Isotopic Control over Self-Assembly in Supramolecular Gels.

It is common to switch between H2O and D2O when examining peptide-based systems, with the assumption being that there are no effects from this change. Here, we describe the effect of changing from H2O to D2O in a number of low-molecular-weight dipeptide-based gels. Gels are formed by decreasing the pH. In most cases, there is little difference in the structures formed at high pH, but this is not universally true. On lowering the pH, the kinetics of gelation are affected and, in some cases, the structures underpinning the gel network are different. Where there are differences in the self-assembled structures, the resulting gel properties are different. We, therefore, show that isotopic control over gel properties is possible.

Low-molecular-weight, or supramolecular, gels are formed by the self-assembly of small molecules into fibers that subsequently entangle.[1−4] The assembly is driven by noncovalent interactions including hydrogen bonding, hydrophobicity, and π-stacking. As such, very small changes in molecular structure often lead to dramatic differences. It is therefore unsurprising that each molecule has solvent-dependent gelation efficiency.[1]

For hydrogels, hydrophobicity and hydrogen bonding are dominant noncovalent interactions.[5] On changing from H2O to D2O, a number of properties change, including density, viscosity, and hydrogen bond strength.[6] Additionally, the hydrophobic effect has also been reported to be more pronounced in D2O than in H2O.[7] In some systems, substituting H2O for D2O can lead to a change in properties. For example, the persistence length of elastic peptides is higher in D2O than that in H2O, ascribed to stronger hydrogen bonding.[8] Slight differences in dimensions have been reported for nanotubes formed from a small peptide in H2O or D2O.[9] For biopolymer-based gels, the melting temperature of gelatin gels is higher in D2O as compared to H2O,[10] and the gels are more rigid in D2O. Similarly, agar-based gels have a higher modulus in D2O compared to H2O,[11] as do κ-carrageenan-based gels.[12] Fibrinogen has been shown to have higher degrees of lateral aggregation in the gel state in D2O as compared to H2O.[13] The higher melting points of gelatin gels in D2O can be ascribed to the enhanced stability of the triple helices[10] and similar increases in melting point have been shown for other biopolymer gels in D2O compared to H2O.[12,14] Structural changes have also been observed in lipid systems when changing from H2O to D2O.[15]

For low-molecular-weight gels, there is very little information as to whether there is an effect of changing from H2O to D2O. Canrinus et al. reported differences in gel strength in some cases when changing from H2O to D2O on the basis of gel melting temperatures, which could differ by as much as 50 °C.[16] The rheological data were stated to be essentially the same. Variations in hydrophobicity were assigned as the dominant reason for changes in the melting point.

In addition to the suggestions that it might be possible to change the gel properties when using D2O instead of H2O, there are also important implications for a number of experimental techniques. It is common, for example, to carry out infrared spectroscopy in D2O instead of H2O to minimize the absorbance of water.[17,18] Likewise, NMR experiments are typically carried out in D2O. Small-angle neutron scattering (SANS) is most often carried out in D2O to allow contrast with the gelators.[19,20] In all cases, the often implicit assumption is that this change has no effect.

Here, we focus on a small library of dipeptide-based gelators (Scheme ).[4,21−26] These form gels in water using a pH-switch. Typically, a solution of one of the gelators is prepared by dispersing the molecule at high pH (pH 10–11) at a concentration of 5 mg/mL. Decreasing the pH results in gelation. The kinetics here control the homogeneity of the gel and so we commonly exploit the hydrolysis of glucono-δ-lactone (GdL) to gluconic acid to lead to a slow, controlled decrease in pH.[27,28] This leads to very reproducible gels.[21] The rate of hydrolysis of GdL has been reported to differ in H2O and D2O.[28]

Scheme 1

Chemical Structures of the Gelators

There are therefore primarily two states to be considered where there might be differences in H2O and in D2O: the high-pH (solution) phase and the low-pH (gel) phase. It can be difficult to probe these states effectively. It is common to use electron microscopy to image the underlying structures. However, there can be drying artifacts for these systems.[29] We, therefore, turned to small-angle X-ray scattering (SAXS).[19] SAXS can be carried out directly on either the solution or gel phase, provides data on the structures of the bulk sample, and can be carried out equally effectively in H2O and D2O. We also note here that even small changes in molecular structure can have a profound effect on the outcome of the self-assembly in both the solution and gel states.[2,4,22]

Initially, we focus on the behavior in the solution state at high pH. We have reported previously the assembly of 1 in both H2O[30] and D2O,[31] with no major difference observed between the systems. At high pH, at a concentration of 5 mg/mL, 1 forms a viscous solution. In line with previous data, at high pH, the SAXS data (Figure a) fit to a flexible cylinder model with radii of 4.1 and 4.3 nm in H2O and D2O, respectively, Kuhn lengths of 50 and 77 nm, respectively, and a length outside the scattering length that is accessible from collecting the data over this Q-range. In line with these data, cryo-TEM of the solutions (Figure b,c) shows long, flexible structures.

Figure 1

SAXS data and fit for solutions of 1–4 in H2O (open symbols) and D2O (closed symbols), with fits as red lines: (a) 1, (d) 2, (g) 3, and (j) 4. Also shown are example cryo-TEM data for solutions of 1–4 in H2O and D2O: (b) and (c) 1 in H2O and D2O, respectively, (e) and (f) 2 in H2O and D2O, respectively, (h) and (i) 3 in H2O and D2O, respectively, and (k) and (l) 4 in H2O and D2O, respectively. All data was collected at a concentration of 5 mg/mL and a pH of 11. For the cryo-TEM data, the scale bars represent 200 nm in each case.

For solutions of 2 at high pH, the best fit to the SAXS data (Figure d) is using a hollow cylinder combined with a power law to take into account the scattering at low Q. The fits to the data imply that the tubes have radii of 1.7 and 1.9 nm in H2O and D2O, respectively, and thicknesses of 3.2 and 2.8 nm, respectively. The cryo-TEM images (Figure e,f) agree with the SAXS data, showing long, anisotropic structures. For 3 at high pH, the SAXS (Figure g) and cryo-TEM data (Figure h,i) again show that very similar structures are formed in H2O and D2O. In both cases, the SAXS data can be fitted to a hollow tube model, with radii of 28.1 and 28.7 nm in H2O and D2O, respectively, and a thickness of 4.3 nm in each case. Cryo-TEM again backs up the fits to the SAXS data. Finally, for 4 at high pH, there is a difference in the SAXS data (Figure j). The model that best fits the SAXS data for the sample in H2O is a flexible elliptical cylinder with a radius of 1.05 nm and an axis ratio of 3.9, whilst the sample in D2O is best fit using a flexible cylinder with a radius of 2.6 nm. The cryo-TEM data (Figure k,l) backs up the fits to the SAXS data, showing that the structures formed in H2O and D2O at high pH are indeed different, with more tapelike structures found in H2O.

Hence, there is generally little difference in H2O and D2O at high pH. There is a general tendency for the radii to be very slightly higher in D2O, which may be due to solvation differences. Nonetheless, the structures formed are very similar in both solvents. However, for 4, the structures formed are different.

We now discuss the gels. Gelation was then induced in all cases by the addition of GdL,[27,28] leading to protonation of the terminal carboxylates. The rate of pH decrease is different in H2O and D2O, being slower in D2O (Figure S3, Supporting Information) in all cases. As a result, the times at which gelation begins (where the storage (G′) begins to deviate strongly from the loss (G″) modulus) as well as the profiles of G′ and G″ are different. In all cases, gelation begins and achieves plateau values at earlier times in H2O as compared to D2O, correlating with the slower hydrolysis of GdL in D2O. The rate of hydrolysis of GdL is catalyzed by many acids and bases, with the relative rate depending on the catalytic species.[28] Since we have a complex solution where aggregates exist and are changing, as well as an evolving pH, the exact species catalyzing the hydrolysis is difficult to determine. Nonetheless, we observe that the hydrolysis in these systems is faster in H2O than in D2O (Figure S3) and this directly links to faster gelation in the H2O compared to that in D2O. The final gels are visually similar in both H2O and D2O (Figure ). For 1, although the underlying structures are very similar at high pH (see the discussion above), the viscosities are different, which may be a result of the higher Kuhn length in D2O as compared to H2O. This manifests in the sample in D2O at early times having a storage modulus (G′) that is higher than the loss modulus (G″) (Figure a). The SAXS data can be used to determine the structures present but will not be easily able to pull out information about interactions between these structures. In H2O, whilst still viscous, G″ dominates at early times. Since the hydrolysis of GdL is faster in H2O, changes in G′ and G″ occur earlier in the sample in H2O compared to that in D2O (Figure a). However, the final rheological values of G′ and G″ are similar in H2O and D2O for the gels formed from 1 (Figure S4). This is expected; we have previously found little differences for gels formed from 1 in both H2O and D2O.

Figure 2

Time-sweep rheology data, photographs of the final gels, and SAXS data for gelation of (a) 1, (b) 2, (c) 3, and (d) 4. For rheology, the data in blue are for H2O and the data in red are for D2O. In all cases, closed symbols show G′ and open symbols show G″. In the photographs, the left vial is for the gel in H2O and the right vial is for the gel in D2O. For the SAXS data, the open symbols are for the gels in H2O and the filled symbols are for the gels in D2O. The fits are red lines. For (d), the data have been manually rescaled as the data were collected on different instruments.

However, the final values of G′ and G″ differ for gels formed from 2, 3, and 4 in H2O and D2O. For 2, the initial solutions are very similar in terms of the values of G′ and G″ (Figure b) and, whilst the rates of change in the moduli differ in H2O and D2O, the moduli for the final gels are relatively similar (Figure S4). For 3, the initial values of G′ and G″ are different, with G′ being higher for the solutions in H2O. The differences in rheological data at early times for 3 show that the interactions between the structures must be stronger in H2O as compared to those in D2O since the SAXS data implies that the structures present at high pH are very similar. The final gels are stiffer in H2O as compared to those in D2O. For 4, the initial solutions have higher values of G′ and G″ in H2O compared to those in D2O, and G′ dominates over G″ from time zero. This correlates with the SAXS data showing that the structures are different at high pH. There are differences in the profile of the changes in G′ and G″ with time for 4 (Figure d), with the sample in H2O showing a steady change in G′ and G″, whilst that in D2O shows a two-stage process. We have previously ascribed such two-stage processes to initial fiber formation and then lateral bundling.[32]

The rheological data are determined from the mechanical properties of the primary self-assembled structures, as well as the degree of lateral association and other entanglements, which combine to give the overall gel network. The similarity in data for gels formed from 1 in H2O and D2O could be coincidental, with the average of very different interactions leading to an overall similar gel.[14,33] Alternatively, the similarity may suggest that the primary structures and networks are not affected by the change in solvent.

Cryo-TEM of the gel phase is problematic due to sampling issues from the stiff networks (see the discussion in the Supporting Information and Figure S5). Hence, to probe the underlying structures, we again turned to SAXS (Figure ). For gels of 1, the SAXS data are very similar. The data can be fitted to a flexible elliptical cylinder. This is as expected from previous work; primary fibers laterally aggregate to lead to structures where the scattering can be best fit to an elliptical shape.[31] From the fitting, the radii were 2.5 and 2.7 nm in H2O and D2O, with axis ratios of 2.1 and 2.2, respectively. There are differences in the Kuhn length, a measure of the structures’ flexibility, with values of 25 and 95 nm for H2O and D2O, respectively. The lengths in both cases are again outside the range that can be probed here. These data imply that the structures in the gel phase are essentially the same in both H2O and D2O, with perhaps some variation in flexibility. The gels are formed at different rates and so the difference in flexibility may represent different degrees of entanglement and lateral packing resulting from how quickly charge is removed from the structures.

For gels formed from 2, the best fit to the SAXS data is again the flexible elliptical cylinder, with the radii being very similar (4.3 and 4.6 nm in H2O and D2O, respectively), as are the axis ratios (3.1 and 3.3, respectively), and the Kuhn lengths (around 25 nm in both cases), with the overall length again being outside the range that can be probed by SAXS. Hence, for 2, the structures in the gel phase are very similar in H2O and D2O despite the small differences in rheology.

For gels formed from 3, the differences in the rheology data are reflected in the SAXS data. The data for the gels in H2O can be best fitted to a hollow cylinder model, with a radius of 22 nm and a thickness of 6.5 nm. A polydispersity in the radius of 0.11 needed to be included to ensure a good fit. Hence, in H2O, the structures in the gel phase are very similar to those in the solution state. In D2O, however, the SAXS data can be best fitted to a flexible elliptical cylinder model with a radius of 3.2 nm and an axis ratio of 3.5. Hence, the differences in rheology can be understood in terms of different underlying structures in the two solvents.

For gels formed from 4, the scattering data are again different from one another. The data from the gels in H2O can be best fitted to a flexible elliptical cylinder model, with a radius of 2.9 nm and an axis ratio of 1.9. The data for the gels formed in D2O fit best to a cylinder model combined with a power law. The cylinders have a radius of 4.0 nm. Hence, again, the differences in the rheology of the gels in H2O and D2O can be ascribed primarily to different structures underpinning the network.

Hence, where the underpinning structures differ, there are concomitant differences in the rheological properties. In all cases, the kinetics of the hydrolysis of GdL and hence the rate of pH decrease, and gelation are different; we cannot, therefore, relate the structural differences where they are present simply to kinetics. The rate of hydrolysis is temperature dependent.[28] However, it is not possible to simply carry out experiments at different temperatures to match the kinetics of hydrolysis in H2O and D2O. For this class of gelator, there can be temperature effects. For example, 1 has a different self-assembled structure at room temperature and above 40 °C, for example.[34] Likewise, it is difficult to suggest that there is a link between a single property such as hydrophobicity and whether there is an effect on changing from H2O to D2O. Nonetheless, we show that there is potential to use isotopic changes to control the properties of gels from a single gelator. This shows that the general assumption that there is no effect in moving between H2O and D2O does not always hold.