Fliposomes: stimuli-triggered conformational flip of novel amphiphiles causes an instant cargo release from liposomes

Trans-2-aminocyclohexanol-based flipids

The first amphiphiles with the pH-triggerable conformational switch were derivatives of trans-2-aminocyclohexanol 1 that performed an acid-induced ring flip (36–41), spreading their lipophilic tails like peacocks (this conformational change can be dubbed a ‘peacock effect’; Figure 1). Previously, we used the trans-2-aminocyclohexanol moiety to construct the conformationally controlled crown ethers and podands (29, 30, 47, 48).

Figure 1

Protonation-induced conformational flip causes spreading and efficient shortening of the lipophilic tails (a ‘peacock effect’) along with a change of the charge, effective size, and shape of the polar head resulting in a quick perturbation of the lipid bilayer, lipid phase separation, and fast content leakage from the fliposomes (30, 36–41).

The driving force of this dramatic acid-triggered transition is a strong, protonation-generated intramolecular hydrogen bond of HO···H-N⁺ type and an electrostatic/dipole-dipole attraction stabilizing a conformer with the gauche form of O-C-C-N fragment (1BH⁺ in Figure 1). This impulse results in a conformational flip of the cycle that moves the ester groups COOR at the other end of the molecule away from each other into axial positions. The relocation of substituents changes their intra- and intermolecular interactions, for example, their ability to form complexes with cations or to pack into lipid bilayers (depending on the nature of substituents).

We examined the chair-chair flip of the cyclohexane ring in flipids 1 (Figure 1) in solution by the proton NMR, ¹H NMR (36–41). The vicinal coupling constants ³J_HH between several protons attached to the cyclohexane moiety and their chemical shifts are strongly conformation dependent, which allows an assignment of a predominant conformation and an estimation of the position of equilibrium. To characterize the acid-induced swing of the conformational equilibrium, the changes of ¹H NMR spectra were monitored during titration of the diluted d₄-methanolic solutions of 1 with d-trifluoroacetic acid (Figure 2). The observation of a single set of well-resolved multiplets in the course of acidification attested to high rates of both the acid-base and the conformational equilibria on the NMR time scale. During the incremental addition of acid, the signal parameters changed significantly, indicating a strong protonation-induced shift of the conformational equilibrium from A to BD⁺ (∼100% in excess acid). This shift did not occur gradually over the whole course of titration, but happened only within a narrow pH range, the value of which depended on the basicity of the amino group. Using the change of ΔG_B-A values for the conformational equilibrium in methanol solution upon addition of acid, we estimated the power of this conformational pH trigger to be ≥11–12 kJ/mol. Remarkably, the intramolecular hydrogen bond HO···H-N⁺ wins a competition with hydrogen bonding to solvent molecules. Similar results were obtained for aqueous solutions of the model trans-2-aminocyclohexanols equipped with ethyl groups instead of long hydrocarbon chains and therefore slightly soluble in D₂O (41).

Figure 2

Change of the signal width (W=ΣJ_HH) for the proton H4 in ¹H NMR due to the conformational switch of the flipids 1a (NRR′=morpholinyl, n=m=1) and 1b (NRR′=CH₃OCH₂CH₂NH, n=m=1) in CD₃OD solution caused by titration with d-TFA (■) and then by backward titration of the resulting acidic solution with DBU (□) (39).

PEGylated liposomes that encapsulated a fluorescent dye (ANTS) and a quencher (DPX) were prepared by the freeze-thawing method. Starting with any particular lipid composition, this method yielded liposomes of reproducible colloidal properties and thus allowed studies on the relationship between the lipid composition of the liposomes and their acid-triggered release of contents (36–41). Noteworthy, after storage at 4°C for >6 months, the optimized liposome formulations (1/POPC/PEG-ceramide, 50:45:5 mol%) did not show noticeable changes in hydrodynamic diameter or polydispersity index, and kept a near-zero ζ-potential. We measured the leakage of ANTS/DPX after injecting a small aliquot of the fliposome preparation into buffer solutions with pH varying from 9.5 to 2.8 as exemplified for the case of two flipids in Figure 3.

Figure 3

pH dependence of the ANTS/DPX leakage from 1/POPC/PEG-ceramide fliposomes made with 50 mol% of flipid 1a, or 1b, or their 1:1 blend (1a+1b).

The black curve (▴) was obtained for the 1:1 mixture of fliposomes 1a and 1b. The dashed line is an average of the curves 1a and 1b (39).

As the initial horizontal part of the titration curves shows, fliposome leakage does not occur in basic (for 1a and 1b) or neutral (for 1a) medium. However, the increase of acidity brings about a dramatic jump of leakage between pH 8.5 and 6.5 in the case of fliposomes containing 50 mol% 1b, and between pH 6.0 and 4.0 in the case of fliposomes with 50 mol% 1a. The relative position of these transitional areas meets the expectation of a higher basicity for the secondary amine 1b as compared with the tertiary amine 1a (in protic solvents). The resulting diagrams are very similar to the NMR titration curves for the conformational flip of amphiphiles 1a and 1b (Figure 2). We consider this similarity as an evidence for the intrinsic dependence of the pH-induced fliposome leakage on the pH-triggered conformational flip of the amphiphiles 1.

Further evidence for the predominant role of conformational flip in the liposome leakage was obtained from the comparison of diastereomeric amphiphiles 1a and 2 (Scheme 1) (36, 37). Compound 2 differs from flipid 1a only in the configuration of one lipid tail substituent; however, for this reason, it is unable to noticeably change its conformation after protonation. Thus, the results for the control 2/POPC/PEG-ceramide liposomes allowed observation of all possible effects of the protonation of morpholine group on the liposome permeability (change of headgroup charge and radius, change of hydrogen bondings, etc.), except for the effects caused by the change of conformation. Therefore, the much larger and faster acid-induced leakage of 1/POPC/PEG-ceramide liposomes than that of 2/POPC/PEG-ceramide liposomes could be attributed to loosening of the liposome membrane caused by the pH-driven conformational change of the lipid tails in 1a. As expected, control liposomes containing neither 1a nor 2 were not responsive to lowered pH (36, 37).

Scheme 1

Absence of protonation-induced conformational flip in amphiphile 2 (36, 37).

FFEM allowed us to gain more insight into the mechanism of triggered release from fliposomes. This powerful technique not only can image structures of lipid colloids at nanometer resolution but also takes snapshots of lipid phase transformation by rapid freezing. The unloaded 1/POPC/PEG-ceramide (50:45:5) fliposome formulation was studied both at pH 7.4 and after exposure to pH 5.5 (Figure 4) (37). At both pH values, the samples contained vesicles, although a large part of them had dissipated in the more acidic medium. The diameters of the fliposomes measured on the electron micrographs ranged from 20 to 230 nm, with an average diameter of about 100 nm. At pH 7.4, the fliposome formulation appeared to be composed mostly of unilamellar vesicles (Figure 4A). Upon decrease of pH from 7.4 to 5.5, the remaining fliposomes underwent drastic morphological transformations (Figure 4B,C), including division, fusion, and budding. It appears that a large number of buds pinched off the fliposomes as freely suspended small spherical particles. Defects of lipid packing that resembled long and deep cracks were also observed on some of the dividing fliposome membranes.

Figure 4

FFEM of the 1/POPC/PEG-ceramide fliposome formulation at pH 7.4 (A) and 5 min after adjusting the pH to 5.5 with diluted acetic acid (B, C).

Examples of division, buds, and stripes are shown with white, black, and yellow arrows respectively. The bars represent 100 nm (37).

On the basis of ¹H NMR studies of the pH-induced conformational flip of compounds 1, the fluorometric and ultraviolet studies of the pH-induced leakage of fliposomes, and the FFEM images of the morphological changes of the 1/POPC/PEG-ceramide fliposomes, we proposed a novel mechanism of the pH-triggered content release from fliposomes (37). The protonation generates a strong intramolecular hydrogen bond between the amine and the neighboring hydroxy group, which flips the chair conformation of the cyclohexane ring and mechanically switches its two remote ester groups from equatorial to axial positions (Figure 1). Thereby, the spatial separation of the two lipid tails of 1 is increased, especially at their proximal end (peacock effect). At the distal end, the conformational freedom and the hydrophobic interactions may allow the chains to partially remain packed or to re-pack in the lipid bilayer, although with a decreased packable length of the hydrophobic moiety. Besides this conformational shortening and widening, as flipid 1 is protonated, its head group assumes a positive charge and increases its hydrodynamic size as a result of higher hydration. This leads to additional electrostatic and steric repulsion. All these changes are very fast, and they directly perturb the lipid bilayer and could induce phase separation of the fliposome membrane into thinner and thicker domains that are rich in 1 or in longer POPC molecules, respectively. The encapsulated contents of the fliposomes would then quickly leak through defects between the domains. Some domains of monolayers would subsequently bud from the fliposome membranes as small micelles, as observed in FFEM (Figure 4) (37).

Peacock effect: heads or tails?

A simple synthetic scheme used for preparation of flipids 1 allows substantial flexibility in the design of these molecular devices (30, 36–41). Their parameters can be conveniently modified by structural variation of all substituents, especially the hydrocarbon tails and the amino group in the relatively polar heads of the molecules.

Because the basicity of amines is determined by the nature of groups attached to the nitrogen atom, the flipid pH sensitivity can be tuned by appropriate modification of the amino group structure. This may allow tailoring of fliposomes for the triggered release at certain pH values. Moreover, by mixing such liposome preparations, each with different pH sensitivity, one could potentially obtain a system for simultaneous administration, but independent delivery, of different drugs that would be released each at a specific pH value. Furthermore, the reversibility of conformational flip upon addition of a base (Figure 2) suggests that fliposomes prepared in acidic medium could be used for targeted delivery to the more basic places, for instance, to the small intestine (49). We explored some of these possibilities (39–41).

We synthesized a series of flipids 1 with a variety of ‘heads’ and ‘tails’, and studied the pH dependence of the ANTS/DPX leakage from the corresponding 1/POPC/PEG-ceramide fliposomes (36–41). The measured pH ranges for the leakage varied from pH 9 to pH 4 depending on the nature of the amino group (as shown in Figure 3 for flipids 1a and 1b with different basicity). Thus, the pH sensitivity of a fliposome preparation can be indeed custom-tailored. Very conveniently, the leakage range for each flipid can be accurately predicted from the results of ¹H NMR titration (Figure 2).

Intriguing results were obtained when we used a 1:1 blend of flipids 1a and 1b for preparation of (1+2)/POPC/PEG-ceramide fliposomes (39, 40). Because the pH ranges for leakage of the fliposomes based on just one of these triggers practically do not overlap (Figure 3), one could expect the mixture of both flipids to produce an additive effect. That would extend the leakage over a pH range from 8.5 through 4.0 (dashed line in Figure 3 representing an average of the curves for 1a and 1b). Instead, in our experiments, the release was complete by pH 5.5. The first (more basic) part of the experimental curve, which can be credited to the conformational flip of 1b, practically coincides with the expectation (the dashed line). However, the second part of the curve, which was supposed to depend on switching of 1a, goes up and achieves the complete leakage much ‘earlier’ on the acidity scale than expected.

Very similar results have been obtained when we mixed two separate fliposome formulations, one based on 1a (50 mol% of lipid composition) and the second based on 1b (also 50 mol%). We supposed that fliposomes in this mixture of formulations were to respond to the change of pH independently – first those containing the more basic 1b and then those containing 1a. However, in this case, the second kind of fliposomes released its content also at less acidic conditions than expected, and the content release was complete by pH 5.0 (Figure 3).

Thus, when the flipids are blended before the preparation of fliposomes, or when the ready fliposome formulations are mixed, those flipids do not respond to acidification independently. There appears to be some complicated interaction between them. A plausible explanation of the observed effects may be as follows. When the more basic molecules 1b become protonated, the fliposomes that contain them are partially or completely destroyed. The ions 1bH⁺ thus released into the bulk aqueous media may play a role of a detergent, which permeates other fliposomes and releases their content. In other words, the acid-induced release of fliposome cargo may sometimes have an autocatalytic character. To verify this hypothesis, we performed the standard experiment by injection of the ANTS/DPX fliposome preparation based on 1a into buffer solutions with pH 7.5 and 5.5. Then, we added the dispersion of 1b in identical buffer. The molar amounts of 1a and 1b were equal. The addition of 1b produced no change when the medium had pH 7.5. However, the fluorescence released by the fliposome preparation of 1a in the pH 5.5 buffer increased upon injection of 1b. This observation seems to confirm the ability of protonated compound 1b to act as a detergent.

As the structure of lipid tails is critically important for the stability of lipid bilayers, the ability to release the liposome content could be tuned by appropriate design of these hydrophobic parts. We prepared and characterized the fliposomes containing new trans-2-morpholinocyclohexanol-based flipids equipped with either longer hydrocarbon chains, bent or branched hydrocarbon chains, and compared them with fliposomes based on flipid 1a (38, 40, 41). The pH-driven conformational flip was studied by ¹H NMR titration. The ¹H NMR and conformational parameters were practically identical between these flipids in all the studied conditions. Thus, the elongation, bending, and branching of the hydrocarbon chains did not produce any noticeable conformational effect in the polar heads of these structures and did not alter their pH sensitivity. (This observation suggests inter alia that in future studies, simpler model compounds with short alkyl tails can be used for a preliminary estimation of conformational properties of the designed flipids.) The pH dependence of the ANTS/DPX leakage from the fliposomes made of these flipids was similar to the NMR titration curve, which confirmed the intrinsic dependence of the pH-induced liposome leakage on the pH-triggered conformational flip.

However, the elongation and branching of the hydrocarbon chains reduced the content release from fliposomes (38, 40, 41). This observation may be considered as an additional confirmation to the proposed mechanism of pH sensitivity, wherein bilayer destabilization starts with an acid-triggered conformational flip of the trans-2-aminocyclohexanol moiety (the polar head group) that increases the spatial separation of the two lipid tails, which results in a set of membrane perturbations that cause the leakage (Figure 1). The tails should be separated the most in the vicinity of the polar head and may partially fold back at the distal end to re-pack in the lipid bilayer. However, the packable length of the tails would decrease (Figure 1), thus inducing the phase separation of the lipid bilayer and liposome leakage. Therefore, the effect of the conformational flip in the polar head on the packing of the tails should relatively decrease as the flipids carry longer tails. On the other hand, membrane perturbation by the conformational flip is generally possible because of a certain degree of order in the hydrophobic part of the lipid bilayer. When some additional disorder (fluidity) is introduced from the very beginning by the asymmetrically branched groups, the effect of the conformational flip on the membrane integrity becomes less significant. These results suggest that a better performance of the flipids may be achieved by making their hydrocarbon chains shorter and/or by introducing additional rigid fragments into their lipid tails. These hypotheses are currently under exploration.

Piperidinol-based flipids

We recently suggested the trans-3,4-bis(acyloxy)-piperidine structure 3 (Scheme 2) as a new platform for the pH-triggered conformational switches (42, 50, 51). The driving force in these devices is the same as in flipids 1: a strong, protonation-generated intramolecular hydrogen bond of O···H-N⁺ type and an electrostatic/dipole-dipole attraction stabilizing a conformer with the gauche form of the O-C-C-N fragment (3AH⁺ in Scheme 2).

Scheme 2

Protonation-induced conformational flip, shortening of the lipid tails, and change of the charge, effective size, and shape of the polar head cause a quick perturbation of the lipid bilayer, lipid phase separation, and fast content leakage from the fliposomes (42).

Applying the approaches and methods developed for the cyclohexane derivatives 1, we studied the conformational behavior of the piperidine-based flipids 3 and their properties as the lipidic components of fliposomes. The ¹H NMR data showed that flipids 3 changed conformation when acidity increased, and this flip occurred between apparent pD 5.5 and 3.5 in d₄-methanol solution. The fliposomes comprising flipid 3, POPC and PEG-ceramide, were tested using the ANTS/DPX fluorescent assay and demonstrated a leakage that started at pH 5.6 and became substantial at pH 4.0. The release results achieved thus far with the flipids 3 are below the efficiency of the fliposomes containing trans-2-aminocyclohexanol-based flipids 1. Perhaps, the perturbation of the lipid bilayer is stronger in the case of flipids 1 because the latter have more substituents, which drastically changed their position during the acid-triggered conformational flip (Figure 1).

Targeted drug and gene delivery

To confirm that fliposomes are capable of releasing encapsulated drugs in response to lowered pH, we constructed the 1a/POPC/PEG-ceramide (50:45:5) and the 1a/POPC/PEG-DPPE fliposomes (50:45:5) containing the anionic anticancer drug MTX (37). Both fliposome formulations were subjected to equilibrium microdialysis as a more versatile method to quantify drug release regardless of its fluorescence. The percentage of MTX release was determined by measuring the UV absorbance (λ=306 nm) of the free drug in the solution outside the dialysis bag after the microdialysis had reached equilibrium. None of the fliposome formulations released a significant amount of MTX after dialysis at pH 7.4 for 4 h. Upon dialysis at pH 5.5, both fliposome formulations containing flipid 1a released most of the encapsulated MTX. The concentration of MTX generated in the buffer chamber by far exceeded the typical levels of MTX in patients’ plasma and serum. For comparison, liposomes containing the diastereomeric amphiphile 2 released much less MTX after microdialysis at pH 5.5 (37). Such observations confirm that pH-triggered release of contents from fliposomes (ANTS, MTX, etc.) results mainly from the conformational change of the lipid tails, which takes place in 1a but not in its diastereomer 2.

Finally, to test whether fliposomes can serve as viable drug delivery systems, selected liposome formulations encapsulating the anticancer drug MTX were applied to HeLa cells followed by an assessment of the cell viability using the MTS assay (37). HeLa cells treated with 1 μm (final concentration) of MTX solution retained 65.0±4.3% viability, consistent with prior reports on the anticancer activity of the free drug. In comparison, HeLa cells treated with MTX-loaded fliposomes built of 1a/POPC/PEG-ceramide (50:45:5) showed a significantly lower viability (39.4±4.3%), indicating the superior anticancer cytotoxicity of such liposomal formulations of MTX compared with the free drug at the same dosage. The HeLa cells treated with MTX-loaded liposomes containing the diastereomeric analogue 2 instead of 1a retained most of the viability, thus highlighting the importance of the conformational flip in enhancing the cytotoxicity of the payload MTX. The cell culture media were buffered at ∼pH 7.4; therefore, discharge of the 1a/POPC/PEG-ceramide/MTX fliposomes most likely took place in the acidic endosomal compartment of the HeLa cells after their cellular uptake. Because MTX is too hydrophilic to passively diffuse across biomembranes, it is likely that the fliposomes not only released MTX in response to the lowered pH in the endosome but also facilitated the diffusion of the released MTX from the endosome to the cytosol. One possible mechanism of such facilitated diffusion could be a destabilization of the endosomal membranes by the micelles generated from the destruction of fliposomes.

It is known that pH sensitivity improves the efficiency of a number of gene delivery systems, including viral vectors, polyplexes, and lipoplexes (10, 52). Inspired by the drug delivery experiments, we started exploring the possibility to use flipids 1 as helper lipids in cationic lipoplexes for gene transfection. The preliminary tests using luciferase assay produced very promising results (41, 53), and this project is currently in progress.