauged by 19F and 1H NMR spectroscopy. When two different cages were added to spatially separated gel layers, the gel–cage composite material enabled the spatial segregation of a mixture of guests that diffused into the gel. Each cage selectively encapsulated its preferred guest during diffusion. We thus present a new strategy for using nested supramolecular interactions to enable the separation of small molecules.


1.2) Mass spectrometry (MS)
Low resolution electrospray mass spectra (LR ESI-MS) were obtained on a Micromass Quattro LC (cone voltage 10-30 eV, desolvation temp. 313 K, ionization temp. 313 K) infused from a Harvard Syringe Pump at a rate of 10 µL per minute. High resolution electrospray mass spectra (HR ESI-MS) were obtained on a Thermofisher LTQ Orbitrap XL hybrid ion trap mass spectrometer. NMR spectra of gels and slice selective experiments were recorded on a Bruker 500 MHz AVIII HD Smart Probe Spectrometer using the previously reported pulse and au program designed by Duncan Howe for the Chemistry Department of the University of Cambridge. [2]

1.5) Circular Dichroism.
A 0.1 mm quartz cell was used in a Jasco J-815 Spectropolarimeter, with 1 s integrations, 1 accumulation and a step size of 1 nm with a bandwidth of 1 nm. The CD signal was monitored from 210 to 700 nm at 25 °C (Peltier).

1.6) Oscillatory rheometry.
Dynamic time sweep rheological analyses were performed on a Malvern Kinexus Ultra Plus Rheometer (Alfatest, Milan, Italy) with a 20 mm stainless steel parallel plate geometry. The system was kept at 25 °C using a Peltier temperature controller. Each gel was prepared in situ and immediately analyzed with a gap of 1 mm. Time sweeps were recorded for 1 hour, using a frequency of 2 Hz and a controlled stress of 2 Pa. After 1 hour, frequency sweeps were recorded from 0.1 to 10 Hz using a controlled stress of 2 Pa. Finally,

SUPPORTING INFORMATION
S4 stress sweeps were recorded using a frequency of 2 Hz until the breaking point for every gel, recognizable by the inversion of G' and G'' values. Each analysis was repeated at least 3 times.

1.7) UV analysis.
UV visible spectra were recorded on a PerkinElmer Lambda 35 UV-Vis spectrophotometer using a 0.1 mm path length cuvette. The samples were freshly prepared for the analysis and the measurements were performed at room temperature.

1.8) Transmission electron microscopy (TEM) and energy-dispersive X-ray spectroscopy (EDXS).
TEM analyses were performed on JEM 2100 (Jeol, Japan) equipped with an Oxford Instruments INCS energy-dispersive X-ray spectroscopy (EDXS) detector operated at 100 kV. TEM grids (copper-grid-supported lacey carbon film) were first exposed to the UV-ozone cleaner (UV-Ozone Procleaner Plus) for 10 mins to make the grid surface more hydrophilic. Five days aged gels were precisely deposited on a TEM grid, dried for 15 mins at room temperature, and contrasted by aqueous tungsten phosphate solution (pH 7.4). TEM micrographs were acquired on at least 15 different spots on TEM grid accompanying corresponding EDXS analyses.
The average size or cross-section diameter of the nanostructures was determined by averaging at least 100 individual nanostructures.

1.9) Raman Spectroscopy.
Raman analysis was acquired in an Invia Renishaw microspectrometer (50) equipped with He-Ne laser at 532 nm. The laser was focused using a 50 x microscope and the power of the laser was set as required for the sample. At least 10 spectra per sample were acquired to assure the homogeneity of the samples. For the characterization of the powders, a small amount of sample was deposited on a glass microscope slide. For the gels, a small piece of gel was deposited on quartz and open air dried before the analysis. Sample preparation was analogous for Raman imaging. An area of around 625 µm was mapped over 676 points. The exposure time was 1 s and the power of the laser was set at 0.08 mW. The data was normalized to the maximum and the relative intensity of the peak at 1448 and 1472 cm -1 were plotted for cage 1 and cage 2, respectively.

2.1) Peptide p-aminobenzoyl-L-Phe-D-Ala-L-Phe-NH2
p-aminobenzoyl-L-Phe-D-Ala-L-Phe-NH2 was synthesized following Fmoc-based SPPS under dry and inert atmosphere. Briefly, the swelling of the resin (2-chlorotrytil chloride, 10 g) was done in dichloromethane (40 mL). Then SOCl2 (1 mL) was added, and the reaction was shaken under an argon flow for 1 h. After that, the resin was washed with DMF (2 x 30 mL) and dicloromethane (2 x 30 mL). Next, a solution of Fmoc-Rink amide linker (5.4 g, 10 mmol), DIPEA (9 mL) in DMF/dichloromethane       The data was consistent with the previously reported data for cage 1 with OTfcounterions. [3]    The data was consistent with the previously reported data for cage 2 with OTfcounterions. HOAt (23 mg, 0.17 mmol, 0.1 eq) were combined in THF (50 ml). The mixture were stirred at room temperature for 18 h. The solvents were removed under reduced pressure and the resulting materials was dissolved in 50 ml of CH2Cl2. The solution was washed with water (2 × 50 ml), saturated aqueous NH4Cl (2 × 50 ml) and saturated aqueous NaHCO3 (2 × 50 ml). The organic phase was dried over MgSO4 and the solvents removed under reduced pressure. The resulting solid was purified by silica gel chromatography (CH2Cl2) giving the desired nitrophenyl compound as a yellow solid. The nitrophenyl compound obtained was dissolved in CH3OH and Pd/C (50 mg) was added.
The mixture was stirred under H2 atmosphere for 3 h. The Pd/C was removed by filtration on celite and the solvents removed under reduced pressure to give the desired aniline as an off-white solid (443.9 mg, 1.48 mmol, 89% over two steps).  These data were consistent with previously reported data. [4] 2.4.2) Cage 3 5,5',5''-(benzene-1,3,5-triyl)tripicolinaldehyde A (3.6 mg, 0.0092 mmol, 1.0 eq.) and aniline C (8.2 mg, 0.0275 mmol, 3.0 eq.) were dissolved in 2.0 mL of CH3CN in a sealed 5 mL Schlenk flask. Oxygen was removed by freeze-pumped-thawing with N2 three times after which Fe(NTf2)2 (6.3 mg, 0.0023 mmol, 1.0 eq.) was added. The solution was freeze-pumped-thawed an additional two times. HR-ESI-MS: m/z calculated for 3(NTf2)3 5+ = 1200.0794, observed = 1200.0829.       The data was consistent with the previously reported data for cage 2 binding FA. [5] SUPPORTING INFORMATION S14  The peptide was dissolved in acetonitrile by heating. Then the other solvent was added to get a final concentration of 20 mM and it was sonicated for the indicated time to get the gel.

4.2) General preparation of peptide/cage gel systems (1⊂Gel and 2⊂Gel)
For the gels containing 1 mM of cage, a stock solution of 5 mM cage was prepared in acetonitrile. Then, the peptide was dissolved in acetonitrile and added to the required volume of the cage solution to get the desired final concentrations of peptide and cage of 50 mM and 1 mM, respectively. The sample was ultrasonicated for 5 minutes and the gel was formed. For the gels containing 5 mM of cage, a stock solution of 20 mM cage was prepared in acetonitrile. Then, the peptide was dissolved in acetonitrile and added to the required volume of the cage solution to get the desired final concentrations of peptide and cage of 50 mM and 5 mM, respectively.
The sample was ultrasonicated for 10 minutes and the gel was formed.

4.3) Circular Dichroism.
The analysis of the samples at the concentration of peptide required for gelation (50 mM) was not viable since the voltage goes above ~ 600 and the detector is saturated (signal too noisy to be reliable). Thus, we selected 10 mM as peptide concentration and SUPPORTING INFORMATION S15 different concentrations of cage (0; 0.05; 0.1; 0.15 and 0.2 mM). Samples were freshly prepared, and the spectra immediately recorded.       32.0 ± 5.6 38.6 ± 4.9 21.9 ± 6.0

S21
This technique is quick and non-destructive, and has also been fruitfully applied to other peptide gel composite materials. [6] In the present case, the native gel displayed intense signals at 1004, 1037, 1570, and 1607 cm -1 . To our knowledge, imine-based supramolecular capsules have not previously been studied by Raman. Spectra of both MOCs displayed characteristic, intense bands between 1450 and 1650 cm -1 , which were attributed to C=N and N-Fe stretching, and to the Raman-active modes of the ligand.
Raman signals attributed to the MOCs dominated the spectra of 1⊂Gel and 2⊂Gel, confirming the integrity of the cages.

5) Host-guest chemistry in gels
All samples were prepared in sealed J-Young NMR tubes to prevent evaporation of the solvent over time. Samples were monitored by 1 H and 19 F NMR for a period of 3 weeks.
The errors on the values of the integrals (reading error) were assumed to be 5%.  No precipitate was observed to form over time and no significant drop in the intensity of the encapsulated TFApeak in the NMR was observed, demonstrating the stability of the cage in solution in the presence of 20 eq. of KTFA. The small shift of the NTf2peak observed upon addition of KTFA was attributed to the change in the environment of the cage due to TFApotentially associating with the outside of the cage and due to small change in the solution's acidity. Precipitate was observed to form over time and a small amount of free FA was observed after 2 weeks, which is consistent with the drop in intensity of the encapsulated FA. The free FA observed was attributed to a slight decomposition of 2 over time, whereas the precipitate and the important drop in the encapsulated FA observed was attributed to 2 precipitating out of solution as the TFAcomplex. This shows that 2 is reasonably stable in the presence of 20 eq. of KTFA in solution on short time scales but precipitation and a small amount of decomposition occurs over longer periods.

5.1.2) Stability of 1 and 2 in the gel
To 0.35 mL of a 5 mM solution of cage 1 or FA ⊂ 2 in CD3CN was added 10 equivalents (12.5 mg, 17.5 µmol) of the peptide. The samples were sonicated for 10 min to promote the formation of the gel. The encapsulated guest peak in the 19 F NMR spectra was monitored over time and integrated against the NTf2peak. The values were normalised so that the highest point (at 336 h for 1 and 12 h for 2) is equal to 100% as only the host-guest complex was observed at this stage.

5.1.3) Subcomponent exchange in 1
Aniline C (5.2 mg, 17.5 µmol, 10 eq.) was added to a 5 mM solution of MOC 1 (8.9 mg, 1.75 µmol, 1 eq.) containing KTFA (5.3 mg, 35 µmol, 20 eq.). The sample was monitored over time and compared against the spectra of TFA -⊂3 to evidence aniline exchange. No signals corresponding to cage 3 were observed after 1 week. This was explained by the small amount of aniline exchanged and the likely desymmetrisation of the signals in the species incorporating C. However, small signals corresponding to free p-toluidine in solution appeared over time. The maximum exchange (6%) was reached after 4 days. We concluded that, due to the electron withdrawing character of the amide, nucleophilic attack of C onto 1 was limited at the concentrations used in our study. Furthermore, we consider this case to be the "worst case scenario" as we expect the equilibriation in the gel to be slower, as has been seen for guest uptake. We also expect the anilines to be less available for nucleophilic attack in the gel than in solution as they are involved in the gel fibrils. By extension, we expect an absolute maximum of 6% of the p-toluidine to be exchanged for the tripeptide in 1, which represents less than one molecule per cage. The maximum encapsulation of ReO4 -(86% ± 4) was reached after four days. The process followed an exponential asymptotic model but was too fast to allow calculation of a rate of encapsulation in this case.

SUPPORTING INFORMATION
S29 Figure  The maximum encapsulation of FA (69% ± 3) was reached within 2 h of the guest addition. The process was too fast to allow calculation of a rate of encapsulation in this case. The subsequent drop in the encapsulated FA followed the same trend as FA ⊂ 2 in solution and was attributed to the precipitation of 2 as the TFAadduct as no increase in the free FA indicative of cage decomposition could be observed.

5.2.2) Guest uptake in gel
To 0.35 mL of a 5 mM solution of cage 1 or 2 in CD3CN was added 10 equivalents (12.5 mg, 17.5 µmol) of the peptide. The samples were sonicated for 10 min to promote the formation of the gel. After 2 h of equilibration for 2 and 3 weeks for 1 (in this case the guest uptake experiment was performed on the same gel sample used to test the stability of cage 1), 1 equivalent (1.75 µmol) of TBA ReO4 or FA in 20 µL of CD3CN was respectively layered on top of the gels containing 1 or 2. The encapsulated TFApeak for 1, and the

SUPPORTING INFORMATION
S30 encapsulated NTf2 -, encapsulated FA and free FA for 2 in the 19 F NMR spectra were monitored over time and integrated against the NTf2peak. The values for 1 were normalised so that the value for the encapsulated TFAis equal to 100% before the addition of TBA ReO4 as only the host-guest complex TFA -⊂ 1 is present at this stage. The drop in the intensity of the encapsulated TFAis attributed to the encapsulation of ReO4and therefore the values for the encapsulated ReO4were assumed the complement of the values for the TFA -. The values for 2 were normalised so that the maximal sum of the signal for the free and encapsulated FA (after 1 week, once all the guest had diffused through the gel) was equal to 100%.  The initial rate of encapsulation was given by the formula: [7] = − * ( ) giving the rate of encapsulation for ReO4 -: kiniReO4 = 1.6 ± 0.05 %.h -1 Figure S54: 1 H NMR spectra (500 MHz, 298 K, CD3CN) (left) and 19 F NMR spectra (471 MHz, 298 K, CD3CN) (right) of 2 ⊂ Gel over three weeks after addition of 1 eq. of FA. The encapsulated NTf2dropped rapidly, with no trace of encapsulation for this compound after 48 h. Maximum encapsulation of FA (59% ± 3) was reached after 72 h. The process followed an exponential asymptotic model. The calculated rate of encapsulation is kiniFA = 7.23 ± 0.15 %.h -1

5.3.1) Diffusion of respective guest
The diffusion of guests was followed over time by following the evolution of their signals in the NMR spectra. The diffusion curve of TBA ReO4 in the gel made of the peptide and cage 1 was obtained by monitoring the signal for the TBA + at 1.6 ppm in the 1 H NMR spectra integrated against the acetonitrile peak. Values were normalized so that the maximum diffusion after 3 weeks was equal to 100%. The diffusion curve of the FA in the gel made of the peptide and cage 2 was obtained by summing the integrals for the free and encapsulated FA relative to the NTf2peak. Values were normalized so that the maximal sum of the signal for the free and encapsulated FA (after 1 week) was equal to 100%.

SUPPORTING INFORMATION
S33 Figure   Diffusion of FA reached a steady regime after 2 days. The process followed an exponential asymptotic model. The calculated rate of diffusion is kiniFA/diff2 = 14.61 ± 0.28 %.h -1 .

5.3.2) Diffusion of FA in 1⊂ Gel
To compare the diffusion of compounds between gels containing 1 and 2, 1 equivalent of FA (1.75 µmol) in 20 µL of CD3CN was layered on top of a sample of gel containing 1. The values of the integral of the free FA peak against NTf2were normalized so that the maximum value after 3 weeks was equal to 100%. The diffusion follows different kinetics for the FA in the gels containing 1 and 2, indicating that the rate of diffusion is influenced by the type of cage incorporated in the gel. This is consistent with the gel having different structures when cage 1 or 2 are incorporated, an observation which is also supported by the rheology results.  Diffusion of TFA⊂1 reached a steady regime after 1 week. The process followed an exponential asymptotic model. The calculated rate of diffusion is kini1/diff2 = 2.11 ± 0.11 %.h -1 . Diffusion of FA⊂2 reached a steady regime after 10 days. The process followed an exponential asymptotic model. The calculated rate of diffusion is kini2/diff1 = 1.31 ± 0.20 %.h -1 . The diffusion follows different kinetics for the guests and the cages in the gel containing 1 or 2. The diffusion of the guest is much faster than of the cage, showing the suitability of this system for separation on timescale shorter than a week. Indeed, the guests will diffuse and become encapsulated in their respective layers before significant leaching of the cages into each other's layer happens.

5.4) Spatial segregation of cages and guest separation
The set up for slice selective NMR was similar to the one previously reported. [2] To 0.120 mL of a 5 mM solution of cage 1 in CD3CN was added 10 equivalents The optimal slices showing most intense signal in the 1 H were identified for each layer of gel and the SPOFFS1 were recorded: Top layer: SPOFFS1 = 9200 Bottom layer: SPOFFS1 = -9200 The SPOFFS1 being dependent of the frequency of excitation and the probe used, the SPOFFS1 for the 19 F NMR were identified by matching the slices, giving: Top layer: SPOFFS1 = 9200 Bottom layer: SPOFFS1 = -18400 In order to keep the performance of the 19 F experiments similar to 1 H, a 20ppm spectral width was acquired and offset selected to observe each region of interest. One region of the 1 H NMR (0 -10 ppm, NS = 8) and two regions of the 19 F NMR (−135 to − 115 ppm and −88 to − 68 ppm, NS = 256) were monitored. It is thus not possible to treat the 19 F data quantitatively, as the peaks could not be

SUPPORTING INFORMATION
S40 integrated against a reference signal. It was also necessary to acquire 256 scans per spectrum to enable detection of the guest signals, which prevented us from acquiring well-defined 2D 19 F NMR maps of the two regions of interest (−135 to − 115 ppm and −88 to − 68 ppm). Such 2D maps would have taken approximately six hours each to acquire and would therefore not have been representative of the state of the system at a single point in time. However, relative values can be obtained by comparing the absolute values of the integrals of peaks between different time points in the 1D slice-selective 19 F NMR as the gain and number of scans was kept identical between runs.
Once the slice selective experimental parameter was optimized, 1 equivalent (0.60 µmol) of a mixture of TBA ReO4 and FA in 20 µL of CD3CN was then layered on top of the sample. The peaks corresponding to encapsulated TFA -, encapsulated FA and free FA in the 19 F NMR spectra were monitored over time. Figure S69: Composition of the sample for the slice selective NMR experiment.  After addition of the mixture of guests on top of the tri-layered sample, we observe the appearance of the signals corresponding to the encapsulated FA in the top layer containing 2. Free FA was observed in both layers 1 and 2, which we infer to be a result of the weaker binding of this guest in cage 2. We note that no comparisons between the intensity of the signals observed in both layers could be made, as the performance of the detection greatly varied from one layer to another, evidenced by the higher signal/noise ratio in layer 2 compared to 1. However, layer 2 additionally presented signals corresponding to FA ⊂ 2, which were only observed after 24 h in layer 1 due to the leaching of FA ⊂ 2. We thus concluded that despite the lack of absolute separation of FA between layers, an enrichment of this compound was achieved in layer 2.
Similarly, we observe the disappearance of the encapsulated TFA − peak in the layer containing 1, indicative of the ReO4 − encapsulation. As ReO4is NMR silent, its presence in layer 2 could not be directly observed. However, 1 eq. of perrhenate per eq. of MOC 1 was added to the tri-layered system, and approximately 80% of ReO4was observed to be encapsulated in 1, as noted by following the displacement of TFA by 19 F NMR by integration against the triflimide peak. We infer that the remaining 20% would be entrapped in the gel (as opposed to encapsulated inside the MOC) and would become spread evenly between both layers. We thus conclude that, even though complete segregation of the ReO4in layer 1 was not achieved, an estimated enrichment of up to a 9:1 ratio between layer 1 and layer 2 was obtained.