Perilipin 3 promotes the formation of membrane domains enriched in diacylglycerol and lipid droplet biogenesis proteins

Abstract

Lipid droplets (LDs) serve as intracellular stores of energy-rich neutral lipids. LDs form at discrete sites in the endoplasmic reticulum (ER) and they remain closely associated with the ER during lipogenic growth and lipolytic consumption. Their hydrophobic neutral lipid core is covered by a monolayer of phospholipids, which harbors a specific set of proteins. This LD surface is coated and stabilized by perilipins, a family of soluble proteins that specifically target LDs from the cytosol. We have previously used chimeric fusion proteins between perilipins and integral ER membrane proteins to test whether proteins that are anchored to the ER bilayer could be dragged onto the LD monolayer. Expression of these chimeric proteins induces repositioning of the ER membrane around LDs. Here, we test the properties of membrane-anchored perilipins in cells that lack LDs. Unexpectedly, membrane-anchored perilipins induce expansion and vesiculation of the perinuclear membrane resulting in the formation of crescent-shaped membrane domains that harbor LD-like properties. These domains are stained by LD-specific lipophilic dyes, harbor LD marker proteins, and they transform into nascent LDs upon induction of neutral lipid synthesis. These ER domains are enriched in diacylglycerol (DAG) and in ER proteins that are important for early steps of LD biogenesis, including seipin and Pex30. Formation of the domains in vivo depends on DAG levels, and we show that perilipin 3 (PLIN3) binds to liposomes containing DAG in vitro. Taken together, these observations indicate that perilipin not only serve to stabilize the surface of mature LDs but that they are likely to exert a more active role in early steps of LD biogenesis at ER subdomains enriched in DAG, seipin, and neutral lipid biosynthetic enzymes.

Keywords: Saccharomyces cerevisiae; diacylglycerol; endoplasmic reticulum; lipid droplets; perilipins; seipin; steryl esters; triacylglycerols.

FIGURE 1

Membrane-anchored PLIN3 induce formation of ER domains in cells lacking LDs (A,B) Schematic drawing of the structure and localization of the membrane-anchored reporter proteins used in this study. The ER residential integral membrane proteins Wbp1, a single spanner and subunit of the glycosyltransferase, or Sec61, a polytopic transmembrane protein and subunit of the translocon, were fused to a GFP reporter (green oval) and the LD-targeting perilipin 3 (PLIN3, red). Native Sec61 and Wbp1, i.e., when not fused to PLIN3, are localized to the ER membrane in cells containing LDs (A). LDs are shown as globular structure filed with triacylglycerol and steryl esters, indicated in yellow. When appended to PLIN3, however, Wbp1-GFP-PLIN3 and Sec61-GFP-PLIN3, induce wrapping of the ER around LDs and thus appear to localize to the rim of LDs (B). (C) ER localization of Wbp1-GFP and Sec61-GFP (not fused to PLIN3) in wild-type cells. Cells coexpressing GFP-tagged Wbp1 or Sec61 together with Erg6-mCherry were grown in medium containing oleic acid, but no glucose, and analyzed by confocal microscopy. Blue arrows point to the peripheral ER, white arrowheads mark LDs. Scale bar, 5 µm. (D) LD-localization of the membrane-anchored PLIN3. Wild-type cells coexpressing Wbp1-GFP-PLIN3 or Sec61-GFP-PLIN3 with Erg6-mCherry were grown in medium containing oleic acid and analyzed by confocal microscopy. White arrowheads and blue arrows indicate colocalization of the markers at the rim of LDs in the merged image. Scale bar, 5 µm. (E,F) In cells lacking LDs, membrane-anchored PLIN3 localize to crescent-shaped ER domains. In quadruple mutant cells lacking LDs (4∆, are1∆ are2∆ dga1∆ lro1∆) Wbp1-GFP-PLIN3 (E) and Sec61-GFP-PLIN3 (F) localize to crescent-shaped and punctate domains in the ER (white arrowheads) and colocalize with the ER luminal marker mCherry-HDEL (blue arrows). Wbp1-GFP (E) and Sec61-GFP (F), lacking PLIN3, display homogenous ER localization (white arrowheads) and colocalize with mCherry-HDEL (blue arrows). Soluble PLIN3-GFP displays granular cytosolic localization in cells lacking LDs. No significant labelling of the ER membrane as visualized by mCherry-HDEL (blue arrows) is observed when PLIN3 is not anchored to an integral membrane protein. Scale bar, 5 µm. (G,H) Colocalization of native and PLIN3-containing versions of Wbp1 and Sec61. The native mScarlet-tagged versions of Wbp1 (G) and Sec61 (H) lacking PLIN3 (blue arrows) are affected to different degrees by the expression of the membrane-anchored version of Wbp1-GFP-PLIN3 or Sec61-GFP-PLIN3 (white arrowheads), respectively. Scale bar, 5 µm. (I,J) Localization of Ist2 in the cortical ER in cells expressing membrane-anchored PLIN3. Cells coexpressing GFP-Ist2 with either Wbp1-mScarlet-PLIN3 (I) or Sec61-mScarlet-PLIN3 (J) were cultivated, and the distribution of the marker proteins was analyzed by confocal microscopy. ER crescents are indicated by white arrowheads, Ist2 localization in the cortical ER is indicated by blue arrows. Scale bar, 5 µm.

FIGURE 2

ER domains formed by membrane-anchored PLIN3 transform upon induction of LD biogenesis (A,B) Cells harboring the TAG synthase Lro1 under the control of a galactose inducible promoter (GAL-LRO1 are1∆ are2∆ dga1∆) and co-expressing Wbp1-GFP-PLIN3 (A), or Sec61-GFP-PLIN3 (B), with Erg6-mCherry were cultivated in glucose medium. At time 0, cells were switched to galactose-containing medium to induce expression of Lro1, and hence TAG synthesis, and LD biogenesis. The subcellular distribution of the marker proteins was recorded by confocal microscopy after the indicated periods of time following galactose induction of Lro1 (0, switch to galactose; 1 h, 1 h in galactose; ON, overnight in galactose). The localization of the GFP-tagged membrane-proximal PLIN3 is indicated by white arrowheads, that of Erg6-mCherry by blue arrows. Scale bar, 5 µm. (C) Quantification of the size and number of structures labelled by membrane-proximal PLIN3 and/or Erg6 during induction of LD biogenesis. Graph on the left-hand side, size of the structures labelled by Wbp1-GFP-PLIN3, or Sec61-GFP-PLIN3, compared to Erg6-mCherry at different time points after induction of Lro1 are plotted. The median is indicated in a box that represents the 25–75th percentile range. The whiskers denote the largest and smallest values with 1.5× of the interquartile range from the hinges of the box. Outliers are depicted by black circles. Right-hand side graph, violin plot of the number of globular-shaped structures containing Erg6-mCherry. Statistical significance (n > 50 cells) was assessed using a Wilcoxon signed-rank test with continuity correction. ***p < 0.001. (D–F) Colocalization between membrane-proximal PLIN3 and LD markers proteins such as Ayr1 (D), Tgl1 (E), or Tgl4 (F). 4∆ mutant cells expressing Wbp1-mScarlet-PLIN3, or Sec61-mScarlet -PLIN3 together with GFP-tagged versions of the indicated LD markers were imaged by confocal microscopy. Localization of the membrane-proximal PLIN3 is indicated by white arrowheads, that of the LD markers by blue arrows. Scale bar, 5 µm. (G,H) Membrane domains induced by the expression of the membrane-proximal PLIN3 are recognized by LD-specific fluorescent dyes Nile Red (G) and BODIPY (H). 4∆ mutant cells expressing GFP- or mScarlet-tagged versions of Wbp1-PLIN3, or Sec61-PLIN3 were incubated with the neutral lipid specific dyes Nile Red or BODIPY and imaged by confocal microscopy. Localization of the membrane-proximal PLIN3 is indicated by white arrowheads, that of the LD-specific fluorescent dyes by blue arrows. Scale bar, 5 µm.

FIGURE 3

PLIN-induced ER domains colocalize with an ER-DAG sensor and are dependent on DAG formation (A) Schematic representation of the ER-DAG sensor. The ER-DAG sensor used here is composed of the DAG-binding tandem C1 domains of Protein Kinase D (C1a/b-PKD; lila box) fused to GFP (green circle), which in turn is fused to the transmembrane domain of Ubc6 (blue cylinder), a tail-anchored ER protein. DAG accumulation is shown in orange. (B,C) Expression of the membrane-proximal PLIN3 induces colocalization of the ER-DAG sensor. Cells coexpressing inducible versions of either Wbp1-mScarlet-PLIN3 (B), or Sec61-mScarlet-PLIN3 (C) with a constitutively expressed GFP-tagged ER-DAG sensor, were imaged after growth in the presence (time 0) or absence of doxycycline (6 h, and overnight, ON) to induce expression of the PLIN3 reporters. ER domains formed by membrane-proximal PLIN3 are indicated by white arrowheads, the localization of the ER-DAG sensor is indicated by blue arrows. Scale bar, 5 µm. (D) Quantification of the area occupied by the DAG-sensor upon induction of expression of membrane-anchored PLIN3. The relative area occupied by the ER-DAG sensor was quantified manually in n > 50 cells by pixel analysis and plotted as box plot. (E,F) PLIN3 membrane domains transform upon inhibition of lipid synthesis. Cells coexpressing the indicated mScarlet-tagged membrane-proximal PLIN3 together with the GFP-tagged ER-DAG sensor were treated with cerulenin (10 μg/mL) and terbinafine (30 μg/mL) for the indicated period of time, and the distribution of the marker proteins was analyzed by confocal microscopy. Localization of the membrane-proximal PLIN3 is indicated by white arrowheads, that of the ER-DAG sensor by blue arrows. Scale bar, 5 µm. (G–H) Formation of PLIN3 membrane domains depends on DAG synthesis. Schematic representation of the activation of the phosphatidate phosphatase Pah1 through dephosphorylation by the ER membrane embedded Nem1/Spo7 phosphatase complex. Dephosphorylation of Pah1 in wild-type cells promotes DAG production, whereas lack of Nem1 results in low DAG levels (G). ER domain formation was analyzed in mutants lacking the activator of Pah1, Nem1 (are1∆ are2∆ dga1∆ lro1∆ nem1∆), panel (G) and in 4∆ nem1∆ mutant cells expressing a constitutive active version of Pah1, Pah1-7A (panel H). Cells co-expressing the membrane-proximal PLIN3 together with the ER-DAG sensor were imaged by confocal microscopy. Localization of the membrane-proximal PLIN3 is indicated by white arrowheads, that of the ER-DAG sensor by blue arrows. Scale bar, 5 µm.

FIGURE 4

Membrane domains formed by PLIN3 are enriched in LD biogenesis proteins (A) Schematic representation of ER-localized LD biogenesis proteins. Seipin (Fld1), which in yeast forms a complex with Ldb16, colocalizes with the reticulon homology domain containing protein Pex30, the lipid droplet organizing proteins and promethin ortholog Ldo16/45, and the activator complex of the phosphatidate phosphatase Pah1, Nem1/Spo7, at the base of neutral lipid (in yellow) filled LDs. (B) In the absence of expression of membrane-anchored PLIN3, LD biogenesis factors are localized to punctate structures, but do not form crescent-shaped membrane domains. Cells bearing plasmids containing Wbp1-mScarlet-PLIN3, or Sec61-mScarlet-PLIN3, were cultivated in the presence of doxycycline to repress expression of the membrane-anchored PLIN3 fusion proteins and the localization of the genomically-tagged versions of the LD biogenesis proteins was assessed by confocal microscopy. Punctate localization within the ER of the LD biogenesis proteins is indicated by pink arrowheads. Scale bar, 5 µm. (C–D) ER crescents formed by the membrane-anchored PLIN3 are enriched in the LD biogenesis proteins seipin (Fld1), Nem1, Spo7, Pex30, Ldb16, and Ldo16/45. Quadruple mutant cells co-expressing either GFP- (signal in green) or mScarlet-tagged (signal in red) versions of Wbp1-PLIN3 (C), or Sec61-PLIN3 (D), with the indicated genomically-tagged LD biogenesis proteins were cultivated to early logarithmic growth phase in rich medium and the subcellular localization of the indicated fluorescent proteins was analyzed by confocal microscopy. Localization of the membrane-proximal PLIN3 is indicated by white arrowheads, that of the LD biogenesis proteins by pink arrowheads. Regions selected for the line scans (yellow line) are shown in the insets and line scans are plotted to the right of the merged images. Scale bar, 5 µm.

FIGURE 5

LD biogenesis proteins affect the size and number of ER domains formed by membrane-proximal PLIN3. (A,B) Colocalization between membrane-proximal PLIN3 and the ER-DAG sensor was analyzed in 4∆ cells (are1∆ are2∆ dga1∆ lro1∆) lacking one of the indicated LD biogenesis genes (fld1∆, ldb16∆, or ldo16/45∆). Localization of the membrane-proximal PLIN3 fused to mScarlet and Wbp1-PLIN3 (A) or Sec61-PLIN3 (B) is indicated by white arrowheads, that of the ER-DAG sensor by blue arrows. Scale bar, 5 µm. (C,D) Quantification of the average number of PLIN3 membrane domains per cell and the relative area occupied by the ER-DAG sensor in 4∆ mutant cells lacking LD biogenesis proteins. The average number of PLIN3 membrane domains and the relative area occupied by the ER-DAG sensor was quantified manually in n > 50 cells and plotted as violin and box plot, respectively. The median was calculated by counting the number of punctate and measuring the area occupied by the ER-DAG sensor. The significance was assessed using a Wilcoxon rank-sum test with continuity correction. ***p < 0.001.

FIGURE 6

Protein domains within PLIN3 exert redundant and synergistic functions in ER membrane domain formation and clustering of DAG (A) Schematic representation of the domain structure of PLIN3. The three-domain architecture of PLIN3 is illustrated: N-terminal PAT domain (blue rectangle, amino acids 1–99), the repetition of 11-mer amphipathic helices (red ovals, amino acids 96–192), and the C-terminal 4-helix bundle (dark blue ellipse, amino acids 192–434). (B) Western blot analysis of membrane-anchored PLIN3 protein domains. Proteins from cells expressing fusions of Wbp1 or Sec61 with either full-length PLIN3 or the indicated single- or double-PLIN3 protein domains were extracted, separated by SDS-PAGE, and probed for the presence of the fusion proteins [Fusion] or for endogenous native Wbp1 or Sec61 [Endo]. (C,D) Any two of the three protein domains of PLIN3 are required for ER membrane domain formation. 4∆ cells (are1∆ are2∆ dga1∆ lro1∆) co-expressing the ER-DAG sensor together with single protein domains of PLIN3, i.e., PAT, 11-mer repeat, or 4-helix bundle, or two-domain fusions, i.e., PAT+11-mer, PAT+4-helix bundle, or 11-mer+4-helix bundle, fused to either Wbp1-mScarlet (C) or Sec61-mScarlet (D) were analyzed by confocal microscopy. Localization of the membrane-proximal PLIN3 domain in crescent shaped ER domains is indicated by white arrowheads, that of the ER-DAG sensor by blue arrows. Scale bar, 5 µm.

FIGURE 7

Membrane-proximal PLIN3 induces proliferation and vesiculation of the perinuclear ER. (A–F) Cells expressing either the soluble GFP-PLIN3 (A,B), the membrane-proximal Wbp1-GFP-PLIN3 (C,D), or Sec61-GFP-PLIN3 (E,F), were fixed by high pressure freezing and processed for CLEM analysis as described in Materials and Methods. Images on the left show tomograms with and without the correlated fluorescence image (green) and fiducial beads (yellow dots and arrows). Ribosome-free, PLIN3-containing membrane domains are outlined in green and vesicular structures are highlighted by green arrowheads. Nucleus, vacuole and ribosomes are indicated. Scale bar, 1 μm in low magnification tomograms and 100 nm in high magnification tomograms.

FIGURE 8

PLIN3 binds DAG-containing liposomes in vitro (A) Purified PLIN3 binds short chain DAG in vitro. An N-terminal polyhistidine-tagged version of PLIN3 was expressed in bacteria and purified by Ni²⁺-affinity chromatography. The protein was fluorescently labelled and binding of short chain DAG (C8:0-DAG) was measured by microscale thermophoresis. The fraction of the ligand-bound protein is plotted against the concentration of DAG and the deduced dissociation constant (K _D ) is indicated. Values represent mean ± S.D. of three independent determinations. (B,C) PLIN3 displays increased binding to liposomes (LUV) containing DAG. Purified and fluorescently labelled PLIN3 was incubated with liposomes lacking (panel B) or containing 5 mol% C16:0-DAG (1,2-dipalmitoyl-sn-glycerol), panel (C) and the binding affinity was determined by microscale thermophoresis. The deduced dissociation constants (K _D ) are indicated. Values represent mean ± S.D. of three independent determinations. (D) A mutant version of PLIN3, which bears a point mutation in the amphipathic helix repeats encoded by the 11-mer repeats, PLIN3^V158D, is stably expressed. A schematic drawing of the structure of PLIN3 and the position of the point mutation is shown on top. Western blot analysis of cells expressing fusions between Wbp1-mScarlet or Sec61-mScarlet and either wild-type (wt) or the V158D mutant version of PLIN3. Kar2 is shown as a loading control. (E) Expression of the point mutant version of membrane-anchored PLIN3 affect the number of membrane domains that are being formed and the relative distribution of the ER-DAG sensor. Cells co-expressing membrane-anchored versions of PLIN3^V158D with the ER-DAG sensor were cultivated and analyzed by confocal microscopy. Localization of the membrane-proximal PLIN3^V158D domain is indicated by white arrowheads, that of the ER-DAG sensor by blue arrows. Scale bar, 5 µm. (F,G) The V158D mutant version of the membrane-anchored PLIN3 affects the numbers of punctate structures and the localization of the ER-DAG sensor. The number of punctuate structures formed by the wild-type and the V158D mutant version of the membrane-anchored PLIN3 is plotted (F) as is the relative area of the ER-DAG sensor (G). The average number of PLIN3 membrane domains and the relative area occupied by the ER-DAG sensor was quantified manually in n > 50 cells and plotted as violin and box plot, respectively. The median was calculated by counting the number of punctates and measuring the area occupied by the ER-DAG sensor. The significance was assessed using a Wilcoxon rank-sum test with continuity correction. ***p < 0.001. (H) The PLIN3^V158D point mutant has reduced affinity to short chain DAG. Purified PLIN3^V158D was fluorescently labelled and binding of short chain DAG (C8:0-DAG) was measured by microscale thermophoresis. The deduced dissociation constant (K _D ) is indicated. Values represent mean ± S.D. of three independent determinations. (I,J) PLIN3^V158D has reduced binding affinity towards liposomes containing DAG. PLIN3^V158D was incubated with liposomes lacking (I) or containing 5 mol% C16:0-DAG (1,2-dipalmitoyl-sn-glycerol), (J) and the binding affinity was determined by microscale thermophoresis. The deduced dissociation constants (K _D ) are indicated. Values represent mean ± S.D. of three independent determinations.