Conformational photoswitching of a synthetic peptide foldamer bound within a phospholipid bilayer

See allHide authors and affiliations

Science  29 Apr 2016:
Vol. 352, Issue 6285, pp. 575-580
DOI: 10.1126/science.aad8352

Synthetic twists among lipids

Proteins embedded in cell membranes perform a wide variety of signaling and transport functions through conformational shifts. De Poli et al. examined how a much smaller, simpler construct might begin to achieve similar aims (see the Perspective by Thiele and Ulrich). Specifically, they designed an artificial peptide with a photosensitive group at one end and embedded it in a phospholipid bilayer akin to a membrane. Nuclear magnetic resonance spectroscopy revealed how light-induced isomerization influenced conformational dynamics at the other end. The results point the way toward development of small-molecule–based switches in membrane environments.

Science, this issue p. 575; see also p. 520


The dynamic properties of foldamers, synthetic molecules that mimic folded biomolecules, have mainly been explored in free solution. We report on the design, synthesis, and conformational behavior of photoresponsive foldamers bound in a phospholipid bilayer akin to a biological membrane phase. These molecules contain a chromophore, which can be switched between two configurations by different wavelengths of light, attached to a helical synthetic peptide that both promotes membrane insertion and communicates conformational change along its length. Light-induced structural changes in the chromophore are translated into global conformational changes, which are detected by monitoring the solid-state 19F nuclear magnetic resonance signals of a remote fluorine-containing residue located 1 to 2 nanometers away. The behavior of the foldamers in the membrane phase is similar to that of analogous compounds in organic solvents.

In the field of synthetic biology, substantial progress has been made in the use of light to control biological function by customization of membrane-bound proteins with artificial chromophores (1). In parallel, synthetic molecular photoswitches have been used to control chemical processes such as ligand binding and catalysis in isotropic solution (2, 3). Here, we report the design and synthesis of a fully synthetic photoresponsive helical molecule that can insert into a phospholipid bilayer. We show that light-induced switching between configurational isomers can be used to induce global conformational change that propagates over several nanometers in a synthetic molecule within a membrane environment. Membrane-bound artificial photoswitchable synthetic structures capable of translating photochemical information into extended conformational changes, in a manner reminiscent of the operation of natural photoswitchable proteins such as rhodopsin (4), could ultimately provide opportunities for controlling chemical processes within membrane-defined compartments.

A detailed understanding of how the phospholipid bilayer affects long-range conformational changes in membrane-bound molecules is impeded by the difficulty of directly observing conformational changes in the membrane phase and by a lack of examples of biomolecules that adopt well-defined structures both in the membrane phase and in solution (57). A simplified yet functional synthetic analog of the membrane-spanning domains of natural proteins, containing in-built spectroscopic handles that are diagnostic of conformation, would be a powerful tool. Dynamic conformational changes in a membrane-bound molecule could then be explored, free of the complexities of protein structure, and compared with analogous changes in isotropic solution.

Given the requirement for a synthetic structure with a tendency to embed into membranes and with well-understood conformational dynamics, we chose to explore the membrane insertion of foldamers [synthetic polymeric molecules with well-defined conformations (8)] built from oligomers of the achiral amino acid Aib (2-aminoisobutyric acid; Fig. 1A, shown in black). These Aib foldamers show a strong preference for helical conformations (9) and therefore have two principal conformational states, in which the helix adopts a global left- or right-handed screw sense. Furthermore, helical Aib-rich peptides have a known tendency to insert into phospholipid bilayers because they occur naturally in the form of membrane-disrupting fungal antibiotics known as peptaibols (10).

Fig. 1 Validation of the photoswitchable foldamer architecture 1 in solution.

(A) Switching the terminal azobenzene chromophore between E and Z configurations changes the population distribution between right- and left-handed conformations of foldamers 1a to 1d with general structure Azo-Val-Aib4-GlyNH2. (B) Portions of 1H NMR spectra showing the change in chemical shift separation Δδ between the signals arising from the glycinamide methylene protons of foldamer 1d in CD3OD upon irradiation with light of 365 nm. (C) Percentage change in population distribution for foldamers 1a to 1d, calculated from Δδ values. (D) The intramolecular hydrogen bonding network (shown in green) in the x-ray crystal structure of Azo-Aib-Ala-Aib4-GlyNH2 2.

A chiral amino acid residue was covalently linked to the N terminus of an (Aib)n foldamer and then N-acylated by an azobenzene motif (Fig. 1A, shown in red) to enable photochemical induction of conformational change (11). This motif manifests well-understood photochemical interconversion between E and Z configurations and thus offers a reversible light-driven means of initiating conformational reorganization from the terminus of the oligomer. The influence of azobenzene geometry on the relative population of conformational states of the (Aib)n helix was first explored in solution using foldamers 1 (Fig. 1A) that carry a C-terminal glycinamide as a solution-state 1H nuclear magnetic resonance (NMR)–compatible reporter of helical conformation (12). An unequal conformational population can result when the first turn of a helical Aib-containing foldamer incorporates a single chiral tertiary amino acid residue, and the magnitude of this bias depends on the detailed structure of the first (N-terminal) β-turn of the helix (12).

1H NMR spectra of valine-containing foldamers 1a to 1d were acquired in deuterated methanol (CD3OD) solution as their thermally equilibrated mixtures of E (major) and Z (minor) geometrical isomers (analogous behavior in phenylalanine-containing foldamers is reported in table S1). Methanol has a polarity similar to that of the interfacial region of the phospholipid bilayer and prevents aggregation of the foldamers in solution (13, 14). Because the left- and right-handed conformational states of an (Aib)n helix interconvert on a submillisecond time scale at ambient temperature, the 1H NMR spectrum that is observed results from a weighted average of both conformational states, reflecting their relative population. The diagnostic feature in the averaged spectra of foldamers 1 is the signal or signals due to the methylene protons of the C-terminal glycinamide residue. A single signal indicates equal populations of the two states; a pair of signals indicates unequally populated states. Furthermore, the magnitude of the chemical shift difference Δδ between these signals is proportional to the excess population of one screw-sense conformation over the other. A change in Δδ therefore indicates a change in population distribution across these two conformational states (12).

A comparison of the chemical shift differences Δδ in the glycinamide residue of the E and Z geometrical isomers of 1a to 1d (Fig. 1B) showed that the E isomers consistently exhibit greater Δδ values than the Z isomers. This indicates that the E isomer of the azobenzene induces a more powerful conformational preference KE than that of the Z isomer KZ, and induces a more unequal distribution of screw-sense populations (Fig. 1, A to C). The sensitivity of the conformational preference of 1 to the geometry of the azobenzene was tuned by varying the substituent in the meta position of the terminal ring (15, 16). Methoxy-substituted 1c and 1d showed the greatest differences in conformational populations between their E and Z isomers (Fig. 1C) and were cleanly photoswitched from E to Z upon irradiation at 365 nm and from Z to E upon irradiation at 455 nm (fig. S2). Both 1c and 1d exhibited particularly slow thermal relaxation from Z to E [t1/2 of 64 hours for 1d (fig. S6)], and the para-fluoro substituent of 1d additionally provided a local 19F NMR reporter of azobenzene geometry.

We assume that the conformational switching of 1 is driven by a difference in electronic properties between the E and Z azobenzene-2-carboxamides. The population distribution between screw-sense conformations across a series of related (Aib)n foldamers is sensitive to the basicity of the carbonyl group linked to the N-terminal residue that initiates the first β-turn of the helix (12). The x-ray crystal structure of an azobenzene-2-carboxamide capped foldamer 2 (Fig. 1D) reveals that the proximal azo nitrogen atom forms an intramolecular hydrogen bond in the solid state with the NH of this N-terminal residue. Given the persistence of the intramolecular hydrogen bond network of Aib foldamers even in polar, hydrogen-bonding solvents (17), we assume that this hydrogen bond is retained in solution. Upon photoisomerization from E to Z, the azobenzene loses planarity and the diazo group becomes more basic (18), likely strengthening this hydrogen bond and in turn altering the geometry of the hydrogen bonding—and hence the conformational preference—within the β-turn. This change in conformational preference is propagated through the helical chain, leading to a detectable shift in the relative populations of left- and right-handed helical conformations, reported by the 1H NMR signals of the C-terminal glycinamide reporter.

Having established that the conformational populations of the foldamers in solution could be perturbed by photochemical switching and detected by solution-state 1H NMR, we needed to devise a means of detecting conformational changes in related foldamers when embedded in a membrane. Solid-state NMR [ss-NMR (19)] is a powerful analytical technique that can determine structures in nondissolved systems with quasi-atomic resolution. It can be used to characterize membrane-bound structures (20, 21) and can yield detailed information on the conformation and dynamics of membrane-bound proteins and peptides (2224) and their artificial mimics (25). It has recently been used to investigate the structure and dynamics of the photoswitchable vision protein rhodopsin (26) and provides evidence that the solution-phase conformational preference of the Aib-rich peptaibol alamethicin is preserved when embedded in phospholipid bilayers (27). Magic angle spinning (MAS) is the most commonly used ss-NMR method (28), and in a lipid bilayer, MAS spin rates of ~10 kHz generate well-resolved 1H ss-NMR signals within an experimental time scale of 1 hour (29). In the H-, C-, and P-rich membrane environment, details of orientation, conformation, folding, and association can be obtained through incorporation of fluorine substituents into natural (30) and artificial molecules (31). Temperature-, drug-, and solvation-dependent conformational changes of labeled membrane-bound proteins have been detected by ss-NMR (32), but the method has rarely been used to quantify conformational ratios or to observe dynamic conformational changes in membrane-bound artificial molecules (33).

To identify the foldamer in the phospholipid bilayer, we incorporated into its structure a 2,2-difluoroAib [Fib (34)] residue (Fig. 2, shown in green). Located at the foldamer’s C terminus, several nanometers remote from the chromophore, Fib allows the use of 19F ss-NMR (29, 35) to detect changes in the conformational population distributions, because chemical shift differences (Δδ) between the 19F NMR signals of the two fluorine atoms of Fib report on the conformational preferences of helical foldamers in solution, in a similar way to the glycinamide probe of 1 (33). Initially, a series of foldamers 3a to 3f (Fig. 2) lacking the azobenzene chromophore was synthesized to validate the use of Fib as a conformational reporter in the membrane phase. To enhance membrane solubility, we ligated a triethylene glycol (TEG) tail to the C terminus of the foldamers, adjacent to the Fib residue. The spatial separation between the FibTEG reporter and the N-terminal chiral residue ensures that the chemical shift difference Δδ between the 19F NMR signals is insensitive to configurational changes in the chromophore, and thus reports only on induced conformational changes in the helical part of the foldamer. The 19F NMR spectra of foldamers 3a to 3f in trideuterated methanol (CD3OH) at 23°C showed pairs of signals with chemical shift differences Δδ that are proportional to the screw-sense population distributions reported for related compounds (12). The Δδ value provides a measure of population distribution between conformations only when exchange between screw senses is fast on the NMR time scale, so variable-temperature 19F NMR (fig. S7) was used to confirm that the conformations of 3a interconvert rapidly (i.e., at a rate of >6000 s–1) in both CD3OH (coalescence temperature Tc of 3a = 0°C) and deuteriochloroform (CDCl3, Tc of 3a = +7°C), a low–dielectric constant solvent that simulates the low polarity at the center of a phospholipid bilayer (36).

Fig. 2 Solution-phase and membrane-phase 19F NMR signals of fluorinated foldamers 3a to 3f along with their corresponding chemical shift separations Δδ.

aMv = l-α-methylvaline. *Achiral residue; †two consecutive aMv residues.

We then used ss-NMR to study the dynamic conformational behavior of the helical foldamers 3 in a membrane environment. Multilamellar vesicles [MLVs (37)] were prepared by codissolving 3 and the phospholipid 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) [peptide/DOPC weight ratio (w/w) of 1:19 to 1:9] in chloroform (38). After removal of solvent under reduced pressure, the residue was rehydrated, freeze-dried, and resuspended in phosphate buffer solution (pH 7.2). Centrifugation and removal of the aqueous supernatant layer afforded a viscous lipid phase, which was loaded into a MAS zirconia rotor (fig. S6). The 31P static NMR spectrum showed an anisotropic signal, whereas the 1H MAS ss-NMR spectrum showed resolved, sharp peaks from both the lipids and foldamer (fig. S7) indicative of a lipidic fluid lamellar phase into which the oligomer is embedded but remains freely mobile (3941).

19F MAS ss-NMR spectra of 3a to 3f embedded in DOPC membranes were acquired with a spinning rate of 10 kHz and were compared with solution-phase 19F NMR spectra of the same compounds 3a to 3f in CD3OH (Fig. 2). In every case, either one or two 19F signals centered around δ –234 ppm were observed in the ss-NMR spectrum, further confirming that the fluorinated foldamers 3a to 3f had partitioned into the lipid bilayer. The single 19F NMR signal of the achiral foldamer 3a in both solution-phase NMR and ss-NMR suggested that the rapid screw-sense inversion observed in solution persists in the membrane phase, allowing use of chemical shift differences Δδ in the 19F ss-NMR spectrum to measure the population distribution of helical conformations in the membrane. The increasing separation of the 19F NMR signals of the FibTEG probe in the series 3b to 3f in both solution and solid state confirms the increasingly strong chiral influence at the remote N terminus in both environments. Strongly helicogenic, Cα-tetrasubstituted residues (α-methylvaline, aMv), and in particular two consecutive aMv residues, induced the greatest bias in the conformational population in both environments. Although some subtle differences are evident, probably due to a weakening of conformational induction in a membrane environment, the similar conformational behavior of 3a to 3f both in isotropic solution and in the liquid crystal (42) environment of the DOPC bilayer suggests that the population of foldamer conformations in the membrane phase may be predicted by studying their population distribution in solution. 19F ss-NMR spectra were concentration-independent: There was no change in Δδ upon varying the loading of foldamer 3d from 3 to 15% w/w (fig. S12), nor upon adding a nonfluorinated chiral foldamer to a lipid bilayer already containing 3d (fig. S13), indicating that screw-sense preferences detected by 19F ss-NMR are not the result of intermolecular communication between helices. The single 19F ss-NMR signal of 3a confirms that the chirality of DOPC has no discernible influence on screw-sense preference.

Having shown that the constituent parts function successfully in isolation, we brought them together in the synthesis of a foldamer designed for photoswitching in the membrane phase. The most effective azobenzene motif 1d (Fig. 1) was ligated to the N terminus of an Aib-containing foldamer carrying the C-terminal FibTEG ss-NMR conformational reporter. The resulting oligomer 4 (Fig. 3) carries 19F labels at both the N and C termini, allowing parallel monitoring by 19F ss-NMR of both configurational E/Z photoswitching in the azobenzene unit and consequent global conformational population switching of the foldamer. (The similar behavior of a Phe-containing analog is described in fig. S17.)

Fig. 3 Conformational switching of 4 by irradiation in a DOPC phospholipid bilayer (5% w/w).

Molecular dimensions, but not necessarily orientation, of 4 are shown in proportion to the thickness of the bilayer, with the bilayer boundary indicated by gray spheres representing the phosphate head groups. Portions of the 19F ss-NMR spectra corresponding to the N-terminal fluoroarene substituent (δ –125 to –140 ppm, left) and the C-terminal FibTEG reporter (δ –230 to –235 ppm, right) are shown as well. (A) Dark-equilibrated sample (>99:1 E/Z azobenzene ratio). (B and C) The same sample after irradiation at 365 nm (14:86 E/Z ratio) (B) and after subsequent irradiation at 455 nm (69:31 E/Z ratio) (C).

Foldamer 4 was equilibrated in the dark to a >99:1 E/Z ratio and embedded into the DOPC membrane phase using the procedure described earlier, but minimizing exposure to light. The 19F ss-NMR spectrum (Fig. 3A) of the membrane-embedded foldamer showed a fluoroarene signal at δ –126 ppm, accompanied by two equally populated but unequally broadened signals (due to chemical shift anisotropy differences) in the region of δ –234 ppm. The single signal at δ –126 ppm confirmed that the azobenzene was still a single geometrical isomer after incorporation into the lipid bilayer. The appearance of two signals at δ –234 ppm, separated by ~1 ppm (Fig. 3A), indicates that the foldamer is in its “on” state, with one of the equilibrating screw-sense conformers of the helical structure preferentially populated. By analogy to the behavior of related Aib foldamers in solution, we infer that the preferred conformational state is left-handed and accounts for approximately 60 to 65% of the equilibrium population (43).

The suspension of MLVs containing foldamer (E)-4 was then transferred to a quartz cuvette and illuminated using an LED of wavelength 365 nm (fig. S11). After 5 min, the phospholipid mixture was spun out from the cuvette and directly loaded into the MAS rotor. The 19F ss-NMR spectrum (Fig. 3B) now revealed relative intensities of 14:86 for the signals corresponding to the E and Z azobenzenes, indicating substantial switching of the azobenzene geometry from E to Z within the phospholipid bilayer. Simultaneously, the pair of signals arising from the FibTEG reporter had collapsed into a broad singlet (Fig. 3B). This indicates that the photoinduced change in geometry at the N-terminal azobenzene had induced a global conformational switch into an “off” state, in which an approximately equal population of left- and right-handed screw-sense conformers is detected by the C-terminal reporter.

This Z-rich sample of 4 was transferred back into a quartz cuvette and illuminated a second time, with light of wavelength 455 nm, and again a ss-NMR sample was prepared using this now twice-illuminated, membrane-bound oligomer. This longer-wavelength illumination changed back the E/Z isomer population to a 69:31 ratio and restored the initial peak separation of the FibTEG reporter (Fig. 3C). This second irradiation at longer wavelength thus switched the foldamer back into its “on” state, reinstating its original conformational preference for a preferred screw sense and demonstrating the reversibility of the photoswitching process (see also fig. S16 for multiple, reversible photoswitching cycles).

No changes were evident in the FibTEG region of the 19F ss-NMR spectrum when a similar illumination procedure was applied to an azobenzene-free foldamer, nor when using an azobenzene-capped but achiral foldamer after irradiation at 365, 405, or 455 nm (figs. S14 and S15). The slow Z to E thermal relaxation exhibited by the fluorinated azobenzene-valine moiety in CD3OD solution (t1/2 = 64 hours for 1d; fig. S6) became only slightly faster in the phospholipid bilayer (t1/2 ≈ 44 hours for both 4 and 5; figs. S19 and S20). This proved particularly helpful for the purposes of this study, given the acquisition time required (0.5 to 1 hour) for each ss-NMR spectrum.

With the aim of transmitting conformational information over multi-nanometer distances comparable to the reach of the photoinduced conformational changes in rhodopsin that underpin the biochemistry of vision (44), we also studied foldamers with a much more extended helical structure. Foldamer 5 was synthesized, in which the azobenzene chromophore was separated from the FibTEG reporter by eight Aib residues, or about three turns of a 310 helix, corresponding to a distance of 2 nm (45), a distance commensurate the hydrophobic thickness of a DOPC bilayer (Fig. 4A) (40). (A related phenylalanine-containing foldamer is shown in fig. S18.) Preliminary 19F NMR studies in CDCl3 (fig. S8) indicated that the rate of exchange between conformational states is slower in these longer homologs [Tc ≈ +35°C, indicating a rate of <6000 s–1 at room temperature (46)]. After insertion into the DOPC bilayer but before irradiation, integration of the 19F ss-NMR signals at ~ δ –130 ppm indicated that 5 exists as a 92:8 mixture of E and Z geometrical isomers. Foldamer 5 showed two FibTEG NMR signals in the region of δ –234 ppm separated by 4.0 ppm (Fig. 4A). The membrane-embedded foldamers were irradiated for 7 min with light of wavelength 365 nm, which switched the azobenzene chromophore to an 18:82 E/Z geometrical ratio and simultaneously caused a marked change in the FibTEG reporter signal, increasing the separation of the peaks to 5.4 ppm (Fig. 4B). A second period of illumination at 455 nm switched back the population of azobenzene geometrical isomers to a 71:29 E/Z ratio and returned the chemical shift separation to its initial value (Fig. 4C). As in CDCl3 solution, the spectroscopic responses of 5 in the membrane phase are characteristic of slower exchange between screw-sense conformations (they are in intermediate exchange on the NMR time scale), which at this stage in the work prevents us from quantifying the relative populations of the conformational states. These findings demonstrate that light-induced configurational switching at one location in extended membrane-bound synthetic molecules induces local changes in structure that propagate through the oligomers to induce global changes in conformation, causing a change in conformational populations that is detected by spatially remote reporters.

Fig. 4 Conformational switching of 5 by irradiation in a DOPC phospholipid bilayer (5% w/w).

Molecular dimensions, but not necessarily orientation, of 5 are shown in proportion to the thickness of the bilayer, with the bilayer boundary indicated by gray spheres representing the phosphate head groups. Portions of the 19F ss-NMR spectra corresponding to the N-terminal fluoroarene substituent (δ –125 to –135 ppm, left) and the C-terminal FibTEG reporter (δ –230 to –240 ppm, right) are shown as well. (A) Dark-equilibrated sample (92:8 E/Z azobenzene ratio). (B and C) The same sample after irradiation at 365 nm (18:82 E/Z ratio) (B) and after subsequent irradiation at 455 nm (71:29 E/Z ratio) (C).

Our results show that the dynamic control over the conformational behavior of foldamers in solution may be replicated in a phospholipid bilayer. We envisage that replacement of the spectroscopic reporter with a catalytic or binding site in related switchable oligomers should allow photoswitchable changes in conformation leading to the stimulated release of a chemical messenger, or to catalytic formation of a secondary messenger to be up- or down-regulated (47). Incorporation of such switchable functional molecules into membranes would allow localized, photoinduced chemical changes to be translated into chemical responses in the lumen of artificial vesicles, or even the controlled release or uptake of a chemical signal in the cytosol of individual live cells. The work demonstrates that simplified synthetic molecules may be designed to display the essential functions of much more complex, evolved biomolecules.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S21

Tables S1 to S9

References (4859)

References and Notes

  1. Electron-donating groups para to the azo linkage were avoided, as they are known to promote fast thermal relaxation to the E isomer [see (16)].
  2. Acknowledgments: Supported by the European Research Council (Advanced Investigator Grant ROCOCO) and UK Engineering and Physical Sciences Research Council grants EP/K039547/1 and EP/N009134/1. We thank B. Gore for assistance with ss-NMR experiments, R. Adams and G. Morris for helpful discussions on solution and solid-state NMR data, and S. Koehler and L. Natrajan for valuable suggestions for preliminary irradiation studies. Detailed experimental procedures and all data supporting this publication are presented in the supplementary materials. Metrical parameters for the structure of 2 are available free of charge from the Cambridge Crystallographic Data Centre under accession number CCDC 1415763. Author contributions: M.D.P., S.J.W., and J.C. conceived the project and designed the experiments; M.D.P. synthesized the compounds with assistance from W.Z. and O.Q.; M.D.P. and M.L. devised and carried out the ss-NMR experiments; M.D.P., S.J.W., and J.C. analyzed the data and wrote the paper.
View Abstract

Navigate This Article