Measuring Lipid Transfer Protein Activity Using Bicelle-Dilution Model Membranes

Abstract

In vitro assessment of lipid intermembrane transfer activity by cellular proteins typically involves measurement of either radiolabeled or fluorescently labeled lipid trafficking between vesicle model membranes. Use of bilayer vesicles in lipid transfer assays usually comes with inherent challenges because of complexities associated with the preparation of vesicles and their rather short “shelf life”. Such issues necessitate the laborious task of fresh vesicle preparation to achieve lipid transfer assays of high quality, precision, and reproducibility. To overcome these limitations, we have assessed model membrane generation by bicelle dilution for monitoring the transfer rates and specificity of various BODIPY-labeled sphingolipids by different glycolipid transfer protein (GLTP) superfamily members using a sensitive fluorescence resonance energy transfer approach. Robust, protein-selective sphingolipid transfer is observed using donor and acceptor model membranes generated by dilution of 0.5 q-value mixtures. The sphingolipid transfer rates are comparable to those observed between small bilayer vesicles produced by sonication or ethanol injection. Among the notable advantages of using bicelle-generated model membranes are (i) easy and straightforward preparation by means that avoid lipid fluorophore degradation and (ii) long “shelf life” after production (≥6 days) and resilience to freeze–thaw storage. The bicelle-dilution-based assay is sufficiently robust, sensitive, and stable for application, not only to purified LTPs but also for LTP activity detection in crude cytosolic fractions of cell homogenates.

Graphical Abstract

Lipids provide the basic bilayer structural platform for membranes that surround and internally compartmentalize cells in various eukaryotic tissues. To help regulate the lipid metabolic and recycling processes needed to maintain membrane integrity during periods of cell growth and proliferation as well as during programmed cell death events, cells rely on nonvesicular lipid transport. This kind of lipid transport is carried out by specific proteins, i.e., lipid transfer proteins (LTPs) that have the ability to bind and transfer specific lipids between membranes. LTPs are amphitropic proteins that use their binding sites for specific lipid types to acquire and release their lipid cargoes during transient interaction with membranes. When bound, the lipid cargo is sufficiently shielded from the cellular aqueous milieu so that LTPs effectively become molecular solubilizers of lipid. Interest in LTPs has enjoyed a renaissance stemming from emerging information indicating that certain LTPs can function in vivo as molecular sensors and/or presentation devices involved in lipid metabolic regulation and signaling processes.^1–5

Current approaches for assaying lipid intermembrane transfer typically rely on phospholipid bilayer vesicles prepared by sonication, ethanol injection, or extrusion. Fresh vesicle preparation is needed to avoid postpreparation, storage-related changes that affect vesicle size uniformity and aggregation state. Such changes can negatively impact the precision, reliability, and reproducibility of lipid transfer assays. To address the issue, we explored the usefulness of bicelle-related model membranes for evaluating lipid transfer protein activity.

Bicelles are lipid aggregates that self-assemble in aqueous environments to form disc-like structures consisting of flat bilayer-like core regions and curved micelle-like edge regions. The planar bilayer core is formed by a long-chain phosphoglyceride, whereas the curved rim consists of detergent or short-chain phosphoglyceride that shields the long-chain lipid tails from water.^6–9 Nanodiscs are also disc-like lipid aggregates consisting of flat bilayer core regions but with their edges stabilized by protein “wrappers”, i.e., amphipathic helical membrane scaffold proteins (MSPs), that shield the long-chain lipid tails from water.^10,11 Although construction of lipoprotein nanodiscs is technically more challenging than bicelle preparation,¹² nanodiscs and bicelles both have both been used as bilayer platforms to study the structure of embedded integral proteins by NMR.^13–16 Yet neither bicelles nor nanodiscs have previously been used to study lipid transfer reactions mediated by proteins with specificity for select lipids.

Fluorescence resonance energy transfer (FRET)-based lipid transfer assays are well-established for tracking the transfer of specific sphingolipids between membrane vesicles by glycolipid transfer protein (GLTP)^17–21 and by other GLTP superfamily members.^22–26 In the assay system reported herein, we have characterized conditions for the effective use of a bicelle-dilution-based lipid transfer assay system and show the advantages compared to other model membranes for in vitro FRET monitoring of fluorescent lipid transfer reactions.

MATERIALS AND METHODS

Materials.

1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) and 1,2-hexanoyl-sn-glycero-3-phosphocholine (DHPC) were purchased from Avanti Polar Lipids and used without further purification. Sphingolipids (SLs) labeled with N-[15-(4,4-difluoro-1,3,5,7-tetramethyl-4-bora-3a,4a-diaza-s-indacene-8-yl)] pentadecanoyl acyl chains (e.g., Me₄-BODIPY-15-ceramide-1-phosphate (C1P) or Me₄-BODIPY-15-galactosylceramide (GalCer) were synthesized by reacylation of sphingosine-1-phosphocholine followed by phospholipase D treatment²⁷ or by reacylation of galactosyl(β)-sphingosine,²⁸ respectively, and then purified.

Bicelle-Based Dilution Preparations.

Lipid mixtures with q-ratios ranging from 0.1 to 2 were prepared by combining the appropriate amounts of BODIPY-15-SL (1 mol %) and DiI-C18 (1.5 mol %) with respect to POPC along with DHPC for stock mixtures as detailed in the Supporting Information.

Vesicle Preparation by Conventional Means.

POPC receiver (acceptor) vesicles and SL source (donor) vesicles were prepared by probe sonication and ethanol injection, respectively, as described by Mattjus et al.¹⁷ SL source vesicles were composed of POPC (97.5 mol %), BODIPY-15-SL (1 mol %), and DiI-C18 (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate) (1.5 mol %). Donor and acceptor vesicle diameters averaged 44 ± 2.7 and 25–30 nm, respectively. Buffer consisted of 10 mM potassium phosphate (pH 6.6), 150 mM NaCl, and 0.2% EDTA.

Fluorescent Lipid Transfer between Membranes.

Real-time intermembrane transfer of fluorescent SLs was assessed by FRET using a SPEX FluoroLog3 spectrofluorometer (Horiba Scientific), with excitation and emission band passes of 2 nm and a stirred (∼100 rpm), temperature-controlled (25 ± 0.1 °C) sample cuvette holder. Assay details appear in the Supporting Information.

Electron Microscopy.

Cryo-EM data was collected on samples prepared as detailed in the Supporting Information using a Titan Krios electron microscope operating at 300 kV and equipped with a Falcon 3EC direct electron detector (Thermo Fisher Sci.). Additional details are found in the Supporting Information. For negative-stain EM, images were recorded on a Technai G2 Spirit BioTwin electron microscope (FEI) with a Gatan Ultrascan camera (Gatan, Pleasanton, U.S.A.) operated at 120 kV at 23 000× magnification using samples described in the Supporting Information.

Light-Scattering Measurements.

Static light scattering was measured at 90° relative to incident light using the SPEX FluoroLog3 spectrofluorometer (Horiba Scientific) as previously described by Mattjus et al.¹⁷ Dynamic light scattering (DLS) was measured with a Zetasizer Nano ZS equipped with Peltier temperature control (Malvern, Worcestershire, UK) using noninvasive backscatter optics. Other details are provided as Supporting Information.

Recombinant LTP Purification.

Cloning, expression, and purification of GLTP,^26,29–31 ACD11,²³ and CPTP²² reported previously are detailed in the Supporting Information.

HeLa Cell Lysate Preparation.

HeLa cells were plated on 100 mm dishes and transfected with different vectors using Lipofectamine 2000 (Invitrogen, 11668–019). At 24 h post-transfection, cells were harvested by trypsinization, washed with phosphate-buffered saline (PBS) by benchtop centrifugation, and then suspended in PBS buffer (200 μL). Disruption of the HeLa cells was accomplished by brief probe sonication (30 s × 2) on ice. Cell supernatants were recovered by benchtop centrifugation (5 min @ 12 000 rpm) at 4 °C, and 10 μL aliquots were assayed for SLTP activity using donor bicelle-dilution vesicles containing either Me4-BODIPY-GalCer or Me4-BODIPY-C1P along with C18-DiI.

RESULTS

FRET Approach for Tracking Lipid Intermembrane Transfer.

Compared to many other fluorophores, BODIPY exhibits photophysical properties that enhance lipid transfer assay performance, i.e., high environmental stability, high photostability, low environmental polarity sensitivity, and high emission intensity.^28,32–34 Thus, we relied on SL carrying a pentadecanoyl acyl chain labeled with tetramethyl-BODIPY (Me₄-BODIPY) as an energy donor and C₁₈-DiI as an energy acceptor for real-time kinetic tracking of the complete lipid transfer reaction using fluorescence resonance energy transfer (FRET) technology, which provides an order of magnitude sensitivity increase compared to the anthrylvinyl-SL/3-perylenoyl-PC energy donor/acceptor pair previously used to track lipid transfer.^2,34 The lipid transfer reaction involves uptake of SL by protein from “SL-source” model membranes and delivery of SL by protein to “receiver” model membranes. Figure 1 illustrates the lipid fluorophore structures and changing FRET response that reflects SL transfer. Wavelength-selective excitation of Me₄-BODIPY-SL initially results in minimal emission but strong emission by C₁₈-DiI via FRET due to the close proximity of both fluorescent lipids in SL-donor (source) model membranes. Addition of excess acceptor (receiver) model membranes (containing no lipid fluorophores) and SL-specific transfer protein (SLTP) triggers a sudden and time-dependent emission increase by Me₄-BODIPY SL due to the loss of FRET resulting from fluorescent SL transfer that creates separation from nontransferable C₁₈-DiI present in the SL-donor (source) model membranes. The signal response is indicative of the SL intermembrane transfer rate.

Figure 1.

Bicelle-dilution lipid transfer measurement by lipid transfer proteins using fluorescence resonance energy transfer (FRET). (A) Structures of FRET energy donors (Me₄-BODIPY-GalCer, Me₄^-BODIPY-C1P) that are transferred by GLTP and CPTP, respectively, and nontransferable FRET energy acceptor (C18-DiI). (B) Excitation and emission spectra of Me₄-BODIPY-SL and C18-DiI. (C) FRET emission changes observed upon mixing POPC/DHPC donor bicelle-dilution model membranes containing Me₄-BODIPY-GalCer and C18-DiI with excess acceptor POPC/DHPC model membranes followed by GLTP. (D) Schematic showing GLTP-mediated transfer of Me₄-BODIPY-GalCer (lime green) out of POPC/DHPC bicelle-dilution “donor” vesicles containing nontransferable C18-DiI (red) to bicelle-dilution POPC/DHPC “acceptor” vesicles. (Bicelles and vesicles are not drawn to scale).

Use of Bicelle-Dilution Model Membranes for Tracking Lipid Intermembrane Transfer.

An advantage of FRET is high sensitivity using relatively low model membrane concentrations. We assessed whether bicelle-based dilution can generate suitable model membranes for determining the complete SL-transfer reaction mediated by SL lipid transfer proteins, i.e., protein loading with SL from “donors” followed by SL delivery to “acceptors” (Figure 1). We initially tested 0.5 q-value mixtures because of their well-characterized properties¹³ and established suitability for high-resolution solution-state NMR studies involving the structure and dynamics of membrane-associated peptides.^35,36 q-Value refers to the molar ratio of long-chain PC to detergent used to construct the bicelles. Adjustment of the q-value alters the aggregate form and discoidal size of bicelles. POPC and DHPC were selected to favorably accommodate BODIPY-labeled SL and C₁₈-DiI.^7,8,37 As shown in Figure 2A, SL-donor and -acceptor model membranes formed by dilution of 0.5 q-value mixtures support robust SL transfer. Notably, almost no increase in BODIPY-SL emission occurs until transfer protein is added, confirming very slow spontaneous SL migration to acceptors in agreement with earlier findings involving conventionally prepared membrane vesicles.^38–40 Also, combining SL-donors with only protein fails to significantly increase BODIPY-SL emission unless followed by addition of acceptors showing that protein binding of BODIPY-SL does not explain the FRET response. Rather, the low “catalytic” amounts of GLTP act in “shuttle-like” fashion to transfer Me₄-BODIPY-GalCer continuously from the SL-donors to the excess acceptors until dynamic equilibrium is approached (∼20 min). Other potential explanations for the observed FRET signal changes such as the aggregation/fusion state of the donors and acceptors before and after GLTP addition were ruled out based on control experiments described in the Supporting Information.

Figure 2.

Bicelle-dilution LTP assay: optimal performance conditions. (A) FRET changes reflect BODIPY-lipid transfer between donor and acceptor model membranes, not BODIPY-lipid binding by lipid transfer protein (GLTP). For black trace, a = donors added, b = acceptors added, and c = GLTP added. For red trace, a = donors added, b = GLTP added, and c = acceptors added. Response signals are nearly identical. (B) Effect of different q-value donors on GLTP transfer activity when mixed with 0.5 q-value acceptors. (C) Effect of different q-value acceptors on GLTP transfer activity when mixed with 0.5 q-value donors. a = donors added, b= acceptors added, and c = GLTP added. (D) Cryo-EM of 0.5 q-value dilution POPC/DHPC donors with FRET fluorphore lipids showing unilamellar nature and 31.5 ± 3.8 nm outer diameter. Bar graph shows vesicle size distribution. (E) Cryo-EM of 0.5 q-value dilution POPC/DHPC acceptors showing unilamellar nature and 36.3 ± 6.1 nm outer diameter. Bar graph shows vesicle size distribution.

As mentioned earlier, changing the q-value alters the size and structure of bicelles. In the low q-value range (<0.3), mixed micelles form.⁴¹ Higher q-values (e.g., 1.0 and 2.0) lead to fewer bicelles but with increased overall diameter, i.e., more bilayer core and proportionally less edge, compared to 0.5 q-value bicelles at equivalent total lipid concentrations. To determine which q-value(s) work best in the lipid transfer assay, SL-donors generated by dilution of various q-value mixtures were assessed when using 0.5 q-value dilution acceptors. Figure 2B shows that 0.5 q-value dilution donors resulted in faster SL transfer compared to donors formed using lower (0.1) or higher (1.0 or 2.0) q-values. Figure 2C shows the effect of acceptor model membranes produced from different q-value mixtures on SL-transfer rates when using donor model membranes produced by dilution of 0.5 q-value mixtures. Under these conditions, SL transfer decreased strongly when the acceptors originated from 0.1 q-value mixtures but not for acceptors originating from 1.0 and 2.0 q-value mixtures. Due to the preceding outcomes, we focused further analyses on donor and acceptor model membranes produced by dilution of 0.5 q-value POPC/DHPC mixtures.

Assay conditions for 0.5 q-value stock mixtures of donors and acceptors typically involve 125-fold dilution to their respective range of ∼192 to 512 μM DHPC, well-below its 14 mM critical micelle concentration (cmc) but not that of POPC (≈ 0.5 nM). In bicelles, the high aqueous solubility of rim-stabilizing amphiphiles such as DHPC requires high total lipid concentrations (e.g., >1–2 wt %) to maintain bicelle structural stability. Dilution with aqueous buffer can trigger re-equilibration of DHPC into the aqueous phase to drive structural changes including bilayer vesicle formation.^41–46 Such changes have been documented when DHPC/DMPC bicelles are diluted below 1–2 wt %. Modeling studies^47,48 suggest a vesicle formation mechanism driven by bicelle coalescence resulting from DHPC loss from the rim. With more DHPC departure, the perimeter (rim) line tension begins to dominate the elastic bending energy, leading to “cup-shaped” hemi-vesicle intermediates that reduce hydrocarbon chain exposure to water. Eventual closure forms unilamellar vesicles. Comprehensive testing of the preceding model by studies of various lipid compositions, formation conditions, and kinetics remains to be carried out.

To determine the nature of the stock model membranes formed from the 0.5 q-value mixtures, we assessed form and size. Cryo-electron microscopy (cryo-EM) imaging, which relies on ultrarapid freezing, indicated formation of highly homogeneous unilamellar vesicles by the 0.5 q-value dilution donors and acceptors (Figure 2D,E). Donor vesicle diameters averaged 32 ± 4 nm, and acceptor vesicle diameters averaged 36 ± 6 nm. Negative-stain EM analyses provided similar size information, although sample dehydration and osmotic stress artifacts appeared to promote vesicle aggregation and affect vesicle shape (data not shown).

Further support for the unilamellar nature of the bicelle-dilution donor vesicles was obtained by testing BODIPY-SL accessibility to soluble GLTP (Figure 3A). The BODIPY-GalCer transfer equilibrium by GLTP approaches ∼70% for bicelle-dilution vesicles. This value represents the BODIPY-GalCer present in the donor vesicle outer surface and accessible to GLTP based on solubilization with excess Tween-20 “infinitely” dispersing and diluting the BODIPY and DiI lipid fluorophores to provide signal estimates for 100% transfer. The unequal BODIPY-SL distribution in the outer and inner surfaces reflects the well-known lipid transbilayer mass imbalance resulting from curvature of small bilayer vesicles. The findings are consistent with the vast majority of the bicelle-dilution vesicles being unilamellar.

Figure 3.

Bicelle-dilution LTP assay robustness. (A) GLTP accessibility to BODIPY-GalCer indicates bd-vesicles are stable and unilamellar. (B) Lack of dilution-induced structure changes to POPC/DHPC model membranes during the assay time course. 0.5 q-value donors (a) were mixed with 0.5 q-value acceptors (b) and equilibrated for various time intervals prior to GLTP addition (c). (C) Effects of different combinations of bd-vesicles versus conventional SUV donors and acceptors on GLTP transfer activity. Vesicles (s) = sonicated small vesicles; vesicles (e) = ethanol-injection small vesicles; donor and acceptor q-value mix = 0.5. (D) Superior stability of bd-vesicles improves FRET lipid transfer assay performance compared to conventional vesicles. Donors prepared in different ways were used in the transfer assay either soon after preparation or subjected to three freeze–thaw cycles (20 to −20 °C) prior to use. a = donors added (bd-vesicles or ethanol-injection vesicles), b = acceptors added (bd-vesicles or sonicated vesicles), and c = GLTP added.

We tested the stability of the stock donor and acceptor vesicles upon dilution into the assay by comparing different mixing equilibration times prior to assay initiation by protein addition. Figure 3B shows that introduction of GLTP at different time intervals after the initial mixing of POPC/DHPC bicelle-dilution SL donors and POPC/DHPC acceptors had minimal effect on the transfer kinetics of Me₄-BODIPY-SL. Moreover, after combining all assay components, no detectable change in 90° static light scattering occurs over the experimental time course (Figure S1). DLS measurements of the 0.5 q-value stock acceptors diluted to FRET assay conditions showed no change in vesicle size over a 20 min interval. Similar dilution measurements with stock 0.5 q-value donors exceeded reliable detection limits for vesicle sizing by DLS. Altogether, the results indicate that the time course for the typical lipid transfer assay is sufficiently fast to avoid major structural changes to the vesicles formed by bicelle dilution during the time course of a typical assay.

SLTP Transfer Rates Achieved with Bicelle-Dilution Vesicles versus Conventional Small Vesicles.

Small PC vesicles have typically been used as model membranes to assess lipid interbilayer transfer because their curvature yields faster transfer rates compared to those obtained using larger vesicles.^2,18,49 We tested donor and acceptor combinations involving either conventional small vesicles and/or 0.5 q-value bicelle-dilution vesicles (bd-vesicles) to determine which combination produced the fastest intermembrane transfer rates (Figure 3C). Notably, the data show that bd-vesicle-to-bd-vesicle conditions result in fast SL-transfer rates comparable with conventional vesicle-to-vesicle conditions so long as the donor vesicles are produced by rapid ethanol injection rather than by sonication. Production of the donor vesicles by sonication markedly slowed SL-transfer rates regardless of whether the acceptors were bd-vesicles or sonicated vesicles presumably because of fluorophore destabilization by probe sonication. Thus, similar high assay sensitivity is achieved using either bd-vesicle-to-bd-vesicles or combinations involving certain conventional vesicles. In a separate control, inclusion of DHPC at similar amounts as present when using bicelle-dilution vesicles minimally affected the GLTP transfer rate of Me₄-BODIPY-GalCer between ethanol-injection donors and sonicated acceptor vesicles (Figure S3).

Improved Storage Capacity of Bicelle-Dilution Vesicles versus Conventional Small Vesicles.

Aside from the technical challenges associated with preparation of uniform populations of bilayer vesicles by sonication, extrusion, or reverse phase evaporation, their quasistable nature often limits “shelf life” and imparts a need for fresh vesicle preparations to carry out high-quality lipid transfer assays. To determine if bd-vesicles exhibit superior stability and storage capacity compared to lipid vesicles prepared by other means, we tested their resistance to freeze–thaw changes that often affect bilayer vesicles by promoting aggregation and/or fusion. Figure 3D shows the comparative effects of three cycles of freeze/thawing stock bd-vesicles versus other vesicles prior to using them in the FRET lipid transfer assay. Me₄-BODIPY-GalCer transfer by GLTP was not affected by freeze/thawing of the bd-vesicles, whereas the transfer rates with the other vesicles were significantly diminished. The findings indicate that the bd-vesicles can be prepared and then stored frozen prior to use without negatively affecting assay performance. In a separate control, stock bd-vesicles kept at room temperature for up to 6 days did not negatively impact on their performance in the FRET assay (Figure 3A).

Bicelle-Dilution LTP Assay Applications.

Lipid Specificity of Different SLTPs.

The GLTP superfamily consists of members with differing SL binding and transfer specificity.^19,24,26 We assessed whether use of the bd-vesicle-to-bd-vesicle transfer conditions enables detection of the SL-transfer specificity by different superfamily members. Figure 4A clearly shows that transfer of Me₄-BODIPY-GalCer, but not Me₄-BODIPY-C1P, by GLTP is detected using bicelle-dilution donor and acceptors. The specific transfer of Me₄-BODIPY-C1P, but not Me₄-BODIPY-GalCer, by the plant CPTP ortholog, ACD11, is also detected (Figure 4B). This outcome agrees with earlier findings involving conventionally prepared vesicle-to-vesicle assay systems.^2,21

Figure 4.

Applications of the bicelle-dilution LTP assay system. (A) Detection of Me₄-BODIPY-GalCer transfer (blue trace) by GLTP but not by ACD11 (red trace). (B) Detection of Me₄-BODIPY-C1P transfer by ACD11 (red trace) but not by GLTP (blue trace). q-Values = 0.5 for both donors and acceptors; a = donor addition, b = acceptor addition, and c = GLTP or ACD11 addition. (C) Nonfluorescent diethyl-C1P fails to slow BODIPY-C1P transfer, indicating no competition with BODIPY-C1P for interaction with ACD11 (i.e., no inhibition effect). (D) Slowing of ACD11 transfer of BODIPY-C1P by increasing levels of nonfluorescent C1P (1, 2, or 4 mol %) in bd-donor vesicles is indicative of competition for interaction with ACD11. (Donor q = 0.5; acceptor q = 0.5.) (E,F) In vivo transfer activity detection using the bd-assay by GLTP and CPTP expressed in HeLa cells after transient transfection. (E) GFP-GLTP or (F) GFP-CPTP: after transfection and growth for 24 h, the HeLa cells were disrupted by brief probe sonication (30 s × 2) on ice. Cell supernatants were prepared and used as described in the Methods section. (Donor q = 0.5; acceptor q = 0.5.)

Assessment of Potential SLTP Inhibitors.

The C1P derivative, diethyl-C1P, reportedly can inhibit cytosolic phospholipase A₂α (cPLA₂α) activity when the C1P acyl chain is short, i.e., C2-diethyl-C1P.⁵⁰ The inhibition presumably occurs via C2-diethyl-C1P interaction with the cPLA₂α C2-domain, which contains a C1P binding site that activates cPLA₂α by enhancing translocation to membranes.^51,52 To date, no inhibitors of C1P transfer proteins (e.g., CPTP and ACD11) have been reported. We used our bicelle-related dilution assay to test whether C2-diethyl-C1P can inhibit the plant CPTP ortholog, ACD11. Figure 4C shows that increasing concentrations of C2-diethyl-C1P in donor bd-vesicles fail to exert a strong competition effect on ACD11-mediated transfer rates of Me₄-BODIPY-C1P. In contrast, increasing concentrations of nonfluorescent C1P in donor bd-vesicles dramatically slow the transfer rates (Figure 4D). Despite the negative outcome for C2-diethyl-C1P, the data clearly show the potential of the bicelle-dilution FRET assay for identifying as of yet undiscovered inhibitors for various LTPs.

SLTP Activity and Specificity Detection Using Cell Lysate.

We also assessed whether the bd-vesicle-to-bd-vesicle transfer assay conditions are suitable for reliable detection of SL transfer specificity by different SLTPs within the cellular milieu. Figure 4E,F shows the results obtained using lysates of HeLa cells overexpressing GFP-GLTP or GFP-CPTP. The sudden sharp increase in emission at 503 nm reflects addition of GFP protein (Em_max = 509 nm). The ensuing slower but steadily increasing emission intensity reflects bd-vesicle-to-bd-vesicle transfer of Me₄-BODIPY-GalCer by GFP-GLTP but not by GFP-CPTP (Figure 4E). In contrast, Me₄-BODIPY-C1P transfer by GFP-CPTP, but not by GFP-GLTP, is shown in Figure 4F. Thus, the bd-vesicle-to-bd-vesicle transfer assay performs reliably in the presence of various other components present in crude cytosolic fractions.

DISCUSSION

In vitro measurement of lipid intermembrane transfer has typically relied on biomembrane mimetics such as unilamellar liposomal bilayer vesicles to model the physiological situation. Inherent challenges associated with conventional bilayer vesicle preparation include time-consuming multistep methodologies^53,54 and a rather short “shelf life” of the resulting small vesicles necessitating fresh preparation to carry out high-quality lipid transfer assays. The current study shows that use of POPC/DHPC bicelle-dilution model membranes provides an effective and efficient way to measure the intermembrane transfer of specific lipids by various LTPs. The highly sensitive FRET-based monitoring of lipid transfer provides real-time kinetics and avoids the need to physically separate and recover the donor or acceptor membranes to assess transfer.

A key aspect of the new lipid transfer assay is the ease and simplicity of preparing donor and acceptor model membranes by diluting 0.5 q-value POPC/DHPC mixtures. Preparation involves lipid mixing, solvent removal, hydration, and mild agitation at room temperature. The ensuing re-equilibration of DHPC into the surrounding aqueous milieu results in spontaneous formation of small unilamellar vesicles that serve as the donor and acceptor model membranes. The finding of such vesicle formation is not surprising based on earlier work showing vesicle formation by detergent removal.^55–57 Yet, a large excess of detergent was typically used to form mixed micelles with long-chain phosphoglycerides prior to detergent removal by dialysis or size exclusion chromatography resulting in somewhat oligolamellar vesicles with greater size heterogeneity. Vesicle formation resulting from bicelle dilution has also been reported^41–46 but has remained narrowly studied with regards to lipid compositions, formation conditions, kinetics, and vesicle uses. Notably, we have found that simple dilution of 0.5 q-value POPC/DHPC mixtures at room temperature along with mild agitation results in formation of unilamellar vesicles with narrow size distributions (32–36 nm diameter) that are stable and ideal for measuring LTP-mediated lipid transfer. Moreover, use of bicelle-dilution unilamellar vesicles in conjunction with the highly sensitive FRET approach requires >100-fold less resources (e.g., lipids, lipid fluorophores) than needed to maintain bicelles.

Bicelles and lipoprotein nanodiscs at high concentrations have been used previously as donor and acceptor model membranes to show spontaneous lipid transfer that is fast, nonspecific, and likely occurs via a collisional-based mechanism.⁵⁸ Interestingly, fast spontaneous lipid transfer via collisions among polymer-bounded nanodiscs has also recently been reported⁵⁹ in agreement with transfer mechanisms reported for bilayer vesicles at high concentrations over three decades ago.^38,39 Lowering the lipid model membrane concentrations or adding small amounts of charged lipid mitigates the model membrane collisional contacts and vastly reduces the rate of nonspecific spontaneous lipid transfer.^38,39 In recent studies of DMPC/DHPC or DPPC/DHPC bicelles containing 5 mol % of negatively charged phosphatidylglycerol,^60,61 the spontaneous transfer rates for the DMPC or DPPC bicelle bilayer-matrix lipid were slow and with their half-times differing from several hours to hundreds of hours, respectively, in agreement with earlier bilayer vesicles studies. The findings suggest a solubility-driven exchange via lipid monomer transfer through the aqueous medium, in agreement with studies of conventional bilayer vesicles^38,39 and lipoprotein nanodiscs.⁵⁸

In our assay, the spontaneous lipid transfer rates are vastly slower than LTP-mediated transfer of specific lipids, i.e., time scales of many hours versus ∼10 min, respectively. Nearly undetectable and extremely slow spontaneous transfer is also observed for Me₄-BODIPY-SL prior to the addition of various SLTPs due to fluorophore connection to the SL via a long pentadecanoyl acyl chain linker chain (Figure S2). The chemical makeup of the FRET energy acceptor fluorophore (C₁₈-DiI) renders it “nontransferable” and keeps it associated with SL-donor bicelle-generated model membranes after assay initiation because of the protein selectivity for specific SLs.

It is noteworthy that certain topological aspects of the physiological situation are accommodated by the assay design. In mammals, simple SLs such as glucosylceramide and C1P are produced anabolicly by glucosylceramide synthase and ceramide kinase, respectively, at specific locations on the cytosolic face of the Golgi, prior to transport by GLTP and CPTP to other intracellular sites.^22,62 To model this physiological situation, the SLs are initially confined to the SL source bd-model membranes and not present in the receiver bd-model membranes prior to transfer by protein.

We find that the use of bd-vesicles provides similar sensitivity and reproducibility as other small vesicle assays but without the need for preparation of fresh model membranes prior to assaying. In fact, bd-model membrane stocks can be repeatedly frozen and thawed without diminishing assay performance. Other advantages over other bilayer vesicle preparation approaches include (i) easy and fast production of bd-vesicles by simple dilution that avoids lipid fluorophore degradation; (ii) stability and relatively long “shelf life” after production including freezer storage without detrimental effects to FRET assay performance.

CONCLUSIONS

The use of bicelle-dilution-generated unilamellar vesicles in conjunction with FRET-based lipid monitoring provides a straightforward and easy lipid transfer assay with exceptional sensitivity, stability, and reproducibility compared to other conventional vesicle systems. The bd-vesicle-to-bd-vesicle assay is sufficiently robust for real-time detection of SL-transfer activity and SL specificity by various LTPs, not only when purified but also when present in crude cytosolic fractions recovered from cell homogenates. A noteworthy advantage is the easy and straightforward preparation by means that avoid lipid fluorophore degradation. The resulting model membranes are highly stable and have a long “shelf life” after production. The new assay offers the potential for adaptation to provide a platform for the development of “designer” LTP assay kits for (1) monitoring the activity of various LTPs with known lipid specificity; (2) identifying the lipid specificity of newly discovered proteins that potentially function as LTPs. Adaptation to high-throughput formats is expected to provide a highly sensitive and robust method for the screening of large libraries of small-molecule compounds to identify inhibitors of various lipid transfer proteins. Identified inhibitors could enable development of new drugs for treatments of inflammation, cardiovascular diseases, and other pathologies associated with abnormal lipid transfer protein expression and activities.