Molecular Machines For The Control Of Transmembrane Transport Authors: Toby G Johnson, Matthew J Langton Published: 2023 ABSTRACT AbstractNature embeds some of its molecular machinery, including ion pumps, within lipid bilayer membranes. This has inspired chemists to attempt to develop synthetic analogues to exploit membrane confinement and transmembrane potential gradients, much like their biological cousins. In this perspective, we outline the various strategies by which molecular machines—molecular systems in which a nanomechanical motion is exploited for function—have been designed to be incorporated within lipid membranes and utilized to mediate transmembrane ion transport. We survey molecular machines spanning both switches and motors, those that act as mobile carriers or that are anchored within the membrane, mechanically interlocked molecules, and examples that are activated in response to external stimuli. INTRODUCTION Molecular machines lie at the heart of almost all biological processes, operating at length scales where random thermal fluctuations dominates their motion. This ubiquity has inspired chemists to strive toward developing artificial analogues, mimicking the roles of their natural cousins, or identifying entirely artificial functions. The initial developments in this field of Sauvage, Stoddart, and Feringa were recognized by the 2016 Nobel Prize in Chemistry.1−3 While numerous sophisticated systems have been reported in solution, due to the random tumbling of molecules, in most cases, the controlled nanomechanical motions of molecular machines are effectively randomized. The spatial organization of molecular machines, such as on a surface, is therefore of particular importance in order to extract useful work from the assembly. Nature immobilizes some of its molecular machinery within lipid bilayer membranes. For example, the membrane-bound ion pump ATP-ase transduces energy from ATP hydrolysis to pump protons across the cellular membrane, in a process accompanied by unidirectional rotation of the protein (or, in the reverse mode, utilizing the energy stored in the ion gradient to catalyze the chemical synthesis of ATP).4 The sophistication and out-of-equilibrium functions of this biological molecular machine have so far remained out of reach of synthetic systems. For an overview of the state of the art in solution phase research, including pumps, motors and chemically fuelled systems, as well as surface bound molecular machines, the reader is directed to a number of recent reviews.5−8 Artificial molecular machines in membranes have remained comparatively rare, and initially were primarily composed of modified biological machines such as those derived from nanopores.9 This field was last reviewed by Watson and Cockcroft in 2015, and in the following eight years, there has been a step-change in developing entirely artificial membrane confined molecular machines. Practically speaking, by incorporating organic molecular machine components into hydrophobic membranes, chemists have at their disposal a versatile method of interfacing lipophilic supramolecular chemistry with aqueous biological environments (noting that the vast majority of functional organic/inorganic molecular machines are not water soluble, limiting their application in biological contexts). Because of this, membrane chemistry is arguable one of the most promising areas in which we can expect to see useful applications of molecular machines through its impact on biology and medicine. Given the rapid recent progress in this regard, in this perspective, we aim to take stock of where the field is at, to categorize the various new mechanisms of ion transport based on the nanomechanical motions of synthetic molecular machines, and provide a perspective on where we envisage the next challenges and developments will be, in terms of both fundamental science and practical applications. Ion Transport by Synthetic Supramolecular Systems: Mechanistic Diversity In nature, ion transport is mediated primarily by transmembrane protein channels or sophisticated biomolecular machine ion pumps and, to a lesser extent, by mobile carrier (also referred to as ionophores). A wide range of synthetic ion channels and mobile carriers have emerged,10,11 including those with stimuli-responsive behavior.12−14 Channels provide a pore through which ions may flow down their concentration gradient (Figure 1.i) while mobile carriers shuttle ions across the membrane, via consecutive binding, translocation, and release steps (Figure 1.ii). A range of different intermolecular interactions have been employed for the binding of ions to promote transport; we direct the reader to recent reviews on metal-organic-based transporters,15,16 hydrogen bonding systems17,18 and transporters utilizing sigma-hole interactions.19,20 For reviews concerning the methods developed for the study of ion transport, we direct the reader to the following.21−24 Mechanisms of transmembrane ion transport. Schematic of transmembrane ion channels (i), mobile carriers (ii), and anchored carriers (iii), which can be either unimolecular or relay-based systems. Recently, new abiotic mechanisms of transport that exploit the nanomechanical motions of molecular machines have emerged, such as relays, swings, and shuttles, vide infra. These novel mechanisms of transport typically rely on an anchoring group to provide some degree of immobilization of the transporter in the bilayer, while a mobile component capable of large amplitude molecular motion (e.g., macrocycle component of a rotaxane or carrier arm of a relay) facilitates transport (Figure 1.iii). We propose to define such ion transport mechanisms as “anchored carriers”, because they incorporate features of channels (anchored to the membrane and immobile) and carriers (requiring motion of a receptor between both membrane interfaces). The work of Gokel and co-workers in the 1990s provides early inspiration for anchored carrier systems, in which channels based on crown ethers connected by flexible linkers were anchored within the membrane.25,26 While these systems operated via a channel mechanism, the structural characteristics of these early transporters provide inspiration for anchored carrier systems. In this perspective, we adopt a broad definition of molecular machine, namely, a molecule or assembly of molecules in which nanomechanical motion is exploited for a function. Within this broad category, we can further subdivide between switches and motors: the former cannot do work, while the latter are able to.5 From a more molecular perspective, a molecular switch is a species capable of reversible and repeated interconversion between two states via some degree of nanomechanical molecular motion, which in the context of ion transport has commonly been utilized to switch activity on and off. Molecular motors, on the other hand, are able to do work by transducing energy from photons or a chemical fuel. These include, but are not limited to, molecular rotors, which are molecules that undergo unidirectional rotation under photoirradiation, and have been exploited to permeabilize membranes or enhance the activity of ion transporters. A key feature of molecular machines is the stimuli-responsive control over motion, with pH changes, temperature, ligand binding, redox chemistry, light, and other stimuli having been reported.27 We highlight these systems within this perspective, alongside interlocked ion transporters, which are themselves an important class of molecules within the rapidly expanding field of molecular machine development. TRANSMEMBRANE CHANNELS Transmembrane protein channels and pumps are the predominant means of regulating ion gradients in biology. Stimuli-responsive opening and closing of gated-channels controls the function of many biological processes within the cell, and engineering of photoresponsive biological machinery in the form of cellular membrane channels is the basis of the field of optogenetics.28 Many sophisticated examples of regulating protein channels using synthetic molecular photoswitches have been reported, but are beyond the scope of this perspective and have been reviewed elsewhere.28−30 We focus our discussion on wholly synthetic examples and direct the reader to the reviews cited for more information about these topics. Mechanically Interlocked Ion Channels In 2020, Leigh and co-workers demonstrated that the complex mechanically interlocked topology of a Star of David [2]catenane was essential to its transport activity.31 Two metalated interlocked structures, (FeII)5-coordinated pentafoil knot 1 and (FeII)6-coordinated Star of David [2]catenane 2, were shown to mediate anion transport when incorporated into large unilamellar vesicle (LUV) membranes (Figure 2). The Star of David channel 2 displayed more than 50-fold greater activity compared to knot 1, which was attributed to the larger cavity size of the Star of David. The Fe(II) cations in the complex were shown to be essential for activity—the demetalated [2]catenane was completely inactive in the 8-hydroxy-1,3,6-pyrenetrisulfonic acid (HPTS) assay. The cations act to both rigidify the channel structure and enhance anion binding. The hexameric cyclic helicate precursor to 2 was found to be inactive, demonstrating that the conformational constraints imposed by interlocking were crucial to the transport activity. While the interlocked topology was shown to be essential for activity, the structure undergoes no nanomechanical motion (and thus cannot be described as a molecular machine). However, given that mechanically interlocked molecules are the basis of a huge array of molecular machine type systems in solution, future developments of such knotted architectures may have the potential for dynamic molecular machine activity within membranes. Interlocked ion channels. A. Schematic of the transmembrane channel transport sequence. B. Pentafoil knot 1 and Star of David [2]catenane 2 chloride ion channels.31 Formation of a pseudorotaxane interlocked structure, in which a blocking ligand threads through an ion channel, has been demonstrated as an approach to reversibly gate the transport activity of channel 3 (Figure 3).32−34 The peptide-appended bis-resorcinarene 3 afforded regular square-like signals in black lipid membrane (BLM) single channel conductance experiments, indicative of transmembrane channel formation. Straight chain alkyl amines (octyl to octadecyl) inhibited the unimolecular channel cation transport activity, where longer alkyl chains demonstrated an improved ability to block channel 3. The blocking ligands could be removed from the pseudorotaxane type host–guest complex by Cu2+ cations which sequestered the amines and restored transport activity. A. Reversible ligand-gated formation of pseudorotaxane architectures controls ion channel activity. B. Peptide-appended bis-resorcinarene unimolecular channel (3) with reversible ligand-gated cation transport activity.33 Pelta and co-workers have also reported α-cyclodextrin nanotube channels,35−37 which were synthesized via Harada’s rotaxane templated synthesis method38 via threading of α-cyclodextrins onto a polymer axle. Employing the interlocked structure to favor the coupling of neighboring α-cyclodextrins gave threaded nanotube channels. These threaded channels showed diminished transport activity, while hydrolysis and dethreading yielded efficient ion channels. Molecular Switch-Based Transmembrane Channels The same principal of regulating transport through reversible blocking of a channel pore, as a pseudorotaxane or inclusion complex, has been demonstrated with azobenzene appended ion channels.39,40 Gin and co-workers reported a β-cyclodextrin-based ion channel that displayed switchable ion transport selectivity in response to the configuration of the appended azobenzene switch (Figure 4).40 When in the E-configuration, the azobenzene forms a self-inclusion complex E-4, partially blocking the pore and only allowing smaller sodium cations to be transported. In contrast, Z-isomer Z-4 does not adopt the threaded β-cyclodextrin conformation to block the pore, enabling the larger chloride anions to be transported. Characterization of the inclusion complex within the bilayer was challenging; however, the demonstration of a change in activity in response to conformational switching supports the likely structural blocking of the pore. A. Reversible photocontrol of channel pore size controls ion channel selectivity. B. Photoswitching of 4 via pseudorotaxane formation controls ion selectivity of this β-cyclodextrin based channel.40 An alternative strategy to control transport activity with molecular switches has been to control the molecular self-assembly of multicomponent channels within the bilayer. Reversible photocontrol of azobenzene43,44 and acylhydrazone41 photoswitches has been demonstrated for the regulation of cation transport in model membranes. The self-assembly of stacks of the acylhydrazone-linked crown ether triad 5 could be controlled in the bilayer to turn transport on and off with light (Figure 5).41 In the Z-configuration, intermolecular hydrogen bonding held the stacks together and resulted in efficient transport of cations through dibenzo-24-crown-8 channels. Irradiation with 320 nm light photoswitched the acylhydrazone to the E-configuration and resulted in disassembly of the channel with a concomitant loss of transport activity, while 365 nm photoirradiation regenerated the Z-configuration and reassembly restored channel activity. A. Photoswitchable control of a self-assembly channel. B. Reversible switching of conformation in response to membrane tension regulates ion channel activity. C. Acylhydrazone-linked crown ether triad 5 self-assembles into hydrogen-bonded stacks of Z-5, but cannot when in the E-configuration.41 D. The BPBP trimer 6 serves as a mechanosensitive K+ ion channel.42 Mechanical stimulation has been applied to the reversible conformational switching of a multipass transmembrane channel.42 Mechano-sensitive ion channels are common in nature; however, synthetic analogues are extremely rare.45 Kinbara and co-workers have reported a transmembrane trimer 6 consisting of repeating oligo(ethylene glycol) chains and 3,3′-dimethyl-5,5′-bis(phenylethynyl)-2,2′-bipyridine (BPBP) units assembled as a multipass channel in the lipid bilayer (Figure 5).42 The hydrophobic BPBP subunits span the hydrophobic portion of the membrane, adopting either an expanded or contracted conformation in the bilayer. With a contracted membrane, the BPBP units are forced to compact, forming a transient pore mediating K+ ion transport. However, with an expanding tension, the pore collapsed, resulting in loss of transport activity. The conformation of this channel could be controlled with membrane tension, with reversible control, thus acting as a molecular switch. Importantly, this conformational switching was the direct result of assembly within the bilayer, demonstrating the efficacy of the bilayer environment to organize molecules into conformations which display more complex functions than would be achievable in bulk solution, and as a means to transduce mechanical force into molecular level effects. β-barrel ion channels reported by Matile and co-workers46,47 have also been shown to display conformational switching within the lipid bilayer, to regulate the flow of ions through the pore.48−50 In these systems, dialkoxynaphthalene ligand binding to naphthalene diimide-derived rigid-rod scaffolds leads to a global conformational change and opening of the channel. This is triggered by intercalation of the electron-rich dialkoxynaphthalene ligands between the electron-deficient naphthalene diimides to form a π-stack architecture which stabilizes the channel. Molecular-Motor-Based Transmembrane Channels Very recently, molecular motors have been incorporated into the structure of ion channels, to regulate the transport activity with controlled unidirectional rotation of Feringa type light-driven rotary motors (Figure 6).51,52 Two such systems have been reported by Barboiu and Giuseppone (7), as well as Qu and Bao (8), which incorporated benzo-18-crown-6 cation receptor units to elicit potassium-cation-selective ion channels. Urea motifs were integrated into 7 to promote the self-assembly of the channel structure in the membrane. Transport activity experiments in HPTS-loaded egg yolk phosphatidylcholine (EYPC) LUVs (100 nm) found the channel to be cation-selective with the following relative rates of transport: Rb+ > K+ > Na+. BLM single channel conductance measurements provided unequivocal evidence of a channel transport mechanism. Under constant 365 nm irradiation, the transport activity was significantly enhanced in both LUV and BLM studies, where covalently coupling the motor to the channel was shown to be essential to the enhanced transport activity. This enhancement was postulated to be due to the local deformations in microphase and increased fluidity due to local heating through energy dissipation.53 Rotation of the molecular motor conjugated to ion channels enhances transport activity. Molecular motor conjugated ion channels reported by Barboiu and Giuseppone (7)51 and Qu and Bao (8),52 which display enhanced transport activity under constant irradiation that induces unidirectional rotation. Interestingly, studies with rotor 8, reported by Qu, Bao and co-workers, demonstrated the same mechanism of light-driven motor rotation enhanced transport activity. The channel was K+-selective and displayed enhanced activity with 365 nm irradiation. A calcein leakage assay demonstrated that even under UV irradiation, 8 did not cause appreciable membrane disruption or nonspecific leakage. Further studies of 8 in cellular membranes showed the same photodependent transport activity for K+, which induced cell apoptosis in HeLa cells demonstrating the potential of these systems for cancer therapy. This new mechanism of regulating ion transport through the continuous out-of-equilibrium rotation of a molecular motor represents a major step toward the sophisticated operation of biological machinery such as ATP synthase. These systems do not yet demonstrate active transport of ions against a concentration gradient, but the stimuli-responsive regulation of activity with molecular motor motion is a significant step forward in the complexity of membrane-bound artificial molecular machines. MOBILE CARRIERS A wide variety of synthetic mobile carriers have been reported, utilizing a range of intermolecular interactions to selectively transport a desired ion, with an increasing emphasis on incorporating stimuli-responsive behavior to exert control over their activity (see the Introduction for references to a range of reviews on this topic). Anion carriers operate via consecutive ion binding, membrane translocation, and ion release steps, and are typically characterized by anion binding-limited transport activity,54,55 although some rare examples of translocation-limited mobile carriers have been reported.56,57 For the development of stimuli-responsive mobile carriers, including those based on molecular machines, it is therefore important to consider the impact of changing receptor interactions on the ion binding step as well as carrier geometry and lipophilicity, in order to influence the translocation rate of the ion–mobile carrier complex. Recently, the interplay of binding and translocation kinetics have also been highlighted to be essential to the selectivity of mobile carriers.58 Mechanically Interlocked Mobile Carriers Mechanically interlocked structures, both naturally derived and synthetically obtained, are beginning to gain greater traction in biomedical applications.59−61 While numerous examples of synthetic mobile carriers have been reported, which employ a variety of intermolecular interactions to facilitate transmembrane transport, those with a mechanically interlocked architecture remain very rare. The first example of an interlocked mobile carrier was reported by Smithrud and co-workers, who utilized [2]rotaxane 9 to facilitate the transport of fluorophores across cellular membranes (Figure 7).62 Rotaxane-mediated transport of dianionic fluorescein and fluorescein–peptide conjugates across cellular membranes enabled visualization of the cargo localization within COS-7 cells by fluorescence microscopy. Interestingly, in these systems the shuttling motion of the [2]rotaxanes was suggested to be key to transport activity:63 control experiments in which the secondary amine on the [2]rotaxane axle was acetylated revealed inhibited shuttling motion and reduced binding affinity for fluorescein in aqueous mediate relative to the dynamic [2]rotaxane, while binding in CHCl3 was unaffected. The ability to adapt the conformation to maintain a high binding affinity for the molecular cargo was critical in these systems, as the polarity of the bilayer environment changes significantly during the transport process. To confirm 9 acted as a mobile carrier which passively diffused across the cellular membrane, experiments were conducted at reduced temperature and with ATP-depleted COS-7 cells.64 Under these conditions, the transport activity was only moderately reduced, confirming that endocytosis was not the major pathway to fluorescein internalization. The cyclophane [2]rotaxane 9 was subsequently investigated for the transport of a range of fluorescein–peptide conjugates of various polarity in COS-7 and ES-2 cells as well as in bulk transport U-tube experiments.64 Studies on related [2]rotaxane derivatives have shown that an elongated polyethylene glycol axle improved aqueous solubility and hence transport activity, while a carboxylic acid or guanidine-functionalized macrocycle tuned transport selectivity for cationic peptide over negative charged or apolar peptide cargos.65,66 The same group have reported benzo-crown ether-appended [2]rotaxanes which displayed anticancer activity as a result of Ca2+ cation transport.67,68 A rhodamine B-derivatized [2]rotaxane allowed visualization of the membrane localization of this transporter, which facilitated the intracellular delivery of the anticancer drug doxorubicin for enhanced cytotoxic activity.69 Interlocked molecular transporters via noncovalent recognition. Top: Smithrud’s cyclophane [2]rotaxane peptide transporter 9.62 Bottom: Schmitzer’s umbrella rotaxane (10a) and pseudorotaxane (10b) Cl– mobile carriers.70,71 Schmitzer and colleagues reported [2]rotaxane 10a which facilitated Cl– transport via a mobile carrier mechanism across a model lipid membrane.70 The rotaxane carrier was reported to have an EC50 value (concentration required to achieve half-maximal activity) of 1.56 mol % with resepct to lipid in EYPC vesicles, as determined using the chloride-sensitive fluorophore lucigenin encapsulated inside. Large amplitude conformational change resembling an umbrella motion was necessary for chloride transport in these systems as the structure translocated across the bilayer. A subsequent study found that self-assembly of pseudorotaxane 10b within the bilayer served as a ligand-gated means of regulating transport activity.71 The unthreaded axle was found to be more active than the pseudorotaxane complex 10b, where addition of the macrocycle component arrested transport. Another strategy for the transport of impermeable molecules into cells has been to covalently link the cargo as part of a rotaxane architecture,72−76 in which the interlocked structure improves membrane permeability and cellular uptake compared to the individual components. This strategy is, however, limited by the fact that interlocked structures must break down within the cell to release the cargo in an irreversible process. [2]Rotaxane 10a (Figure 7), in addition to its noncovalent Cl– transport capability, was also shown to be able to deliver macrocyclic cargo into EYPC vesicles via covalent attachment.70 In this work, the authors demonstrated the enzyme-mediated cleavage of amide bonds on the axle of 10a by the protease α-chymotrypsin encapsulated within the vesicle lumen to release the macrocycle cargo. The same strategy was demonstrated with [2]rotaxane Pt prodrug 11 in U2OS osteosarcoma cells (Figure 8), which showed enhanced cell permeability and reduced cytotoxicity compared to the parent noninterlocked metallodrug.77 Three interlocked prodrugs were developed including noncleavable 11a to serve as a control, which had little background cytotoxicity. Protease enzyme responsive 11b and photoresponsive derivative 11c enabled spatiotemporal control over the dethreading of the macrocycle, to reveal the active G-quadruplex DNA binder PtII-axle which was responsible for cytotoxicity. Interlocked molecular transporters using covalent cargo attachment. The [2]rotaxane architecture acts to improve cellular delivery of the cytotoxic PtII-axle, while enzyme- (11b) and photo- (11c) responsive stoppers enable spatiotemporal control of activity.77 Recently, Langton, Beer, and co-workers reported the first catenane ion carriers, in which halide-selective binding via halogen bonds translated into high, therapeutically relevant Cl– > OH–/NO3– selectivity in anion transport studies.78 The activity of the interlocked [2]catenanes was improved > 3-fold relative to the most active macrocyclic component, despite the increase in molecular size, and demonstrates the potential for mechanically interlocked transporters which show improved activity and selectivity relative to their topologically simple components. The interlocked systems described above have employed the mechanical bond to improve the analyte binding strength and membrane permeability, but these examples do not yet constitute true molecular machines. We anticipate that the future of interlocked mobile carriers likely will include transporters displaying enhanced selectivity resulting from the mechanical bond effect, activity dependent on the co-conformational motion of the interlocked components, and out-of-equilibrium ion pumps. Indeed, Stoddart and colleagues have already postulated that the incorporation of pseudorotaxane molecular pumps into lipid bilayers would enable the active transport of molecular cargo across membranes.79 Mobile Ion Carriers Derived from Molecular Switches Molecular switches have been incorporated into mobile carriers to enable stimuli-responsive transport activity. Typically, the molecular switching is utilized to control the proximity of two binding groups to improve the binding affinity and analyte encapsulation in the active state. Distinct from irreversible systems which allow the stimuli-responsive initiation or inhibition of transport activity, molecular switches facilitate fully reversible activation and deactivation of ion transport activity.14 While many different molecular switch architectures have been developed,80 the majority of photoswitchable mobile carriers reported to date have incorporated azobenzenes.81−84 Shinkai et al. initially established the application of molecular switches to control ion extraction between two liquid phases.85 Mobile carriers which transport cations across lipid bilayers have also been reported based on spiropyran,86 a bis-anthracene couple,87 and azobenzene.84 In 2014, Jeong and co-workers were the first to demonstrate that anion transport activity in lipid bilayer membranes could be regulated through the use of a molecular switch to modulate the geometry of the mobile carriers 12a–g.88 Their bis(thio)urea appended azobenzene transporter 12 could be photoswitched between a compact active Z-isomer and an open shape inactive E-isomer. In the Z-isomer, the receptor units are brought into close proximity to cooperatively bind a Cl– anion, with binding constants found to be an order of magnitude greater compared to those of the corresponding open E-isomer. These transporters were also investigated within FRT cells, where Cl– transport was visualized with the halide-responsive yellow fluorescent protein (YFP) expressed within the cells. The same activity trend in model membranes was seen across the cellular membrane, where the Z-isomer of the mobile carrier was more active than the E-isomer. However, reversible in situ switching of transport activity was not achieved in this system. Langton and co-workers subsequently utilized red-shifted tetra-ortho-chloroazobenzenes functionalized with squaramide anion binding groups 13a–i (Figure 9) to facilitate reversible visible light (red and blue) responsive control of transport activity in a fully reversible, in situ fashion.82,83 Hydrazone switches have also been incorporated into anion transporters, which rely on different stimuli to switch between each state.89,90 Photoirradiation switches the hydrazone to a closed off state, while acid isomerizes the switch back to the active form. Photoswitchable mobile carriers. Bis-(thio)urea azobenzene (12)88 and red-shifted bis-squaramide tetra-chloroazobenzene (13) mobile carriers.82 Beyond systems demonstrating stimuli-responsive switching between inactive and active states, those that also display high selectivity for a specific ion are highly desirable. pH-Regulated molecular switches, which change conformation upon protonation, have been incorporated within mobile carrier design91 to give highly selective transporters.92−94 Gale and co-workers reported a phenylthiosemicarbazone molecular switch 14 which facilitated H+/Cl– symport exclusively in acidic conditions and diminished transport at neutral pH (Figure 10).93 Under neutral conditions, the intramolecular H-bond locks the structure into an anti-conformation with reduced anion binding affinity. In acidic conditions, the structure is protonated and adopts the syn-conformation, where both thiourea NH hydrogen bond donors are able to participate in anion binding. This pH-responsive molecular switch regulated transport activity and ensured highly selective cotransport of H+ and Cl– ions, because only the syn-14 HCl complex was neutral and therefore membrane-permeable. More recently, Wezenberg and co-workers reported stiff-stilbene-based photoswitchable anion mobile carriers.95,96 Bis(thio)urea derivative 15a–g displayed photomodulated and selective Cl– transport in 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) vesicles.95 The Z-isomers of 15a–g were found to have significantly improved transport activities compared to the respective E-isomers, with carrier 15e displaying a 568-fold difference in activity between the isomers. The Z-isomers were highly selective for Cl–, functioning as electrogenic uniporters, and were able to photomodulate a membrane potential. Selective mobile carriers based on molecular switches. pH-Responsive phenyl thiosemicarbazone H+/Cl–-selective symporter 14(93) and stiff-stilbene photoresponsive Cl–-selective electrogenic transporter 15.95 Recently, the photoregulated transport of oligoarginine peptides across model and cellular membranes was reported.97 A sulfonatocalix[4]arene receptor unit served to bind the cationic peptides, while a pendant azobenzene molecular switch modulated the lipophilicity of the transporter, which regulated the translocation step in the transmembrane transport process. In the Z-configuration, the azobenzene was more polar, and therefore transport was inhibited. Irradiation with 500 nm light isomerized the azobenzene to the more hydrophobic E-configuration, which promoted peptide transport. ANCHORED ION CARRIERS In recent years, the scope of transporters has been broadened beyond the archetypal classification of either a mobile carrier or membrane-spanning channel.9,98 These new transporters can be described as “anchored ion carriers”: operating via molecular machine-inspired nanomechanical motion within the bilayer, but without the entire construct translocating through the membrane (as for mobile carriers). These are entirely abiotic mechanisms, distinct from their biological counterparts. Two clear classes of anchored carriers have emerged, namely unimolecular anchored transporters (where a single binding group facilitates the transport of the analyte through the entirety of the membrane) and relay transporters (where at least two binding groups are required to mediate the passing of the analyte between leaflets of the bilayer via an exchange step to enable transport). Unimolecular Anchored Transporters A number of systems that act as unimolecular anchored transporters have been reported in recent years under a plethora of different names, which fall under this description. Zeng and co-workers have been pioneers in this area,98 developing transport system 16 (Figure 11), a so-called “molecular ion fisher.” The cholesterol unit was used to anchor the system, while the flexible linker to the benzo-18-crown-6 K+ binding motif facilitated transport by this unconventional mechanism.99 Transporters with different anchoring architectures have been reported, based on a single cholesterol lock,99 as well as systems with a membrane-spanning cross-beam containing two cholesterol units 17.100,102 A computational study found that H-bonding between the cholesterol and the phosphate of a lipid restricted lateral translocation, rationalizing the efficiency of this anchor group.103 Interestingly, after prolonged investigations, no single-channel current traces were obtained for the moncholesterol 16 or a bis-cholesterol anchored system in BLM conductance experiments, indicating that ions were not transported through well-defined channels.99,102 Ill-defined and transient single-channel current traces were observed for the molecular swing transporter 17 in BLM experiments.100 This highlights that there is a broad spectrum of possible mechanisms of ion transport, varying from the ill-defined pathway of mobile carriers to the well-defined path of ion translocation through an ion channel with the unimolecular anchored carrier mechanism positioned somewhere between the two. Unimolecular anchored carriers. A. Schematic of the unimolecular anchored carrier transport mechanism. B. Structures of monocholesterol anchored 16,99 bis-cholesterol anchored 17,100 and tris- and bis-indolyl anchored 18a–b.101 Exchanging the substituent on a tetrasubstituted anchored carrier provided insight into the effect of increasing the number of carrier arms.101 Derivative 18a with three indolyl anchoring groups and one benzo-18-crown-6 receptor displayed so-called “molecular fishing” activity, while the anchored carrier 18b (two anchors, two receptor) displayed >30 times greater activity. The carrier arms therefore operated cooperatively, but whether both receptors bound the ion simultaneously to transport via a molecular fishing mechanism, or whether an exchange process occurred within the membrane to transport via a relay mechanism was difficult to confirm. Since no single channel behavior was observed in BLM experiments, transport activity was not due to a channel mechanism and thus can be attributed to an anchored carrier mechanism. In a similar approach, the same group reported an octa-substituted C60 buckyball with crown ether receptors which operated via an anchored carrier mechanism.106 The selectivity for the cation transported could be tuned based on the size of the crown ether macrocycle. In these anchored carrier systems, the flexible linker between the receptor and anchoring group enabled transport by allowing mobility of the pendant receptor. In contrast, crown ethers appended directly to a membrane-spanning anchor with short rigid linkers have previously been shown to follow a strict channel-type mechanism of transport.107 Employing a mechanical bond to link the carrier unit to the anchoring group was demonstrated by Bao, Zhu, and co-workers in 2018, who reported a [2]rotaxane shuttle 19 able to transport K+ via a shuttle-type mechanism (Figure 12).104 The activity of 19 was found to be linearly dependent on the concentration of the transporter in the bilayer, indicative of a single molecule being responsible for transport. Regular square-like signals with long opening times and ohmic I–V profiles in BLM experiments suggested a stable channel may form within the bilayer. However, a unimolecular dependence of transport activity and the reduced transport activity with decreasing shuttling rate implied that the mobility of the carrier arm along the axle of the rotaxane was responsible for the transport activity.108 Subsequent interlocked unimolecular anchored carriers have been reported, which have unique axle structures and an amide-linked crown ether macrocycle carrier arm.109,110 Interestingly, the rigid para-tetraphenylene axle derivative (EC50 = 1.75 mol %)109 displayed enhanced activity relative to the flexible [2]rotaxane 19 (EC50 = 3 mol %).104 Increasing the number of carrier arms in the interlocked transporter was shown to improve the transport activity, with EC50 values of 4.3 mol % and 2.0 mol % for the [2]rotaxane and [3]rotaxane, respectively.110 As with the unimolecular anchored carrier system 18b, determination of whether both receptors bind the ion simultaneously to transport via a molecular fishing mechanism, or whether an exchange process occurred within the membrane to transport via a relay mechanism, was not confirmed. Unimolecular anchored carriers based on a [2]rotaxane architecture. A. Structure of [2]rotaxane cation transporter 19.104 B. Schematic of [2]rotaxane ion transporters where shuttling motion facilitates ion transport. C. Azobenzene isomerization in the axle of [2]rotaxane 20 allows photoswitching of transport activity between the off state (Z-20) and on state (E-20).105 A later related system combined an interlocked architecture with a molecular switch to facilitate photogated shuttling motion for the regulation of transport activity.105 An azobenzene was incorporated within the membrane-spanning axle, such that shuttling of the macrocycle, and hence mobility of the anchored carrier arm, could be restricted through photoswitching the azobenzene moiety to Z-20 to diminish cation transport rates. The system was shown to be fully reversible, with in situ light-gated transmembrane activity measured in LUVs and BLM experiments. Single-channel conductance experiments showed regular square-like signals after irradiation with 450 nm visible light forming E-20, while the Z-20 displayed no current signal after long detection times. This unique molecular-machine-inspired anchored carrier transporter combines the properties of molecular switches with the dynamic motion of interlocked components to control transport across lipid membranes. The structure of the axle is key to its function as an anchor, orienting the molecule as a bolaamphiphile to span the membrane, and enabling the shuttling motion of the [2]rotaxane to transport ions across the membrane Relay Transporters Relay transporters operate via the exchange of ions between transporters located in opposite leaflets of a lipid bilayer membrane. The relay transport mechanism was first reported by Smith and co-workers in 2008, who demonstrated that a urea-appended phospholipid derivative 21 facilitated ion transport by passing anions between transporters immobilized in opposite leaflets of the bilayer (Figure 13).111 They established the key principles of this unique transport mechanism, demonstrating that relay transporters must be present in both leaflets of the bilayer with no activity observed when loaded into only one leaflet of the membrane. The electron-withdrawing nitro substituent of 21a significantly improved transport activity relative to the t-butyl group (21b), as a result of the increased binding strength for anions. An extensive survey of different phospholipids of varying alkyl chain lengths was also conducted to explore the effect of bilayer hydrophobic thicknesses on transport rates. Activity was found to incrementally decrease with increasing phospholipid alkyl chain length from 14 to 18 carbons and was completely suppressed for lipid alkyl chain lengths of 20 to 24 carbons. Notably, this trend contrasts with that typically observed for ion channels which display a bell-shaped relationship, where optimal activity is observed for channel structures that matched the thickness of the bilayer,113 and unlike that of mobile carriers, which is either invariant with membrane thickness when ion binding is rate-limiting,55 or linearly decreased with increasing hydrophobic thickness when translocation of the carrier–ion complex is rate-limiting.114,115 Relay transporters. A. Smith’s urea relay transporter 21,111 Langton’s bis-iodotriazole hydrogen (22HB), and halogen bonding (22XB) relay transporters.112 B. Schematic of the relay transport mechanism. Langton and co-workers subsequently reported a relay transporter with a bis-iodotriazole halogen bonding (XB) anion receptor.112 The relay transporter 22XB showed enhanced transport activity with an EC50 value of 0.18 mol % with respect to lipid in POPC LUVs, compared to ∼ 2 mol % in POPC/cholesterol (7:3) for Smith’s urea derivative 21a.111 In the presence of the protonophore FCCP, the activity of the halogen-bonding relay transporter 22XB was significantly enhanced (EC50 = 0.036 mol %), as a result of appreciable Cl– > H+/OH– selectivity. The prototriazole derivative 22HB exhibited efficient transport activity (EC50 = 0.18 mol %, in POPC LUVs) but no transport enhancement with FCCP, due to the hydrogen bond donors showing no Cl– > H+/OH– selectivity. Phospholipid-based ion transporters have the added benefit of preferential solubility in physiological environments to improve the likelihood of pharmaceutical success. Relay 22XB demonstrates that otherwise highly hydrophobic halogen bonding anion receptors can be incorporated into amphiphilic ion transporters, while maintaining the intrinsic Cl– > OH– selectivity which is vital for the application of synthetic transporters as therapeutics. The same group also demonstrated that the incorporation of a molecular switch into the carrier arm of a relay transporter could be employed to reversibly control the length of the relay arm, in turn regulating the transport activity (Figure 14).116 A visible light photoswitchable tetra-fluoroazobenzene117 was incorporated into a phospholipid-based relay transporter, which could be efficiently switched between the E- and Z-isomers using green and blue light within the membrane, with the same photostationary state distribution as was observed in solution. This system demonstrated the critical role of the “telescopic” arm length for controlling the relay transport process: the gap between the thiourea anion-binding sites of transporters on opposite sides of the membrane must be sufficiently small to mediate transport for E-23, but also sufficiently large for Z-23 to suppress transport in the off state. Notably, this mechanism requires the unprecedented control of multiple molecular machine-like components positioned on opposite sides of a membrane, which work together in a cooperative manner to facilitate transport. Photoswitchable relay transporter. Schematic representation of photoregulated relay transport activity, and Langton’s tetra-fluoroazobenzene thiourea relay transporter 23.116 CONTROLLED RELEASE OF CARGO FROM LIPID BOUND COMPARTMENTS In the previous sections, we have described how molecular machine components and nanomechanical motions of anchored carriers have been used to control transport of cargo across lipid bilayer membranes through selective binding interactions. Molecular machines have also been shown to control the disruption of membranes to facilitate the nonspecific release of cargo from lipid-bound compartments.118 Indeed, extensive research has been conducted into the development of compounds which disrupt cellular membranes for antiviral and antimicrobial activity,119,120 as well as for targeted drug delivery.121−125 This perspective will focus on the molecular switches and motors that exploit controlled nanomechanical motion to disrupt both cellular and model membranes. Membrane Disruption with Molecular Switches The field of photolipids is well established126,127 for the light-controlled release of cargo from the lumen128 or the membrane of vesicles.122 In general, molecular switches in the linear E-configuration (E-25) align well with other lipids in the bilayer, while photoswitching to the bent Z-configuration (Z-25) disrupts lipid packing and causes the release of the cargo (Figure 15).129 In recent years, to improve the biocompatibility of these vesicle drug delivery systems (which are typically activated with UV light with poor biocompatibility and tissue penetration), there has been a drive to red-shift the photoswitching process for operation within the therapeutic window. Photoswitches for lipid membrane disruption. Top: structure of DSPC lipid 24, UV-responsive azoPC 25,125 red-light-responsive red-azoPC 26,130 and ionizing-radiation-responsive gadolinium complex 27.133 Bottom: schematic of phototriggered release of cargo from a lipid compartment. Trauner and co-workers introduced the tetra-chloroazobenzene phospholipid 26 which has been shown to mediate the photocontrolled release of doxorubicin from distearoylphosphatidylcholine (DSPC)/cholesterol vesicles with 630 nm/465 nm light in vivo in zebrafish embryos.125,130 Encapsulating lanthanide upconversion nanoparticles within the lumen of vesicles is another method that has been shown to enable long-wavelength photoswitching of azobenzene photolipids.131,132 These nanoparticles absorb 980 nm near-infrared photons and re-emit shorter wavelengths of light, including UV (∼360 nm) and blue (∼450 nm), which was shown to isomerize the azobenzene, releasing doxorubicin. A gadolinium–azobenzene conjugate 27 has recently been reported, which can be switched with ionizing radiation.133 The lanthanide initially absorbs high-energy photons and mediated switching through energy transfer to the azobenzene core. This radio-switch amphiphile demonstrated ionizing radiation could induce molecular rearrangement of the azobenzene to disrupt cellular membranes and trigger a cytotoxic effect. Molecular switches responsive to pH have also been demonstrated to control the release of cargo from vesicles.134 The Leblond group have developed a bis(methoxyphenyl)-pyridine-based pH-responsive synthetic lipid 28 which displayed large amplitude conformational switching (Figure 16).135 In the neutral state, the molecule adopted a closed conformation where the long alkyl chains of syn-28 were in a parallel arrangement, which allowed efficient incorporation and packing within a DSPC membrane. Protonation of the pyridine nitrogen at low pH induced molecular switching to the open anti-28 conformation, which significantly disrupted the bilayer of LUVs and triggered the release of encapsulated cargo. The pH-triggered release of sulforhodamine B from 28b-loaded DSPC LUVs was shown to be actuated by the acidic interior of HeLa cells, which affects the rapid endosomal release of cargo into live cells. A subsequent report demonstrated that the same strategy could be used for the delivery of small interfering RNA (siRNA) into HeLa cells.136 This technology has promise for treating a range of diseases, due to the ability of siRNA to regulate gene expression.137 pH-Responsive bis(methoxyphenyl)-pyridine switch for triggered membranes disruptions. The various R groups of 28a–c control the pH at which conformational switching from syn-28 to anti-28 occurs.135 Membrane Release Triggered by Molecular Motors Amphiphilic molecular motors have been shown to reversibly reorganize between different self-assembled structures in aqueous media, controlled by the isomerization of the motor.138−140 However, only a few examples have been interfaced with membranes to perform a function.141 Tour and co-workers first demonstrated that membrane-anchored molecular motors could disrupt cellular membranes for the targeted killing of cancer cells.142 Their results suggested that the unidirectional rotation of the molecular motors could be exploited to form transient pores in both model and cellular membranes. Under 355 nm irradiation, 29a increased the permeability of cellular membranes, leading to necrosis of human prostate adenocarcinoma cell (PC-3). The demethylated derivative 29b, incapable of unidirectional rotation, displayed a reduced cytotoxic effect. However, a recent study from Pohl, Antonenko, and co-workers suggested that the mechanism of action of motor 29a is the generation of singlet oxygen by the photoexcited state of the motor, which caused oxidative damage of unsaturated lipids and destabilized the membrane.143 This was explored by conducting experiments in giant unilamellar vesicles—those prepared with dipalmitoylphosphatidylcholine (DOPC, an unsaturated lipid) were susceptible to motor-induced membrane disruption after irradiation; however, those prepared with 1,2-diphytanyl-sn-glycero-3-phosphocholine (DPhPC, a saturated lipid) were inert to molecular motor action. Tour and co-workers have also explored hemithioindigo-based visible-light-activated molecular motors for their antibacterial activity (Figure 17).144 The sulfoxide hemithioindigo motor 30 displayed some antibacterial activity upon 455 nm irradiation, which promoted 360° unidirectional rotation. However, the related sulfide hemithioindigo switch 31 that is incapable of complete rotation was found to be more potent, again pointing to the generation of reactive oxygen species (ROS) playing a role in the mechanism of action of these biologically active molecular machines. Molecular motors used to disrupt cellular membranes. A. Schematic of motor-induced membrane rupture through generation of ROS or via the proposed nanomechanical motion drilling holes in membranes. B. Tour’s fast rotating motor (29a), switch (29b).142 The hemithioindigo motor (30) and switch (31) disrupt bacterial membranes through oxidative damage.144 Feringa and co-workers have investigated the use of molecular motors for the controlled release of cargo from lipid-bound compartments, with the view to develop targeted drug delivery vehicles.145 Incorporating motors within amphiphilic molecules has been shown to generate micelle and vesicle self-assembled structures, to allow the investigation of their dynamic behavior in biologically relevant media.146,147 Hydrophobic motors have also been incorporated within lipid membranes, where 365 nm light irradiation led to the release of molecular cargo from DOPC (unsaturated lipid) vesicles.148 Tour and co-workers have also demonstrated that molecular motors can be used in combination with therapeutic agents to improve their cellular internalization.149 The action of 29a increased the sensitivity of Klebsiella pneumoniae to Meropenem, thereby improving the biological effect of the antibiotic. While the mechanism of action of cellular targeting molecular motors may not have been fully elucidated,150 their apparent efficacy within a biological environment is certainly promising and worthy of detailed investigation, and provides inspiration for the development of further biocompatible molecular rotors in the future. STRATEGIES AND CHALLENGES FOR MEMBRANE TARGETING AND TOWARD APPLICATIONS IN CELLS In cellular biology, membrane labeling for imaging is typically achieved using lipophilic fluorophores with pendant aliphatic chains, which spontaneously intercalate into the lipid bilayer.151 Uptake of hydrophobic molecular machines into artificial membranes remains the main method to ensure membrane confinement of these systems and is often achieved with bioinspired anchoring groups, such as cholesterol, farnesyl, or functionalized lipids themselves. Alternatively, in many cases a sufficiently hydrophobic core scaffold of a molecular machine derived from entirely artificial structures—a Feringa type rotor for instance—is more than sufficient for facilitating membrane uptake.51 Influencing the orientation and position of a molecular machine within the lipid bilayer, for example, where it sits in relation to the interface, has been typically achieved by careful control over the positioning of hydrophilic and hydrophobic motifs within the system. For instance, Kinbara’s mechanosensitive multipass channel (Figure 5D) uses a combination of a hydrophobic core to embed within the membrane interior, and polyethylene glycol chains which reside at the membrane interface, in order to control the required membrane-spanning orientation.42 Similarly, Zeng has made use of hydroxylated cholesterol motifs to preferentially locate a molecular ion fisher at the membrane interface (Figure 11).99 Controlling the orientation of molecular transporter machines in membranes has typically been achieved through designing amphiphilic systems, such that when added to the membranes of preformed vesicles, directional insertion of the hydrophobic portion of the machine occurs preferentially. This is demonstrated in Smith’s and Langton’s relay transporter systems (Figure 13 and 14), in which the amphiphilic lipid anchor ensures directional uptake into the membrane.111,112,116 More generally, there are a range of challenges to be overcome to interface molecular machines with living cells. Indeed, some of the recent progress in the field of ion transport systems has arguably been motivated by the potential longer-term translation of these synthetic molecules as therapeutics, yet the application of synthetic ion transporters has rarely progressed beyond model systems within the lab. Further work is required to understand how results in model vesicle or BLM membranes translate into comparable behavior in cells, animal models, and in vivo. Deliverability is likely to be a key challenge to address, and cell membrane uptake of ionophores in general is currently underexplored. Indeed, a study by Sheppard and co-workers investigated a wide range of potent hydrogen-bonding anionophores in vesicles and Fischer rat thyroid (FRT) cells,152 and found that many of the synthetic transporters had poor deliverability to the membrane. They also demonstrated that cellular uptake could be enhanced by adding transporters in the presence of lipid to form water-soluble aggregates. Liposome-based delivery systems have also shown some promise in enhancing uptake into giant unilamellar vesicles,153 and this approach is certainly worthy of investigation for the delivery of molecular machines to cells. In general, a thorough investigation into whether the current strategies for facilitating uptake into model membranes, as discussed above, translate to cells is now required. Such studies will also need to explore the time residence of such machines within the bilayer, and methods to slow down endocytic elimination.154 To date there are few studies in which molecular machine ion transporters, including switchable transporters,88,97 interlocked carriers,62 or molecular rotors52,142 have been studied in cellular systems, and thus there is little current understanding of how such systems may be usefully employed in this context. Jeong and co-workers investigated photoswitchable anion mobile carriers 12 (Figure 9) in FRT cells.88 These investigations revealed that the para-substituted transporters 12f and 12g displayed no activity as either the E- or Z-isomers (unlike in POPC vesicles), while in contrast, the analogous meta-substituted derivatives displayed transport activity in the compact Z-isomer and not in the extended E-isomer in FRT cells, in line with observation in LUVs. The lack of activity for the para-substituted transporters was speculated to be due to protein binding inhibiting activity but could also plausibly arise from low deliverability to the cellular membrane and highlights the challenges in translating experiments in model systems through to cells. The disruption of cellular homeostasis through perturbation of ion concentration gradients by synthetic transporters has been shown to induce cell death,155 and the measurement of cell viability has been used to provide an indirect measure of transport activity. Such experiments have been conducted with the anchored carrier 17 and a benzo-18-crown-6 analogue in human primary glioblastoma cancer cells.100 The efficacy of the anchored carrier transport system in inducing cell death suggests that the abiotic anchored carrier transport mechanism can operate in cellular membranes. This is the only example to date of an anchored carrier having been tested in cellular membranes but showcases the potential of abiotic mechanisms of molecular machine-like transporters for facilitating transport in cellular systems. CONCLUSIONS AND OUTLOOK Nature confines its protein molecular machinery within lipid bilayer membranes to orient and exploit compartmentalization and transmembrane ion gradients. Numerous different approaches to incorporating molecular machines, in the broadest sense, within lipid bilayers to control ion transport have been developed. Many of these are derived from molecular machines first explored in the solution phase, such as switches, rotors, and rotaxanes. The lipid bilayer environment provides a means of confining molecular machines in an organized manner, with a high local concentration promoting intermolecular interactions and hence the potential for cooperative function. As such, molecules which in bulk solution may not display any function, such as relay transporters, can be immobilized within the bilayer and work cooperatively to control transmembrane transport. The biophysical properties of lipid bilayer membranes, such as membrane tension—a property that is evidently absent in the solution phase—have been demonstrated to regulate the operation of membrane-bound molecular machines and provide a further level of control to these systems. More generally, embedding molecular machines within the hydrophobic interior of the bilayer provides a useful method for interfacing organic molecules with aqueous, biologically relevant environments. Indeed, employing molecular machines for biomedical applications already appears to be a fruitful area of research, with fundamental studies in living cells demonstrating the benefit of their abiotic mechanisms of action. Beyond controlling transmembrane transport, molecular machines in lipid bilayers may also serve to mediate other functions, including signal transduction,156−159 catalysis,160−162 and sensing applications.163,164 Advancements in these related fields are likely to serve as inspiration for the development of molecular-machine-based transmembrane transporters. Indeed, the control of membrane incorporation, orientation, flip-flop, and residence time are common challenges within all of these areas of study. For example, Matile’s membrane tension responsive fluorescent flipper probes, which change conformation and hence fluorescent output within the membrane,165−167 provide strategies to internalization and targeting of specific cellular membranes which could in future be adapted to molecular transport machines. How best to interface synthetic molecular machines with membranes, with control over membrane targeting, orientation, and retention, is one of the most key current challenges to be addressed for the translation of molecular machines into biological environments. The future directions for this field of research will surely also include the development of out-of-equilibrium function, active transport, and chemically fuelled membrane-bound molecular machines. The complexity of biological membrane-bound machines, such as ATP synthase, will continue to serve as a benchmark for inspiring chemists to develop synthetic systems of increased complexity and sophistication. So far, no synthetic systems have been developed in which nanomechanical motion is driven by a concentration gradient across the membrane. Nor have any molecular machines been reported which facilitate active transport of an analyte against its concentration gradient, driven by a chemical fuel,168,169 in a manner reminiscent of the mode of action of ATP-ases. Indeed, active transport across lipid membranes using synthetic transporters in general remains extremely challenging, with only two examples using a synthetic photoredox carrier170 or channel48 system reported. Finally, interfacing synthetic molecular machines with biological components to access novel biohybrid systems171 is likely to be the focus of research to come, and is set to yield exciting opportunities for new biomedical applications. Acknowledgments T.G.J. thanks the Royal Society and Exeter College, Oxford, for funding. M.J.L. is a Royal Society University Research Fellow. Author Contributions The manuscript was written through contributions of both authors. The authors declare no competing financial interest. References