Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 Jan 8.
Published in final edited form as: Dev Cell. 2018 Jan 8;44(1):73–86.e4. doi: 10.1016/j.devcel.2017.12.011

Mechanism and Determinants of Amphipathic Helix-Containing Protein Targeting to Lipid Droplets

Coline Prévost 1,2,3,*, Morris E Sharp 4,*, Nora Kory 1,2,3,$, Qingqing Lin 1,2,3,&, Gregory A Voth 4,#, Robert V Farese Jr 1,2,3,#, Tobias C Walther 1,2,3,5,6,#
PMCID: PMC5764114  NIHMSID: NIHMS927865  PMID: 29316443

Summary

Cytosolic lipid droplets (LDs) are the main storage organelles for metabolic energy in most cells. They are unusual organelles that are bounded by a phospholipid monolayer and specific surface proteins, including key enzymes of lipid and energy metabolism. Proteins targeting LDs from the cytoplasm often contain amphipathic helices, but how they bind to LDs is not well understood. Combining computer simulations with experimental studies in vitro and in cells, we uncover a general mechanism for targeting of cytosolic proteins to LDs: large hydrophobic residues of amphipathic helices detect and bind to large, persistent membrane packing defects that are unique to the LD surface. Surprisingly, amphipathic helices with large hydrophobic residues from many different proteins are capable of binding to LDs. This suggests that LD protein composition is additionally determined by mechanisms that selectively prevent proteins from binding LDs, such as macromolecular crowding at the LD surface.

Keywords: Lipid droplets, amphipathic helices, protein targeting, phospholipid bilayers, phospholipid monolayers, phospholipid packing defects, all-atom molecular dynamics simulations, cell biology, reconstitution assay

eTOC Blurb

Prévost, Sharp et al. explore how cytosolic proteins target the surface of lipid droplets (LDs). A distinctive surface structure makes LDs suited to recruiting large hydrophobic amino acid-enriched protein motifs. Rather than specific LD protein targeting mechanisms, regulation is likely at the level of preventing promiscuous association by non-LD proteins.

graphic file with name nihms927865u1.jpg

INTRODUCTION

Lipid droplets (LDs) are cytoplasmic organelles that store neutral lipids, such as triacylglycerols (TGs) or sterol esters, which serve as reservoirs of lipids for cell proliferation or energy (Pol et al., 2014; Walther and Farese, 2012). The number and size of LDs vary greatly in cells, depending on the metabolic state, and LD accumulation in tissues, such as the adipose tissue and liver, is the hallmark of obesity. In addition to their function in energy metabolism, LDs are involved in a host of other cellular functions, such as maintaining proteostasis and as signaling hubs (Welte, 2015). Knowledge of basic LD biology is thus a fundamental aspect of cell biology, with important implications for our understanding of the pathogenesis of metabolic diseases and for engineering storage of neutral lipid oils in LDs for industrial purposes (Gluchowski et al., 2017; Krahmer et al., 2013a)

LDs comprise a neutral lipid core bounded by a phospholipid monolayer, which sets them apart from most other organelles that instead have a bilayer membrane. The phospholipid monolayer is derived from the cytosolic face of the ER, and its composition is similar to the ER lipid composition (Bartz et al., 2007; Choudhary et al., 2015). Many proteins target to the LD monolayer where they carry out crucial functions. For example, correct targeting of proteins is key for synthesizing or hydrolyzing TGs at the organelle surface (Kory et al., 2016). LD proteins differ between organisms and cell types, but generally include structural and regulatory proteins and enzymes of lipid metabolism (Brasaemle et al., 2004; Fujimoto et al., 2004; Kory et al., 2016; Krahmer et al., 2011; Krahmer et al., 2013b; Wilfling et al., 2013). How these proteins recognize and target to the LD surface is an important unanswered question of cell biology.

Current models posit two distinct pathways by which proteins target to the surface of LDs (Kory et al., 2016). In one pathway (Class I), proteins are initially inserted into the ER membrane and subsequently re-localize as membrane-embedded proteins from the ER to the LD surface using physical continuities between the two organelles, known as membrane bridges (Jacquier et al., 2011; Thiam et al., 2013; Wilfling et al., 2013). Consistent with their biochemical behavior as integral membrane proteins, cargoes of this pathway are characterized by stretches of highly hydrophobic amino acids with the potential to form membrane-embedded domains, possibly hairpin loops. Examples of Class I proteins include GPAT4 and DGAT2, enzymes involved in TG synthesis (Wilfling et al., 2013), and the methyltransferase ALDI (Zehmer et al., 2008).

In a second pathway (Class II), proteins are translated in the cytoplasm and subsequently directly bind the LD surface. Members of this class include the structural and regulatory perilipin proteins, as well as important metabolic enzymes (e.g., CCT enzymes 1 and 2) (Bickel et al., 2009; Kimmel and Sztalryd, 2016; Krahmer et al., 2011; Miyanari et al., 2007; Ohsaki et al., 2016; Rowe et al., 2016). These proteins typically contain sequences with a propensity to form amphipathic helices or short stretches of hydrophobic residues that are required for LD binding (Kory et al., 2016).

How these cytoplasmic proteins recognize and bind to the LD surface is largely unknown. The LD targeting pathway is likely unique as LDs do not have dedicated protein targeting/import machinery, such as the translocon or the TOM/TIM import channel for targeting of proteins to the ER or mitochondria, respectively. Nor do LDs apparently have specific lipids that serve as spatial organelle identity markers, such as phosphoinositides, that could be bound by targeting domains, such as PH domains. Given this, it remains unclear how LDs achieve their specific protein composition.

Here we combined computational and experimental approaches to unravel how LD surfaces differ from membrane bilayers surfaces and to identify mechanisms for targeting proteins to LD monolayers. Our data indicate that LDs exhibit distinctive surface properties that distinguish them from membrane bilayers and that these features of LD monolayers are recognized by cytosolic proteins with amphipathic helices containing large hydrophobic residues, thus providing a general mechanism for LD targeting of class II proteins. Surprisingly, many amphipathic helices are able to bind LDs in isolation, indicating promiscuity in binding and highlighting the need for mechanisms to prevent erroneous targeting and to maintain the LD proteome.

RESULTS

Molecular Dynamics Simulations Reveal that LD Surfaces Exhibit Numerous, Large Phospholipid Packing Defects

Because targeting of class II proteins to LDs does not seem to involve specific protein machinery or lipid markers, we hypothesized that LD proteins with AH motifs might detect global surface properties that are specific to LDs. To identify such features, we performed all-atom molecular dynamic (MD) simulations of LD surfaces (a summary of all the simulations run is presented in Table S1). Because the relatively large size of LDs prohibits the study of entire LDs in MD simulations due to limitations in computing power, we modeled the behavior of LDs by inserting a layer of neutral lipids, composed of 1:1 triolein (TRIO) and cholesteryl oleate (CO), inside of a flat bilayer, composed of 65:27:8 palmitoyl-oleoyl phosphatidylcholine (POPC): dioleoylphosphatidylethanolamine (DOPE): stearoyl-arachidonoyl phosphatidylinositol (SAPI) (Figure 1A), a previously demonstrated technique for simulating LD surfaces (Bacle et al., 2017; Hennere et al., 2005; Hennere et al., 2009; Koivuniemi et al., 2009). To determine the smallest system that exhibits LD surface properties, we varied the thickness of the neutral lipid layer. These analyses revealed that a 4-nm thick layer of neutral lipids was sufficient to change the properties of the phospholipid monolayers on each side, and that increasing this layer further did not lead to additional changes (see also below and Figure S1). Specifically, modeling LDs with these conditions revealed that inclusion of neutral lipids increased the surface area per phospholipid from 67 Å2 for bilayer membranes to 71 Å2 for monolayers of the same phospholipid composition. This increase results from the significant mixing between the neutral lipids and the phospholipids chains, illustrated by the z-density profile of the 4-nm-thick LD in Figure 1B. This finding is in agreement with the results of a recent report (Bacle et al., 2017). A snapshot of the simulation (Figure 1A) shows an example of a neutral lipid that has inserted all the way to the solvent exposed region, causing a packing defect on the surface of the membrane. Our simulations, thus, support a model in which the physical surface properties of LDs are distinct from that of bilayer membranes as LDs expose a rougher interface towards the aqueous cytosol, with a looser packing of surface phospholipids.

Figure 1. Molecular Dynamics Simulations Reveal that LDs Have Larger and More Persistent Surface Packing Defects than Bilayer Membranes.

Figure 1

Open in a new tab

(A) Simulation snapshots of the monolayer system (phospholipids: 65:27:8 POPC:DOPE:SAPI, neutral lipids: TG:SE 1:1, neutral lipid thickness: 4nm). Arrow indicates a neutral lipid that has inserted all the way to the surface.

(B) z-Density profile for the 4-nm LD. The relative density is normalized by the maximum number density. Note the mixing between the neutral lipids and the tails of the phospholipids. See also Table S1 for a summary of all simulations run.

(C) Distribution of packing defect sizes for bilayer and LD surfaces. The monolayer exhibits increases in the frequency of larger defects. Normalized frequency: number of defects for a given size range is normalized by the total number of defects over the simulation time frame. Solid lines are least-square fits to exponential decays. See also Figure S1.

(D) Distribution of packing defect sizes taking into account only the defects with characteristic lifetimes,τ, longer than 5 ns. Compared with the defects on the bilayer, the defects on the monolayer are roughly 10 times as likely to last longer than 5 ns for all defect sizes. Solid lines are least-square fits to exponential decays.

To further explore this possibility, we calculated the probability of observing hydrophobic lipid packing defects on the LD and on the bilayer surface based on our simulation data. We defined packing defects by mapping the phospholipid hydrocarbon tails exposed to the solvent and projecting them on the plane of the membrane (Cui et al., 2011). We refer to the area of these defects as “defect size”. Using this metric, LD monolayers show significantly more large packing defects than bilayer membranes (Figure 1C), as reported (Bacle et al., 2017). When we analyzed the persistence of packing defects, we found that defects with a life-time of more than 5 ns, the time scale of initial protein interaction with the membrane (Vanni et al., 2013), were more frequent on the monolayer than on the bilayer surface (Figure 1D). This suggests that the probability of a peptide productively encountering a large packing defect is much greater in such monolayer membranes than in bilayer membranes.

Initial Binding of the CCTα Amphipathic Helix to the LD Surface in Molecular Dynamics Simulations

We next used MD simulations to analyze the initial binding of amphipathic helices to the LD surface. For bilayer membranes, such simulations suggest a pathway that has several distinct steps (Tang et al., 2007; Ulmschneider et al., 2010; Vanni et al., 2013; Voth, 2013; White and Wimley, 1998). In the first step, the peptide, which is unfolded in solution, is often attracted to the membrane due to electrostatic forces. Next, hydrophobic residues of the targeting sequence adsorb onto hydrophobic packing defects of the phospholipid bilayer. In the last step of binding, the sequence around the inserted residue folds into an amphipathic α-helix, resulting in stable binding to the bilayer surface.

We hypothesized that targeting sequences of class II proteins have a high likelihood to similarly interact with packing defects at the surface of monolayer membranes. To test this hypothesis, we performed MD simulations of peptide-monolayer interactions, using the amphipathic helix of the LD-binding metabolic enzyme CCTα (Krahmer et al., 2011) as a model peptide. In four independent simulations, we placed this peptide (referred to as the M-domain) at least 2 nm from the membrane in a random coil conformation. We considered that binding was successful if hydrophobic residues inserted into the membrane below the plane of the phospholipids phosphates (an example is shown on the simulation snapshot Figure 2A) and did not dissociate from the membrane within the timeframe of our simulation (750 ns). If no insertion occurred, or if a residue inserted but subsequently dissociated, the simulation was scored as an instance where binding was unsuccessful (see an example for each case in Figure S2). In three of the four simulations, we observed binding of the M-domain to the LD surface within the timeframe of our simulations (Figure 2C). In each case where binding occurred, a large, but not necessarily the same, hydrophobic residue (F53 twice and L34 once) was first to insert into a packing defect. Additionally, in one of the four simulations, we observed the initial stages of amphipathic helix folding within the time window of our simulations (data not shown).

Figure 2. Molecular Dynamics Simulations Show that Amphipathic Helices Insert Bulky Hydrophobic Residues into Large Lipid Packing Defects.

Figure 2

Open in a new tab

(A) Representative sequence of insertion of a stretch of residues from the M-domain into the LD surface. The inserting residues are depicted in a space-filling representation, and the rest of the peptide is depicted by a ribbon. The horizontal line in the side view represents the phosphate level. Residues labeled are the hydrophobic residues that have inserted below the phosphate plane. See also Movie S1.

(B) Amino acid sequence of the M-domain and helical plots of both halves (P1 and P2). Dashed lines indicate the ends of both peptides within the full M-domain sequence.

(C) Binding success of AHs with (ArfGAP1 ALPS, M-domain, P2) or without (P1) large hydrophobic residues at the LD surface in MD simulations. Binding success is defined as the number of simulations in which at least one residue inserted in the lipid monolayer and remained inserted for the duration of the simulation. See also Figure S2 and Table S2. A total of four simulations were run for each peptide.

(D) Binding success of P2 to different surfaces. Monolayer: phospholipids: POPC:DOPE:SAPI 65:27:8, neutral lipids: TG: SE 1:1, bilayer: POPC: DOPE: SAPI 65: 27: 8, bilayer large defects: DOPC: DOG 85: 15. See also Tables S1 and S2.

Insertion of Large Hydrophobic Residues into Packing Defects Mediates LD Protein Binding

Based on our initial studies, we hypothesized that large hydrophobic residues (i.e., I, F, L, M, W, Y) in the sequences forming amphipathic helices interact with packing defects and, thus, are responsible for recognizing the LD surface. Within the M-domain, 10 such large hydrophobic residues are found in the second half of the sequence that we designated “P2” (Figure 2B; similar to “PEPC22” of (Dunne et al., 1996)). In the computational studies, initial binding of P2 to LDs occurred in each of four simulations (Figure 2C). In contrast, the N-terminal half of the M-domain, “P1” (Figure 2B; similar to “PEPNH1” of (Dunne et al., 1996)), with only two large hydrophobic residues, did not bind LDs in our simulations (Figure 2C). To further test our hypothesis, we analyzed the AH of another LD-targeting protein, ArfGAP1 (Gannon et al., 2014), which contains eight large hydrophobic residues. This peptide, known as the ALPS domain (Bigay et al., 2005), bound in each of four simulations (Figure 2C).

These findings suggest that specific targeting of amphipathic helices with large hydrophobic residues to LDs results from the higher prevalence of packing defects on monolayer than bilayer membranes. To test that, we analyzed P2 binding to lipid bilayer membranes and found that it occurred in only one of four simulations (Figure 2D). Consistent with previous experimental data (Dunne et al., 1996), this finding indicates that P2 can bind bilayer membranes, but does so with a lower probability than that for binding to monolayers. To confirm that differences in the amount of packing defects caused the difference in binding to monolayers and bilayers, we included dioleoylglycerol ((DOG), a conical lipid that increases packing defects (Vamparys et al., 2013)) in the simulation of binding to a bilayer membrane. In this case, P2 binding occurred in all instances of the simulation (Figure 2D).

Amphipathic Helix Binding to Phospholipid Surfaces In Vitro Depends on Large Hydrophobic Residues Recognizing Membrane Packing Defects

To test the MD findings experimentally, we assayed binding of labeled peptides to monolayer and bilayer membranes in several in vitro assay systems. We first generated artificial LDs, mixed them with synthesized, fluorescently labeled peptides and assayed binding to the surface by fluorescence microscopy. In these assays, and consistent with our simulations, P2, but not P1, bound to a large fraction of LDs (Figure 3A). When large hydrophobic residues (F, L or I) were introduced into P1 (“P1 Large Hydrophobics (LH)”, see Table 1), this peptide also bound to LDs (Figure 3A). Conversely, when we mutated most of the large hydrophobic residues in P2 to V (“P2 Small Hydrophobics (SH)”, see Table 1), binding was abolished (Figure 3A). In these experiments, we sometimes found large variations in the LD binding signal, most likely due to heterogeneous phospholipid coverage among different LDs in the sample. Supporting this interpretation, only the P2-binding population persisted in experiments in which the LD preparation lacked phospholipids (i.e., artificial TG droplets without a phospholipid monolayer; data not shown).

Figure 3. Amphipathic Helix Sequences with Large Hydrophobic Residues Bind LDs and Packing Defect-Rich Membranes in In Vitro Systems.

Figure 3

Open in a new tab

(A) Binding of AHs with (P2, P1 LH) or without (P1, P2 SH) large hydrophobic residues to artificial lipid droplets. Emulsion droplets prepared from a mixture of triolein and phospholipids (POPC: DOPE: liver phosphatidylinositol (liver PI) 65: 27: 8) were incubated with Alexa488-labeled synthetic peptides and imaged by fluorescence confocal microscopy. Upper panel: bright field and confocal images of the droplets after incubation with Alexa488-P1 and -P2. The inset highlights the ring-shaped protein signal. Lower panel: boxplot representation of the fluorescence signal on droplets. Over 200 droplets per condition were quantified in each of two independent experiments. Scale bar, 50 μm (larger field), and 10 μm (inset).

(B) Binding of AH peptides to a membrane-LD system. GUVs (POPC: DOPE: liver PI 65: 27: 8) were incubated with an emulsion of triolein in buffer, and the resulting TG-containing GUVs were incubated with labeled AHs and imaged with fluorescence confocal microscopy. Top: schematic of the experimental protocol. Middle: Representative image of a TG-loaded GUV after incubation with Alexa488-P2. Confocal images show the peptide and phospholipid signals. Two membrane-embedded droplets are visible in the bright field image, however only the one in focus is visible in the confocal images. Bottom left: representative images for each peptide. The protein channel is shown (see also Figure S3A for more examples); Bottom right: quantification (mean and standard deviation) of the fluorescence signal at the surface of the LD and membrane. 10–40 GUVs were imaged in each of two independent experiments. Scale bar, 5 μm.

(C) Binding of AH peptides to liposomes of increasing curvature. Liposomes (POPC: DOPE: liver PI 65: 27: 8) of different curvatures were incubated with NBD-labeled peptides and fluorescence emission spectra of the resulting mixtures were recorded. Left: normalized fluorescence emission spectra of NBD-P2 in the presence and absence of liposomes. Right: normalized fluorescence emission signal at 540 nm as a function of the extrusion pore diameter. Mean and standard deviation from three independent experiments with duplicate measurements are shown. Fluorescence values at each wavelength were normalized by the fluorescence at 540 nm in buffer averaged over the six measurements.

(D) Binding of Alexa488-P2 to TG-loaded GUVs of varying surface tensions. Same as (B) except GUVs were aspirated in micropipettes to change their tension. Top left: schematic of the experimental protocol. Arrows represent aspiration pressure. Bottom left: representative images of a GUV aspirated to two different tensions. Scale bar, 5 μm (merge), 2 μm (inlay). Right: quantification of the fluorescence in the protein channel on the monolayer and bilayer parts of the GUV as a function of membrane tension. Each different marker corresponds to a different GUV (N=5). Quantification of fluorescence in the lipid channel is shown in Figure S3B.

Table 1. Amphipathic Helices Studied in this Manuscript.

Shown are the amino acid sequence, a summary of cellular binding data, and references to publications where each sequence was investigated.

# Protein (Domain) Organism Sequencea Binding Indexb Reference
Mean Standard deviation
1 CCTα (M-domain) H. Sapiens HLQERVDKVKKKVKDVEEKSKEFVQKV……EEKSIDLIQKWEEKSREFIGSFLEMFG 2.6 1.5 (Krahmer et al., 2011)
2 CCTα (M-SH) HLQERVDKVKKKVKDVEEKSKEVVQKV……EEKSVDLVQKVEEKSREVVGSVLEMVG 1.2 0.2
3 CCTα (M-LH) HLQERFDKLKKKFKDFEEKSKEFFQKI……EEKSIDLIQKWEEKSREFIGSFLEMFG 11.7 10.8
4 CCTα (P1) HLQERVDKVKKKVKDVEEKSKEFVQKV 1.0 0.1
5 CCTα (P2) VEEKSIDLIQKWEEKSREFIGSFLEMFG 1.9 0.5
6 CCTα (P1 LH) HLQERFDKLKKKFKDFEEKSKEFFQKI 2.3 0.7
7 CCTα (P2 SH) VEEKSVDLVQKVEEKSREVVGSVLEMVG 1.0 0.1
8 CCTα (M-CR1) HLQERVDAVKGAVKDVEAGSKEFVQKV……EAGSIDLIQKWGAKSRAFIGSFLEMFG 5.6 2.9
9 CCTα (M-CR2) HLQERVAGVGAAVGAVEGASGAFVQAV……AAGSIGLIQKWGGASREFIGSFLAMFG 5.2 2.6
10 Lsd1 (H6-7) D. Melanogaster MSKEAIHVLFYAAELIATDPKQAVQKA……KELWVYLS 3.7 1.4 (Arrese et al., 2008; Lin et al., 2014)
11 Lsd1 (H6-7 SH) MSKEAVHVLVVAAELVATDPKQAVQKA……KELVVVLS 1.2 0.1
12 ArfGAP1 (ALPS) H. Sapiens FLNNAMSSLYSGWSSFTTGASRFAS 1.9 0.3 (Bigay et al., 2005)
13 ArfGAP1 (ALPS SH) VLNNAMSSLVSGVSSVTTGASRVAS 1.0 0.1
14 Core (D2) HCV NLGKVIDTLTCGFADLMGYIPLVGAPL……GGAARALAHGVRVLEDGVNY 4.5 1.8 (Boulant et al., 2006)
15 GMAP 210 H. Sapiens MSSWLGGLGSGLGQSLGQVGGSLASLT……GQISNFTKDML 3.2 1.0 (Drin et al., 2007)
16 Magainins (Magainin-2) X. Laevis GIGKFLHSAKKFGKAFVGEIMNS 2.4 0.5 (Seelig, 2004)
17 Mastoparan X V. Xanthoptera INWKGIAAMAKKLL 2.1 0.4 (Seelig, 2004)
18 Arf1 H. Sapiens MGNIFANLFKGLFGKKE 1.8 0.4 (Kahn et al., 1992)
19 Sar1p S. Cerevisiae MAGWDIFGWFRDVLASLGLWNKH 2.2 0.5 (Lee et al., 2005)
20 Kes1p S. Cerevisiae SSSWTSFLKSIASFNGDLSSLSA 1.2 0.1 (Drin et al., 2007)
21 Nup133 H. Sapiens LPQGQGMLSGIGRKVSSLFGILS 1.3 0.2 (Drin et al., 2007)
22 Epsin H. Sapiens TSSLRRQMKNIVH 1.0 0.1 (Ford et al., 2002)
23 Endophilin A-I N.Norvegicus MSVAGLKKQFHKATQKVSEKV 0.9 0.1 (Farsad et al., 2001)
24 Atlastin D. Melanogaster FGGKLDDFATLLWEKFMRPIYHGCMEK 2.1 0.4 (Liu et al., 2012)
25 Yop1p S. Cerevisiae ALPQTGGARMIYQKIVAPLTDRYILR 1.1 0.1 (Brady et al., 2015)
26 Curt1A A. Thaliana SIDTNELITDLKEKWDGL 1.0 0.1 (Armbruster et al., 2013)
27 MinD B. Subtilis VLEEQNKGMMAKIKSFFGVRS 1.6 0.2 (Szeto et al., 2003)
28 SpoVM B. Subtilis MKFYTIKLPKFLGGIVRAMLGSFRKD 2.3 0.4 (Gill et al., 2015)
29 FtsA E. Coli GSWIKRLNSWLRKEF 1.6 0.2 (Pichoff and Lutkenhaus, 2005)
30 Apolipoprotein A-I H. Sapiens VLESFKVSFLSALEEYTKKLNTQ 1.4 0.1 (Mitsche and Small, 2013)
31 Apolipoprotein C-I H. Sapiens ALDKLKEFGNTLEDKARELISRI 1.2 0.1 (Meyers et al., 2012)
Open in a new tab
a

Mutated residues are underlined.

b

Binding index is a measure of the enrichment of a construct on LDs in cells and is defined in the Methods section.

We next established an assay system in which peptides could bind either bilayer or monolayer membranes. For this, we generated giant unilamellar vesicles (GUVs) and subsequently incorporated TG droplets into the GUV bilayers (Figure 3B, (Ben M’barek et al., 2017; Thiam et al., 2013)). We then incubated the TG-loaded GUVs with fluorescently labeled peptides. Fluorescence imaging of the system revealed that P2 bound to TG-GUVs and was enriched on the portion of the GUV consisting of a monolayer covering an embedded TG droplet (Figure 3B). In contrast, P1 exhibited only very weak binding to the monolayer part (and did not bind to the bilayer part). P2 SH did not bind the monolayer nor the bilayer in these in vitro reconstitutions, indicating that large hydrophobic residues are necessary for amphipathic helix binding. Conversely, P1 LH bound in the same way as P2, further indicating that large hydrophobic residues are sufficient to mediate LD binding in the context of an amphipathic helix (Figure 3B; and see more images in Figure S3A).

Another prediction from our simulation models is that increases in phospholipid packing defects in bilayer membranes will increase protein binding to bilayers (Figure 2D). To test this model experimentally, we modulated packing defects in liposomes and tested binding of synthesized peptides harboring environmentally sensitive fluorophores. For these experiments, we varied the liposome diameters to generate packing defects. Previous studies showed that the curvature of small liposomes (<200 nm diameter) leads to phospholipid packing defects due to a mismatch between the curvature of the vesicle and the curvature preferred by the membrane lipids (Bigay and Antonny, 2012). When testing binding of P2 to bilayer membranes, we found low-level binding to flat membranes, such as large diameter vesicles (200 nm radius; Figure 3C). However, smaller liposomes, with higher number of packing defects, bound fluorescently labeled P2 peptide more efficiently (Figure 3C). We also found that, in addition to P2, the P1 mutant with large hydrophobic residues bound preferentially to highly curved membranes, whereas P1 and the P2 mutant in which the large hydrophobic residues were replaced with smaller residues did not bind to membranes of any curvature (Figure 3C).

Finally, our model predicts that increasing the density of packing defects on the monolayer surface results in increased protein binding. To test this prediction, we incubated LD-loaded GUVs with fluorescently-labeled P2, and aspirated these GUVs in a micropipette to increase their surface tension (Evans and Rawicz, 1990) and, as a result, the surface tension of the monolayer (Figure 3D). This increase in tension is expected to increase the area per phospholipid of the monolayer, i.e. to increase packing defect density. In these experiments, the peptide fluorescent signal on the monolayer increased with membrane tension (Figure 3D). Additionally, the fluorescent signal on the GUV membrane, although always lower than the signal on the monolayer, also increases with surface tension. This is consistent with our model, as stretching the GUV membrane also increases the density of bilayer packing defects. However, while the monolayer tensions in this experiment are close to the estimated tensions of cellular LDs (Ben M’barek et al., 2017), the bilayer tensions are orders of magnitudes above physiological tensions (Gauthier et al., 2012). As a consequence, the observed binding of the P2 peptide to the membrane bilayer is likely non-physiological. As a control, we measured the fluorescent signal of a lipid tracer on the monolayer and observed slightly decreased signal with increased membrane tension (Figure S3B).

Large Hydrophobic Residues Are Critical for Amphipathic Helix Binding to LDs in Cells

We next tested whether large hydrophobic residues are also crucial to detect and bind the phospholipid packing defects on the LD surface in cells. We expressed a series of variants of the human CCTα M-domain fused to mCherry in Drosophila S2 cells and assayed these proteins for LD binding. We first mutated eight of the large hydrophobic residues to valines in the context of the full-length M-domain (“M-SH”, see Table 1). Consistent with our hypothesis, these mutations abolished LD binding (Figures 4A and 4B). Conversely, increasing the number of large hydrophobic residues by mutating all six valines to larger hydrophobic residues (F, L or I, resulting in “M-LH”, see Table 1) increased LD binding (Figures 4A and 4B). Furthermore, and consistent with our simulation and in vitro data, the P1 fragment by itself did not bind LDs in Drosophila S2 cells, whereas P2 did bind LDs (Figures 4A and 4B). The binding of P2 depended on its large hydrophobic residues, as P2 SH did not bind LDs in cells (Figures 4A and 4B). In contrast, the P1 LH protein did bind LDs, indicating that large hydrophobic residues in the context of this amphipathic helix are sufficient to mediate LD binding in cells. To assess if electrostatic interactions play a role in LD binding, we also mutated charged residues of the CCTα M-domain (“M-Charge Reduction (CR) 1”, i.e. five positively and five negatively charged residues mutated to A and G and “M-Charge Reduction (CR) 2”, i.e. 10 positively and 10 negatively charged residues mutated to A and G; see Table 1) and found no differences in LD binding (Figure 4A and 4B).

Figure 4. Large Hydrophobic Residues Are Crucial for LD Targeting of the M-domain in Cells.

Figure 4

Open in a new tab

(A) Analysis of the LD-targeting ability of a series of M-domain mutants. Drosophila cells were transfected with mCherry-tagged constructs and incubated 14-18 hours with 0.5 mM oleic acid. LDs were stained with BODIPY. Representative images are shown. Scale bar, 5 μm (merge), 1 μm (inlay).

(B) Quantification of the protein signal on droplets. Data are represented as mean + SD. At least 10 cells were analyzed in each independent experiment.

To test whether large hydrophobic residues are a general feature of amphipathic helices that mediate LD binding, we mutated these residues in the amphipathic helices of the perilipin-like Lsd1 (Arrese et al., 2008; Lin et al., 2014) and in the ArfGAP1 ALPS motif. Similar to the findings with the CCTα M-domain, we found that the large hydrophobic residues were required for LD binding for both helices (Figure 5A and B).

Figure 5. Binding of Amphipathic Helices to Cellular Lipid Droplets Correlates with the Number of Large Hydrophobic Residues.

Figure 5

Open in a new tab

(A) Analysis of the LD-targeting ability of WT and “small hydrophobic residues” (SH) mutants of ArfGAP1 ALPS and Lsd1 H6-7.

(B) Quantification of the protein signal on droplets. Data are represented as mean + SD. Each construct was included in at least two independent experiments, and least 10 cells were analyzed in each experiment.

(C) Analysis of the LD-targeting ability of a range of amphipathic helices from non-LD proteins. More examples are shown in Figure S4. ApoA-I: Apolipoprotein A-I.

(A) and (C) Drosophila cells were transfected with mCherry-tagged constructs and incubated 14-18 hours with 0.5 mM oleic acid. LDs were stained with BODIPY. Representative images are shown. Scale bar, 5 μm (merge), 1 μm (inlay). (D) Correlation between binding index and a range of amphipathic helix physicochemical properties. Each data point corresponds to a different construct. Data for the full set of amphipathic helices analyzed in this paper. The list of amphipathic helices and associated binding indices can be found in Table 1.

Promiscuous binding of amphipathic helices containing large hydrophobic residues to cellular lipid droplets

Amphipathic helices are common bilayer membrane-targeting motifs and are found in many proteins that do not normally target to LDs (Drin and Antonny, 2010). We tested whether our findings more broadly applied to such amphipathic helices by expressing a variety of amphipathic helices, derived from different proteins, in S2 cells and analyzing their binding to LDs (constructs 14 to 31 in Table 1). Most of these AHs (constructs 15 to 29) normally bind to bilayer membranes, HCV core (construct 14) binds to LDs (Boulant et al., 2006), and apolipoproteins A-I and C-I bind to lipoproteins (Segrest et al., 1992). Representative fluorescence images for these experiments are shown in Figure 5C. These constructs displayed a range of localization patterns, with some of them localizing to LDs to varying extents, whereas some (e.g., the AHs from Yop1 and endophilin) were not detectable at the LD surface (Figure 5C). We analyzed various features of the LD targeting sequences that were tested in cells to determine how they correlated with the amount of LD binding. These features included the amount of net charge, the hydrophobic moment (Eisenberg et al., 1982), and the number of large hydrophobic residues in a targeting domain. Among these variables, only the number of large hydrophobic residues correlated with LD binding in cells (Figure 5D). Specifically, amphipathic helices with more than 5 large hydrophobic residues bound LDs, with their binding index being proportional to the number of these residues, independent of the length of the targeting motif (Figure 5D).

DISCUSSION

We show here that the monolayer surfaces of LDs possess distinctive properties that enable the binding of cytosolic amphipathic helix-containing proteins. A combination of MD simulation and in vitro and cell-based experimental approaches consistently validate a model in which relatively large and frequent packing defects on the LD surface are recognized by large hydrophobic residues of AH-containing proteins to allow LD targeting and binding. Specifically, our simulations revealed that the underlying neutral lipids enable monolayer packing defects (defined as regions where the hydrocarbon groups of phospholipids or neutral lipids are accessible to the aqueous cytoplasm) to occur relatively frequently, thereby imparting distinctive surface properties to LDs. In particular, neutral lipids are seen to intercalate between the phospholipid side chains. The observation of such mixing between neutral lipids and phospholipid chains in the present and other MD simulations (Bacle et al., 2017) is consistent with prior studies by Hamilton and others that showed that neutral lipids solubilized into phospholipid liposomes are present in the membrane leaflets, aligning their acyl chains with those of the phospholipids (Hamilton and Small, 1981).

Our studies and prior work suggest a model for how class II (cytosolic) proteins interact with LDs (see Figure 6). In step I, an unfolded peptide sequence interacts with the LD surface and partitions to the interfacial region. For initial association, charges in the peptide are likely important, as our simulations indicate that the number of positively charged residues in a peptide somewhat correlates with the speed of association with the negatively charged membranes (Table S2). Consistent with this, supra-physiological concentrations of negatively charged phospholipids (30% phosphatidic acid) promote binding of the P1 and P1 LH peptides which bear positive net charges (data not shown). However, this attraction is not enough to enable binding to LDs with physiologically relevant surface charges. Instead, the insertion of large, hydrophobic residues into large packing defects on the LD surface appears to be most critical. Consistent with this, we found that mutating the large hydrophobic residues to small ones abolished binding of the AH proteins in cells, whereas charge reductions had no significant effect. This step of protein binding to the LD surface is fast, and hence was detectable on the timescale of our MD simulations. It is also likely reversible, as we found cases in which the peptide initially adsorbed, but subsequently fell off the LD surface (Figure S2). Evidence suggests that large hydrophobic residues are also relevant for LD binding for class II proteins that do not possess AH sequences. For example, the binding of CGI-58 to LDs requires tryptophan residues, although not in the context of an AH (Boeszoermenyi et al., 2015). This mechanism (association of hydrophobic residues in AHs with phospholipid packing defects) is similar to the previously described mechanism of AH binding to highly-curved membranes (Cui et al., 2011; Vanni et al., 2013), as discussed below.

Figure 6. Proposed Thermodynamic Cycle for the Partitioning of Amphipathic Helices at the LD Surface.

Figure 6

Open in a new tab

The amphipathic helix sequence is initially unfolded in the cytosol. Following insertion of a large hydrophobic residue into a packing defect at the lipid droplet surface (I), the amphipathic helix folds in the interfacial region that separates the bulk aqueous phase from the hydrocarbon core (II). High surface pressure at the crowded droplet surface promotes desorption of the folded helix (III), which subsequently unfolds in the cytosol (IV).

In step II of the binding reaction, the amphipathic helix folds in the interfacial region between the hydrophobic LD and the aqueous cytoplasm. The main driving force for this step is likely the need to satisfy the hydrogen bonding requirements of the peptide backbone. Folding into an α-helix lowers the free energy of peptide bond partitioning to the hydrophobic phase due to the intramolecular hydrogen bonds (Fernandez-Vidal et al., 2007; Ladokhin and White, 1999). This step is much slower than step I, and we did not observe complete folding in the microsecond timescale of our MD simulations. However, in one instance, we observed the beginning of the folding reaction, and this example suggests that folding starts with the inserted residues and spreads from there to form the full AH (data not shown). This slow step provides a large amount of free energy (Almeida et al., 2012; Cui et al., 2011; Fernandez-Vidal et al., 2007; Ladokhin and White, 1999) and essentially renders the pathway for binding irreversible in biological systems. Consistent with this notion, we previously observed in FRAP experiments that AH-bound proteins, such as CCT1, have a long half-life on the LD surface (Krahmer et al., 2011). Thus, displacing proteins from the LD surface (step III of the interaction cycle) does not follow a simple reversal of the binding reaction, but instead is due to “bumping-off” when proteins collide in conditions where the LD surface is more crowded (Kory et al., 2015). As described in (Kory et al., 2016), proteins fall off the LD surface if the energy of the collision is greater than the binding energy of the folded AH to the LD surface (Kory et al., 2015). Therefore weaker binders are more easily displaced from the LD surface. Once the protein falls off the LD surface, the peptide unfolds, as the entropic preference for the unfolded configuration dominates. The enthalpic driving force to fold through forming 1-4 peptide backbone hydrogen bonds is significantly decreased, as water can also satisfy these backbone hydrogen-bonding sites (Adrover et al., 2012; Bianco and Franzese, 2015; Camilloni et al., 2016).

Consistent with the notion that AHs bind LDs and highly-curved membranes in a similar fashion, we found that sequences from proteins with various cellular localization, such as GMAP210 that is localized primarily to the Golgi (Doucet et al., 2015), are sufficient to mediate binding to LDs in cells. While this finding is surprising, it shows that there is no strict targeting specificity for LDs encoded in the amphipathic helix by itself. Instead, the LD surface is promiscuous for binding proteins with amphipathic helices, provided that they contain large hydrophobic residues. Such residues are commonly found in most amphipathic helices, including the ones of CCTα, GMAP210 and ArfGAP1. This raises an interesting question: how are some amphipathic helix-containing proteins, such as GMAP210, excluded from LDs in a normal cellular context? The answer to this is currently unknown, but it likely involves combining amphipathic helices with specificity determinants, such as other membrane binding domains or protein-protein interaction motifs. Examples of this include CCTα, which has a nuclear localization domain in addition to a LD-binding amphipathic helix and which in the absence of LDs localizes to the nucleus in Drosophila cells, arguing both signals compete with each other in determining the localization of the protein (Krahmer et al., 2011; Lagace and Ridgway, 2005). Also consistent with this hypothesis, GMAP210 is a long coiled-coil protein, which contains several domains that can interact with membranes or proteins, and that influence its localization in cells (Drin et al., 2008; Sinka et al., 2008). Removing these additional specificity determinants in our experiments revealed the promiscuity of LD binding by AHs. In addition to these mechanisms directing proteins away from LDs, binding strength and thus resistance to molecular crowding likely play a role in determining subcellular localization (Kory et al., 2015). Many instances of genuine LD proteins, such as perilipin proteins, rely on a string of amphipathic helices. Addition of multiple amphipathic helices leads to higher avidity, stronger binding, and thus a higher resistance to molecular crowding (Kory et al., 2015). Also consistent with this, CCTα’s M-domain is an unusually long amphipathic helix (54 amino acids).

Our findings on the promiscuity of amphipathic helices in binding the LD surface may have implications for physiology and disease. For example, human mutations of the ER protein seipin/BSCL2, involved in LD formation, apparently alters properties of the LD surface, leading to abnormal, enhanced recruitment of AH-containing proteins (Grippa et al., 2015). This suggests that seipin deficiency promotes increased packing defects on LD surfaces. Moreover, in pathological conditions where LDs over-accumulate, for instance during obesity-related hepatic steatosis or during the development of foam cells in atherosclerotic plaques, a large amount of LD surfaces with packing defects may effectively compete for amphipathic helix containing proteins normally not binding LDs, thus possibly contributing to cellular dysfunction (e.g., via mistargeting of proteins). Therefore, a better understanding of the mechanisms for protein targeting to LD surfaces is predicted to lead to an enhanced molecular understanding of disease pathogenesis.

STAR Methods

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Tobias C. Walther (twalther@hsph.harvard.edu).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Cell Lines and Cell Culture

The Drosophila cells used in this study are from the S2R+ cell line (sex: male) and were provided by Prof. Norbert Perrimon (Harvard Medical School). Cells were cultured at 26°C in Schneider’s Drosophila Medium (LifeTechnologies) containing 10% fetal bovine serum, 50 units/ml penicillin and 50 mg/ml streptomycin.

METHOD DETAILS

Molecular Dynamics Simulations

General simulation parameters

All molecular dynamics (MD) simulations were performed using the CHARMM36 force field (Best et al., 2012; Klauda et al., 2010; MacKerell et al., 1998). Simulations were performed using the GROMACS (Abraham et al., 2015) MD simulation engine unless otherwise noted. Simulations were integrated with a 2 fs timestep. Van der Waals (VDW) interactions were switched to zero between 1.0 and 1.2 nm, and the Particle Mesh Ewald (Essmann et al., 1995) was used to evaluate electrostatic interactions. All bonds to hydrogen were constrained using the LINCS algorithm (Hess, 2008). The Nose-Hoover thermostat (Hoover, 1985) was used to maintain temperature at 310 K with a coupling time constant of 1.0 ps. For simulations with constant pressure, the Parrinello-Rahman barostat (Parrinello and Rahman, 1981) was used to maintain a pressure of 1.0 bar with a compressibility of 4.5 × 10−5 and a coupling time constant of 5.0 ps. The pressure was maintained semi-isotropically in all membrane simulations, while isotropic pressure coupling was used in protein only simulations. A list of simulations performed is provided in Table S1. The cumulative length of all the simulations was 41.1 μs.

Building of bilayer and lipid droplet membranes

Bilayer systems were built using the CHARMM-GUI membrane builder (Jo et al., 2008; Wu et al., 2014) with the CHARMM36 lipid force field (Klauda et al., 2010). Bilayers of composition 65:27:8 POPC:DOPE:SAPI and 85:15 DOPC:DOG were built of size 11×11 nm2 and solvated with a 2.2-nm layer of TIP3P water on both sides of the membrane as well as a salt concentration of 150 mM NaCl. The system was relaxed and equilibrated using the equilibration procedure (Wu et al., 2014), designed by the CHARMM-GUI using the GROMACS simulation engine. Briefly, harmonic positional restraints were placed on select lipid atoms, and harmonic dihedral restraints were placed to maintain the cis orientation of double bonds in the lipid tails. These restraints were slowly relaxed through multiple steps of equilibration. For lipid droplet simulations, the neutral lipid core was constructed from a box 1:1 cholesteryl oleate:triolein. The neutral lipid box was built using Packmol (Martinez et al., 2009) to match the x-y dimensions of the POPC:DOPE:SAPI bilayer, with different z values (4, 6, and 8 nm). The neutral lipids were equilibrated using the same equilibration procedure described above. The neutral lipid box was then inserted between the two leaflets of the bilayer, after which another round of equilibration was performed.

Determining peptide structure

Each peptide structure (ALPS, P1, P2, M-domain) was first generated as an idealized alpha-helix using Molefacture in VMD (Humphrey et al., 1996). Each peptide was solvated using GROMACS in a water box with at least 1 nm of padding on each side of the peptide. The GROMACS genion tool was used to add ions to the system such that the system was neutralized and had a concentration of 150 mM NaCl. The system was then minimized using steepest-descent until the maximum force was less than 1000.0 kJ/mol/nm. A total of 100 ps of equilibration was then performed in the constant NVT ensemble. The system was then heated up to 500 K, after which 10 ns of simulation was performed to unfold the protein. The system was then simulated at 310 K in the constant NPT ensemble for 5 ns.

Building peptide-membrane simulations

Using VMD, the TIP3P molecules and salt ions were removed from the membrane simulation box. To build each of the peptide-membrane systems, a random structure from the final peptide equilibration run was chosen. A different random structure was chosen for each of the four replicas of each peptide-membrane simulation. The randomly chosen peptide conformation was then placed at least 2 nm away from the membrane. The system was then solvated in VMD with TIP3P water and 150 mM NaCl. The system was then equilibrated using the CHARMM-GUI procedure described above. The systems were then simulated for 750–1000 ns for production runs.

Production Simulations in ANTON

Simulations were also run on the ANTON (Shaw et al., 2009) supercomputer, which is highly specialized for long time MD simulations. Simulations were integrated with a 2-fs timestep using RESPA (Tuckerman et al., 1992), where long range electrostatic interactions were evaluated every three timesteps. VDW interactions were switched to zero between 1.0 and 1.2 nm, and Particle Mesh Ewald (Essmann et al., 1995) was used to evaluate electrostatic interactions. All bonds to hydrogen were constrained using the SHAKE algorithm (Ryckaert et al., 1977). The Nose-Hoover thermostat (Hoover, 1985) was used to maintain temperature at 310 K with a coupling time constant of 1.0 ps. The MTK barostat (Martyna et al., 1994) maintained a pressure of 1.0 atm semi-isotropically in all membrane simulations.

Lipid packing defects

Packing defects were calculated using a described procedure (Cui et al., 2011). In brief, the solvent accessible surface area of the hydrophobic tails was calculated using VMD (Humphrey et al., 1996) using a probe radius of 0.3 nm. The points composing this surface are projected onto the membrane plane. The surface is divided into bins of size 0.16 nm2, and each of these points are placed into the bins. The binned points are considered to be part of the same defect cluster if they are within 0.25 nm of each other. Defect lifetimes were calculated as follows. A defect was considered to have lived from one frame to the next if any of the bins composing the defect were the same in frame i and frame i+1. If a defect split into two or more defects, each smaller defect was followed in time, and the longest lifetime of the split defects was considered to be contributing to the lifetime of the original large defect. A histogram of defect sizes was tabulated for defects with lifetimes longer than 5 ns. This histogram was then fit to an exponential decay = where x is the defect size, p(x) is the probability of finding a defect of size x with a lifetime greater than 5 ns, c is a constant, and a is the decay rate in units of nm−2.

Peptides Design

Peptides were synthesized by Bio-Synthesis Inc. An N-terminal GG linker was included in the sequence of all peptides and peptides were conjugated with fluorescent dyes at their N-terminus. Conjugation with NBD-Chloride was performed by Bio-Synthesis. Peptides conjugated with Alexa Fluor 488 C5 Maleimide (Alexa488) had an extra C at their N-terminus. The reaction was performed according to the manufacturer’s instructions. Unreacted peptide and dye were removed by reverse phase chromatography.

Plasmids Construction and Transfection

Synthetic gene fragments were purchased from Integrated DNA technologies and cloned into either the pENTR™/SD/D-TOPO or pENTR™/D-TOPO vector. The entry clone was subcloned into the pACherryW (all constructs except for the Flag sequence) or pAWCherry (Flag sequence) vector (Guo et al., 2008).

Transfections were performed with Effectene Transfection Reagent (Qiagen) according to the manufacturer’s instructions. Transfections were done 24–36 hr before the experiment. Only the AH fragment was expressed except in the case of ArfGAP1, where the construct comprised amino acids 137-237 of human ArfGAP1 (isoform 1). Constructs 18 to 31 (see Table 1) had N-terminal G-S linkers to bring them to the same size (27 amino acids).

LDs induction

Cells were incubated 14-18 hr with 0.5 mM oleic acid (complexed to bovine serum albumin at a 3:1 molar ratio).

Microscopy

Live cell imaging was performed as described using high numerical aperture 60× or 100× objectives (Wang et al., 2016). For in vitro experiments, imaging was performed with a spinning disk confocal (Yokogawa CSU22) set up on a Nikon Eclipse Ti inverted microscope. Illumination was performed with 488 and 561 nm laser lines, and detection with an imagEM EM-CCD camera (Hamamatsu). TG-loaded GUVs: imaging was performed with a 60X ApoTIRF 1.49 NA objective (Nikon). Artificial LDs: LDs float at the top of the observation chamber and were imaged using a lower magnification/ longer working distance objective (Plan Apo VC 20X, 0.75 NA, Nikon).

In Vitro Assays

For all in vitro experiments we used a phospholipid composition mimicking the composition of the mammalian LD monolayer (1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC): 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE): L-α-phosphatidylinositol from liver, bovine (liver PI) 65: 27: 8 (Bartz et al., 2007)). In experiments with GUVs, 0.05-0.1 mol % 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) (Rhodamine PE) was included in the lipid composition. All experiments were performed at room temperature.

Artificial LDs

Phospholipids in chloroform were added to triolein at a 0.5% molar ratio. The solvent was evaporated under a stream of nitrogen and further by placing the vial in a desiccator for 3–4 h. This phospholipids and oil mixture was added to buffer (20 mM Tris pH 7.5, 100 mM NaCl) at a 0.2% ratio (vol/vol), and the solution was vortexed for a few seconds and sonicated for about a minute in a bath sonicator. This emulsion was incubated with 1 μM Alexa488-labeled peptide for 5 min. before imaging by confocal microscopy.

Liposomes

Phospholipids in chloroform were dried under a stream of nitrogen and the vial was placed in a desiccator for about half an hour to remove traces of solvent. The lipid film was rehydrated in buffer (20 mM Tris pH 7.5, 100 mM NaCl) for 1 h (final lipid concentration: 2.5 mM), and the sample was processed through five cycles of freezing and thawing and subsequently extruded 11 times through polycarbonate filters with pore diameters of either 30, 50, 100 or 200 nm (Avanti Polar Lipids). The resulting liposome solutions were incubated with 1 μM NBD-labeled peptide for 5 min, and fluorescence emission spectra were recorded on an Infinite M200 microplate reader (TECAN), with the following settings: excitation wavelength: 490 nm and bandwidth: 10 nm, emission range: 520–650 nm, bandwidth: 20 nm and step size: 5 nm.

TG-loaded GUVs

Phospholipids in chloroform were spread on indium tin oxide–coated glass slides. GUVs were grown by electroformation (Angelova et al., 1992) in 600 mM sucrose using a Vesicle Prep Pro (Nanion) with the following settings: frequency 10 Hz, amplitude 1.4V, temperature 23°C, for 1 h. GUVs were incubated with an oil-in-water emulsion (5% (vol/vol) triolein in buffer (20 mM Tris pH 7.5, 100 mM NaCl, 400 mM glucose), prepared as described above) for 10 min. The mixture was allowed to settle for about 2 h to separate most of the oil droplets from the GUVs. The TG-loaded GUVs were collected at the bottom of the tube and incubated with 1 μM Alexa488-labeled peptide for 5 min. before imaging by confocal microscopy.

Micropipette experiments

TG-loaded GUVs were prepared as described above and incubated with 1 μM Alexa488-labeled P2 for 5 min. Micropipette aspiration was performed following the protocol described in (Prévost et al., 2017). The GUV tension was calculated using the following expression (Kwok and Evans, 1981): = Δ pip⁄2 1 − pip⁄ GUV, where Δ is the aspiration pressure, pip is the radius of the pipette opening and GUV is the radius of the GUV.

QUANTIFICATION AND STATISTICAL ANALYSIS

The statistical details of the experiments can be found in the figure legends.

Artificial LDs

Images were quantified using CellProfiler software (Carpenter et al., 2006). For each droplet, the integrated intensity (background-subtracted) over the droplet-occupied region in the Alexa488 channel was normalized by the surface area of the droplet.

TG-loaded GUVs

Images were quantified using custom Matlab (MathWorks) software. The mean intensity (background-subtracted) in the protein and lipid channels along a portion of the contour of the monolayer and bilayer part of the GUVs were quantified.

Cell experiments

Images were quantified using FIJI software (Schindelin et al., 2012). An automatic threshold was applied to the BODIPY channel to identify the LD-occupied region. The mean intensity of the mCherry signal in that region was measured. This value was divided by the mean intensity of the signal in a distinct region of the cytoplasm to account for differences in expression levels. Finally, for each construct the average of this ratio was divided by the equivalent average for Flag-mCherry, considered a baseline value. The final quantity is referred to as “binding index” in Figures 4 and 5.

Supplementary Material

1

Movie S1. Related to Figure 3: M-domain binding the monolayer surface. As the peptide approaches the monolayer surface, a number of residues insertions are attempted. Eventually, F53 inserts below the phosphate plane and anchors the peptide to the surface. White residues are hydrophobic, purple non-hydrophobic. As the residues approach the phosphate plane, they are shown in space filling representation.

Download video file (75.5MB, mp4)
2

Highlights.

  • Lipid droplet surfaces are characterized by phospholipid packing defects

  • Packing defects recruit amphipathic helices to the lipid droplet surface

  • Large, hydrophobic residues of amphipathic helices initially bind packing defects

  • In isolation, many amphipathic helices accumulate on lipid droplets

Acknowledgments

We thank Guan-Yue Chen, Jeremy Furtado and Maria-Jesus Olarte for help with experiments, Mijo Simunovic and Patricia Bassereau for providing micropipettes, members of the Farese & Walther and Voth laboratories for comments on the manuscript, and Gary Howard for editorial assistance. This work was supported by 2R01GM097194 (to T.C.W.) the Mathers foundation (to T.C.W). T.C.W is an investigator of the Howard Hughes Medical Institute. Anton computer time was provided by the Pittsburgh Supercomputing Center (PSC) through grant R01GM116961 from the National Institutes of Health. The Anton machine at PSC was generously made available by D.E. Shaw Research. The MD simulation in this research were also performed in part on the Comet supercomputer at the San Diego Supercomputer Center (SDSC) with resources provided by the Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by National Science Foundation grant number ACI-1548562. The MD simulations were also completed in part with resources provided by the University of Chicago Research Computing Center.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Author contributions: CP, MES, NK, TCW, RVF and GAV designed the research, CP, MES, NK and QL conducted the experiments, CP and MES analyzed the data and CP, MES, GAV, RVF and TCW wrote the paper.

Declaration of Interests: The authors declare no competing interests.

References

  1. Abraham MJ, Murtola T, Schulz R, Páll S, Smith JC, Hess B, Lindahl E. GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX. 2015;(1–2):19–25. [Google Scholar]
  2. Adrover M, Martorell G, Martin SR, Urosev D, Konarev PV, Svergun DI, Daura X, Temussi P, Pastore A. The role of hydration in protein stability: comparison of the cold and heat unfolded states of Yfh1. J Mol Biol. 2012;417:413–424. doi: 10.1016/j.jmb.2012.02.002. [DOI] [PubMed] [Google Scholar]
  3. Almeida PF, Ladokhin AS, White SH. Hydrogen-bond energetics drive helix formation in membrane interfaces. Biochimica et biophysica acta. 2012;1818:178–182. doi: 10.1016/j.bbamem.2011.07.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Angelova MI, Soléau S, Méléard P, Faucon JF, Bothorel P. Preparation of giant vesicles by external AC electric fields . Kinetics and applications. Progress in Colloid & Polymer Science. 1992;89:127–131. [Google Scholar]
  5. Armbruster U, Labs M, Pribil M, Viola S, Xu W, Scharfenberg M, Hertle AP, Rojahn U, Jensen PE, Rappaport F, et al. Arabidopsis CURVATURE THYLAKOID1 proteins modify thylakoid architecture by inducing membrane curvature. The Plant cell. 2013;25:2661–2678. doi: 10.1105/tpc.113.113118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Arrese EL, Rivera L, Hamada M, Mirza S, Hartson SD, Weintraub S, Soulages JL. Function and structure of lipid storage droplet protein 1 studied in lipoprotein complexes. Arch Biochem Biophys. 2008;473:42–47. doi: 10.1016/j.abb.2008.02.036. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Bacle A, Gautier R, Jackson CL, Fuchs PF, Vanni S. Interdigitation between Triglycerides and Lipids Modulates Surface Properties of Lipid Droplets. Biophys J. 2017;112:1417–1430. doi: 10.1016/j.bpj.2017.02.032. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Bartz R, Li WH, Venables B, Zehmer JK, Roth MR, Welti R, Anderson RG, Liu P, Chapman KD. Lipidomics reveals that adiposomes store ether lipids and mediate phospholipid traffic. Journal of lipid research. 2007;48:837–847. doi: 10.1194/jlr.M600413-JLR200. [DOI] [PubMed] [Google Scholar]
  9. Ben M’barek K, Ajjaji D, Chorlay A, Vanni S, Foret L, Thiam AR. ER Membrane Phospholipids and Surface Tension Control Cellular Lipid Droplet Formation. Developmental cell. 2017 doi: 10.1016/j.devcel.2017.05.012. [DOI] [PubMed] [Google Scholar]
  10. Best RB, Zhu X, Shim J, Lopes PE, Mittal J, Feig M, Mackerell AD., Jr Optimization of the additive CHARMM all-atom protein force field targeting improved sampling of the backbone phi, psi and side-chain chi(1) and chi(2) dihedral angles. J Chem Theory Comput. 2012;8:3257–3273. doi: 10.1021/ct300400x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Bianco V, Franzese G. Contribution of Water to Pressure and Cold Denaturation of Proteins. Physical review letters. 2015;115:108101. doi: 10.1103/PhysRevLett.115.108101. [DOI] [PubMed] [Google Scholar]
  12. Bickel PE, Tansey JT, Welte MA. PAT proteins, an ancient family of lipid droplet proteins that regulate cellular lipid stores. Biochimica et biophysica acta. 2009;1791:419–440. doi: 10.1016/j.bbalip.2009.04.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Bigay J, Antonny B. Curvature, lipid packing, and electrostatics of membrane organelles: defining cellular territories in determining specificity. Developmental cell. 2012;23:886–895. doi: 10.1016/j.devcel.2012.10.009. [DOI] [PubMed] [Google Scholar]
  14. Bigay J, Casella JF, Drin G, Mesmin B, Antonny B. ArfGAP1 responds to membrane curvature through the folding of a lipid packing sensor motif. Embo J. 2005;24:2244–2253. doi: 10.1038/sj.emboj.7600714. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Boeszoermenyi A, Nagy HM, Arthanari H, Pillip CJ, Lindermuth H, Luna RE, Wagner G, Zechner R, Zangger K, Oberer M. Structure of a CGI-58 motif provides the molecular basis of lipid droplet anchoring. The Journal of biological chemistry. 2015;290:26361–26372. doi: 10.1074/jbc.M115.682203. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Boulant S, Montserret R, Hope RG, Ratinier M, Targett-Adams P, Lavergne JP, Penin F, McLauchlan J. Structural determinants that target the hepatitis C virus core protein to lipid droplets. The Journal of biological chemistry. 2006;281:22236–22247. doi: 10.1074/jbc.M601031200. [DOI] [PubMed] [Google Scholar]
  17. Brady JP, Claridge JK, Smith PG, Schnell JR. A conserved amphipathic helix is required for membrane tubule formation by Yop1p. Proceedings of the National Academy of Sciences of the United States of America. 2015;112:E639–648. doi: 10.1073/pnas.1415882112. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Brasaemle DL, Dolios G, Shapiro L, Wang R. Proteomic analysis of proteins associated with lipid droplets of basal and lipolytically stimulated 3T3-L1 adipocytes. The Journal of biological chemistry. 2004;279:46835–46842. doi: 10.1074/jbc.M409340200. [DOI] [PubMed] [Google Scholar]
  19. Camilloni C, Bonetti D, Morrone A, Giri R, Dobson CM, Brunori M, Gianni S, Vendruscolo M. Towards a structural biology of the hydrophobic effect in protein folding. Sci Rep. 2016;6:28285. doi: 10.1038/srep28285. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Carpenter AE, Jones TR, Lamprecht MR, Clarke C, Kang IH, Friman O, Guertin DA, Chang JH, Lindquist RA, Moffat J, et al. CellProfiler: image analysis software for identifying and quantifying cell phenotypes. Genome biology. 2006;7:R100. doi: 10.1186/gb-2006-7-10-r100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Choudhary V, Ojha N, Golden A, Prinz WA. A conserved family of proteins facilitates nascent lipid droplet budding from the ER. The Journal of cell biology. 2015;211:261–271. doi: 10.1083/jcb.201505067. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Cui H, Lyman E, Voth GA. Mechanism of membrane curvature sensing by amphipathic helix containing proteins. Biophys J. 2011;100:1271–1279. doi: 10.1016/j.bpj.2011.01.036. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Doucet CM, Esmery N, de Saint-Jean M, Antonny B. Membrane Curvature Sensing by Amphipathic Helices Is Modulated by the Surrounding Protein Backbone. PloS one. 2015;10:e0137965. doi: 10.1371/journal.pone.0137965. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Drin G, Antonny B. Amphipathic helices and membrane curvature. FEBS Lett. 2010;584:1840–1847. doi: 10.1016/j.febslet.2009.10.022. [DOI] [PubMed] [Google Scholar]
  25. Drin G, Casella JF, Gautier R, Boehmer T, Schwartz TU, Antonny B. A general amphipathic alpha-helical motif for sensing membrane curvature. Nature structural & molecular biology. 2007;14:138–146. doi: 10.1038/nsmb1194. [DOI] [PubMed] [Google Scholar]
  26. Drin G, Morello V, Casella JF, Gounon P, Antonny B. Asymmetric tethering of flat and curved lipid membranes by a golgin. Science (New York, NY) 2008;320:670–673. doi: 10.1126/science.1155821. [DOI] [PubMed] [Google Scholar]
  27. Dunne SJ, Cornell RB, Johnson JE, Glover NR, Tracey AS. Structure of the membrane binding domain of CTP:phosphocholine cytidylyltransferase. Biochemistry. 1996;35:11975–11984. doi: 10.1021/bi960821+. [DOI] [PubMed] [Google Scholar]
  28. Eisenberg D, Weiss RM, Terwilliger TC. The helical hydrophobic moment: a measure of the amphiphilicity of a helix. Nature. 1982;299:371–374. doi: 10.1038/299371a0. [DOI] [PubMed] [Google Scholar]
  29. Essmann U, Perera L, Berkowitz ML, Darden T, Lee H, Pedersen LG. A smooth particle mesh Ewald method. The Journal of Chemical Physics. 1995;103:8577–8593. [Google Scholar]
  30. Evans E, Rawicz W. Entropy-driven tension and bending elasticity in condensed-fluid membranes. Physical review letters. 1990;64:2094–2097. doi: 10.1103/PhysRevLett.64.2094. [DOI] [PubMed] [Google Scholar]
  31. Farsad K, Ringstad N, Takei K, Floyd SR, Rose K, De Camilli P. Generation of high curvature membranes mediated by direct endophilin bilayer interactions. The Journal of cell biology. 2001;155:193–200. doi: 10.1083/jcb.200107075. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Fernandez-Vidal M, Jayasinghe S, Ladokhin AS, White SH. Folding amphipathic helices into membranes: amphiphilicity trumps hydrophobicity. J Mol Biol. 2007;370:459–470. doi: 10.1016/j.jmb.2007.05.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Ford MG, Mills IG, Peter BJ, Vallis Y, Praefcke GJ, Evans PR, McMahon HT. Curvature of clathrin-coated pits driven by epsin. Nature. 2002;419:361–366. doi: 10.1038/nature01020. [DOI] [PubMed] [Google Scholar]
  34. Fujimoto Y, Itabe H, Sakai J, Makita M, Noda J, Mori M, Higashi Y, Kojima S, Takano T. Identification of major proteins in the lipid droplet-enriched fraction isolated from the human hepatocyte cell line HuH7. Biochimica et biophysica acta. 2004;1644:47–59. doi: 10.1016/j.bbamcr.2003.10.018. [DOI] [PubMed] [Google Scholar]
  35. Gannon J, Fernandez-Rodriguez J, Alamri H, Feng SB, Kalantari F, Negi S, Wong AH, Mazur A, Asp L, Fazel A, et al. ARFGAP1 is dynamically associated with lipid droplets in hepatocytes. PloS one. 2014;9:e111309. doi: 10.1371/journal.pone.0111309. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Gauthier NC, Masters TA, Sheetz MP. Mechanical feedback between membrane tension and dynamics. Trends in cell biology. 2012;22:527–535. doi: 10.1016/j.tcb.2012.07.005. [DOI] [PubMed] [Google Scholar]
  37. Gill RL, Jr, Castaing JP, Hsin J, Tan IS, Wang X, Huang KC, Tian F, Ramamurthi KS. Structural basis for the geometry-driven localization of a small protein. Proceedings of the National Academy of Sciences of the United States of America. 2015;112:E1908–1915. doi: 10.1073/pnas.1423868112. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Gluchowski NL, Becuwe M, Walther TC, Farese RV., Jr Lipid droplets and liver disease: from basic biology to clinical implications. Nat Rev Gastroenterol Hepatol. 2017 doi: 10.1038/nrgastro.2017.32. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Grippa A, Buxo L, Mora G, Funaya C, Idrissi FZ, Mancuso F, Gomez R, Muntanya J, Sabido E, Carvalho P. The seipin complex Fld1/Ldb16 stabilizes ER-lipid droplet contact sites. The Journal of cell biology. 2015;211:829–844. doi: 10.1083/jcb.201502070. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Guo Y, Walther TC, Rao M, Stuurman N, Goshima G, Terayama K, Wong JS, Vale RD, Walter P, Farese RV. Functional genomic screen reveals genes involved in lipid-droplet formation and utilization. Nature. 2008;453:657–661. doi: 10.1038/nature06928. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Hamilton JA, Small DM. Solubilization and localization of triolein in phosphatidylcholine bilayers: a 13C NMR study. Proceedings of the National Academy of Sciences of the United States of America. 1981;78:6878–6882. doi: 10.1073/pnas.78.11.6878. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Hennere G, Havard L, Bonan B, Prognon P. Stability of cidofovir in extemporaneously prepared syringes. Am J Health Syst Pharm. 2005;62:506–509. doi: 10.1093/ajhp/62.5.506. [DOI] [PubMed] [Google Scholar]
  43. Hennere G, Prognon P, Brion F, Nicolis I. Molecular dynamics study of a phospholipid monolayer at a water/triglyceride interface: towards lipid emulsion modelling. Chem Phys Lipids. 2009;157:86–93. doi: 10.1016/j.chemphyslip.2008.10.002. [DOI] [PubMed] [Google Scholar]
  44. Hess B. P-LINCS: A Parallel Linear Constraint Solver for Molecular Simulation. J Chem Theory Comput. 2008;4:116–122. doi: 10.1021/ct700200b. [DOI] [PubMed] [Google Scholar]
  45. Hoover WG. Canonical dynamics: Equilibrium phase-space distributions. Phys Rev A Gen Phys. 1985;31:1695–1697. doi: 10.1103/physreva.31.1695. [DOI] [PubMed] [Google Scholar]
  46. Humphrey W, Dalke A, Schulten K. VMD: visual molecular dynamics. J Mol Graph. 1996;14:33–38. 27–38. doi: 10.1016/0263-7855(96)00018-5. [DOI] [PubMed] [Google Scholar]
  47. Jacquier N, Choudhary V, Mari M, Toulmay A, Reggiori F, Schneiter R. Lipid droplets are functionally connected to the endoplasmic reticulum in Saccharomyces cerevisiae. Journal of cell science. 2011;124:2424–2437. doi: 10.1242/jcs.076836. [DOI] [PubMed] [Google Scholar]
  48. Jo S, Kim T, Iyer VG, Im W. CHARMM-GUI: a web-based graphical user interface for CHARMM. J Comput Chem. 2008;29:1859–1865. doi: 10.1002/jcc.20945. [DOI] [PubMed] [Google Scholar]
  49. Kahn RA, Randazzo P, Serafini T, Weiss O, Rulka C, Clark J, Amherdt M, Roller P, Orci L, Rothman JE. The amino terminus of ADP-ribosylation factor (ARF) is a critical determinant of ARF activities and is a potent and specific inhibitor of protein transport. The Journal of biological chemistry. 1992;267:13039–13046. [PubMed] [Google Scholar]
  50. Kimmel AR, Sztalryd C. The Perilipins: Major Cytosolic Lipid Droplet-Associated Proteins and Their Roles in Cellular Lipid Storage, Mobilization, and Systemic Homeostasis. Annual review of nutrition. 2016;36:471–509. doi: 10.1146/annurev-nutr-071813-105410. [DOI] [PubMed] [Google Scholar]
  51. Klauda JB, Venable RM, Freites JA, O’Connor JW, Tobias DJ, Mondragon-Ramirez C, Vorobyov I, MacKerell AD, Jr, Pastor RW. Update of the CHARMM all-atom additive force field for lipids: validation on six lipid types. The journal of physical chemistry B. 2010;114:7830–7843. doi: 10.1021/jp101759q. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Koivuniemi A, Heikela M, Kovanen PT, Vattulainen I, Hyvonen MT. Atomistic simulations of phosphatidylcholines and cholesteryl esters in high-density lipoprotein-sized lipid droplet and trilayer: clues to cholesteryl ester transport and storage. Biophys J. 2009;96:4099–4108. doi: 10.1016/j.bpj.2009.01.058. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Kory N, Farese RV, Jr, Walther TC. Targeting Fat: Mechanisms of Protein Localization to Lipid Droplets. Trends in cell biology. 2016;26:535–546. doi: 10.1016/j.tcb.2016.02.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Kory N, Thiam AR, Farese RV, Jr, Walther TC. Protein Crowding Is a Determinant of Lipid Droplet Protein Composition. Developmental cell. 2015;34:351–363. doi: 10.1016/j.devcel.2015.06.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Krahmer N, Farese RV, Jr, Walther TC. Balancing the fat: lipid droplets and human disease. EMBO Mol Med. 2013a;5:973–983. doi: 10.1002/emmm.201100671. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Krahmer N, Guo Y, Wilfling F, Hilger M, Lingrell S, Heger K, Newman HW, Schmidt-Supprian M, Vance DE, Mann M, et al. Phosphatidylcholine synthesis for lipid droplet expansion is mediated by localized activation of CTP:phosphocholine cytidylyltransferase. Cell Metab. 2011;14:504–515. doi: 10.1016/j.cmet.2011.07.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Krahmer N, Hilger M, Kory N, Wilfling F, Stoehr G, Mann M, Farese RV, Jr, Walther TC. Protein correlation profiles identify lipid droplet proteins with high confidence. Mol Cell Proteomics. 2013b;12:1115–1126. doi: 10.1074/mcp.M112.020230. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Kwok R, Evans E. Thermoelasticity of large lecithin bilayer vesicles. Biophys J. 1981;35:637–652. doi: 10.1016/S0006-3495(81)84817-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Ladokhin AS, White SH. Folding of amphipathic alpha-helices on membranes: energetics of helix formation by melittin. J Mol Biol. 1999;285:1363–1369. doi: 10.1006/jmbi.1998.2346. [DOI] [PubMed] [Google Scholar]
  60. Lagace TA, Ridgway ND. The rate-limiting enzyme in phosphatidylcholine synthesis regulates proliferation of the nucleoplasmic reticulum. Molecular biology of the cell. 2005;16:1120–1130. doi: 10.1091/mbc.E04-10-0874. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Lee MC, Orci L, Hamamoto S, Futai E, Ravazzola M, Schekman R. Sar1p N-terminal helix initiates membrane curvature and completes the fission of a COPII vesicle. Cell. 2005;122:605–617. doi: 10.1016/j.cell.2005.07.025. [DOI] [PubMed] [Google Scholar]
  62. Lin P, Chen X, Moktan H, Arrese EL, Duan L, Wang L, Soulages JL, Zhou DH. Membrane attachment and structure models of lipid storage droplet protein 1. Biochimica et biophysica acta. 2014;1838:874–881. doi: 10.1016/j.bbamem.2013.12.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Liu TY, Bian X, Sun S, Hu X, Klemm RW, Prinz WA, Rapoport TA, Hu J. Lipid interaction of the C terminus and association of the transmembrane segments facilitate atlastin-mediated homotypic endoplasmic reticulum fusion. Proceedings of the National Academy of Sciences of the United States of America. 2012;109:E2146–2154. doi: 10.1073/pnas.1208385109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. MacKerell AD, Bashford D, Bellott M, Dunbrack RL, Evanseck JD, Field MJ, Fischer S, Gao J, Guo H, Ha S, et al. All-atom empirical potential for molecular modeling and dynamics studies of proteins. The journal of physical chemistry B. 1998;102:3586–3616. doi: 10.1021/jp973084f. [DOI] [PubMed] [Google Scholar]
  65. Martinez L, Andrade R, Birgin EG, Martinez JM. PACKMOL: a package for building initial configurations for molecular dynamics simulations. J Comput Chem. 2009;30:2157–2164. doi: 10.1002/jcc.21224. [DOI] [PubMed] [Google Scholar]
  66. Martyna GJ, Tobias DJ, Klein ML. Constant pressure molecular dynamics algorithms. The Journal of Chemical Physics. 1994;101:4177–4189. [Google Scholar]
  67. Meyers NL, Wang L, Small DM. Apolipoprotein C-I binds more strongly to phospholipid/triolein/water than triolein/water interfaces: a possible model for inhibiting cholesterol ester transfer protein activity and triacylglycerol-rich lipoprotein uptake. Biochemistry. 2012;51:1238–1248. doi: 10.1021/bi2015212. [DOI] [PMC free article] [PubMed] [Google Scholar]
  68. Mitsche MA, Small DM. Surface pressure-dependent conformation change of apolipoprotein-derived amphipathic alpha-helices. Journal of lipid research. 2013;54:1578–1588. doi: 10.1194/jlr.M034462. [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. Miyanari Y, Atsuzawa K, Usuda N, Watashi K, Hishiki T, Zayas M, Bartenschlager R, Wakita T, Hijikata M, Shimotohno K. The lipid droplet is an important organelle for hepatitis C virus production. Nature cell biology. 2007;9:1089–1097. doi: 10.1038/ncb1631. [DOI] [PubMed] [Google Scholar]
  70. Ohsaki Y, Kawai T, Yoshikawa Y, Cheng J, Jokitalo E, Fujimoto T. PML isoform II plays a critical role in nuclear lipid droplet formation. The Journal of cell biology. 2016;212:29–38. doi: 10.1083/jcb.201507122. [DOI] [PMC free article] [PubMed] [Google Scholar]
  71. Parrinello M, Rahman A. Polymorphic transitions in single crystals: A new molecular dynamics method. Journal of Applied Physics. 1981;52:7182–7190. [Google Scholar]
  72. Pichoff S, Lutkenhaus J. Tethering the Z ring to the membrane through a conserved membrane targeting sequence in FtsA. Molecular microbiology. 2005;55:1722–1734. doi: 10.1111/j.1365-2958.2005.04522.x. [DOI] [PubMed] [Google Scholar]
  73. Pol A, Gross SP, Parton RG. Review: biogenesis of the multifunctional lipid droplet: lipids, proteins, and sites. The Journal of cell biology. 2014;204:635–646. doi: 10.1083/jcb.201311051. [DOI] [PMC free article] [PubMed] [Google Scholar]
  74. Prévost C, Tsai FC, Bassereau P, Simunovic M. Pulling Membrane Nanotubes from Giant Unilamellar Vesicles. J Vis Exp. 2017:e56086. doi: 10.3791/56086. [DOI] [PMC free article] [PubMed] [Google Scholar]
  75. Rowe ER, Mimmack ML, Barbosa AD, Haider A, Isaac I, Ouberai MM, Thiam AR, Patel S, Saudek V, Siniossoglou S, et al. Conserved Amphipathic Helices Mediate Lipid Droplet Targeting of Perilipins 1-3. The Journal of biological chemistry. 2016;291:6664–6678. doi: 10.1074/jbc.M115.691048. [DOI] [PMC free article] [PubMed] [Google Scholar]
  76. Ryckaert JP, Ciccotti G, Berendsen HJC. Numerical integration of the cartesian equations of motion of a system with constraints: molecular dynamics of n-alkanes. Journal of Computational Physics. 1977;23:327–341. [Google Scholar]
  77. Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, Pietzsch T, Preibisch S, Rueden C, Saalfeld S, Schmid B, et al. Fiji: an open-source platform for biological-image analysis. Nature methods. 2012;9:676–682. doi: 10.1038/nmeth.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  78. Seelig J. Thermodynamics of lipid-peptide interactions. Biochimica et biophysica acta. 2004;1666:40–50. doi: 10.1016/j.bbamem.2004.08.004. [DOI] [PubMed] [Google Scholar]
  79. Segrest JP, Jones MK, De Loof H, Brouillette CG, Venkatachalapathi YV, Anantharamaiah GM. The amphipathic helix in the exchangeable apolipoproteins: a review of secondary structure and function. Journal of lipid research. 1992;33:141–166. [PubMed] [Google Scholar]
  80. Shaw DE, Dror RO, Salmon JK, Grossman JP, Mackenzie KM, Bank JA, Young C, Deneroff MM, Batson B, Bowers KJ, et al. Millisecond-scale molecular dynamics simulations on Anton. Proceedings of the Conference on High Performance Computing Networking, Storage and Analysis (Portland, Oregon: ACM) 2009:1–11. [Google Scholar]
  81. Sinka R, Gillingham AK, Kondylis V, Munro S. Golgi coiled-coil proteins contain multiple binding sites for Rab family G proteins. The Journal of cell biology. 2008;183:607–615. doi: 10.1083/jcb.200808018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  82. Szeto TH, Rowland SL, Habrukowich CL, King GF. The MinD membrane targeting sequence is a transplantable lipid-binding helix. The Journal of biological chemistry. 2003;278:40050–40056. doi: 10.1074/jbc.M306876200. [DOI] [PubMed] [Google Scholar]
  83. Tang J, Signarvic RS, DeGrado WF, Gai F. Role of helix nucleation in the kinetics of binding of mastoparan X to phospholipid bilayers. Biochemistry. 2007;46:13856–13863. doi: 10.1021/bi7018404. [DOI] [PubMed] [Google Scholar]
  84. Thiam AR, Antonny B, Wang J, Delacotte J, Wilfling F, Walther TC, Beck R, Rothman JE, Pincet F. COPI buds 60-nm lipid droplets from reconstituted water-phospholipid-triacylglyceride interfaces, suggesting a tension clamp function. Proceedings of the National Academy of Sciences of the United States of America. 2013;110:13244–13249. doi: 10.1073/pnas.1307685110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  85. Tuckerman M, Berne BJ, Martyna GJ. Reversible multiple time scale molecular dynamics. The Journal of Chemical Physics. 1992;97:1990–2001. [Google Scholar]
  86. Ulmschneider MB, Doux JPF, Killian JA, Smith JC, Ulmschneider JP. Mechanism and Kinetics of Peptide Partitioning into Membranes from All-Atom Simulations of Thermostable Peptides. Journal of the American Chemical Society. 2010;132:3452–3460. doi: 10.1021/ja909347x. [DOI] [PubMed] [Google Scholar]
  87. Vamparys L, Gautier R, Vanni S, Bennett WFD, Tieleman DP, Antonny B, Etchebest C, Fuchs, Patrick FJ. Conical Lipids in Flat Bilayers Induce Packing Defects Similar to that Induced by Positive Curvature. Biophysical Journal. 2013;104:585–593. doi: 10.1016/j.bpj.2012.11.3836. [DOI] [PMC free article] [PubMed] [Google Scholar]
  88. Vanni S, Vamparys L, Gautier R, Drin G, Etchebest C, Fuchs PF, Antonny B. Amphipathic lipid packing sensor motifs: probing bilayer defects with hydrophobic residues. Biophys J. 2013;104:575–584. doi: 10.1016/j.bpj.2012.11.3837. [DOI] [PMC free article] [PubMed] [Google Scholar]
  89. Voth GA. New and notable: key new insights into membrane targeting by proteins. Biophys J. 2013;104:517–519. doi: 10.1016/j.bpj.2012.12.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
  90. Walther TC, Farese RV., Jr Lipid droplets and cellular lipid metabolism. Annu Rev Biochem. 2012;81:687–714. doi: 10.1146/annurev-biochem-061009-102430. [DOI] [PMC free article] [PubMed] [Google Scholar]
  91. Wang H, Becuwe M, Housden BE, Chitraju C, Porras AJ, Graham MM, Liu XN, Thiam AR, Savage DB, Agarwal AK, et al. Seipin is required for converting nascent to mature lipid droplets. Elife. 2016;5 doi: 10.7554/eLife.16582. [DOI] [PMC free article] [PubMed] [Google Scholar]
  92. Welte MA. Expanding roles for lipid droplets. Curr Biol. 2015;25:R470–481. doi: 10.1016/j.cub.2015.04.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  93. White SH, Wimley WC. Hydrophobic interactions of peptides with membrane interfaces. Biochimica et Biophysica Acta (BBA) - Reviews on Biomembranes. 1998;1376:339–352. doi: 10.1016/s0304-4157(98)00021-5. [DOI] [PubMed] [Google Scholar]
  94. Wilfling F, Wang H, Haas JT, Krahmer N, Gould TJ, Uchida A, Cheng JX, Graham M, Christiano R, Frohlich F, et al. Triacylglycerol synthesis enzymes mediate lipid droplet growth by relocalizing from the ER to lipid droplets. Developmental cell. 2013;24:384–399. doi: 10.1016/j.devcel.2013.01.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  95. Wu EL, Cheng X, Jo S, Rui H, Song KC, Davila-Contreras EM, Qi Y, Lee J, Monje-Galvan V, Venable RM, et al. CHARMM-GUI Membrane Builder toward realistic biological membrane simulations. J Comput Chem. 2014;35:1997–2004. doi: 10.1002/jcc.23702. [DOI] [PMC free article] [PubMed] [Google Scholar]
  96. Zehmer JK, Bartz R, Liu P, Anderson RG. Identification of a novel N-terminal hydrophobic sequence that targets proteins to lipid droplets. Journal of cell science. 2008;121:1852–1860. doi: 10.1242/jcs.012013. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1

Movie S1. Related to Figure 3: M-domain binding the monolayer surface. As the peptide approaches the monolayer surface, a number of residues insertions are attempted. Eventually, F53 inserts below the phosphate plane and anchors the peptide to the surface. White residues are hydrophobic, purple non-hydrophobic. As the residues approach the phosphate plane, they are shown in space filling representation.

Download video file (75.5MB, mp4)
2
NIHMS927865-supplement-2.pdf (5.2MB, pdf)

RESOURCES