Reconstitution of ATP-dependent lipid transporters: gaining insight into molecular characteristics, regulation, and mechanisms

Abstract Lipid transporters play a crucial role in supporting essential cellular processes such as organelle assembly, vesicular trafficking, and lipid homeostasis by driving lipid transport across membranes. Cryo-electron microscopy has recently resolved the structures of several ATP-dependent lipid transporters, but functional characterization remains a major challenge. Although studies of detergent-purified proteins have advanced our understanding of these transporters, in vitro evidence for lipid transport is still limited to a few ATP-dependent lipid transporters. Reconstitution into model membranes, such as liposomes, is a suitable approach to study lipid transporters in vitro and to investigate their key molecular features. In this review, we discuss the current approaches for reconstituting ATP-driven lipid transporters into large liposomes and common techniques used to study lipid transport in proteoliposomes. We also highlight the existing knowledge on the regulatory mechanisms that modulate the activity of lipid transporters, and finally, we address the limitations of the current approaches and future perspectives in this field.


Introduction
Cellular membranes are vital structures that maintain cell integrity and perform critical functions such as organelle compartmentalization, signaling, and membrane trafficking. These complex structures are composed of a hydrophobic matrix consisting of a double layer of lipids [1,2]. Embedded or attached to the membrane are proteins that dynamically interact with lipids to ensure proper membrane integrity and functionality [3]. While lipids can move between the two leaflets of the membrane, the rate of transbilayer lipid movement varies depending on the physical properties of the lipid and the environment. Lipids with small headgroups such as diacylglycerol, ceramide, and cholesterol can move from one leaflet to the other in artificial membranes in seconds or minutes. In contrast, lipids with polar headgroups such as phosphatidylcholine (PC) and sphingomyelin, and glycolipids with bulky hydrophilic carbohydrate moieties move more slowly, with half-times ranging from hours to days, depending on the size and charge of their headgroup [4,5]. Overall, these rates are insufficient to support vital cellular processes such as membrane assembly, control of transbilayer lipid asymmetry, and vesicular trafficking. Therefore, cellular membranes are equipped with lipid transporters that facilitate the movement of lipids across cellular membranes [6].
There are three types of lipid transporters: scramblases, flippases, and floppases ( Figure 1). Scramblases allow bidirectional transport and work by moving lipids down their concentration gradient at an impressive rate of over 10 4 lipids per second [7], thereby rapidly randomizing the distribution of lipids across the bilayer. They can be constitutively active or regulated by physiological stimuli, such as increase in intracellular Ca 2+ or proteolytic cleavage [8][9][10]. In contrast, lipid flippases and floppases are transmembrane proteins that use the energy of ATP hydrolysis to transport specific lipids unidirectionally against their concentration gradient to the cytoplasmic or exoplasmic/luminal leaflet of cellular membranes, respectively, at a rate of approximately 10-100 lipids per second. The interplay between members of these two ATP-dependent primary transport systems is thought to be responsible for the maintenance of membrane lipid asymmetry. Membrane asymmetry is critical for cell viability and is involved in key cellular processes such as signaling events to initiate blood coagulation [11], phagocytes' recognition of cells undergoing apoptosis and subsequent clearance by macrophages [12], and host-virus interactions where it might be important to ensure viral infection [13].
Recent advances in structural biology and computational biology have made it possible to capture the structures of several lipid transporters in multiple functional states (reviewed in [14][15][16][17]). With the increasing number of resolved structures of lipid transporters and the development of molecular dynamics simulations [18-20], we now have a more comprehensive understanding of how these transporters may move lipids. At the same time, advances in the field of mass spectroscopy and quantitative lipidomics have paved the way for the study of intact membrane mechanism. Examples include the 'outward-only' mechanism with outward-occluded and outward-open conformations, proposed for lipid-linked oligosaccharide flopping by PglK, which does not invoke an inward-facing cavity to interact with the substrate [35]. This bacterial ABC transporter catalyzes the translocation of lipid-linked oligosaccharides, which are essential for the bacterial protein glycosylation machinery [36]. Similarly, it has been proposed that ABCA1 does not form an inward-facing transmembrane cavity, which appears to be required for the alternating access model [37]. This mammalian transporter effluxes cholesterol and phospholipids to lipid free apolipoprotein A-I, when the latter binds to the membrane bilayer and generates nascent high-density lipoprotein [38,39]. In addition, based on simulations, a recent study proposes that ABCA1 extracts phospholipids from the outer leaflet of a model plasma membrane into the outward-facing cavity. The lipid then diffuses towards an elongated hydrophobic tunnel and is finally extruded out of the transporter [40]. In this case, the transporter would not promote transbilayer movement, but instead the release of the lipid out of the membrane. To understand the precise mechanisms of lipid translocation by the various transporters, further high-resolution structures with bound lipid substrates in combination with biochemical assays are required.

Studying lipid transporter activities: from cell-based assays toward functional reconstitution
Although ATP-driven lipid transporters have been extensively studied, much of our current understanding has come from cell-based studies. The first discovery of ATP-dependent lipid flippase activity dates back to the 1980s, based on pioneering work using spin-labeled lipid probes [41]. Subsequently, lipid uptake assays using exogenously applied fluorescent-labeled lipids were developed [42][43][44] and have since been widely used to study ATP-driven lipid transporters in various organisms, including mammalian cells, plants, parasites, and yeast. In recent years, heterologous expression in Saccharomyces cerevisiae has become an increasingly popular tool for the functional analysis of candidate lipid flippases. This is due to the ease of handling and the availability of numerous deletion mutants [45][46][47][48].
Cell-based assays to study lipid transport are typically based on short-chained nitrobenzoxadiazol (NBD)-labeled lipids (Figure 2, top panel). The flippase-mediated transport of these probes is usually monitored by extracting the residual fraction of analogues not transported across the membrane with bovine serum albumin (BSA). Since BSA extracts all analogues from the exoplasmic monolayer of the plasma membrane, the inaccessible fraction reflects analogues that have been redistributed across the plasma membrane into cells [49,50]. Alternatively, NBD-labeled lipids on the exoplamic monolayer can be selectively destroyed with the water-soluble quencher dithionite, followed by quantification of the intracellular fluorescence [56,135]. To study floppase-mediated transport across the plasma membrane, cells can be incubated with lipid precursors. Under these conditions, for example, short-chain NBD-phosphatidic acid (NBD-PA) is partially converted into NBD-diacylglycerol, which rapidly crosses the plasma membrane and becomes available for the intracellular synthesis of NBD-PC and NBD-phosphatidylethanolamine (NBD-PE) [51][52][53]. The continuous incubation of the cells in the presence of BSA allows then monitoring of the transport of the newly synthesized analogues to the cell surface by lipid analysis.
Such cell-based studies conducted on P4-ATPases have contributed significantly to our understanding of their lipid specificity (Figure 2, upper panel). Although originally classified as aminophospholipid flippases, individual members of this protein family exhibit distinct substrate preferences and can transport a diverse range of lipid substrates. In general, P4-ATPases can be categorized into three groups: those that preferentially transport PS and to a lesser extent PE, those that preferentially transport PC and PE, and those with a wide range of lipid substrates, including sphingolipids, lysophospholipids, and synthetic alkyl phospholipids [54]. Similarly, cell-based studies have helped to elucidate the lipid substrate specificities of some ABC transporters. For example, studies using short-chain lipids have revealed that human ABCB4 exhibits specificity for PC, whereas the closely related multidrug transporter ABCB1 translocates a variety of short-chain lipids [52,53,55] including platelet-activating factor, also known as acetyl-glyceryl-ether-phosphorylcholine [56]. Studies using (radiolabeled) cholesterol and phospholipids have demonstrated the role of ABCA1 in mediating the efflux of cholesterol and phospholipids such as PC [57][58][59][60].
A major challenge in cell-based studies is that cells typically express multiple lipid transporters, some of which have overlapping transport functions. In addition, the expression pattern of these transporters as well as the amount of lipid substrate transported can vary depending on the cell type and the specific expression approach used. This variability may explain, for example, the conflicting substrate specificities reported for the human P4-ATPase ATP8B1. Although initially characterized as a translocase for aminophospholipids [61][62][63] and potentially for cardiolipin [64,65], other studies found that ATP8B1 translocates PC rather than PS [66] or in some cases, might be inactive [67] upon overexpression in cell lines with low endogenous phospholipid flippase activity. Furthermore, cell-based assays typically monitor the lipid internalization at the plasma membrane, which is influenced by endocytic and exocytic membrane trafficking events. It is therefore important to either minimize these processes, e.g., by lowering the temperature, or to quantify them by an independent measurement before drawing conclusions from the experimental data.
Advances in expression and purification techniques have allowed the biochemical characterization of an increasing number of lipid transporters at the solubilized stage and after reconstitution into proteoliposomes. This approach provides a well-defined system that allows investigating defined conditions outside the complex cellular environment. Parameters, such as lipids, salt concentration, presence of substrates, and cofactors, can be tested and then tailored to meet the requirements of the protein under study. Liposomal reconstitution offers in addition a two-compartment system that permits tracking of the transport of substrates. First, the transporters are usually overexpressed in a suitable expression host and then solubilized from the membrane using a mild detergent, followed by purification and concentration. The choice of detergent is primarily based on empirical testing, and therefore, detergent screenings are a common practice for ATP-dependent lipid transporters [68]. The purified transporters are then reconstituted into large liposomes, also referred to as large unilamellar vesicles, with diameters of 100-200 nm. However, reconstitution remains a challenging task due to difficulties in purifying membrane proteins to sufficient yields, the 'trial and error' approach required to successfully reconstitute active transporters, and the potential need for co-reconstitution of accessory proteins [69,70].
Several methods have been developed for the rapid and effective reconstitution of membrane proteins into lipid membranes. The most commonly used method for the preparation of proteoliposomes is based on detergent-mediated reconstitution [71][72][73]. In this approach, the solubilized membrane protein is supplemented with an excess of either preformed large unilamellar vesicles or a lipid suspension, together with detergent, resulting in a mixture of lipid-protein-detergent and lipid-detergent micelles. The detergent is then removed, resulting in the progressive formation of a vesicle bilayer, thereby embedding the protein (Figure 2, lower panel). ATP-dependent lipid transporters are typically reconstituted using detergents, with octyl glucoside and Triton X-100 being the most frequently used (Table 1). Of note, the use of C 12 E 9 for reconstitution of a P4-ATPase resulted in vesicles that were too leaky for lipid transport studies [74]. The method used to remove the detergent depends on the physicochemical properties of the detergent employed [73]. To reconstitute the ATP-dependent transporters summarized in this review (Table 1), dialysis followed by gel filtration was used to remove octyl glucoside, whereas hydrophobic beads (Bio-beads) were used to remove Triton X-100 and polyoxyethylene detergents. After proteoliposome formation, a centrifugation step is sometimes added to concentrate the proteoliposomes [32,75,76] and/or to remove unincorporated protein from the proteoliposomes [18,77]. Membrane proteins reconstituted in liposomal model membrane systems have either the cytosolic side (inside-out orientation) or the extracellular side (outside-out orientation) exposed. ATP-dependent transporters appear to prefer an inside-out orientation in the vesicle, due to bulky domains and the curvature of the vesicle [78] but the exact protein orientation is hardly predictable, which limits the quantitative and functional analysis of vesicle preparations [79,80].

Assaying of lipid transporter activities in proteoliposomes
Various assays are available for measuring ATP-dependent lipid transporter activities in proteoliposomes ( Figure 3 and Table 1). A common method is to use labeled lipid reporter molecules carrying a fluorescent group (Figure 1, lower panel) or, less commonly, a spin-labeled group. The transbilayer transport of these probes can be monitored over time by chemical modification of the (non-)translocated lipid probes with membrane-impermeable reagents such as dithionite for e.g. NBD-labeled lipids and ascorbate for spin-labeled lipids. Another approach to assess lipid transport based on fluorescent-labeled lipids is the collisional quenching assay ( Figure 3A). Collisional quenching utilizes membrane-impermeable contact quenchers such as iodide or cobalt that generate a nonradiative transition to the ground state of the fluorescent group upon contact [81,82].
Isotopic modification of lipids has also been utilized to assess lipid transport in liposomal systems. For example, the flippase activity of the bacterial ABC transporter PglK was tested using a truncated form of its natural substrate (GlcGalNAc 5 Bac-PP-undecaprenyl). The method is based on the use of a soluble glucosyltransferase acting in the presence of an excess of nucleotide-bound radiolabeled sugars to convert the truncated substrate upon its appearance in the outer leaflet of the liposome. The change in radioactivity over time provides a quantitative measure of the rate at which PglK flips the truncated substrate ( [35]; Table 1).
Recently, a novel approach has been employed to study the energetics of lipid transport by the ABC transporter MsbA [76]. This transporter accepts as substrate PE carrying a biotin moiety covalently linked to the headgroup. This allows measuring the amount of accessible biotin-labeled lipid in proteoliposomes from the fluorescence emission of fluorescence-tagged avidin, initially bound to a quencher, which becomes displaced when avidin binds to the biotin moiety.
When studying charged lipid substrates, such as PS, electrophysiological measurements can be a useful approach. In this method, proteoliposomes are loaded onto a solid-supported membrane and a transient electrical current is measured upon the addition of ATP. As an ATP-dependent transporter moves charged lipids across the membrane, the capacitive coupling between the proteoliposomes and the solid-supported membrane allows the measurement of the resulting current transient [18,77] ( Figure 3B). This current can provide two types of information: firstly, the decay time constant, i.e., the rate at which charges are moved, can be calculated by fitting the decay of the current signal with a first-order exponential function. Secondly, the amplitude of the current signals, which represents the total number of charges moved, can be calculated by integrating the current signals. These parameters can then be used to assess the activity of the transporter and the kinetics of the lipid transfer process. Notably, this assay does not require lipid probes and can be used with endogenous lipids.
Another type of assay is based on the use of donor vesicles that are added to the proteoliposomes. In this approach, donor liposomes are prepared containing, e.g. short-chain NBD-lipids and rhodamine-labeled PE (N-Rho-PE), which cannot translocate from the donor liposomes to the proteoliposomes. During the exchange and translocation of the NBD-lipid, a decrease in the energy transfer between the NBD-lipid and the N-Rho-PE occurs, resulting in an increase in NBD fluorescence and enabling continuous measurement of lipid translocation ( Figure 3C). This method has been used to assess lipid transport by an ABC transporter [83]. Studies conducted by Wang and co-workers [84,85] have shown that the donor vesicle approach is also an effective method for observing the transfer of radiolabeled sterols. The results of donor-acceptor vesicle approach, however, should be interpreted with care. Since this method involves     ∼0.08 mol% tLLO ∼25% tLLO [35] Abbreviations: CL, cholesterol; DDM, n-Dodecyl β-D-maltoside; DM, n-decyl-β-D-maltopyranoside; FL, full-length; GalCer, galactosylceramide; Glc-Cer, glucosylceramide; LacCer, lactosylceramide; MDR1, multidrug resistance protein 1; MRP1, multidrug resistance protein 1; NBD, nitrobenzoxadiazol; N-Rho-PE, head-labeled rhodamine phosphatidylethanolamine; OG, n-octyl-β-D-glucopyranoside; PC, phosphatidylcholine; PE, phosphoethanolamine; PG, phosphatidylglycerol; Pgp, P-glycoprotein multidrug transporter; PI(4)P, phosphatidylinositol-4-phosphate; PI, phosphatidylinositol; PS, phosphatidylserine; SL, spin-labeled; SM, sphingomyelin; SSM, solid supported membranes; TAPL, transporter associated with antigen; tLLO, trisaccharide lipid-linked-oligosaccharide. 1 "Method" refers to the reconstitution approach (starting material: Lipid suspension or preformed liposomes, also known as large unilamellar vesicles; detergent used for reconstitution; detergent removal; method for monitoring lipid transport/transfer). 2 Lipid composition of lipososomes used for reconstitution. 3 The NBD-group is attached to an acyl chain (NBD-) or headgroup (N-NBD-). 4 Amplitude = % of lipids transported. 5 Amplitude range indicates differences depending on substrate tested. 6 The Mg 2+ -ATPase activity is likely based on ATP11C [129].
the transfer of lipids between vesicles, the assay may measure the ability of the transporter to insert or extrude lipids into or out of the membrane rather than to transport them across the bilayer.

Insights from in vitro studies
To date, only a small number of P4-ATPases and ABC transporters have been successfully reconstituted (see Table  1 and references therein). In most cases, the assignment of lipid translocation activity has been based on the use of fluorescent lipid probes, with only a few studies attempting to measure the transport of natural lipids. Nevertheless, the reconstitution experiments performed so far provide the best evidence that these transporters directly catalyze lipid transport.

P4-ATPases: lipid flippases with different substrate specificities
One of the first purified P4-ATPases studied upon reconstitution was ATP8A2, a P4-ATPase present in the disc membranes of rod and cone photoreceptors. Upon reconstitution into chemically-defined liposomes, the purified enzyme was found to flip fluorescent-labeled PS [86]. Similarly, reconstitution of the yeast P4-ATPase Drs2 showed that this enzyme catalyzes PS transport [87]. Substrate competition assays have been used here to verify that natural lipids are indeed substrates. The natural lipid substrate competes with the fluorescent probe, resulting in a decrease in the transport activity toward the lipid probe. Such inhibition has been demonstrated, for example, for the P4-ATPases In the dithionite assay, the membrane-impermeable quencher dithionite reduces the NBD group into a nonfluorescent derivative (ABD-lipid, gray). In the collisional quenching assay, potassium iodide (KI) or cobalt chloride (CoCl 2 ) generate a non-radiative transition to the ground state of the NBD-group (Transiently quenched lipid, brown). In case of spin-labeled lipids, chemical reduction by aqueous ascorbate is used to generate reduced spin-label (rSL-lipid, black). (B) In the electrophysiological method, proteoliposomes are adsorbed on a solid supported membrane and subjected to ATP activation. Upon transport of charged lipids, a current signal is detected (adapted from [77]). RE, reference electrode; V, potential difference; SSM, solid supported membrane. (C) Donor-acceptor approaches require two sets of liposomes, e.g., donor vesicles containing a FRET-pair, here an NBD-lipid together with a headgroup labeled N-rhodamine lipid such as PE (N-Rho-PE) and acceptor proteoliposomes without fluorescent lipids. Exchange and translocation of the NBD-lipid can be quantified by measuring the decrease of the energy transfer between the NBD lipid and the N-Rho-PE or by measuring NBD-fluorescence after separating donor from acceptor vesicles.
ATP8A2 [86] and ATP11A, B, C [88]. Further biochemical characterization of P4-ATPases at the solubilized level [89][90][91] has shown that the lipid headgroup is the key structural element for substrate recognition by P4-ATPases. However, the structure of the lipid tail may influence the correct positioning of the glycerol backbone and thereby affect the recognition of mono-and di-acyl lipids. Another important finding is the identification of phosphoinositides as activators of some P4-ATPases, as shown for the P4-ATPases Drs2 [74,[92][93][94], ATP8B1 [95], and ATP2 [96]. The C-terminal extension of Drs2, together with the transmembrane domain 10, contains amino acids that bind phosphatidylinositol-4-phosphate [94], and the presence of this phosphoinositide stimulates the flippase activity of the enzyme in proteoliposomes [74]. This finding suggests that phosphoinositides may relieve the autoinhibition imposed by the C-terminus of Drs2 that can bind in between the P-and the N-domains, thereby restricting potential domain movements required for pumping [93,94].
A non-exclusive possibility is that phosphoinositides act as lipid cofactors, for example, by potentially regulating substrate access to lipid transport pathway in the protein by inducing conformational changes in the transmembrane domain. This notion is supported by the observation that the activity of the purified human P4-ATPase ATP8B1 is stimulated by phosphoinositides only when the C-terminus is removed [95]. The N-terminus of Drs2 and ATP8B1 also seem to participate in the autoinhibition of these transporters and together with the C-terminus might have a synergistic effect on autoinhibition [92,95]. In addition, some P4-ATPases share the (G/A)(Y/F)AFS motif at the C-terminus, which is involved in autoinhibition through interaction with the N-domain [48,95,96]. Thus, regulation by the C-terminus may be a common feature of this transporter family. Furthermore, the activity of some P4-ATPase in cells is regulated by interacting proteins such as kinases and thus, phosphorylation [97][98][99] and small GTP-binding proteins [100] that might participate in autoinhibition release. However, it has not yet been possible to recapitulate this requirement in vitro.

ABC transporters: a family of exporters and importers
One of the first ABC transporters to be reconstituted for analysis of lipid transport activity was the human multidrug resistance transporter ABCB1 [78]. After reconstitution into chemically-defined liposomes, it was shown to act as a floppase for a variety of fluorescent-labeled phospholipids and sphingolipids [78,101], including simple glycolipids. Similarly, the ABCB4 and the glutathione-dependent multidrug transporter ABCC1 were shown to transport fluorescent-labeled PC after reconstitution into proteoliposomes [31,75], while ABCA1 and ABCA7 transported fluorescent-labeled PC, PS, and SM with a preference for PC and PS, respectively [82]. In contrast, ABCA4 was shown to function as an importer flipping fluorescent-labeled PE in liposomes and radiolabeled N-retinylidene-PE in photoreceptor disc membranes and proteoliposomes [82,102]. The ABC transporter PglK from Campylobacter jejuni catalyzed the translocation of lipid-linked oligosaccharides in vivo [36] and upon reconstitution into liposomes, PglK translocated an analog of the natural lipid-linked oligosaccharide [35]. Functional reconstitution allowed also demonstration of the stereospecific sterol transfer activity of co-purified ABCG5/G8 [84]. Thus, similar to P4-ATPases, individual ABC transporters differ in their substrate specificities. Some ABC transporters may have a dual substrate specificity, as recently uncovered for the lysosomal ABC transporter ABCB9 [32]. In addition to its role as a peptide transporter, ABCB9 was found to transport fluorescent-labeled PS after reconstitution into proteoliposomes.
As much of the evidence is based on the transport of fluorescent-labeled lipids, the selectivity for labeled substrates may not necessarily apply to the natural substrates in each case. In some instances, substrate competition data provide evidence that natural lipids act as substrates. For example, the transport of fluorescent-labeled PE by ABCA4 is inhibited by the presence of PE, whereas the transport activity of ABCA1 for fluorescent-labeled PC is suppressed by PC [82]. In the case of MsbA, the presence of physiologically relevant substrates such as lipid A was found to inhibit the translocation of PE probes in proteoliposomes [101]. Interestingly, a new twist to the story came when recent work identified the requirement of a chemical proton gradient for the transport of PE by MsbA ( [76]; Table 1). Notably, flopping of the large hexa-acylated Lipid-A required only ATP, pointing out that the energetic requirements of MsbA as a lipid transporter are substrate dependent. Future studies will need to define if this property is shared by other members of the ABC transporter and possibly, of the P4-ATPase families.
Reconstitution studies have also been instrumental in elucidating the dependence of the ABC transporters on specific lipids for their activity. In particular, phospholipids and sterols have been shown to influence the activity of several ABC transporters involved in lipid translocation. One example is the yeast ABC transporter Aus1, which localizes primarily to the plasma membrane and, together with Pdr11, is required for sterol uptake under anaerobic conditions. Upon reconstitution into proteoliposomes, the ATPase activity of Aus1 was specifically stimulated by PS in a stereoselective manner [103]. Presence of cholesterol in liposomes reconstituted with the mammalian ABC transporters ABCA1 and ABCA4 reduced the transport of fluorescent-labeled PC and PE, respectively [82]. However, it is important to note that in these experiments it is not possible to distinguish whether lipids modulate the protein indirectly by changing the properties of the lipid membrane or by acting as co-factors.

Current limitations and future avenues
Despite their widespread use as a model membrane system, proteoliposomes have some limitations. Liposomes are closed bilayer structures of small dimensions with high curvature (diameter 100-200 nm). Therefore, unidirectional lipid transport in these vesicles, even at small amplitudes, can lead to the accumulation of bilayer stress [104]. This mechanical stress could feedback to inhibit transporter activity, indicating substrate saturation of the recipient leaflet [105]. It is therefore crucial to conduct careful controls to ensure that changes in the signal can be attributed to transporter activity. These might include experiments carried out in the presence of non-hydrolyzable ATP derivatives, inhibitors, in the absence of either ATP or Mg 2+ and/or experiments carried out on reconstituted catalytically inactive mutants of the transporter under investigation. Another important aspect of liposome reconstitution is ensuring the absence of leakiness, which can be caused by residual detergent [73]. To detect potential leakiness in reconstituted liposomes, various approaches have been employed. One approach involves entrapping fluorescent probes, such as a fluorescent glucose analog, within liposomes, followed by dithionite quenching to assess leakage [106]. Alternatively, a self-quenching probe like calcein has been used, where leakage is monitored over time in a calcein leakage assay by recording changes in fluorescence [107].
As mentioned above, current cell-based and reconstitution assays commonly use lipid probes. These probes have a reporter group attached to a short fatty acid chain at the sn-2 position, while a long fatty acid chain is at the sn-1 position (Figure 1, lower panel). However, a major drawback of these tagged probes is the substantial alterations of steric bulk and polarity with respect to the endogenous lipids, which may affect their recognition as substrates and affect the biophysical properties of membranes. In the case of ABC transporters that mediate the efflux of fluorescent dyes, the lipid probe may even be perceived as a drug analog rather than a lipid. In the future, new approaches will be needed to measure transbilayer lipid movement using non-labeled lipids. Promising steps have been taken for lipid scramblases, including shape-change visualization experiments in giant unilamellar vesicles (GUVs) [108], phospholipase-based assays [109], and detection of specific lipids from each membrane leaflet using lipid-binding protein domains [110] (Figure 4). GUVs, ranging in size from 1 to 100 μm, provide a valuable platform for functional analysis of lipid transporters using light microscopy techniques without requiring labeled lipids. In this setup, unidirectional lipid transport by energy-coupled flippases or floppases creates a mass imbalance between the two membrane leaflets, inducing shape changes in the giant vesicle ( Figure 4A). However, the reconstitution of membrane transporters into GUVs has proved challenging due to the inherent instability of these large liposomes and the potentially unfavorable conditions used during the reconstitution process. As a result, successful application of this approach has been limited up to date to scramblases and the erythrocyte lipid flippase [108,[111][112][113]. Furthermore, phospholipase-based assays have been used to detect lipid scrambling in proteoliposomes, and should also be useful for studying energy-coupled flippases and floppases. For example,  used a phosphoinositol (PI)-specific phospholipase C and radiolabeled PI to detect phosphatidylinositol scrambling [109]. Matoba et al. [2020] investigated lipid scrambling in proteoliposomes by using a phosphoinositide 3-kinase and ATP to convert PI in the outer leaflet to phosphatidylinositol 3-phosphate (PI3P). The redistribution of this lipid was then analysed using a rapid freezing and freeze-fracture replica method together with a PI3P-recognizing protein domain [110].
Finally, techniques such as measuring the zeta potential using the mechanisms of electrophoresis [107], nuclear magnetic resonance spectroscopy [114] and small angle neutron scattering [115] have been tested to characterize liposomes. Zeta potential measurements exploit the correlation between the zeta potential and the surface charge density, which depends exclusively on the anionic lipid content in the outer leaflet of liposomes (Figure 4). Indeed, these measurements have been successfully employed to study the asymmetry of anionic lipids and their successful incorporation and stability in large unilamellar vesicles (LUVs) and GUVs [116][117][118]. However, the aforementioned approaches have not yet been applied to study lipid transport in reconstituted model membrane systems. Combined with advanced reconstitution protocols that enable the generation of proteoliposomes with an asymmetric lipid arrangement [119], such techniques may prove useful for measuring the transbilayer movement of natural lipids.
Another problem encountered in liposome reconstitution is sample heterogeneity, which can bias the quantitative analysis of lipid and protein content as well as protein activity. To overcome this challenge, a promising approach is to characterize the intrasample compositional variation at the single vesicle level. This approach is based on the visualization of single liposomes by microscopy, using fluorescently labeled lipids and reconstituted proteins with corresponding fluorescent labels. By using site-specific quenching of the fluorophores, it is possible to obtain a detailed characterization of the reconstituted proteoliposomes, including parameters such as size, tightness, lamellarity and the number of proteins per liposome as well as their orientation [120]. In addition, this approach allows functional characterisation of the reconstituted enzyme, providing insights into important features such as the transition between active and inactive states and the role of autoinhibitory domains in these two states, as well as intrinsic transport rates [121,122]. Watanabe et al. (2018) have recently developed an elegant method for the single-molecule analysis of membrane transporters using femtoliter chamber arrays. The femtoliter chambers are enclosed by asymmetrical membrane bilayers, with fluorescently labeled phospholipids present exclusively in one leaflet. These bilayers also contain individual membrane proteins, enabling the monitoring of lipid translocation events. This method offers a high level of sensitivity, allowing for detailed investigations into the mechanisms and kinetics of lipid transport processes.

Summary
Lipid transporters are inherently difficult to study. On the one hand, this can be attributed to the difficulties in handling the integral membrane proteins and their delicacy in production, purification, and characterization of their assembly with membrane lipids. On the other hand, sophisticated techniques are required to study membrane proteins at the molecular level. Despite these challenges, researchers continue to make progress in understanding the mechanisms underlying transporter function and in developing new techniques for studying these critical cellular components. In this review, we included current published reports on ATP-dependent lipid transporters and their functional analysis using liposomal reconstitution. Such functional assays, coupled with structural analysis and computational studies, including both molecular dynamics and coarse-grained simulations, will be essential in unravelling the inner working of these important transporter classes. In addition, we highlighted recent progress in uncovering key molecular and regulatory mechanisms of these transporters, the role of which remains to be further defined by systematic and detailed studies in model membrane systems and in vivo.

Competing Interests
The authors declare that there are no competing interests associated with the manuscript.

Funding
The research in the author's laboratory was funded by grants from the Deutsche Forschungsgemeinschaft [grant numbers GU 1133/11-1 and GU 1133/13-1].