Carbohydrate-based surfactants as photocontrollable inhibitors of ice recrystallization

Abstract

We report the synthesis and photocontrollable ice recrystallization inhibition (IRI) activity of a panel of carbohydrate-based surfactants. Such materials should attract broad interest as on-demand, photocontrollable cryoprotectants and responsive tools for investigating the dynamic mechanisms of ice recrystallization.

---

Biological antifreezes, including the well-studied antifreeze glycoproteins (AFGPs), are potent inhibitors of ice recrystallization that have long attracted considerable interest as potential cryoprotectants with industrial and biotechnological applications.¹ However, biological antifreezes are considered poor cryoprotectants owing to the propensity of these compounds to irreversibly bind ice crystals and promote thermal hysteresis (TH).² TH activity is undesirable for tissue preservation applications since it leads to premature cryoinjury and cell death during thawing and freezing cycles. The rational development of non-natural analogues that display potent ice recrystallization inhibition (IRI) activity without TH activity has been actively pursued, yet despite intensive efforts, has been met with limited success.^3,4 Furthermore, analogues that were shown to display IRI activity are often complex, high molecular weight materials that are not amenable to large-scale synthesis required for cryopreservation applications. An alternative strategy is based on the observation that small molecule carbohydrate-based surfactants and hydrogelators are potent inhibitors of ice recrystallization, yet do not possess appreciable TH activity.⁵ Despite the considerable potential of this class of cryoprotectant, the precise mechanisms of IRI activity are not well understood, although the degree of hydration of the carbohydrate head group and the presence of hydrophobic moieties are important features.^6,7 IRI activity of these molecules has also been shown to be highly sensitive to changes in the hydrophilic–lipophilic balance (HLB) and is independent of micelle or hydrogel formation.^5,6 To aid the rational design of safe, economical and effective cryoprotectants, further structure–activity studies are required. The ability to tune the physicochemical properties of amphiphilic carbohydrates using an environmental trigger has attracted considerable interest for the development of supramolecular materials with promising applications in synthesis and catalysis,⁸ environmental remediation,⁹ controlled gelation,¹⁰ macromolecular self-assembly,¹¹ and drug delivery.¹² However, the application of an external stimulus for modulating the colloidal properties of amphiphilic carbohydrates in order to control IRI activity has not yet been explored, and would enable access to responsive tools for probing the dynamic mechanisms of ice recrystallization.

Azobenzene trans–cis photoisomerization has emerged as a powerful strategy for studying the structural dynamics of biomolecules,¹³ supramolecular self-assembly of amphiphiles,¹⁴ and the biological activity of low molecular weight drugs.¹⁵ In the presence of UV light, azobenzene undergoes reversible changes in molecular geometry and polarity, as a result of the rapid photoisomerization of the planar and hydrophobic trans isomer to the non-planar and less hydrophobic cis isomer. This process occurs in high quantum yields and can be repeated multiple times without loss of function.¹⁶ The well-studied photochemical properties of this ubiquitous dye molecule and the ease of molecular synthesis have also enabled development of responsive tools with potential in vivo applications.¹⁷ UV-visible light is an ideal environmental trigger since it allows clean, spatiotemporal control over molecular function. Our group has recently reported the photocontrollable interfacial activity and self-assembly properties of a family of light-addressable carbohydrate-based surfactants.^18,19 During this work, we demonstrated the dramatic effect of the head group size, stereochemical configuration and polarity on the physicochemical properties of surfactants, which could be modulated using UV-visible light. In the present study, we report the synthesis and photocontrollable antifreeze activity of a panel of carbohydrate-based surfactants (Fig. 1). We describe the first example of the antifreeze activity of azobenzene containing amphiphiles, as well as the application of azobenzene trans–cis photoisomerization for modulating the IRI activity of carbohydrate-based surfactants.

Fig. 1 Cartoon representation of surfactants 1–3 as photocontrollable inhibitors of ice recrystallization.

To facilitate our strategy, studies commenced with the parallel synthesis of three photoswitchable carbohydrate-based surfactants 1–3 (Fig. 2).¹⁸ Surfactants incorporated a variable monosaccharide head group, including D-glucose (GlcAzo, 1), D-mannose (ManAzo, 2), and D-galactose (GalAzo, 3), which were tethered to a hydrophobic n-butylazobenzene tail group as the photoswitchable unit. Carbohydrate head groups were selected based on the difference in stereochemical configuration, polarity and degree of hydration.⁷ A triethylene glycolate spacer was incorporated to provide additional water solubility. The synthesis of 1–3 proceeded by the Lewis acid-promoted glycosidation of known trichloroacetimidates 4–6 with n-butylazobenzene 7 alcohol acceptor,²⁰ followed by deprotection of the crude intermediates using methanolic sodium methoxide (see ESI† for details). The surfactants 1–3 were obtained as single diastereomers in acceptable yields following purification by reversed-phase preparative HPLC (20–39%).

Fig. 2 Synthesis of photoswitchable carbohydrate-based surfactants 1–3.¹⁸

We next evaluated the photocontrollable interfacial activity of surfactants 1–3 using pendant drop tensiometry (see Table 1 and ESI, Fig. S1†). In order to measure the surface tension of surfactants in the respective trans- and cis-dominated photostationary states, surfactants were first irradiated with UV-light (361 nm) for 10 minutes. Azobenzene trans–cis photoisomerization was monitored by UV-vis spectroscopy by observing the diminished intensity of the peak at 350 nm corresponding to the π–π* transition and an increase in the intensity of the peak at 440 nm corresponding to the n–π* transition, along with two isobestic points at 320 nm and 420 nm.²¹ The ratio of the trans and cis isomers in both photostationary states was estimated by integration of a resolved proton signal in the ¹H NMR spectrum of 2 (ESI, Fig. S2†). In the resting trans-dominated photostationary state, <10% of molecules existed in the cis configuration, while in the cis-dominated photostationary state approximately 70% of molecules existed as the cis isomer.^16,21 The rate of thermal relaxation to the more stable trans isomer was then determined by measuring the change in UV absorption for dark-adapted cis isomers of ManAzo 2 (ESI, Fig. S3†). Encouragingly, little thermal cis–trans relaxation was observed at 20 °C within the 24 h time period. Given the very low temperatures employed for the IRI activity assay, we reasoned that the rate of thermal relaxation of 1–3 would be negligible under these conditions.

From the surface tension data, an increase in the critical micelle concentration (CMC) was observed following photoisomerization, which is in good agreement with previous observations concerning photoswitchable surfactants (Table S1, ESI†). This can be attributed to the increased polarity and altered molecular geometry of the cis isomer relative to the hydrophobic, planar trans isomer.¹⁶ Due to the very poor solubility of GalAzo 3 in water, accurate determination of the CMC was not possible for this compound. The comparable CMC values of GlcAzo 1 and ManAzo 2 in the trans and cis photostationary states indicated a very similar HLB for both photoisomers of these surfactants. In order to evaluate the relative hydrophobicity of the cis and trans isomers of 1 and 2, we then measured the 1-octanol-water partition coefficient (log P) (ESI, Fig. S4†). This technique provides a convenient measure of the partitioning of surfactant monomers between a non-polar oily phase and a polar aqueous phase, and thus the relative affinity for these phases, with larger values indicating greater solubility in the non-polar phase.²² The large negative log P values obtained for trans 1–3 indicates high preference for the aqueous phase, although the variable values indicate a strong dependency on the head group employed. In all cases, the log P values decreased following photoisomerization owing to the formation of the more polar cis azobenzene, which was in good agreement with the CMC data. Unfortunately, due to the significant hydrophilicity of cis GlcAzo 1 and cis ManAzo 2, an accurate log P value could not be determined for these photoisomers.

The ability of these carbohydrate-based surfactants to inhibit ice recrystallization was then investigated using a “splat-cooling” assay (Fig. 3).²³ Previous studies have focused on assessing the IRI activity of ‘classical’ amphiphiles and investigating the effect of modifying alkyl chain length and functionalization. We were interested in investigating the IRI activity of amphiphiles incorporating azobenzene groups, since this moiety readily undergoes hydrophobic π–π stacking interactions in water and also provides a mechanism for tuning amphiphilicity with light. In order to determine the photocontrollable IRI activity of surfactants 1–3, aqueous solutions were irradiated with UV light (361 nm) for 10 minutes prior to performing the IRI assay. Surfactants 1–3 were dissolved in a phosphate buffered saline (PBS) solution and the area of ice crystals was measured after annealing for 30 minutes at −6.4 °C. The ice crystals were represented as a percent mean grain size (% MGS) relative to the PBS positive control for ice recrystallization. Similar to the control, both the trans- and cis-dominated photostationary states of GlcAzo 1 showed weak IRI activity. No photomodulation of IRI activity was observed for this compound, as the trans- and cis-isomers showed near identical IRI activity. Unfortunately, owing to the very poor solubility in PBS buffer, both photoisomers of GalAzo 3 at 5 μM were found to be IRI-inactive. Notably, ManAzo 2 exhibited potent IRI activity in the trans-dominated photostationary state. This striking difference in IRI activity between 1 and 2 is surprising, given the similar properties of these surfactants and the near identical hydration index (HI) of D-glucose and D-mannose. Previously, D-gluco-configured analogues have been shown to possess stronger IRI activity than compounds incorporating D-mannose.²⁴ However, in the case of the photoswitchable amphiphiles, the opposite trend is observed, which highlights the sensitivity of IRI activity toward changes in head group stereochemistry and polarity, as well as tail group hydrophobicity and geometry. Photoisomerization of the trans isomer 2 to the corresponding cis isomer appeared to reduce the IRI activity. The attenuation in IRI activity relative to the trans form may be attributed to the lowered hydrophobicity of the tail group, which is also reflected by the decrease in the log P value. The inhibitory ability of the trans isomer of 2 was further quantified by a modified “splat-cooling” assay where the ice crystals in a high-ice fraction are “binned” based upon size at a five-minute time point (ESI).²⁵ This analysis provides a means of addressing the non-uniform crystal sizes observed during the recrystallization process. By incrementally testing concentrations of ManAzo 2 up to the maximum solubility of 30 mM, a dose–response curve was generated (Fig. 4a) and provided a half maximal inhibitory concentration (IC₅₀) value of 7.0 mM. This kinetic analysis reveals the effective IRI ability of ManAzo 2 at low millimolar concentrations.

Fig. 3 IRI activity of 1–3 at 22 mM represented as a % MGS (mean grain size) of ice crystals relative to the PBS positive control. Values represent the average of three runs ± % SEM. Asterisk indicates significant difference calculated using unpaired Student's T test (P < 0.05).

Fig. 4 (a) Dose–response curve generated for trans ManAzo 2. Normalized rate constants obtained from three experiments ± SEM. A two-parameter sigmoidal curve was fit to the data (see ESI†). (b) Wafer images obtained after 5 minute annealing period during the modified “splat-cooling” assay. Left image corresponds to PBS control while the image on the right corresponds to 30 mM trans ManAzo.

Given the potent IRI activity of ManAzo 2, TH activity was then investigated using nanoliter osmometry (ESI, Fig. S5†). In this assay, a single droplet of a solution of 2 in water is enclosed within an oil-filled well in a sample holder. Using a thermoelectrically-controlled microscope stage, the sample is frozen and then slowly thawed until a single ice crystal remains. At this point, the morphology of the single ice crystal is monitored, as the temperature of the sample is gradually decreased. TH activity is measured as the depression of the freezing point in relation to a static melting point, with non-uniform ice crystal growth being indicative of interaction of the compound with the ice crystal lattice. Due to the amphiphilic nature of ManAzo 2, we were unable to examine its activity at 10 mg mL⁻¹, the concentration used to assess TH activity of the previously investigated C-linked AFGP analogues and carbohydrate-based surfactants and hydrogelators.^5,7 Importantly, when tested at 0.5 mg mL⁻¹, no TH activity was observed for ManAzo 2; ultimately suggesting that the mechanism of ice recrystallization inhibition is not due to the interaction of the compound with the ice lattice.

Since ManAzo 2 exhibited potent IRI activity with no observable TH activity, we investigated the cryoprotective ability of 2 in vitro. During cryopreservation, it is well known that substantial cellular damage is associated with ice recrystallization during thawing.³ Tf-1α cells were cryopreserved with 30 mM 2 in 0–10% DMSO solutions. Samples were thawed under fast-thaw conditions and cell viability and recovery were assessed using flow cytometry. Preliminary results from the cryopreservation of Tf-1α cells with the relatively high concentration of 2 resulted in significantly lower cell recovery than with DMSO alone (ESI, Fig. S6†). Greater cell viability was only achieved with 2 in solutions containing up to 2% DMSO in comparison to the corresponding controls. Further investigation revealed that incubation of Tf-1α cells with 2 prior to cryopreservation led to substantial cell death. These results are not surprising since high concentrations of surfactants are known to solubilize cell membranes.²⁶ While the toxicity of azobenzene functionalized molecules against mammalian cells has been documented, several reports have described non-toxic azobenzene derivatives for in vivo applications.^17,27,28 Therefore, at this stage it is difficult to determine whether the toxicity of 2 arises from membrane damage and lysis and/or the build-up of toxic azobenzene metabolites (e.g. from reduction). Based on the premise that at lower concentrations some surfactants appear to possess a protective ability against cell membrane lysis,²⁶ further studies should be conducted to investigate the cryoprotective ability of 2 at its IC₅₀. Additionally, while in vivo applications may be limited, ManAzo 2 may have potential for materials science applications.

Conclusions

In summary, we report the stereoselective synthesis, photocontrollable interfacial activity, and unprecedented IRI activity of a panel of non-ionic carbohydrate-based surfactants. A decrease in the log P values accompanying photoisomerization could be attributed to the lowered amphiphilicity (higher HLB) of the cis-configured surfactant. The IRI activity of the trans and cis isomers of 1–3 was determined using a “splat-cooling” assay. From this panel, a surfactant molecule incorporating a D-mannoside head group (2) was shown to display potent, light-modulated IRI activity. Furthermore, the trans isomer of 2 did not show any TH activity, thus strongly suggesting that this molecule was not directly binding to the ice lattice to exert antifreeze activity. Following trans–cis photoisomerization of 2, a decrease in IRI activity was observed, which suggests that the formation of the more polar cis azobenzene was detrimental to IRI activity. This supports previous reports that highlight the importance of hydrophobic moieties and increasing hydrophobicity for invoking potent IRI activity. In contrast to surfactant 2, GlcAzo 1 possessing a D-glucoside head group, showed very weak IRI activity that could not be modulated using UV light. To the best of our knowledge, this work represents the first example of an amphiphilic, IRI-active compound incorporating an azobenzene moiety, and the first time a D-manno-configured compound has shown greater IRI activity than a D-glucose analogue. The results presented herein also underscore the importance of head group stereochemistry, polarity and hydration for tuning IRI activity. Our results also demonstrate the viability of using light as an external stimulus for tuning the IRI activity of amphiphilic molecules. Whilst the trans isomer of surfactant 2 resulted in poor cell recovery and viability in cryopreservation studies at high concentrations, these materials show significant promise as a new class of photocontrollable cryoprotectants and as responsive tools for studying ice recrystallization. Work is on going in our laboratories toward the development of non-amphiphilic analogues as well as amphiphilic molecules with tuneable photochemical properties.

Acknowledgements

B. W. thanks the Australian Research Council for the Discovery Early Career research award (DE130101673) for funding. R. N. B. acknowledges the Natural Sciences and Engineering Research Council of Canada (NSERC), Canadian Blood Services (CBS) and Canadian Institutes of Health Research (CIHR) for financial support. M. K. A. and J. S. P. thank NSERC for Canada Graduate Scholarships (CGS M). The authors thank Mathieu Morin (University of Ottawa) for assistance with photoisomerization experiments.

Notes and references

1. M. M. Harding, P. I. Anderberg and A. D. J. Haymet, Eur. J. Biochem., 2003, 270, 1381.
2. J. A. Raymond and A. L. DeVries, Proc. Natl. Acad. Sci. U. S. A., 1977, 74, 2589.
3. A. K. Balcerzak, C. J. Capicciotti, J. G. Briard and R. N. Ben, RSC Adv., 2014, 4, 42682.
4. R. N. Ben, A. A. Eniade and L. Hauer, Org. Lett., 1999, 1, 1759.
5. C. J. Capicciotti, M. Leclére, F. A. Perras, D. L. Bryce, H. Paulin, J. Harden, Y. Liuc and R. N. Ben, Chem. Sci., 2012, 3, 1408.
6. A. K. Balcerzak, M. Febbraro and R. N. Ben, RSC Adv., 2013, 3, 3232.
7. R. Y. Tam, S. S. Ferreira, P. Czechura, J. L. Chaytor and R. N. Ben, J. Am. Chem. Soc., 2008, 130, 17494.
8. N. Drillaud, E. Banaszak-Leonard, I. Pezron and C. Len, J. Org. Chem., 2012, 77, 9553.
9. R. Rajaganesh, A. Gopal, T. Mohan Das and A. Ajayaghosh, Org. Lett., 2012, 14, 748.
10. S. Matsumoto, S. Yamaguchi, S. Ueno, H. Komatsu, M. Ikeda, K. Ishizuka, Y. Iko, K. V. Tabata, H. Aoki, S. Ito, H. Noji and I. Hamachi, Chem.–Eur. J., 2008, 14, 3977.
11. Y. Ogawa, C. Yoshiyama and T. Kitaoka, Langmuir, 2012, 28, 4404.
12. J. Li, M. Huo, J. Wang, J. Zhou, J. M. Mohammad, Y. Zhang, Q. Zhu, A. Y. Waddad and Q. Zhang, Biomaterials, 2012, 33, 2310.
13. A. A. Beharry and G. A. Woolley, Chem. Soc. Rev., 2011, 40, 4422.
14. J. Eastoe and A. Vesperinas, Soft Matter, 2005, 1, 338.
15. W. A. Velema, J. P. van der Berg, M. J. Hansen, W. Szymanski, A. J. Driessen and B. L. Feringa, Nat. Chem., 2013, 5, 924.
16. H. Fliegl, A. Kohn, C. Hattig and R. Ahlrichs, J. Am. Chem. Soc., 2003, 125, 9821.
17. M. Dong, A. Babalhavaeji, S. Samanta, A. A. Beharry and G. A. Woolley, Acc. Chem. Res., 2015, 48, 2662.
18. R. F. Tabor, D. D. Tan, S. S. Han, S. A. Young, Z. L. E. Seeger, M. J. Pottage, C. J. Garvey and B. L. Wilkinson, Chem.–Eur. J., 2014, 20, 13881.
19. R. F. Tabor, M. J. Pottage, C. J. Garvey and B. L. Wilkinson, Chem. Commun., 2015, 51, 5509.
20. M. Billamboz, F. Mangin, N. Drillaud, C. Chevrin-Villette, E. Banaszak-Léonard and C. Len, J. Org. Chem., 2014, 79, 493.
21. (a) C. T. Lee Jr, K. A. Smith and T. A. Hatton, Langmuir, 2009, 25, 13784–13794; (b) A.-L. LeNy and C. T. Lee Jr, J. Am. Chem. Soc., 2006, 128, 6400.
22. Y. Hu, J. B. Marlow, R. Ramanathan, W. Zou, H. G. Tiew, M. J. Pottage, V. Bansal, R. F. Tabor and B. L. Wilkinson, Aust. J. Chem., 2015, 68, 1880.
23. C. A. Knight, J. Hallett and A. L. DeVries, Cryobiology, 1988, 25, 55.
24. P. Czechura, R. Y. Tam, E. Dimitrijevic, A. V. Murphy and R. N. Ben, J. Am. Chem. Soc., 2008, 130, 2928.
25. S. Abraham, K. Keillor, C. J. Capicciotti, G. E. Perley-Robertson, J. W. Keillor and R. N. Ben, Cryst. Growth Des., 2015, 15, 5034.
26. M. Manaargadoo-Catin, A. Ali-Cherif, J.-L. Pougnas and C. Perrin, Adv. Colloid Interface Sci., 2016, 228, 1.
27. S. Samanta, A. A. Beharry, O. Sadovski, T. M. McCormick, A. Babalavaeiji, V. Tropepe and G. A. Woolley, J. Am. Chem. Soc., 2013, 135, 9777.
28. A. A. Beharry, L. Wong, V. Tropepe and G. A. Woolley, Angew. Chem., Int. Ed., 2011, 50, 1325.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental procedures, supplementary tables, figures and spectra. See DOI: 10.1039/c6ra07030b

---